This item is only available as the following downloads:

( PDF )

PAGE 1

SUPERSPACEor Onethousandandone lessonsinsupersymmetryS.JamesGates,Jr. MassachusettsInstituteofTechnology,Cambridge,Massachusetts (Present a ddress: UniversityofMaryland,CollegePark,Maryland) gatess@wam. umd.edu MarcusT.Grisaru BrandeisUniversity,Waltham,Massachusetts (Present a ddress: McGillUniversity,M ontreal, Que bec) grisaru@physics.mcgill.ca MartinRo cek StateUniversityofNewYork,StonyBrook,NewYork rocek@insti.physics.sunysb.edu Wa rrenSiegel UniversityofCalifornia,Berkeley,California (Present a ddress: StateUniversityofNewYork) wa rren@wcgall.physics.sunysb.edu

PAGE 2

LibraryofCongressCataloginginPublicationData Mainentryundertitle: Superspace:onethousandandonelessonsinsupersymmetry. (Frontiersinphysics;v.58) Includesindex. 1.Supersymmetry.2.Quantumgravity. 3.Supergravity.I.Ga tes,S.J.II .Series. QC 174.17.S9S971983530.1283-5986 ISBN0-8053-3160-3 ISBN0-8053-3160-1(pbk.)

PAGE 3

Superspaceisthegreatestinventionsincethewheel[1]. Preface Saidto,,and:Letswritear eviewpaper.Saidand:Great idea!Said:Naaa. Butafewdayslaterhadproducedatableofcontentswith1001items. ,,andwrote.Thendidntwrite.Thenwroteagain.Thereviewgrew; andgrew;andgrew.Itbecameanoutlinef orabook;itbecamea“rstdraft;itbecame as econddraft.Itbecameaburden.Itbecameagony.Temperswerelost;andhairs; an da fewpounds(alas,quicklyregained).Theyarguedabout;vs..,about whichvs.that,vs.,  vs .   , +  vs .  .M ad eb ad puns,drewpicturesonthe blackboard,wererudetotheircolleagues,neglectedtheirduties.Bemoaned thepaucityoflettersintheGreekandRomanalphabets,ofhoursintheday,daysin thew eek,weeksinthemonth.,,andwroteandwrote. *** Thismuststop;wewanttogetbacktoresearch,toourfamilies,friendsandstude nt s.Wewanttolookattheskyagain,goforwalks,sleepatnight.Writeasecond volume?Nev er!Well, inaco upleofyears? We be go urreadersindulgence.Wehavetriedtopresentasubjectthatwelike, thatwethinkisimportant.Wehavetriedtopresentourinsights,ourtoolsandour knowledge.Alongtheway,someerrorsandm iscon ceptionshavewithoutdoubtslipped in.Theremustbewrongstatements,mispri nts,mi stakes,awkwardphrases,islandsof incomprehensibility(buttheystartedoutascontinents!).Wecould,probablywe should,improveandimprove.Butwecannolongerwait.Likeclimberswithinsightof thesummitwearerushing,castingasideca ution,reachingtowardsthemomentwhenwe canshoutitsbehindus. Thisisnotapolishedwork.Withoutdoubtsometopicsaretreatedbetterelsewhere.Withoutdoubtwehaveleftouttopicsthatshouldhavebeenincluded.Without doubtwehavetreatedthesubjectfromapersonalpointofview,emphasizingaspects thatwearefamiliarwith,andneglectingsomethatwouldhaverequiredstudyingothers work.Nev ertheless,wehopethisbookwillbeuseful,bothtothosenewtothesubject andtothosewhohelpeddevelopit.Wehavepresentedmanytopicsthatarenotavailableelsewhere,andmanytopicsofinterestalsooutsidesupersymmetry.Wehave [1 ]. A. Oop,Asupersymmetricversionoftheleg,Gondwanalandpredraw(January10,000,000 B.C.), tobediscovered.

PAGE 4

includedtopicsw hosetreatmentisincomplete,andpresentedconclusionsthatarereally onlyconjectures.Insomecases,thisre”ectsthestateofthesubject.Fillinginthe holesandprovingtheconjecturesmaybegoodresearchprojects. Supersymmetryisthecreationofmanyta lentedphysicists.Wewouldliketo thankallourfriendsinthe“el d,wehavemany,fortheircontributionstothesubject, andbegtheirpardonfornotpresentingalistofreferencestotheirpapers. Mostofthew orkonthisbookwasdo newh ilethefourofuswereattheCalifornia InstituteofTechnology,duringthe1982-83academicyear.Wewouldliketothankthe InstituteandthePhysicsDepartmentfortheirhospitalityandtheuseoftheircomputer facilities,theNSF,DOE,theFleischmannF oundationandtheFairchildVisitingScholarsProgramfortheirsupport.SomeoftheworkwasdonewhileM.T.G.andM.R.were visitingtheInstituteforTheoreticalPhysicsatSantaBarbara.Finally,wewouldliketo thankRichardGrisaruforthemanyhourshedevotedtotypingtheequationsinthis book,HyunJeanKimfordrawi ngthedi agrams,andAndersKarlhedeforcarefullyreadinglargepartsofthemanuscriptandforhisu sefulsuggestions;andalltheotherswho helpedus. S.J.G.,M.T .G.,M.R.,W.D.S. Pa sadena,January1983 Au gust2001: Freeversionr eleasedonweb;correctionsandbookmarksadded.

PAGE 5

Contents Preface 1.Introduction 1 2.Atoysuperspace 2.1.Notationandconventions7 2.2.Supersymmetrya ndsuper“elds9 2.3.Scalarmultiplet15 2.4.Vectormultiplet18 2.5.Otherglobalgaugemultiplets28 2.6.Supergravity34 2.7.Quantumsuperspace46 3.Representationsofsupersymmetry 3.1.Notation54 3.2.Thesupersymmetrygroups62 3.3.Representationsofsupersymmetry69 3.4.Covariantderivatives83 3.5.Constraine dsuper “elds89 3.6.Componentexpansions92 3.7.Superintegration97 3.8.Superfunctionaldierentiationandintegration101 3.9.Physical,auxiliary,andgaugecomponents108 3.10.Compensators112 3.11.Projectionoperators120 3.12.On-shellrepresentationsandsuper“elds138 3.13.O-shell“eldstrengthsandprepotentials147 4.Classical,global,simple( N =1)super “elds 4.1.Thescalarmultiplet149 4.2.Yang-Millsgaugetheories159 4.3.Gauge-invariantmodels178 4.4.Superforms181 4.5.Othergaugemultiplets198 4.6. N -extendedmultiplets216 5.Classical N =1superg ravity 5.1.Reviewofgravity232 5.2.Prepotentials244

PAGE 6

5.3.Covariantapproach267 5.4.SolutiontoBianchiidentities292 5.5.Actions299 5.6.Fromsuperspacetocomponents315 5.7.DeSittersupersymmetry335 6.Quantumgloba lsuper “elds 6.1.Introductiontosupergraphs337 6.2.Gauge“xingandghosts340 6.3.Supergraphrules348 6.4.Examples364 6.5.Thebackground“eldmethod373 6.6.Regularization393 6.7.AnomaliesinYang-Millscurrents401 7.Quantum N =1superg ravity 7.1.Introduction408 7. 2. Ba ck gr o und-quantumsplitting410 7.3.Ghosts420 7.4.Quantization431 7.5.Supergravitysupergraphs438 7.6.CovariantFeynmanrules446 7.7.Generalpropertiesoftheeectiveaction452 7.8.Examples460 7.9.Locallysupersymmetricdimensionalregularization469 7.10.Anomalies473 8.Breakdown 8.1.Introduction496 8.2.Explicitbreakingofglobalsupersymmetry500 8.3.Spontaneousbreakingofglobalsupersymmetry506 8.4.Traceformulaefromsuperspace518 8.5.Nonlinearrealizations522 8.6.SuperHiggsmechanism527 8.7.Supergravityandsymmetrybreaking529 Index 542

PAGE 7

1.INTROD UCTION Thereisa“fthdimensionbeyondthatwhichisknowntoman.Itisa dimensionasvastasspaceandast imelessasin“nity.Itisthemiddle groundbetweenlightandshadow,betweenscienceandsuperstition;anditlies be tweenthepitofmansfearsandthesummitofhisknowledge.Thisisthe dimensionofimagination.Itisanareawhichwecall,theTwilightZone. RodSer ling 1001:Asuperspaceodyssey Symmetryprinciples,bothglobalandlocal,areafundamentalfeatureofmodern particlephysics.Attheclassicalandphenomenologicallevel,globalsymmetriesaccount formanyofthe(approximate) regularitiesweobserveinnature,whilelocal(gauge) symmetries explainandunifytheinteractionso fthe basicconstituentsofmatter.At thequantumlevelsymmetries(viaWardidentities)facilitatethestudyoftheultraviolet behavior of“eldtheorymodelsandtheirrenormaliz ation.In particular,theconstructionofmodelswithlocal(internal)Yang-Millssymmetrythatareasymptoticallyfree hasincreasedenormouslyourunderstandingofthequantumbehaviorofmatteratshort distances.Ifthisunderstandingcouldbeextendedtothequantumbehaviorofgravitationalinteractions(quantumgravity)wewouldbeclosetoasatisfactorydescriptionof micronatureintermsofbasicfermioniccons tituentsformingmultipletsofsomeuni“cationgroup,andbosonicgaugeparticlesrespo nsib lefortheirinteractions.Evenmore satisfactorywouldbetheexistenceinnatureofasymmetrywhichuni“esthebosons andthefermions,theconstituentsandtheforces,intoasingleentity. Supersymmetryisthesupremesymmetry:Ituni“esspacetimesymmetrieswith internalsymmetries,fermionswithbosons,an d(localsup ersymmetry)grav itywithmatter.U nderquitegeneralassumptionsitisthelargestpossiblesymmetryoftheSmatrix.Atthequantumlevel,renormalizablegloballysupersymmetricmodelsexhibit improvedultravioletbehavior:Becauseofcancellationsbetweenfermionicandbosonic contributionsquadraticdivergencesareabsen t;somesupersymmetricmodels,inparticularmaximallyextendedsuper-Yang-Millstheory,aretheonlyknownexamplesoffourdimensional“eldtheoriesthatare“nitetoallordersofperturbationtheory.Locally

PAGE 8

21.INTRODUCTIONsupersymmetricgravity(supergravity)m aybetheonlywayinwh ichnaturecanreconc ileEinsteingravityandquantumtheory.Althoughwedonotknowatpresentifitisa “nitetheory,quantumsupergravitydoesexhibitlessdivergentshortdistancebehavior thanordinaryquantumgravi ty.Outsi detherealmofstandardquantum“eldtheory,it isbelievedthattheonlyreasonablestringtheories(i.e.,thosewithfermionsandwithout quantuminconsistencies)aresupersymme tric;thesein cludemodelsthatmaybe“nite (themaximallysupersymmetrictheories). Atthepresent timethereisnodir ectexperimentalevidencethatsupersymmetryis af undamentalsymmetryofnature,butthecurrentlevelofactivityinthe“eldindicates thatmanyphysicistsshareourbeliefthats uchevidencewilleventuallyemerge.Onthe theoreticalside,thesymmetrymakesitpossibletobuildmodelswith(super)natural hierarchies.Onestheticgrounds,theideaofasuperuni“edtheoryisveryappealing. Evenifsupersymmetryandsupergravityar enottheult imatetheory,th eirstudyhas in creasedourunderstandingofclassicalandquantum“eldtheory,andtheymaybean im po rtantstepintheunderstandingofsomeyetunknown,correcttheoryofnature. Wemeanby(P oincar e)supersymmetryanextension ofordinaryspacetimesymmetriesobtainedbyadjoining N spinorialgenerators Q whose an ticommutator yieldsa translationgenerator: { Q Q } = P .Thissy mmetrycanberealizedonordinary“elds (functionsofspacetime)bytransformationsthatmixbosonsandfermions.Suchrealizationssucetostudysupersymmetry(onecanwriteinvariantactions,etc.)butareas cumbersomeandinconvenientasdoingvectorcalculuscomponentbycomponent.A compactalternativetothiscomponent“eldapproachisgivenbythe superspace--super“eld approac h.Superspaceisanextensionofordinaryspacetimetoinclude extra anticomm uting coordinatesintheformof N two-comp onentWeylspinors Super“elds( x )aref unctionsde“nedoverthisspace.Theycanbeexpandedina Taylorserieswi threspecttotheanticommutingcoordinates ;b ecausethesquareofan anticommutingquantityvanishes,this serieshasonlya“nitenumberofterms.The coecientsobtainedinthiswayaretheordinarycomponent“eldsmentionedabove.In superspace,supersymmetryismanifest:Th esupersy mmetryalgebraisrepresentedby translationsandrotationsinvolvingboththespacetimeandtheanticommutingcoordinates.Thetransformationsofthecomponent“eldsfollowfromtheTaylorexpansionof thetranslatedandrotatedsup er“elds.Inparticular,thetransformationsmixingbosons

PAGE 9

1.INTROD UCTION3andfermionsareconstanttranslationsofthe coordinates,andrelatedrotationsof intothespacetimecoordinate x Afurtheradvant ageofsuper“eldsisthattheyautomaticallyinclude,inaddition tothedynamicaldegreesoffreedom,certain unphysical“elds:(1)a uxiliary“elds(“elds withnonderivativekineticterms),neededclassicallyfortheo-shellclosureofthesupersymmetryalgebra,and(2)compensating“elds(“eldsthatconsistentirelyofgauge degreesoffreedom),whichareusedtoenlarg ethe usualgau getransformationstoan entiremultipletoftransformationsforming arepresentationofsup ersymmetry; together withtheauxiliary“elds,theyallowthealgebratobe“eldindependent.Thecompensatorsareparticularlyimportantforquantization,sincetheypermittheuseofsupersymmetricgauges,ghosts,Feynmangraphs ,andsupersy mmetricpower-counting. Unfort unately,ourpresentknowledgeofo-shell extended ( N > 1)supersymmetry is solimitedthatformostextendedtheoriestheseunphysical“elds,andthusalsothe correspondingsuper“elds,areunknown.Onecouldhopeto “nd th e unphysicalcomponentsdirectlyfromsuperspace;theessentiald icultyisthat,ingeneral,asuper“eldisa highlyreduciblerepresentationofthesupersymmetryalgebra,andtheproblembecomes on eo f “nding which representationspermittheconstructionofconsistentlocalactions. Therefore,exceptwhendiscussingthefeatu reswhicharecommontogeneralsuperspace, werest rictourselves inthisvolume toadiscussionof simple ( N =1)sup er“eldsupersymmetry.Wehopetotreatextendedsuperspaceandothertopicsthatneedfurther developmentinasecond(andhopefullylast)volume. Weintro ducesuper“eldsinchapter2forthesimplerworldofthreespacetime dimensions,wheresuper“eldsareverysimilartoordinary“elds.Weskipthediscussion ofno nsuperspacetopics(background“elds,gravity,etc.)whicharecoveredinfollowing ch apters,andconcentrateonapedagogicaltreatmentofsuperspace.Wereturntofour dimensionsinchapter3,wherewedescribehowsupersymmetryisrepresentedonsuper“elds,anddiscussallgeneralpropertiesoffreesuper“elds(andtheirrelationtoordinary “elds).Inchapter4wediscusssimple( N =1)super “eldsinc lassicalgloba lsupersymmetry.Weincludesuchtopicsasgauge-covaria ntderivati ves, supersymmetricmodels, extendedsupersymmetrywithunextendedsuper“elds,andsuperforms.Inchapter5we extendthediscussiontolocalsupersymmetry( supergravity),relyin gheav ilyonthecompensat orapproach.Wediscussprepotentialsandcovariantderivatives,theconstruction

PAGE 10

41.INTRODUCTIONofactions,andshowhowtogofromsuperspacetocomponentresults.Thequantum aspectsofglobalt heoriesisthetopicofchapter6,whichincludesadiscussionofthe background“eldformalism,supersymmetric regularization,anomalies,andmanyexamplesofsupergraphcalculations.Inchapter7wemakethecorrespondinganalysisof quantumsupergravity,includingmanyofthenovelfeaturesofthequantizationprocedure(varioustypes ofghosts).Chapter8describess upersymmetrybre aking,explicit andspontaneous,inc l udingthesuperHiggsmechanismandtheuseofnonlinearrealizations. Wehave notdiscussedcomponentsupersymme tryandsupergravity,realistic superGUTmodelswithorwithoutsupergravity,andsomeofthegeometricalaspectsof classicalsupergravity.Forthe“rsttopicthereadermayconsultmanyoftheexcellent reviewsandlecturenotes.Thesecondisoneofthecurrentareasofactiveresearch.It isourbeliefthatsuperspacemethodseventuallywillprovideaframeworkforstreamliningthe phenomenology,oncewehavebettercon trolofourtools.Thethirdtopicis attractingincreasedattention,buttherearestillmanyissuestobesettled;thereagain, superspacemethodsshouldproveuseful. Wea ssumethereaderhasaknowledgeofstandardquantum“eldtheory(sucient todoFeynmangraphca lculationsinQCD).Wehavetriedtomakethisbookaspedagogicalandencyclopedicaspossible,buthave omittedsomestraightforwardalgebraic detailswhicharelefttothereaderas(necessary!)exercises.

PAGE 11

1.INTROD UCTION5Ahitc hhikersguide Weareh oping,ofcourse,thatthisbookwillbeofinteresttomanypeople,with dierentinterestsandbackgrounds.Thegraduatestudentwhohascompletedacourse inquantum“eldtheoryandwant stost udysuperspaceshould: (1) Read anarticleortworeviewingcomponentglobalsupersymmetryandsupergravity. (2) Read chapter2foraquickand easy(?)introductiontosuperspace.Sections1, 2,and3arestraightforward.Section4introduces,inasimplesetting,theconceptof constrainedcovariantderivatives,andthesolutionoftheconstraintsintermsofprepotentials.Section5couldbes kippedat“rstreading.Sect ion6doesfors upergravity whatsection4didforYang-Mills;super“eld supergravityinthreedimensionsisdeceptivelysimple.Section7introducesquantizationandFeynmanrulesinasimplersituationthaninfour dimensions. (3) Study subsections3.2.a-donsupersymme tryalgebras,andsections3.3.a, 3.3.b.1-b.3,3.4.a,b,3.5and3.6onsuper“e lds,covariantd erivatives,andcomponent expansions. Study section3.10oncompensators;weus ethemextensive lyinsupergravity. (4) Study section4.1aonthescalarmultiplet,andsections4.2and4.3ongauge theori es,theirprepotentials,covariantderivativesandsolutionoftheconstraints.A re ading ofsections4.4.b,4.4.c.1,4.5.aand4.5.emightbepro“table. (5) Takeadee pbreath and slowly studysection5.1,which isourfavoriteapproach togravity,andsections5.2to5.5onsupergravity;thisiswheretheactionis.Foran i nductiveapproachthatstartswiththeprepotentialsandconstructsthecovariant derivati vess ection5 .2issucient,andonecanthengodirectlytosection5.5.Alternatively,onecouldstartwithsection5.3,and ade ductiveapproachbas edonconstrained covariantderivatives,gothroughsection5.4andagainendat5.5. (6) Study sections6.1through6.4onquantizationandsupergraphs.Thetopicsin theses ectionsshou ldbefairlyaccessible. (7) Study sections8.1-8.4. (8)Gobackt othebegi nningand skipnothing thistime.

PAGE 12

61.INTRODUCTIONOu rp articlephysicscolleagueswhoarefamiliarwithglobalsuperspaceshould skim3.1fornotation,3.4-6and4.1,read4.2(no,youdontknowitall),andgetbusy onchapter5. Theexpertsshouldlookforseriousmistakes.Wewouldappreciatehearingabout them. Abriefguidetot heliterature Acomplete listofref erencesisbecomingincreasinglydiculttocompile,andwe havenotattemp tedtodoso.However,thefollowing(incomplete!)listofreviewarticles andproceedingsofvariousschoolsandconfere nces,andthereferencestherein,areuseful andshouldprovideeasyacces st ot hejournalliterature: Forglobals upersymmetry,thestandardreviewarticlesare: P. Fa ye ta nd S. Fe rrara,Supersymmetry,PhysicsReports32C(1977)250. A.SalamandJ.Strathdee,FortschrittederPhysik,26(1978)5. Forcomponent supergravity,thestandardreviewis P. va nN ieuwenhuizen,Supergravity,PhysicsReports68(1981)189. ThefollowingProceedingscontainextensiveandup-to-datelecturesonmany supersymmetryandsupergravitytopics:  RecentDevelopmentsinGravitation(Carges` e 1978),eds.M.LevyandS.Deser, PlenumPress,N.Y.  Sup ergravity(StonyBrook1979),eds.D.Z.FreedmanandP.vanNieuwenhui zen,North-Holland,Amsterdam. TopicsinQuantumF ieldTheoryandGaugeTheories(Salamanca),Phys.77, SpringerVerl ag,Berlin.  Sup erspaceandSupergravity(Cambridge1980),eds.S.W.HawkingandM. Ro cek,CambridgeUniversityPress,Cambridge.  SupersymmetryandSupergravity81(Trieste),eds.S.Ferrara,J.G.Taylorand P.vanNieuwen hui zen,CambridgeUniversityPress,Cambridge.  SupersymmetryandSupergravity82(Trieste),eds.S.Ferrara,J.G.Taylorand P.vanNieuwen hui zen,WorldSc ienti“cPublishingCo.,Singapore.

PAGE 13

Contentsof 2.ATOYSUPERSPACE 2.1.Notationandconventions7 a.Indexcon ventions7 b.Superspace8 2.2.Supersymmetrya ndsuper“elds9 a.Representations9 b.Componentsbyexpansion10 c.Actionsandcomponentsbyprojection11 d.Irreduciblerepresentations13 2.3.Scalarmultiplet15 2.4.Vectormultiplet18 a.Abeliangaugetheory18 a.1.Gaugeconnections18 a.2.Components19 a.3.Constraints20 a.4.Bianchiidentities22 a.5.Mattercouplings23 b.Nonabeliancase24 c.Gaugeinvar iantma sses26 2.5.Otherglobalgaugemultiplets28 a.Superforms:generalcase28 b.Super2-form30 c.Spinorgaugesuper“eld32 2.6.Supergravity34 a.Supercoordinatetransformations34 b.Lorentztransformations35 c.Covariantderivatives35 d.Gaugechoices37 d.1.Asupersy mmetricgauge37 d.2.Wess-Zu minogauge38 e.Fieldstrengths38 f.Bianchiide ntit ies39 g.Actions 42 2.7.Quantumsuperspace46 a.Scalarmultiplet46

PAGE 14

a.1.Generalformalism46 a.2.Examples49 b.Vectormu ltiplet52

PAGE 15

2.ATOYSUPERSPACE 2.1.Notationandconventions Thischapterpresentsaself-containedtreatmentofsupersymmetryinthree spacetimedimensions.Ourmainmotivationfor consideringthiscasei ssimp licity.Irreduciblerepresentationsofsimple( N =1)globals upersymmetryareeasiertoobtain thaninfour dimensions:Scalarsuper“elds(single,realfunctionsofthesuperspacecoordinates)provideonesuchrepresentation,andallothersareobtainedbyappending Lorentzorinternalsymmetryindices.Inaddition,thedescriptionoflocalsupersymmetry(supergravity)iseasier. a.Indexconventions Ourthree-dimensionalnotationisasfollows:Inthree-dimensionalspacetime (withsignature Š ++)theLore ntzgroupis SL (2, R )(insteadof SL (2, C ))andthecorrespondingfundamentalrepresentationactsona real (Majorana)two-componentspinor =( +, Š).IngeneralweusespinornotationforallLorentzrepresentations,denotingspi norindicesbyGreekletters , ... , ... .Thusav ector(thethree-dimensionalrepresentation)willbedescri bedbyasy mmetricsecond-rankspinor V=( V++, V+ Š, VŠŠ)oratra celesssecond-rankspinor V .(Forcompari son,infour dimensionswehavespinors €andavectorisgivenbyahermitianmatrix V€.) Al lo urspinorswillbeanticommuting(Grassmann). Spinorindicesareraisedandloweredbythesecond-rankantisymmetricsymbol C,whichisalso usedtode“nethesquareofaspinor: C= Š C= 0 i Š i 0 = Š C, CC= [ ] Š ; = C, = C, 2=1 2 = i +Š.(2. 1.1) Werepres entsymmetrizat ionandantisymmetrizationof n i ndicesby()and[],respectively(withoutafactorof1 n ).Weoftenmakeuseoftheidentity A[ B ]= Š CAB,(2. 1.2)

PAGE 16

82.ATOYSUPERSPACEwhichfollowsfrom(2.1.1).Weuse C(insteadofthecustomaryreal )tosim p lify therulesforhermiti anconjugatio n.Inparticular,itmakes 2hermitian (r ecall and anticommute)andgivestheconventionalhermiticitypropertiestoderivatives(see below).Notehowev erthatwhereas isreal, isimaginary. b.Superspace Superspaceforsimplesupersymmetryisl abeledbythreespacetimecoordinates xandtwoanticommutin gspi norcoordinates ,denot edcollectivelyby zM=( x, ). Theyhavethehermiticityproperties( zM)= zM.Wede “nederivativesby { , } x [ , x] 1 2 ( ) ,(2. 1.3a) sothat themomentumoperatorshavethehermiticityproperties ( i )= Š ( i ),( i )=+( i ).(2.1 .3b) andthus( i M)= i M.(De “nite)integra tionoverasingleanticommutingvariable is de“nedsothattheintegralistranslat ionallyinvariant(seesec.3.7);henced 1=0,d =aconstantwhichw etaketobe1.Weobservethatafunction f ( )hasaterminatingTaylorseries f ( )= f (0)+ f(0)since { } =0imp lies 2=0.Thusd f ( )= f(0)sotha tintegrationis equivalenttodierentiation.Forourspinorial coordinatesd = andhence d = .(2. 1.4) Thereforethedoubleintegral d22= Š 1,(2.1 .5) andwecan de“nethe -function 2( )= Š 2= Š1 2 *** Weoftenu sethenotation X | toindicatethequantity X evaluatedat =0.

PAGE 17

2.2.Supersymmetryandsuper“elds92.2.Supersymmetryandsuper“elds a.Representations Wede “nefunctionsoversuperspace:...( x )wherethedotss tandforLorentz (s pinor)and/orinternalsymmetryindices.Theytransformintheusualwayunderthe Poincar egroupwith generators P(translations)and M(Lorentzrotations).We grade(ormakesuper)thePoincar ealgebraby introducin ga dditi onal spinor supersymmetrygenerators Q,satisfy ingthe supersymmetryalgebra [ P, P]=0,(2 .2.1a) { Q, Q} =2 P,(2. 2.1b) [ Q, P]=0,(2 .2.1c) aswellastheusualcommutationrelationswith M.Thisalgebra isrealizedon super“elds ...( x )inte rmsofderivativesby: P= i , Q= i ( Š i );(2.2 .2a) ( x, )= exp [ i ( P+ Q)] ( x+ Š i 2 ( ), + ).(2.2 .2b) Thus P+ Qgeneratesasupercoord inatetransformation x = x+ Š i 2 ( ), = + .(2. 2.2c) withreal,constantparameters Thereadercanverifythat(2.2.2)providesarepresentationofthealgebra(2.2.1). Were markinparticularthatiftheanticommutator(2.2.1b)vanished, Qwoulda nnihilateallphysicalstates(seesec.3.3).W ealsonotetha tb ecauseof(2.2.1a,c)and (2.2.2a),notonlyandfunctionsof butalsothespace-timederivatives ca rrya representationofsupersymmetry(aresuper“elds).However,becauseof(2.2.2a),thisis notthecaseforthesp inorialderivatives .Supersymmetricallyinvariantderivatives canbede“nedby DM=( D, D)=( , + i ).(2.2 .3)

PAGE 18

102.ATOYSUPERSPACETheset DM(anti)commut eswiththegenerators:[ DM, P]=[ DM, Q} =0.Weuse [ A B } todenoteagradedcommutator:anticommutatorifboth A and B arefermionic, commutatorotherwise. Thecovariantderivativescanalsobede“nedbytheirgradedcommutationrelations { D, D} =2 iD,[ D, D]=[ D, D]=0;(2 .2.4) or,moreconcisely: [ DM, DN} = TMN PDP; T = i ( ) rest =0.(2. 2.5) Thus,inthelanguageofdierentialgeometry,globalsuperspacehas torsion. The derivati vessatisfythefur theridentities = DD= i + CD2, DDD=0, D2D= Š DD2= i D,( D2)2= .(2. 2.6) TheyalsosatisfytheLeibnitzruleandcanbeintegratedbypartswheninside d3xd2 integrals(sincetheyareacombinationof x and derivati ves).Thefo llowingidentityis useful d3xd2 ( x )= d3x 2( x )= d3x ( D2( x )) | (2.2.7) (whererecallthat | meansevaluationat =0).Theextra space-timederivativesin D(ascomparedto )dropoutafter x -integration. b.Componentsbyexpansion Super“eldscanbeexpandedina(terminating)Taylorseriesin .Forex ample, ...( x )= A ...( x )+ ...( x ) Š 2F ...( x ).(2.2 .8) A B F arethe component “eldsof.Thesupersymmetr ytransfo rmationsofthecomponentscanbederivedfromt hoseofthesuper“eld.Forsimplicityofnotation,weconsiderascalarsuper“eld(noLorentzindices)

PAGE 19

2.2.Supersymmetryandsuper“elds11( x )= A ( x )+ ( x ) Š 2F ( x ),(2.2 .9) Thesupersymmetrytransformation( =0, i n“nitesimal) ( x )= Š ( Š i )( x ) A + Š 2 F ,(2. 2.10) gives,uponequa tingpowersof A = Š ,( 2.2.11a) = Š ( CF + i A ),(2.2 .11b) F = Š i .(2. 2.11c) Itiseasytoverifythatonthecomponent“eldsthesupersymmetryalgebraissatis“ed: Thecommutatoroftwotransformationsgivesatranslation,[ Q( ), Q( )]= Š 2 i etc. c.Actionsandcomponentsbyprojection Theconstructionof(integral)invariantsisfacilitatedbytheobservationthat supersymmetrytransformation sarecoordi natetransformationsinsuperspace.Because wecanign oretotal -derivat ives(d3xd2f=0,whichfo llowsfrom( )3=0)and totalspacetim ederivat ives,we “ndthat any superspaceintegral S = d3xd2 f (, D, ... )(2. 2.12) th at doesnotdependexplicitlyonthecoordinatesisinvariantunderthefullalgebra.If thesuper “eldexpansionintermsofcomponentsissubstitutedintotheintegralandthe -integrationiscarriedout,theresultingcomponentintegralisinvariantunderthe transformationsof(2.2.11)(theintegrandingeneralchangesbyatotalderivative).This alsocanbeseenfromthefactthatthe -integrationpicksoutthe F componentof f whichtransformsasaspacetimederivative(see(2.2.11c)). Wenowdes cribeatechnicaldevicethatcanbeextremelyhelpful.Ingeneral,to obtaincomponentexpressionsbydirect -expansionscanbecumbersome.Amore

PAGE 20

122.ATOYSUPERSPACEecientprocedureistoobservethatthecomponentsin(2.2.9)canbede“nedby projection: A ( x )=( x ) | ( x )= D( x ) | F ( x )= D2( x ) | .(2. 2.13) Thiscanbeused,forexample,in(2.2.12)byrewriting(c.f.(2.2.7)) S = d3xD2f (, D, ... ) | .(2. 2.14) Afterthe derivativesareevaluated(usingtheLeibnitzruleandpayingduerespectto theanticommutativityofthe D s),theresultisdir ectlyexpressibleintermsofthecomponents(2. 2.13).Thereadershouldverifyinafewsimpleexamplesthatthisisamuch moreecientprocedurethandirect -expansionandintegration. Finally,wecanalsoreobtainthecomponenttransformationlawsbythismethod. We“rst note theidentity iQ+ D=2 i .(2. 2.15) Thuswe“nd,forexample A = i Q | = Š ( D Š 2 i ) | = Š .(2. 2.16) Ingeneralwehave iQf | = Š Df | .(2. 2.17) Thisissucienttoobtainallofthecomponent“eldstransformationlawsbyrepeated applicationof(2.2.17),where f is, D, D2andwe use(2. 2.6)and(2.2.13).

PAGE 21

2.2.Supersymmetryandsuper“elds13d.Irreduciblerepresentations Ingeneralatheoryisdescribedby“eldswhichinmomentumspacearede“ned forarbitraryvaluesof p2.For any“xedvalueof p2the“eldsarearepresentationofthe Poincar egro up.Wecallsuch“elds,de“nedfor arbitrary valuesof p2,an o-shell representationoft hePoincar egro up.Similarly,whenasetof“eldsisarepresentationofthe supersy mmetry algebrafor any valueof p2,wecallitano-shellrep resentationofsupersymmetry.Whenthe“eldequationsareimposed,aparticularvalueof p2(i.e., m2)is pi ck ed out.Someofthecomponentsofthe“elds(auxiliarycomponents)arethenconstrainedtovanish;theremaining(physical)componentsformwhatwecallan on-shell representationofthePoincar e(orsupersy mmetry)group. Asuper “eld ...( p )isani rreduciblerepresentationoftheLorentzgroup,with regardtoitsexternalindices,ifitistotallysymmetricintheseindices.Forarepresentationofthe(super)Poincar egroupweca nre duceitfurther.Since inthreedimensions thelittlegroupis SO (2),anditsirreduciblerepresen tationsareonecomponent(complex),thisreductionwillgiveone-componentsuper“elds(withrespecttoexternal i ndices).Suchsuper“eldsareirreduciblerepresentationsofo-shellsupersymmetry, whenarealityconditionisimposedin x -space(butthesuper“eldisthenstillcomplexin p -space,where( p )= ( Š p )). Inanappropriatereferenceframewecanassignhelicity(i.e.,theeigenvalueof the SO (2)gen erator) 1 2 tothespino ri ndices,andtheirreduciblerepresentationswill belabeledbythesupe rhelicity(thehelicityofthesuper“eld):halfthenumberof+ externalindicesminusthenumberof Š s.Inthisframewecanalsoassign 1 2 he licity to .Expa ndingthesuper“eldofsuperhelicity h intocomponents,weseethatthese componentsha vehe licities h h 1 2 h .Forex ample,a scalarmultiplet, consistingof spin s( i.e., SO (2,1)representations)0,1 2 (i.e.,helicities0, 1 2 )isdes cribedbya super“eldofsuperhelicity0:ascalarsuper“eld.A vectormultiplet, consistingofspins1 2 ,1(he licities0,1 2 ,1 2 ,1)isdes cribedbyasuper“eld ofsuperhelicity+1 2 :the+componentofas pinorsuper“eld;the Š compon entbeinggaugedaway(inalight-cone gauge).Ingeneral,thesuperhelicitycontentofasuper“eldisanalyzedbychoosinga gauge(thesupersymmetriclight-conegauge)whereasmanyaspossibleLorentzcomponentsofasuper“eldhavebeengaugedto0:the superhelicitycontentofanyremaining

PAGE 22

142.ATOYSUPERSPACEcomponentissimply1 2 thenumberof+sminus Š s.Unlesso therwise stated ,wew ill automaticallyconsider all thr ee-dimensionalsuper“eldstobe real.

PAGE 23

2.3.Scalarmultiplet152.3.Scalarmultiplet Thesimplestrepresentationofsupersymmetryisthescalarmultipletdescribed bythereals uper“eld( x ),andcontainingthescalars A F andthetwo-component spinor .From(2. 2.1,2)weseethat hasdimension( mass )Š1 2 .Also,thec anonical dimensionsofcomponent“eldsinthreedimensionsare1 2 lessthaninfo urdimensions (b ecauseweused3x insteadofd4x inthekineticterm).There fore,ifthismultiplet istodescribe physical“elds,wemustassigndimension( mass )1 2 tosothat has canonicaldimension( mass )1.(Althou ghitisnotimmediatelyobviouswhichscalar shouldhavecanonicaldimension,thereisonlyonespinor.)Then A willhavedimension ( mass )1 2 andwillbethephysicalscalarpartnerof ,whereas F hastoohighadimensiontodescribeacanonicalphysicalmode. Sincea integralisthesameasa deriva tive,d2 hasdimension( mass )1. Therefore,ondimensionalgroundsweexpectthefollowingexpressiontogivethecorrect (massless)kineticactionforthescalarmultiplet: Skin= Š1 2 d3xd2 ( D)2,(2. 3.1) (recallthatforanyspinor wehave 2=1 2 ).Thisex pressionisreminiscentof thekineticactionforanordinaryscalar“eldwiththesubstitutions d3x d3xd2 and D.Thecompon entexpressioncanbeo btainedb yexp licit -expansionand integration.However,wepr efertousethealternativeprocedure(“rstintegrating Dby parts): Skin=1 2 d3xd2 D2 =1 2 d3xD2[ D2] | =1 2 d3x ( D2 D2+ D DD2+( D2)2) | =1 2 d3x ( F2+ i + A A ),(2.3 .2)

PAGE 24

162.ATOYSUPERSPACEwherewehaveusedtheidentities(2.2.6)andthede“nitions(2.2.13).The A and kinetictermsareconventional,while F isclearlynon-propagating. Theauxiliary“eld F canbeeliminatedfromtheact ionbyusingitsequationof motion F =0(or,inaf unction alintegral, F canbetriviallyintegratedout).The resultingactionisstillinvariantunderthe bo se-fermitransforma tions(2.2.11a,b)with F =0;howev er,thesearenotsupersymmetrytrans formations(notare presentationof thesupersymmetryalgebra)exceptonshell.Thecommutatoroftwosuchtransformationsdoesnotclose(doesnotgiveatranslation)exceptwhen satis“esits“eldequation.Thiso-shellnon-closureofthealgebraistypicaloftransformationsfromwhich auxiliary“eldshavebeeneliminated. Massandinter actiontermscanbeaddedto(2.3.1).Aterm SI= d3xd2 f (),(2 .3.3) leadstoacomp onentaction SI= d3xD2f () | = d3x [ f()( D)2+ f() D2] | = d3x [ f( A ) 2+ f( A ) F ].(2.3 .4) Inarenormalizablemodel f ()canbe atmostquartic.Inparticular, f ()=1 2 m 2+1 6 3givesmassterms,Yukawaandc ubicinterac tionterms.Together withthekineticterm,weobtain d3xd2 [ Š1 2 ( D)2+1 2 m 2+1 6 3] = d3x [1 2 ( A A + i + F2) + m ( 2+ AF )+ ( A 2+1 2 A2F )].(2.3.5) F canagainbeeliminatedusingits(algebraic)equationofmotion,leadingtoa

PAGE 25

2.3.Scalarmultiplet17conventionalmasstermandquarticinteractionsforthescalar“eld A .Moreexotic kineticactionsarepossiblebyusinginsteadof(2.3.1) S kin= d3xd2 ( ,), = D,(2.3 .6) whereissomefunctionsuchthat2 | ,=0= Š1 2 C.Ifweint roducemorethan onemultipletofscalarsuper“elds,then,forexample,wecanobtaingeneralizedsupersymmetricno n linearsigmamodels: S = Š1 2 d3xd2 gij()1 2 ( Di)( Dj)(2. 3.7)

PAGE 26

182.ATOYSUPERSPACE2.4.Vectormultiplet a.Abeliangaugetheory Inaccordancewiththediscussioninsec.2.2,arealspinorgaugesuper“eldwithsuperhelicity h =1 2 ( h = Š1 2 canbegaugedaway)willconsistofcomponentswith he licities0,1 2 ,1 2 ,1.Itcanbeused todescribeamasslessgaugevector“eldandits fermionicpartner.(Inthreedimensions,agaugevectorparticlehasonephysicalcomponentofde“nitehelicity.)Thesuper“eldcan beintrodu cedbyanalogywithscalarQED (thegeneralizationtothenonabeliancaseisstraightforward,andwillbediscussed below).Consideracomp lexscalarsuper“eld(a doubletofrealscalarsuper“elds)transformingundera cons tant phasero tation = eiK, = eŠ iK.(2. 4.1) ThefreeLagrangian | D |2is in va riantunderthesetransformations. a.1.Gaugeconnections Weextend thistoa local phaseinvariancewith K areals calarsuper“elddependingon x and ,bycovaria ntiz ingthespinorderivatives D: D= DŠ+ i ,(2. 4.2) whenactingonor ,respectively.Thespinorgaugepotential(orconnection)transformsintheusualway = DK ,(2. 4.3) toensure = eiKeŠ iK.(2. 4.4) Thisisrequiredby( )= eiK( ),andguaranteesthattheLagrangian | |2is locallygaugeinvariant.(Thecouplingconstantcanberestoredbyrescaling g ).

PAGE 27

2.4.Vectormultiplet19Itisnowstraightforward,byanalogywithQED,to“ndagaugeinvariant“eld strengthandactionforthemultipletdescribedbyandtost udyitscomponentcoup lingstothecomplexscalarmultipletcontainedin | |2.H ow ever,bothtounderstand itsstructureasanirreduciblerepresentati onofsupersymmetry,andasanintroduction tomorecomplicatedgaugesuper“elds(e.g.insupergravity),we“rstgiveageometrical presentation. Althoughtheactionswehave considereddonotcontainthespacetimederivative ,inother contextsweneedthecovariantobject = Š i = K ,(2. 4.5) introducingadistinct(vector)gaugepotentialsuper“eld.Thetransformation of thisconn ectionischosentogive: = eiKeŠ iK.(2. 4.6) (Fromageometricviewpoint,itisnaturaltointroducethevectorconnection;thenandcanberegardedasthecomponentsofasuper1-formA=(,);se es ec. 2.5).However,wewill“ndthatshouldnotbeindependent,andcanbeexpressedin termsof. a.2.Comp onents Togetoriented, weexaminethe componentsofintheTaylorseries -expansion. Theycanbede“neddirectlybyusingthespinorderivatives D: =| B =1 2 D| V=Š i 2 D( )| =1 2 DD| ,(2. 4.7a) and W=| = D| = D( )| T= D2| .(2. 4.7b) Wehavese paratedthecomponentsintoirreduciblerepresentationsoftheLorentzgroup, th atis,traces(orantisymmetrizedpieces,see(2.1.2))andsymmetrizedpieces.Wealso

PAGE 28

202.ATOYSUPERSPACEde“nethecomponentsofthegaugeparameter K : = K | = DK | = D2K | (2.4.8) Thecomponentgaugetransformationsforthecomponentsde“nedin(2.4.7)arefound byrepeatedlydi erentiating(2.4.3-5)withspinorderivatives D.W e “nd: = B = V= =0,(2. 4.9a) and W= = = ( ), T= .(2. 4.9b) Notethat and B suerarbitraryshiftsasaconsequenceofagaugetransformation, and,inparticular,canbegaugedcompletelyaway;thegauge = B =0isca lled WessZumino gauge,and explicitly breakssupersymmetry.However,thisgaugeisusefulsince itrevealsthe physicalcontentofthemult iplet. Examinationofthecomponentsthatremainrevealsseveralpeculiarfeatures: Thereare two componentgaugepotentials Vand Wforonly one gaugesymmetry, andthereisahighdimensionspin3 2 “eld .These problemsw illberesolvedbelow whenweexpressintermsof. Wecanalso “ndsupersymmetricLorentzgaugesby“xing D;such gaugesare usefulforquantization(seesec.2.7).Furthermore,inthreedimensionsitispossibleto chooseasuper symmetriclight-conegauge+=0.(Intheabe liancasethegaugetransformationtakesthesimpleform K = D+( i ++)Š 1+.)Eq.(2.4.14)belowimpliesthatin thisgaugethesuper“eld++alsovanishes.Theremainingcomponentsinthisgaugeare Š, V+ Š, VŠŠ,and Š,with V++=0and + ++Š. a.3.Constraints To understandhowthevectorconnectioncanbeexpressedintermsofthe spinorconnection,r ecallthe(an ti)commutationrelationsfortheordinaryderivatives are:

PAGE 29

2.4.Vectormultiplet21[ DM, DN} = TMN PDP.(2. 4.10) Forthecovarian tderiv atives A=( )thegr adedcommutationrelationscanbe written(from(2.4.2)and(2.4.5)weseethatthetorsion TAB Cisunmodi“ed): [ A, B} = TAB CCŠ iFAB.(2. 4.11) The“eldstrengths FABareinvariant( F AB= FAB) duetothecovarianceofthederivatives A.Observethat the“eldstrengthsareantihermitianmatrices, FAB= Š FBA,so thatthesymmetric“eldstrength Fisimaginarywhiletheantisymmetric“eld strength F is real.Examiningaparticularequationfrom(2.4.11),we“nd: {, } =2 i Š iF=2 i +2Š iF.(2. 4.12) Thesuper“eldwasintrodu cedtocovariantizethespace-timederivative .However,itisclearthatanalternativechoiceis =Ši 2 Fsince Fiscovariant(a “eldstrength).Thenewcovariantspace-time derivati vew illthensatisfy(wedropthe primes) {, } =2 i ,(2. 4.13) withthenewspace-timeconnectionsatisfying(aftersubstitutingin2.4.12theexplicit forms A= DAŠ i A) = Š i 2 D( ).(2. 4.14) Thusthe con ventionalconstraint F=0,(2. 4.15) imposedonthesystem(2.4.11)hasallowedthevectorpotentialtobeexpressedinterms ofthespinorpotential.Thissolvesboththeproblemoftwogauge“elds W, Vand theproblemoftheh ighers pinanddimensioncomponents T:The gauge“elds areidenti“edwitheachother( W= V),andtheextracompone ntsareexpr essedas de rivativesoffamiliarlowerspinanddimension“elds(see2.4.7).Theindependentcomponentst hatremaini nWe ss-Zuminogaugeaftertheconstraintisimposedare Vand .

PAGE 30

222.ATOYSUPERSPACEWestre sstheimportanceoftheconstraint (2.4.15)ontheobjectsde“nedin (2.4.11).Unconstrained“eldstrengthsingeneralleadtoreduciblerepresentationsof supersymmetry(i.e.,thespinorandvectorpo tentials),andtheconstraintsareneededto en su reirreducibility. a.4.Bianchiidentities Inordinary“eldtheories,the“eldstrengthssatisfyBianchiidentitiesbecausethey areexpressedintermsofthepotentials;theyare identi ties andcarrynoinformation. For gaugetheoriesdescribedbycovariantderivatives,theBianchiidentitiesarejust Jacobiidentities: [ [ A,[ B, C )}} =0,(2. 4.16) (where[)isthe graded antisymmetrizationsymbol,identicaltotheusualantisymmetrizationsymbolbutwithanextrafactorof( Š 1)foreachpairofinterchanged fermionicindices).However,onceweimpose constraintssuchas(2.4.13,15)onsomeof the“eldstrengths,theBianch iidentit iesimplyconstraintsonother“eldstrengths.For example,theidentity 0=[ {, } ]+[ {, } ]+[ {, } ] =1 2 [ ( {, )} ](2. 4.17) gives(usingtheconstraint(2.4.13,15)) 0=[ ( )]= Š iF( ).(2. 4.18) Thusthetotallysymmetricpartof F vanishes.Ing eneral,wecandecompose F into irreduciblerepresentati onsoftheLo rentzgroup: F =1 6 F( )Š1 3 C ( |F | )(2.4.19) (where i ndicesbetween| ... |,e.g .,inthiscase ,arenoti ncludedinthesymmetrization).Hencetheonlyremainingpieceis: F = iC ( W ),( 2.4.20a) whereweintroducethesuper“eldstrength W.Wecanco mpute F intermsof

PAGE 31

2.4.Vectormultiplet23and “nd W=1 2 DD.(2. 4.20b) Thesuper“eld Wistheonlyindependentgaugeinvariant“eldstrength,andis constrainedby DW=0,whichfo llowsfromtheBianchiidentity(2.4.16).This impliesthato nlyoneLorentzcomponentof Wisindepe ndent.The“eldstrength describesthephysicaldegreesoffreedom:onehelicity1 2 andonehelicit y1mode.Thus Wisasuita bleobjectforconstructinganaction.Indeed,ifwestartwith S =1 g2 d3xd2 W2=1 g2 d3xd2 (1 2 DD)2,(2. 4.21) wecancom putethecomponentaction S =1 g2 d3xD2W2=1 g2 d3x [ WD2WŠ1 2 ( DW)( DW)] | =1 g2 d3x i Š1 2 ff .(2. 4.22) Here(cf.2.4.7) W| while f= DW| = DW| isthespi norformoftheusual “eldstrength F | =( Š ) | =1 2 ( ( f ) )= Š i1 2 [ D( )Š D( )] | .(2. 4.23) Toderivethea bovecompon entactionwehaveusedtheBianchiidentity DW=0,and itsconse quen ce D2W= i W. a.5.Matterc ouplings Wenowexami nethecomponent Lagrangiandescribingthecouplingtoacomplex scalarmultip let.Wecouldstartwith S = Š1 2 d3xd2 ( )( )

PAGE 32

242.ATOYSUPERSPACE= Š1 2 d3xD2[( D+ i ) ][( DŠ i )],(2 .4.24) andworkouttheLagrangianintermsofcomponentsde“nedbyprojection.However,a moreecientprocedure,whichleadstophysicallyequivalentresults,istode“ne covariantcomponents ofby covariant projection A =( x ) | = ( x ) | F = 2( x ) | .(2. 4.25) Thesecomponentsarenotequaltotheordinaryonesbutcanbeobtainedbya(gauge“elddependent)“eldrede“nitionandprovideanequallyvaliddescriptionofthetheory. Wecanalsouse d3xd2 = d3xD2| = d3x 2| ,(2. 4.26) whenactingonaninvariantandhence S = d3x 2[ 2] | = d3x [ 2 2+ 2+ ( 2)2] | = d3x [ FF + ( i + V ) +( i A + h c .)+ A ( Š iV)2A ].(2.4.27) Wehave usedthecommu tationrelationsofthecovariantderivativesandinparticular 2= i + iW, 2= Š i Š 2 iW,( 2)2= Š iW,where is the covariant dAlembertian(covariantizedwith). b.Nonabeliancase Wenowbrie”ycon siderthenonabeliancase:Foramu ltipletofscalarsuper“elds transformingas= eiK,where K = KiTiand TiaregeneratorsoftheLiealgebra, weintro ducecovariantspinorderivatives preciselyasfortheab eliancase(2.4.2). Wede “ne= iTisothat

PAGE 33

2.4.Vectormultiplet25= DŠ i = DŠ i iTi.(2. 4.28) Thespinorconnectionnowtransformsas = K = DK Š i [, K ],(2.4 .29) leaving(2.4.4)unmodi“ed.Thevectorconnectionisagainconstrainedbyrequiring F=0;inotherwo rds,wehave =Š i 2 {, } ,( 2.4.30a) = Š i1 2 [ D( )Š i { ,} ].(2.4 .30b) Theformoftheaction(2.4.21)isunmodi“ed(exceptthatwemustalsotakeatraceover groupindices).Theconstraint(2.4.30)impliesthattheBianchiidentitieshavenontrivialconsequences,andallowsustosolve(2.4.17)forthenonabeliancaseasin (2.4.18,19,20a).Thus,weobtain [ ]= C ( W )(2.4.31a) intermsofthenonabelianformofthe covariant “eldstrength W : W=1 2 DDŠi 2 [, D] Š1 6 [, { ,} ].(2.4 .31b) The“eldstrengthtransformscovariantly: W = eiKWeŠ iK.There mainingBianchi identityis [ {, } ] Š{( ,[ ), ] } =0 .( 2.4.32a) Contractingindiceswe“nd[ {, } ]= {( ,[ ), ] } .Howev er, [ {, } ]=2 i [ ]= 0a nd he n ce,using(2.4.31a), 0= {( ,[ ), ] } = Š 6 {, W} .(2. 4.32b) ThefullimplicationoftheBianchiidentitiesisthus: {, } =2 i (2.4.33a) [ ]= C ( W ), {, W} =0(2.4 .33b) [ ]= Š1 2 i ( ( f ) ), f1 2 {( W )} .(2. 4.33c)

PAGE 34

262.ATOYSUPERSPACEThecomponentsofthemultipletcanbede“nedinanalogyto(2.4.7)by projections of: =| V=| B =1 2 D| = W| (2.4.34) c.Gaugeinvar iantmasses Acurio usfeaturewhichthistheoryhas,andwhichmakesitratherdierentfrom fo ur di mensionalYang-Millstheory,istheexistenceofagauge-invariantmassterm:In theabeliancasetheBianchiidentity DW=0canbeuse dtoprovetheinvarianceof Sm=1 g2 d3xd2 1 2 m W .(2. 4.35) Incomponentsthisactioncontainstheusualgaugeinvariantmasstermforthree-dimensionalelectrodynamics: m d3xVV = m d3xVf,(2. 4.36) whichisgaugeinvariantasaconsequenceoftheusualcomponentBianchiidentity f=0. Thesuper“eldequationswhichresultfrom(2.4.21,35)are: i W+ mW=0,(2. 4.37) whichdescribesanirreduciblemultipletofmass m .TheBianch iidentity DW=0 impliesthato nlyoneLorentzcomponentof W isindepe ndent. Forthe nonabeliancase,themas stermissom ewhatmorecomplicatedbecausethe “eldstrength W iscovariantratherthaninvariant: Sm= tr1 g2 d3xd21 2 m (W+i 6 { ,} D+1 12 { ,}{ ,} )

PAGE 35

2.4.Vectormultiplet27= tr1 g2 d3xd21 2 m ( WŠ1 6 [,]).(2. 4.38) The“eldequations,however,arethecovariantizationsof(2.4.37): i W+ mW=0.(2. 4.39)

PAGE 36

282.ATOYSUPERSPACE2.5.Otherglobalgaugemultiplets a.Superforms:generalcase Thegaugemultipletsdiscussedinthelastsectionmaybedescribedcompletelyin termsofgeometricquantities.ThegaugepotentialsA (,)whichcova riantize thederivatives DAwithrespecttolocalphaserotationsofthemattersuper“eldsconstituteasuper1-fo rm.Wede“nesuper p -formsastensorswith p covariantsupervector indices(i.e.,supervectors ubscripts)thathavetotal graded antisymmetrywithrespectto theseindices(i.e.,aresymmetricinanypairofspinorindices,antisymmetricinavector pairor inamixedpair).Forexample,the“eldstrength FAB ( F , F , F )constitutesasuper2-form. IntermsofsupervectornotationthegaugetransformationforA(from(2.4.3)and (2.4.5))takestheform A= DAK .(2. 5.1) The“eldstrengthde“nedin(2.3.6)whenexpressedintermsofthegaugepotentialcan bewri ttenas FAB= D[ AB )Š TAB CC.(2. 5.2) Thegaugetransformationlawcertainlytakesthefamiliarform,butevenintheabelian case,the“eldstrengthhasanunfamiliarn onderivativeterm.Onewaytounderstand howthistermarisesistomak eachan geofbasisforthecomponentsofasupervector. Wecanexpand DAintermsofpartialderivativesbyintroducingamatrix, EA M,such that DA= EA MM, M ( , ), EA M= 01 2 i ( )1 2 ( ) .(2. 5.3) Thismatrixisthe ”atvielbein; itsinverseis

PAGE 37

2.5.Otherglobalgaugemultiplets29EM A= 0 Š1 2 i ( )1 2 ( ) .(2. 5.4) Ifwede“neMbyA EA MM,then M= MK .(2. 5.5) Similarly,ifwede“ne FMNby FAB ( Š )A ( B + N )EB NEA MFMN,(2. 5.6a) then FMN= [ MN ).(2. 5.6b) (IntheGrassmannparityfactor( Š )A ( B + N )thesuper scripts A B ,and N areequalto onewhent heseindicesrefertospinorialindicesandzerootherwise.)Wethusseethat thenonderivativeterminthe“eldstrengthisabsentwhenthecomponentsofthis supertensorarereferredtoadierentcoordi natebasis.Furthermore,inthisbasisthe Bianchiidentitiestakethesimpleform [ MFNP )=0.(2. 5.7) Thegeneralizationtohigher-rankgradedantisymmetrictensors(superforms)is nowevident.Thereisabasis inwhichthegaugetransformation,“eldstrength,and Bianchiidentitiestaketheforms M1... Mp=1 ( p Š 1)! [ M1KM2... Mp), FM1... Mp +1=1 p [ M1M2... Mp +1), 0= [ M1FM2... Mp +2).(2. 5.8) Wesimplymu ltiplythesebysuitablepowersofthe ”atvielbeinandappropriateGrassmannparityfactorstoobtain A1... Ap=1 ( p Š 1)! D[ A1KA2... Ap)Š1 2( p Š 2)! T[ A1A2| BKB | A3... Ap),

PAGE 38

302.ATOYSUPERSPACEFA1... Ap +1=1 p D[ A1A2... Ap +1)Š1 2( p Š 1)! T[ A1A2| BB | A3... Ap +1), 0=1 ( p +1)! D[ A1FA2... Ap +2)Š1 2 p T[ A1A2| BFB | A3... Ap +2).(2. 5.9) (The | sindicatethatalloft heindicesaregradedantisymmetricexceptthe B s.) b.Super2-form Wenowdiscuss indetailthecaseofasuper2-formgaugesuper“eldABwith gaugetransformation = D( K )Š 2 iK, = DKŠ K, = KŠ K.(2. 5.10) The“eldstrengthforABisasuper3-form: F , =1 2 ( D( )+2 i ( )), F , = D( ), + Š 2 i , F , = D + Š , F , = + + .(2. 5.11) Alloftheseeq uationsarecontainedintheconcisesupervectornotationin(2.5.9). Thegaugesuper“eldAwassubj ecttoconstraintsthatallowedonepart( )to beexpr essedasafunctionoftheremainingpart.Thisisageneralfeatureofsupersymmetricgaugetheories;constraintsareneededtoensureirreducibility.Forthetensor gaugemultipletweimposetheconstraints F , =0, F , = i ( ) G = T G ,(2. 5.12) which,asweshowbelow,allowustoexpressallcovariantquantitiesintermsofthesinglerealscalarsuper“eld G .Theseco nstraintsc anbesolvedasfollows:we“rstobserve thatinthe“eldstrengths alwaysappearsinthecombination D( )+2 i ( ).

PAGE 39

2.5.Otherglobalgaugemultiplets31Therefore,withoutchangingthe“eldstrengthswecanrede“ne byabsorbing D( )intoit.Thus disappearsfromthe“eldstrengthswhichmeansitcouldbe settozerofromthebeginning(equivalently,wecanmakeitzerobyagaugetransformation).The“rstconstraintnowimpliesthatthetotallysymmetricpartof is zero andhencewecanwrite = iC ( )intermsofas pinorsuper“eld.The remainingequationsandconstraintscanbeusednowtoexpress andtheother “eldstrengthsintermsof.We “ndasolution =0, = iC ( ), =1 4 ( ( [ D ) )+ D ) )], G = Š D.(2. 5.13) ThustheconstraintsallowABtobeexpr essedintermsofaspinorsuper“eld.(The generalsolutionoftheconstraintsisagaugetransform(2.5.10)of(2.5.13).) Thequantity G isbyde “nitiona“eldstrength;hencethegaugevariationofmustleave G invariant.Thisimpliesthatthegaugevariationofmustbe(see (2.2.6)) =1 2 DD,(2. 5.14) whereisanarbitraryspinorgaugeparameter.Thisgaugetransformationisofcourse consistentwithwhatremainsof(2.5.10)afterthegaugechoice(2.5.13). Weexp ectthephysicaldegreesoffreedomtoappearinthe(onlyindependent) “eldstrength G .Sinceth isisascalarsuper“eld,itmustdescribeascalarandaspinor, and(orAB)providesa variantrepresentation ofthesupersymmetryalgebranormallydescribedbythescalarsuper“eld.Infactcontainscomponentswithhelicities0,1 2 ,1 2 ,1just likethevectormultiplet,butnowthe1 2 ,1 co mponentsareauxiliary “elds.(= + A + vŠ 2).Forwithcanonicaldimension( mass )1 2 ,on dimensionalgroundsthegaugeinvariantactionmustbegivenby S = Š1 2 d3xd2 ( DG )2.(2. 5.15) Wri tteninthisformweseethatintermsofthecomponentsof G ,theacti onha sthe

PAGE 40

322.ATOYSUPERSPACEsameformasin(2.3 .2).Theonlydier encesarisebecause G isexpr essedintermsof .W e “nd thatonlytheauxiliary“eld F ismodi “ed;itisreplacedbya“eld F.An exp licitcomputationofthisquantityyields F= Š D2D| = i D| V| V1 2 iD( ).(2. 5.16) Inplaceof F thedivergenceofavectorappears.Toseethatthisvector“eldreallyisa gau ge “eld,wecomputeitsvariationunderthegaugetransformation(2.5.14): V=1 4 ( [ D )+ D )].(2.5 .17) Thisisnotthetransformationofanordinarygaugevector(see(2.4.9)),butratherthat ofasecond-ra nkantisymmetrictensor(inthreedimensionsasecond-rankantisymmetric tensoristhesameLorentzrepresentationasavector).Thisisthecomponentgauge “eldthat appearsatlowestorderin in ineq.(2.5.13).A“eldofthistypehasno dynamicsinthreedimensions. c.Spinorgaugesuper“eld Superformsarenottheonly gaugemultipletsonecanstudy,butthepatternfor othercasesissimilar.Ingeneral,(nonvari ant)supersymmetricgaugemultipletscanbe describedbyspinorsuper“eldscarryingadditionalinternal-symmetrygroupindices.(In aparticularc ase,theadditionalindexcanbeaspi norindex:seebelow.)Suchsuper“e ldscontaincomponentgauge“eldsand,asintheYang-Millscase,theirgaugetransformationsaredeterminedbythe =0partofth esuper “eld gaugeparameter(cf. (2.4.9)).Thegaugesuper“eldthustakesth eformoftheco mponent“eldwithavector i ndexreplacedbyaspinorindex,andthetransformationlawtakestheformofthecomponenttransf ormationlawwiththevectorderivati verepla cedbyaspinorderivative. Forexample,todes cribeamu ltipletcontainingaspin3 2 componentgauge“eld,we introduceaspinorgaugesuper“eldwithanadditionalspinorgroupindex: = DK.(2. 5.18) The“eldstrengthhasthesameformasthevectormultiplet“eldstrengthbutwitha spinorgroupindex:

PAGE 41

2.5.Otherglobalgaugemultiplets33W =1 2 DD .(2. 5.19) (Wecan,ofcourse,introduceasupervectorpotentialM inexactanalogywiththe abelianvectormultiplet.The“eldstrengthheresimplyhasanadditionalspinorindex. Theconstraintsareexactlythesameasforthevectormultiplet,i.e., F =0.) Inthreedimensionsmassless“eldsofspingreaterthan1havenodynamical de greesoffreedom.Thekinetictermforthismultipletisanalogoustothe massterm forthevector mult iplet: S d3xd2 W.(2. 5.20) Thisactiondescribescomponent“eldswhichareallauxiliary:aspin3 2 gauge“eld ( ) ,av ector,andascalar,ascanbeveri“edbyexpandingincomponents.The invarianceoftheactionin(2.5.20)isnotmanifest:ItdependsontheBianchiidentity DW=0.Theex p licitappearanceofthesuper“eldisagen eralfeatureofsupersymmetricgaugetheories;itis not alwayspossibletowritethesuperspaceactionfora gaugetheoryintermsof“eldstrengthsalone.

PAGE 42

342.ATOYSUPERSPACE2.6.Supergravity a.Supercoordinatetransformations Supergravity,thesupersymmetricgenera lizationofgravity,isthegaugetheoryof thesupertranslations.Theglobaltransformationswithconstantparameters , generatedby Pand Qarereplacedbylocalonesparametrizedbythesupervector KM( x )=( K, K).Forascalarsuper“eld( x )wede “nethetransformation ( z ) ( z )= eiK( z )= eiK( z ) eŠ iK,(2. 6.1) where K = KMiDM= Ki + KiD.(2. 6.2) (Toexhibittheglobalsupersymmetry,itisconvenienttowrite K intermsof Drather than Q(or ).Thisamountstoarede“nitionof K).Thesecondformofthe transformationofcanbeshowntobeequivalenttothe“rstbycomparingtermsina powerseriesexpan sionofthetwoformsandnotingthat iK =[ iK ,].It iseasy tosee that(2.6.1)isageneralcoordin atetransformationinsuperspace: eiK( z ) eŠ iK=( eiKzeŠ iK);de “ning z eŠ iKzeiK,(2. 6.1)becomes( z)=( z ). We mayexpect,byanalogytotheYang-Millscase,tointroduceagaugesuper“eld H Mwith(linearized)transformationlaws H M= DKM,(2. 6.3) (weint roduce H Maswell,butaconstraintwillrelateitto H M)and de“necovariant derivati vesbyana logyto(2.4.28): EA= DA+ HA MDM= EA MDM.(2. 6.4) EA Misthe vielbein. Thepotentials H , H havealar genumberofcomponents amongwhichweidentify,accord ingtothe discussionfollowingequation(2.5.17),aseco nd-ranktensor(thedreibein,minusits”at-spacepart)describingthegravitonanda spin3 2 “elddescribingthegravitino,whosegaugeparametersarethe =0part softhe v ectorandspinorgaugesuperparameters KM| .Other componentswilldescribegauge de greesoffreedomandauxiliary“elds.

PAGE 43

2.6.Supergravity35b.Lorent ztransfo rmations Thelocalsupertranslationsintroducedso farincludeLorentztransformationsofa scalarsuper“eld,actingonthecoordinates zM=( x, ).Tode “netheiractionon spinorsuper“eldsitisnecessarytointroducetheconceptoftangentspaceandlocal framesattachedateachpoint zMandlocalLorentztransformationsactingonthe i ndicesofsuchsuper“elds ...( zM).(Inchapter5wediscuss thereasonsfo rthisprocedure.)Theenlargedfulllocalgroupisde“nedby ...( x ) ...( x )= eiK ...( x ) eŠ iK,(2. 6.5) wherenow K = KMiDM+ K iM .(2. 6.6) Herethes uper“eld K parametrizesthelocalLorentzt ransformationsandtheLorentz generators M actoneachtangentspaceindexasindicatedby [ X M ,]= X ,(2. 6.7) forarbitrary X Missymmetric,i.e., M istra celess(whichmakesitequivalentto av ectorinthreedimensions).Thus, X isanel ementoftheLorentzalgebra SL (2, R ) (i.e., SO (2,1)).Therefore,th epar ametermatrix K isalsotr aceless. Fromnowonwem ustdistinguishtangentspaceandworldindices;todothis,we denotetheformerbyle ttersfromthebeginningofthealphabet,andthelatterbyletters fromthemiddleofthealphabet.Byde“nition,theformertransformwith K whereas thelattertransformwith KM. c.Covariantderivatives Havingintro ducedlocalLorentztransformationsactingonspinorindices,wenow de“necovariantspinorderivativesby = E MDM+ M ,(2. 6.8) aswellasvectorderivatives .H ow ev er ,j us ta si nt he Ya ng -M illscase,weimposea conventionalconstraintthatde“nes = Š i1 2 {, } ,(2. 6.9)

PAGE 44

362.ATOYSUPERSPACEThe co nnectioncoecients A ,whichap pearin A= EA MDM+A M ,(2. 6.10) andactasgauge“eldsfortheLorentzgroup,willbedeterminedintermsof H Mby imposingfurt hersuitableconstraints.Thecova riantderivativestransformby AA = eiKAeŠ iK.( 2.6.11a) All “elds ...(asopposedtotheoperator )trans formas ...= eiK...eŠ iK= eiK...(2.6.11b) whenallindicesare”at(tangentspace);wealwayschoosetouse”atindices.Wecan usethevielbein EA M(anditsinverse EM A)toconvertbe tw eenworldandtangentspace i ndices.Forexample,ifMisaworldsupervector,A= EA MMisatangentspace supervector. Thesuperderivative EA= EA MDMis to be understoodasatangentspacesuperv ector.Ontheotherhand, DMtr an sformsunderthelocaltranslations(supercoordinate transformations),andthisinducestransformationsof EA Mwithrespecttoitsworld index(inthiscase, M ).Wecanexhibitthis,andverifythat(2.6.6)describesthefamilia rl oc al Lorentzandgeneralcoordinatetransformations,byconsideringthein“nitesimal versionof( 2.6.11): A=[ iK A],(2.6 .12) whichimplies EA M= EA NDNKMŠ KNDNEA MŠ EA NKPTPN MŠ KA BEB M, A = EAK Š KMDMA Š KA BB Š K A + K A = AK Š KMDMA Š KA BB ,(2. 6.13) where TMN Pisthetorsionof ”at,global superspace(2.4.10),and K 1 2 K( ( ) ). The“rstthreetermsinthetransformationlawof EA Mcorrespondtotheusualformof thegeneralcoordinatetransformationofaworldsupervector(labeledbyM),whilethe lasttermisalocalLorentztransformationonthetangentspaceindex A .Therelation betw een K and K impliestheusualreducibilityoftheLorentztransformationson

PAGE 45

2.6.Supergravity37thetangentspace,correspondingtothede “nitionofvectorsassecond-ranksymmetric spinors. d.Gaugechoices d.1.Asupersymmetricgauge Aswehavementio nedabove,thegauge“e lds(orthevielbein EA M)containa largenumberofgaugedegreesoffreedom,andsomeofthemcanbegaugedawayusing the K transformations.Forsimplicitywediscussthisonlyatthelinearizedlevel(where wen eednotdistinguishworldandtangentspaceindices);wewillreturnlatertoamore complete treatment.From(2.6.13)thelinearizedtransformationlawsare E = DKŠ K , E = DKŠ i ( K ).(2. 6.14) Thus K canbeusedtogaugeawayallof E exceptitstrace(recallthat K is tra celess)and Kcangaugeawaypartof E .Intheco rrespondinggaugewecan write E = , E =0;(2. 6.15) this globallysupersymmetric gaugeismaintainedbyfurthert ransformationsrestricted by K =1 2 D( K ) DKŠ1 2 DK, K= Ši 3 DK.(2. 6.16) U ndertheserestrictedtr ansformationswehave =1 6 K, E( )= D( K ).(2. 6.17)

PAGE 46

382.ATOYSUPERSPACEInthisgaugethetracelesspart h( )oftheordinarydreibein(thephysicalgraviton “eld)appearsin E( ).Thetrace h = h iscontaine din(the =0partof )and hasanidentical(linearized)transformationlaw.(Insuperconformaltheoriesthevielbe in al soundergoesasuperscaletransformationwhosescalarparametercanbeusedto gaugeto1,stillinagloballysupersymmetricway.Thus E( )containstheconformalpartofthesupergravitymultiplet,whereascontainsthetraces.) d. 2.Wess-Zuminogauge Theabovegaugeisconvenientforcalculationswherewewishtomaintainmanifest globalsupersymmetry.HoweverjustasinsuperYang-Millstheory,wecan“ndanonsupersymmetricWess-Zuminogaugethatexhibitsthecomponent“eldcontentofsupergravitymostdirectly.Insuchagauge = h + Š 2a E( )= h( )Š 2( ),(2. 6.18) where h and h( )aretheremainingpartsofthedreibein, and ( )ofthegravitino,and a isascalarauxiliary“eld.Theresidualgaugeinvariance(whichmaintains theaboveform)isparametrizedby K= + ( ),(2. 6.19) where ( x )par ametrizesgeneralspacetimecoo rdinatetransformationsand ( x ) parametrizeslocal(component) supersymmetrytransformations. e.Fieldstrengths Wenowret urntoastudyofthegeometricalobjectsofthetheory.The“eld strengthsforsupergravityaresupertorsions TAB Candsupercurvatures RAB ,de “nedby [ A, B} TAB CC+ RAB M .(2. 6.20) Ourdetermi nati onof intermsof (see(2.6.9)),isequivalenttotheconstraints T = i ( ) T = R =0.(2. 6.21) Wen eedonefurtherconstraint torelatetheconnection (thegauge“eldforthe

PAGE 47

2.6.Supergravity39localLorentztransformations)tothegaugepotential H M(orvielbein E M).Itturns outtha tsucha constraintis T =0.(2. 6.22) Tosolvethisc onstraint,andactually“ndintermsof E Mitisconvenienttomake someadditionalde“nitions: E E, EŠi 2 { E, E} [ EA, EB} CAB C EC.(2. 6.23) Theconstraint(2.6.22)isthensolvedfor asfollows:First,express[ ]in termsof andthecheckobjectsof(2.6.23)using(2.6.9).Then,“ndthecoecientof Einthisexpression .Theco rrespondingcoecientoftheright-handsideof (2.6.20)is T .Thisgiv esustheequation T = C Š1 2 ( ) ( )+1 2 ( ( ) )= C Š1 2 C ( ( ) )=0.(2. 6.24) (FromtheJaco biidentity[ E( { E, E )} ]=0,wehave,i ndependentof(2 .6.21,22), C( ) =0.)Wethen solvefor :Wemulti ply(2. 6.24)by Candusetheidentity =1 2 (( ) Š C( ) ).We “nd =1 3 ( C , Š C ( ) ),(2.6 .25) the C sbeingcalculablefrom(2.6.23)asderivativesof E M. f.Bianchiidentities Thetorsionsandcurvaturesarecovariantandmustbeexpressibleonlyinterms ofthephysicalgaugeinvariantcomponent“eldstrengthsforthegravitonandgravitino andauxiliary“elds.Weproceedintwosteps:First,weexpressallthe T sand R sin (2.6.20)intermsofasmallnumberofindepe ndent“eldstrengths;then,weanalyzethe contentofthesesuper“elds.

PAGE 48

402.ATOYSUPERSPACETheJacobiidentitiesforthecovariantderivativesexplicitlytaketheform: [[ [ A, B} C )} =0.(2. 6.26) Thepresenceoftheconstraintsin(2.6.21,22)allowsustoexpressallofthenontrivial torsionandcurvaturetensorscomp letelyintermsoftwosuper“elds R and G(where Gistotallysymmetric),andtheirspinorialderivatives.Thisisaccomplishedbyalgebraica llysolvingtheconstraintsplusJacobiidentities(whicharetheBianchiidentities fo rt he torsionsandcurvatures).WeeitherrepeatthecalculationsoftheYang-Mills case,orwemakeuseoftheresultsthere,asfollows: Weobservetha ttheco nstraint (2.6.21) {, } =2 i isidenticaltotheYangM illsconstraint(2.4.13,30a).TheJacobiidentity[ ( {, )} ]=0hasthes amesolutionasin(2.4.17-20a,31a): [ ]= C ( W ),(2. 6.27) where Wisexpa ndedoverthesupergravitygenerators i and iM (thefactor i is introducedtomakethegeneratorshermitian): W= W i + W i + W iM .(2. 6.28) ThesolutiontotheBianchiidentitiesisthus(2.4.33),withtheidenti“cation(2.6.28). Theconstraint(2.6.22)implies W =0,andwecansolve {, W} =0(see (2.4.33b))explicitly: W= Š CR W= G+1 3 C ( )R G= Š2 3 i R ,(2. 6.29) wherewehaveintroducedascalar R andatotallysymmetricspinor G.Thefull so lutionoftheBianchiidentitiesisthustheYang-Millssolution(2.4.33)withthesubstitutions Š iW= Š R +2 3 ( R ) M + G M G= Š2 3 i R Š if= Š1 3 ( ( R ) )+ G Š 2 Ri +2 3 ( 2R ) M

PAGE 49

2.6.Supergravity41+1 2 ( i ( R ) M ) + W M (2.6.30) where W1 4! ( G ).Wehaveused = i Š C2to“nd(2.6. 30).Individualtorsion sandcurvat urescanbereaddirectlyfromtheseequationsbycomparing withthede“nition(2.6.20).Thus,forexample,wehave R , =1 2 ( ( r ) ) r W Š1 3 ( ) 2R +1 4 ( ( i ) )R .(2. 6.31) The -independentpartof r istheRiccitensorinaspacetimegeometrywith( -indepe ndent)tor sion. Insec.2.4.a.3wediscussedcovariantshiftsofthegaugepotential.In any gauge theorysuch shiftsdonotchangethetransformationpropertiesofthecovariantderivati ve sa ndthusareperfectlyacceptable;theshiftedgauge“eldsprovideanequallygood descriptionofthetheory.Insec.2.4.a.3weusedtherede“nitionstoeliminatea“eld strength.Herewerede“netheconnection toeliminate T by = Š iRM.(2. 6.32) (Thiscorrespo ndstoshifting a b cbyaterm a b cR toca ncel T a b c;wetempora rilymake useofvectorindices a torepres enttracelessbispinorssincethismakesitclearthatthe shift(2.6.32)ispossibleonlyinthreedimensions.)Theshifted r ,dro ppingprimes, is r = W Š1 4 ( ) r r 4 3 2R +2 R2.(2. 6.33) Thisrede“nitionof isequivalenttoreplacingtheconstraint(2.6.9)with {, } =2 i Š 2 RM.(2. 6.34) Wew ill“ndthatthea nalogofthenewtermappearsintheconstraintsforfour dimensionalsupergravity(seechapter5).Thisisbecausewecanobtainthethree di mensionaltheoryfromthefourdimensionalone,andthereisnoshiftanalogousto (2.6.32)possibleinfourdimensions. Thesuper“elds R and Garethevariationsofthesupergravityaction(see below) withrespecttothetwounconstrainedsuper“eldsand E( )of(2.6.15-17).

PAGE 50

422.ATOYSUPERSPACEThe“eldequationsare R = G=0;theseares olvedonlyby”atspace(justasfor ordi narygravityinthree-dimensionalspacetime),sothree-dimensionalsupergravityhas no dy na mics(all“eldsareauxiliary). g.Actions Wenowt urntotheconstructionofactionsa ndtheirexpansi onintermsofcomponent “elds.Asweremarkedearlier,in”atsuperspacetheintegralof any (scalar) super“eldexpressionwiththe d3xd2 measureis globally supersymmetric.Thisisno longertrueforlocallysupersymmetrictheo ries.(Thenewfeaturesthatarisearenot speci“callylimitedtolocalsupersymmetry,butareageneralconsequenceoflocalcoordinateinvariance). Wer ecallthatinourfo rmalismanarbitrary matter super“eldtransforms accordingtotherule = eiK eŠ iK= eŠ iK eiK, K= KMiD M+ K iM ,(2. 6.35) where D Mmeansthatweletthedierentialoperatoractoneverythingtoitsleft.(The variousf ormsofthetransformationlawcanbes eentobeequivalentafterpowerseries expansionoftheexponentials,orbymultiplyingbyatestfunctionandintegratingby parts).L agrangiansarescalarsuper“elds,andsinceanyLagrangian IL isconstr ucted fromsuper“ eldsand operators,aLagrangiantransformsinthesameway. IL= eiKILeŠ iK= eŠ iKILeiK.(2. 6.36) Thereforetheintegral d3xd2 IL is not invariantwithrespecttoourgaugegroup.To “ndinvariants,weconsiderthevielbeinasasquaresupermatrixinitsindicesandcomputeitssuperdeterminant E .Thefo llowingresultwillbederi vedinour discussionof four-dimension s(sees ec.5.1): ( EŠ 1)= eiKEŠ 1eŠ iK(1 eiK) = EŠ 1eiK.(2. 6.37)

PAGE 51

2.6.Supergravity43Thereforetheproduct EŠ 1IL transformsinexactlythesamewayas EŠ 1: ( EŠ 1IL )= EŠ 1ILeiK.(2. 6.38) Sinceeverytermbutthe“rstoneinthepowerseriesexpansionofthe eiKisatotal derivative,weconcludethatuptosurfaceterms S = d3xd2 EŠ 1IL ,(2. 6.39) isinvariant.Wethereforehaveasimpleprescriptionforturninganygloballysupersymmetricactionintoalocallysupersymmetricone: [ IL ( DA,)]glob al EŠ 1IL ( A,),( 2.6.40) inanalogytoordinarygravity.Thus,theactionforthescalarmultipletdescribedbyeq. (2.3.5)takesthecovariantizedform S= d3xd2 EŠ 1[ Š1 2 ( )2+1 2 m 2+ 3! 3].(2.6 .41) Forv ectorgaugemultipletsthes impleprescriptionofreplacing”atderivatives DAbygravit ationallycovariantones Aissucienttoconvertglobalactionsintolocal actions,ifweincludetheYang-Millsgeneratorsinthecovariantderivatives,sothatthey arecovariantwithrespecttobothsupergravityandsuper-Yang-Millsinvariances.However,suchaprocedureisnotsucientformoregeneralgaugemultiplets,andinparticularthesuperformsofsec.2.5.Ontheo therha nd,itispossibl etoformulate all gauge theorieswithinthesuperform framew ork,atleastattheabelianlevel(whichisallthat isrelevantfor p -formsfor p > 1).Additionaltermsduetothegeometryofthespace willautomaticallyappearinthede“nitionsof“eldstrengths.Speci“cally,thecurvedspaceformulationofsuperformsisobtainedasfollows:Thede“nitions(2.5.8)holdin arbitrarysuperspaces,independentofanymetricstructure.Converting(2.5.8)toatangent-space basiswiththecurvedspace EA M,weobtaine quationsthatdierfrom(2.5.9) onlybythereplacementofthe”at-spacecovariantderivatives DAwiththecurved-space ones A. To illustratethis,letusreturntotheabelianvectormultiplet,nowinthepresence ofsupergravity.The“eldstrengthforthevectormultipletisa2-form:

PAGE 52

442.ATOYSUPERSPACEF= + Š 2 i F = ŠŠ T , F = ŠŠ T EE.(2. 6.42) We againimposetheconstraint F=0,which implies F = iC ( W ), W=1 2 + R ;(2. 6.43) wherewehaveused(2.6.30)substitutedinto(2.4.33).Comparingthistotheglobal“eld strengthde“nedin(2.4.20),weseethatanewtermproportionalto R appears.The extratermin Wisnecessaryforgaugeinvarianceduetotheidentity = i2 3 [ ].Inthegloballimitthecommutatorvanishes,butinthelocal caseitgivesacontributionthatispreciselycanceledbythecontributionofthe R term. Theseresultscanalsobeobtainedbyuseofderivativesthatarecovariantwithrespect tobothsupergravityandsuper-Yang-Mills. Weturnnowtot heacti onforthegauge“eldsoflocalsupersymmetry.Weexpect toconstructitoutofthe“eldstrengths Gand R .Bydimensi onalanalysis(noting that hasdimen sions( mass )Š1 2 inthreedimensions),wededuceforthePoincar esupergravityactionthesupersymmetricgeneralizationoftheEinstein-Hilbertaction: SSG= Š 2 2 d3xd2 EŠ 1R .(2. 6.44) Wecancheckthat (2.6.44)leadstothe correctcomponentactionasfollows:d2 EŠ 1R 2R 3 4 r (see(2.6.33)),andthusthegravitationalpartoftheactionis correct.Wecanalsoaddasupers ymmetriccosm ologicalterm Scosmo= 2 d3xd2 EŠ 1,(2. 6.45) whichleadstoanequationsofmotion R = G=0.Theonly solution tothisequation(inthreedimensions)isemptyanti-deSitterspace:From(2.6.33), r =2 2, W=0. Higher-derivativeactionsarepo ssiblebyusingotherfunctionsof Gand R .For ex ample,theanalogofthegauge-invariantmasstermfortheYang-Millsmultipletexists

PAGE 53

2.6.Supergravity45hereandisobtainedbythereplacementsin(2.4.38)(alongwith,ofcourse, d3xd2 d3xd2 EŠ 1): A iTi A iM W iTi G iM +2 3 ( R ) iM .(2. 6.46) Thisgives ILmass= d3xd2 EŠ 1 ( G +2 3 R Š1 6 ( ) ).(2.6 .47)

PAGE 54

462.ATOYSUPERSPACE2.7.Quantumsuperspace a.Scalarmultiplet InthissectionwediscussthederivationoftheFeynmanrulesforthree-dimensionalsuper“eldperturbationtheory.Sincethestartingpoint,thesuper“eldaction,is somuchlikeacomponent(ordinary“eldtheory)action,itispossibletoreadothe rulesfordoingFeynmansupergraphsalmostbyinspection.However,asanintroduction tothefour-dimensionalcaseweusethefullmachineryofthefunctionalintegral.After derivingtherulesweapplythemtosomeone-loopgraphs.Themanipulationsthatwe pe rformonthegraphsaretypicalandillustratethemannerinwhichsuper“eldshandle thecancellationsandothersimpli“cations duetosupersymmetry.Formoredetails,we referthereadertothefour-dime nsionaldiscussi oninchapter6. a.1.Genera lforma lism TheFeynmanrulesforthescalarsuper“eldcanbereaddirectlyfromthe Lagrangian:Thepropagatorisde“nedbythequadraticterms,andtheverticesbythe interactions.Thepropagatorisanoperatorinboth x and space,andatthevertices weintegr ateoverboth x and .ByFouriertr ansformationwechangethe x integration toloop-momentumintegration,butweleavethe integrationalone.( canalsobe Fouriertra nsformed,butthiscauseslittlechangeintherules:seesec.6.3.)Wenow derivetherulesfromth ef unction alintegral. Webeginbyc onsideringthegeneratingfunctionalforthemassivescalarsuper“eld witharb itraryself-interaction: Z ( J )= ID exp d3xd2 [1 2 D2+1 2 m 2+ f ()+ J ] = ID exp [ S0()+ SINT()+ J ] = exp [ SINT( J )] ID exp [ 1 2 ( D2+ m )+ J ].(2.7.1) Intheusualfashionwecompletethesquare,dothe(functional)Gaussianintegralover ,andobtain

PAGE 55

2.7.Quantumsuperspace47Z ( J )= exp [ SINT( J )] exp [ Š d3xd21 2 J 1 D2+ m J ].(2.7 .2) Usingeq.(2.2.6)wecanwrite 1 D2+ m = D2Š m Š m2 .(2. 7.3) (Note D2behavesj ustas /inconventional“eldtheory.)Weobtain,inmomentum space,thefollowingFeynmanrules: Propagator: J ( k ) J ( Š k ) d3k (2 )3 d21 2 J ( k ) D2Š m k2+ m2 J ( Š k ) = D2Š m k2+ m2 2( Š ).(2.7 .4) Verti ces:Aninteractionterm,e.g.d3xd2 D D ... ,gives avertexwith linesleavingit,withtheappropriateoperators D, D,etc.actingont hecorresponding lines,andanintegralover d2 .Theoper ators Dwhichappearinthepropagators,or arecomingfromavertexandactonaspeci“cpropagatorwithmomentum kleaving thatvertex,d ependonthatmomentum: D= + k.(2. 7.5) Inadditionwehaveloop-momentumintegralstoperform. Ingeneralwe“nditconvenienttocalculatetheeectiveaction.Itisobtainedin standardfashionbyaLegendretransformationonthegeneratingfunctionalforconnectedsupergraphs W ( J )anditco nsistsofasumofone-particle-irreduciblecontributionsobtainedbyamputatingexternallinepropagators,replacingthembyexternal“eld factors( pi, i),andintegratingover pi, i.Therefo re,itwillhavetheform ()=n1 n d3p1... d3pn(2 )3 n d21... d2n( p1, 1) ... ( pn, n) (2 )3 ( pi)loops d3k (2 )3 internal vertices d2 propagators vertices (2.7.6)

PAGE 56

482.ATOYSUPERSPACEAswehavealreadymentioned,allofthiscanbereaddirectlyfromtheaction,byanalogywiththederivationoft heusualFeynmanrules. Theintegrandintheeectiveactionis apriori anonloca lf unction ofthe x s(nonpolynomi alinthe p s)andofthe 1, ... n.Howev er,wecanmanipulatethe -integrationssoastoexhibititexplicitlyasafunctionalofthesallevaluatedatasinglecommon asfollows:Ageneralmultiloopintegralconsistsofverticeslabeled i i +1,conn ectedbypropagatorswhichcontainfactors ( iŠ i +1)withope rators Dactingon them.Consideraparticularloopinthediagramandexamineonelineofthatloop. Thefactorsof D canbecombinedbyusingtheresult(transferrule): D( i, k ) ( iŠ i +1)= Š D( i +1, Š k ) ( iŠ i +1),(2.7 .7) aswellastherulesofeq.(2.2.6),afterwhichwehaveatmosttwofactorsof D actingat oneendoftheline.Atthevertexwherethisendisattachedthese D scanbeint egrated bypartsontotheot herlines(orexternal“elds)usingtheLeibnitzrule(andsomecare withminussignssincethe D santicommut e).Thentheparticular -functionnolonger hasanyderivativesactingonitandcanbeusedtodothe iintegration,th useectively shrinking the( i, i +1) linetoapointin -space.Wecanrepeatthisprocedureon eachlineoftheloop,integratingbypartso neatatimeandshrink ing.Thiswillgenerateasumofterms,fromtheintegrationb yparts.The procedurestopswhenineach termweareleftwithexactlytwolines,onewith ( 1Š m)whichisfree ofanyderivatives,andonewith ( mŠ 1)whichmayca rryzero,one,ortwoderivatives.Wenow usetherules(whichfollowfromthede“nition 2( )= Š 2), 2( 1Š m) 2( mŠ 1)=0, 2( 1Š m) D2( mŠ 1)=0, 2( 1Š m) D22( mŠ 1)= 2( 1Š m).(2.7 .8) Thus,inthosetermswhereweareleftwithno D orone D weget zero,while inthe termsinwhichwehavea D2actingononeofthe -functions, mult ipliedbytheother -function,weusetheaboveresult.Weareleftwiththesingle -function,whichwecan usetod oone more integration,thus “nallyreducingthe -spacelooptoapoint.

PAGE 57

2.7.Quantumsuperspace49Theprocedurecanberepeatedloopbyloop,untilthewholemultiloopdiagram hasbeenreducedto one pointin -space,givingacontributiontotheeectiveaction ()= d3p1... d3pn(2 )3 n d2 G ( p1, ... pn)( p1, ) ... D( pi, ) ... D2( pj, ) ... ,(2. 7.9) where G isobtainedbydoingordinaryloop-mo mentumintegrals,withsomemomentum factorsinthenumeratorscomi ngfrom anticommutatorsof D sarisingintheprevious manipulation. a.2.Examples Wegivenowtwoex amples,inamasslessmodelwith3interactions,toshowhow the manipulationworks.The“rstoneisthecalculationofaself-energycorrection representedbythegraphinFig.2.7.1 k k + p ( Š p ) ( p ) Fig.2.7.1 2= d3p (2 )3 d2 d2( Š p )( p ) d3k (2 )3 D2 ( Š ) k2 D2 ( Š ) ( k + p )2 .(2. 7.10) Thetermsinvolving canbemanipulatedasfollows,usingintegrationbyparts: D2 ( Š ) D2 ( Š )( p ) = Š1 2 D ( Š )[ DD2 ( Š )( p )+ D2 ( Š ) D( p )]

PAGE 58

502.ATOYSUPERSPACE= ( Š )[( D2)2 ( Š )( p )+ DD2 ( Š ) D( p ) + D2 ( Š ) D2( p )].(2.7.11) However,using( D2)2= Š k2and DD2= kDweseethata ccordingtotherulesin eq.(2.7.8)onlythelasttermcontributes.We“nd 2= d3p (2 )3 d2 ( Š p ) D2( p ) d3k (2 )3 1 k2( k + p )2 .(2. 7.12) Doingtheintegrationbypartsexplicitlycanbecomerathertediousanditis preferabletoperformitbyindicating D sandmovingthemdirectlyonthegraphs.We showthisinFig.2.7.2: D2D2D2D2D2D2D2DDFig.2.7.2 Onlythelastdiagramgivesacontribution.Onefurtherruleisusefulinthisprocedure: Ingeneral,afterintegrationbyparts,various D -factorsendupindierentplacesinthe “nalexpressionandonehastoworryaboutminussignsintroducedinmovingthempast eachother.Theoverallsigncanbe“xedattheendbyrealizingthatwestartwitha particularorderingofthe D sandwecanexaminewhathappenedtothisorderingat theendofthecalculation.Forexample,wemaystartwithanexpressionsuchas D2... D2... D2... =1 2 DD...1 2 DD...1 2 DD... andend upwith D... D... D... D... D... D... wherethevarious D sactondier ent“elds.The overallsignc anobviouslybedeterminedbyjustcountingthenumberoftranspositions. Forexample,inthecaseabov ewewouldendu pwithaplussign.N otethatt hisrule alsoappliesiffactorssuchas karise,providedonepaysa ttentiontot hemannerin whichtheywereproduced(e.g.,atwhichendofthelinewerethe D sacting?Didit comefrom DDorfrom DD?). Oursecondexampleisthethree-pointd iagrambelow,whichwemanipulateas showninthesequenceofFig.2.7.3:

PAGE 59

2.7.Quantumsuperspace51 D2 DDD2D2D2D2D2D2D2D2D2D2D2D2D2 DDDDFig.2.7.3 Atthe“rsts tagewehaveintegratedbypartsthe D2othebottomlineandimmediatelyreplaced( D2)2by = Š k2.Atthes econdstagewehaveintegratedbypartsthe D2otherightside,butkeptonlythosetermsthatarenotzero:Thebottomlinehas alreadybeenshrunktoapointbythecorresponding -function(butweneednotindicate thisexplicitly;anylinethathasno D sonitcanbeconsideredashavingbeen shrunk)andintheendwekeeponlytermswith exactly twofac torsof D intheloop.For themiddlediagramthismeansusing DD2D= DkD= Š kD2+aterm withno D swhichmaybedropped.Theintegrandintheeectiveactioncanbewrittenthenas d3k (2 )3 1 k2( k + p1)2( k Š p3)2 ( p3, )[ Š ( p1, )( p2, ) k2Š D( p1, ) D( p2, ) k+ D2( p1, ) D2( p2, )],(2.7.13) andonl ythe k -momentumintegralremainstobedone. Ingeneral,theloop-momentumintegralsmayhavetoberegularized.Theprocedurew euse, whichisguaranteedtopreservesupersymmetry, istodo allthe D -manipulations“rst,untilwere ducetheeectiveactiontoanintegraloverasingle ofan expressionthatisaproductofsuper“elds,andthereforemanifestlysupersymmetric. Theremainingloop-momentumintegralsmaythenberegularizedinanymanner

PAGE 60

522.ATOYSUPERSPACEwhatsoever,e.g.,byusingdimensionalregularization.Weshalldiscusstheissues involvedinthiskindofregularizationinsec.6.6.Analternativeprocedure,somewhat cumbersomeinitsapplicationbutbetterunde rstood,issupersymmetricPauli-Villars regularization.Inthreedimensionsthisisapplicableeventogaugetheories,sincegauge invariantmasstermsexist. b.Vectormultiplet Nothingnewisen counteredinthederivationorapplicationoftheFeynman rules.However,thederivationmustbeprecededbyquantization,i.e.,introductionof gauge-“xingtermsandFaddeev-Popovghosts,whichwenowdiscuss. Webeginwit hthecla ssicalaction SC=1 g2 tr d3xd2 W2.(2. 7.14) Thegaugeinvarianceis = K and,bydirectanalogywiththeordinaryYang-Mills case,wecanchoosethegauge-“xingfunction F =1 2 D.Weuseanaver aging procedurewhichleadstoa gauge-“xingtermwithoutdimensionalparameters, FD2F ,and obtain,forthequadraticaction, S2=1 g2 tr d3xd2 [1 2 (1 2 DD)(1 2 DD) Š1 (1 2 D) D2(1 2 D)] =1 2 1 g2 tr d3xd2 [1 2 (1+1 ) +1 2 (1 Š1 )i D2].(2.7 .15) Variouschoi cesofthegaugeparameter arepossible: Thechoice = Š 1givesthe kineticterm1 2 i D2,wh ilethechoice =1gives1 2 ,whichresu ltsinthe simplestpropagator. TheFaddeev-Popovactionissimply SFP=1 g2 tr d3xd2 c( x )1 2 Dc ( x ),(2.7 .16) withtwoscalarmultipletghosts.(Notethatinabackground-“eldformulationofthe

PAGE 61

2.7.Quantumsuperspace53theory,similartotheonewediscussinsec.6.5,onewouldreplacetheoperator D2in th e gau ge “xingtermbythebackground-covariantoperator 2,andthis wouldgiv erise toa third, Nielsen-Kallosh,ghostaswell.) TheFeynmanrulesarenowstraightforwardtoobtain.Theghostpropagatoris conventional,followingfromthequadraticghostkineticterm cD2c ,wh ilethegauge “e ldpropagatoris k2 2( Š ).(2.7 .17) Verti cescanbereadofromtheinteractionterms.Thegauge-“eldself-interactions(in thenonabeliancase)are g2LINT= Ši 4 DD[, D] Š1 12 DD[, { ,} ] Š1 8 [, D][, D]+i 12 [, D][, { ,} ] +1 72 [, { ,} ][, { ,} ],(2.7 .18) thoseoftheghostsare g2LINT= Š i1 2 cD[, c ],(2.7 .19) whilethoseofacomplexscalar“eldare g2LINT= ( 2Š D2)= [ Š i DŠ i1 2 ( D) Š 2].(2. 7.20)

PAGE 62

Contentsof 3.REPRESENTATIONSOFSUPERSYMMETRY 3.1.Notation 54 a.Indexconventions54 b.Superspace56 c.Symmetriz ationandantisymmetrization56 d.Conjugation57 e.Levi-Civitatensorsandindexcontractions58 3.2.Thesupersymmetrygroups62 a.Liealgebras62 b.Super-Lie algebras63 c.Super-Poincar ealgebra63 d.Positivityof theenergy64 e.Superconformalalgebra65 f.Super-deSitteralgebra67 3.3.Representationsofsupersymmetry69 a.Particlerepresentations69 a.1.Masslessrepresentations69 a.2.Massiverepresentationsandcentralcharges71 a.3.Casimiroperators72 b.Representationsonsuper“elds74 b.1.Superspace74 b.2.Acti onofgeneratorsonsuperspace74 b.3.Acti onofgeneratorsonsuper“elds75 b.4.Extendedsupersymmetry76 b.5.CPTinsuperspace77 b.6.Chir alrepresentationsofsupersymmetry78 b.7.Superconformalrepresentations80 b.8.Super-deSitterrepresentations82 3.4.Covariantderivatives83 a.Construction83 b.Algebraicr elations84 c.Geometryof”atsuperspace86 d.Casimiroperators87 3.5.Constraine dsuper “elds89 3.6.Componentexpansions92

PAGE 63

a. -expansions92 b.Projection94 c.Thetransformationsuper“eld96 3.7.Superintegration97 a.Berezinintegral97 b.Dimensions99 c.Superdeterminants99 3.8.Superfunctionaldierentiationandintegration101 a.Dierentiation101 b.Integr ation103 3.9.Physical,auxiliary,andgaugecomponents108 3.10.Compensators112 a.Stueckelbergformalism112 b.CP(1)m odel 113 c.Cosetspaces117 3.11.Projectionoperators120 a.General120 a.1.Poincar eproj ectors121 a.2.Super-Poincar eproj ectors122 b.Exam ples 128 b.1.N= 0 128 b.2.N= 1 130 b.3.N= 2 132 b.4.N= 4 135 3.12.On-shellrepresentationsandsuper“elds138 a.Fieldstrengths138 b.Light-cone formalism142 3.13.O-shell“eldstrengthsandprepotentials147

PAGE 64

3.REPRESENTATIONSOFSUPERSYMMETRY 3.1.Notation Anifor ani,anda2fora2. Wenowt urntofourdimensi ons.Ourtreatmentwillbeentirelyself-contained;it willnotassumefamiliaritywithourthree-dimensionaltoy.Althoughsupersymmetryis morecomplicatedinfourdimensionsthaninthree,becausewegiveamoredetaileddiscussion,somegeneralaspectsofthetheorymaybeeasiertounderstand.Webeginby givingthenotationandconventionsw eusethro ughouttherestofthework. a.Indexconventions Ourindexconventionsareasfollows:Thesimplestnontrivialrepresentationof theLorentzgroup,thetwo-componentco mplex(Weyl)spinorrepresentation(1 2 ,0)of SL (2, C ),islabeledbyatwo-valued(+or-)lower-caseGreekindex(e.g., =( +, Š)), andthecomplex-conjugaterepresentation(0,1 2 )islabel edbyado ttedindex ( €=( €+, €Š)). Af ou r-componentDiracspinoristhecombinationofanundotted spinorwithad ottedone(1 2 ,0)+(0,1 2 ),andaMajoranaspinorisaDiracspinorwhere th ed o tte ds pinoristhecomplexconjugateoftheundottedone.Anarbitraryirreduciblerepresentation( A B )isthenconve nientlyrepresentedbyaquantitywith2 A undottedindicesand2 B dottedindices,totallysymmetricinitsundottedindicesandin itsdottedindices.Anexampleistheself-dualsecond-rankantisymmetrictensor(1,0), whichisrepresentedbyasecond-ranksymmetricspinor f.(Thechoiceo fselfdualvs. anti-self-dualfollowsfromWickrotationfromEuclideanspace,wherethesignisunambigu ous.) Anotherexampleisthevector(1 2 ,1 2 ), la be le dw it ho neundottedandonedotted i ndex,e.g., V€.Arealv ectorsatis“esthehermiticitycondition V€= V€= V€.As as horthandnotation,weoftenuseanunderlinedlower-caseRomanindextoindicatea v ectorindexwhichisacompositeofthecorrespondingundottedanddottedspinor indices:e.g., V a V€.Weconsiders uchanindexmerelyasanabbreviation:Itmay appearononesideofanequationwhiletheexplicitpairofspinorindicesappearsonthe

PAGE 65

3.1.Notation55other,oritmaybecontractedwithanexplicitpairofspinorindices.Whendiscussing Lorentznoncovariantquantit ies(as,e.g.,inlight-coneformalisms),wesometimeslabel thevaluesofavector indexasfollows: V a=( V+€+, V+€Š, VŠ€+, VŠ€Š) ( V+, VT, VT, Š VŠ),(3.1 .1) where VTisthecomplex conjugateof VT,and VarerealinMinkowskispace(but V+isthecomplex conjugateof VŠinWick-rotatedEuclideans pace).Moregenerally,we canrelateavectorlabel ainan arbitrary basis,where a = € ,tothe € basisbyaset ofClebsch-Gordancoecients,thePaulimatrices:Wede“ne forfields : V€=1 2 b €Vb Vb =1 2 b €V€; forderivatives : €= b €b b =1 2 b €€; forcoordinates : x€=1 2 b €xb xb = b €x€.(3. 1.2a) ThePaulimatricessatisfy b €c €=2 b c b €b €=2 €€.(3. 1.2b) TheseconventionsleadtoanunusualnormalizationoftheYang-Millsgaugecoupling constant g ,since € €Š igV€= b €b Š ig1 2 b €Vb b €( b Š i gVb ) andhenceour g is 2timestheu sualone g .(Weuset hesummationconvention:Any indexofanytypeappearingtwiceinthesameterm,oncecontravariant(asasuperscript)andoncecovariant(asasubscript),issummedover.) NexttoLorentzi ndices,thetypeofindiceswemostfrequentlyuseare isospin i ndices:internalsymmetryindices,usuallyforthegroup SU ( N )or U ( N ).Theseare representedbylower-caseRomanletters,withoutunderlining.Weusean underlined i ndexonlytoindicateacompositeindex,anabbreviationforapairofindices.Inadditiontothevectorindexde“nedabove,wede “neacompositespinor-isospinorindexby an underlinedlower-caseGreekindex(undottedordotted): a € a€,

PAGE 66

563.REPRESENTATIONSOFSUPERSYMMETRY a € a€. b.Superspace Wede “ne N -extendedsuperspacetobeaspacewithboththeusualrealcommutingspa cetimecoordinates x€= x a= x a,andantico mmutingcoordinates a = (andtheircompl exconjugates € =( ))whichtran sformasaspinorandan N -componentisos pinor.Todenotethesecoordinatescollectivelyweintroduce supervector i ndices,usingupper-caseRomanletters: zA=( x a, € ),(3.1 .3a) andthecorresp o ndingpartialderivatives A=( a, € ), AzB A B,(3. 1.3b) wherethenonvanishingpartsof A Bare a b, a b ,and € € b a€€.The derivativesare de“nedtosatisfya graded Leibnitzrule,givenbyexpressingdierentiationas gradedcommutation: ( AXY ) [ A, XY } =[ A, X } Y +( Š )XAX [ A, Y } ,(3. 1.4a) where( Š )XAis Š whenboth X and Aareanticommuting,and +otherwise ,andthe graded commutator[ A B } AB Š ( Š )ABBA istheanticommutator { A B } when A and B arebothoperatorswithfermistatistics,andthecommutator[ A B ]otherwise. Eq.(3.1.4a)followsfromwritingeach(anti)commutatorasadierence(sum)oftwo terms.Thepartialderivativesalsosatisfygradedcommutationrelations: [ A, B} =0.(3. 1.4b) c.Symmetrizationand antisymmetrization Ournot ationforsymmetrizingandantisymmetrizingindicesisasfollows:Symmetrizationisindicatedbyparentheses(),whileantisymmetrizationisindicatedby brackets[].Bysymme trizationwemeansimplythesumoverallpermutationsof indices,withoutadditionalfactors(andsimilarlyforantisymmetrization,withtheappropriatepermutationsigns).Allindicesbetweenparentheses(brackets)aretobe (anti)symmetrized exceptthosebetweenverticallines || .Forex ample, A( | B | )=

PAGE 67

3.1.Notation57AB+ AB.Ina ddition,justasitisconvenienttode“nethegradedcommutator [ A B } ,wede “negr adedantisymmetrization[)tobeasumofpermutationswithaplus signforanytranspositionoftwospinorindices,andaminussignforanyotherkindof pair. d.Conjugation Whenworkingwithoperatorswithfermist atistics,theonlytypeofcomplexconjugationthatisusuallyde“nedishermitianconjugation.Itisde“nedsothatthehermitianconjugateofapr o ductistheproductofthehermitianconjugatesofthefactorsin reverseorder.Foranticommutingc-numbershermitianconjugationagainisthemost naturalformofcomplexconjugation.Wedenotetheoperationofhermitianconjugation by ad agger,andi ndicatethehermitianconjugateofagivenspinorbyabar: ( ) €,or( €) .Inparticular, thisappliestot hecoordinates and introducedabove.Hermitianconjugationofanobjectwithmany(upper)spinorindicesis de“nedasforaproductofspinors: ( 1 1... j j 1€1... k€k)= k k... 1 1 j€j... 1€1=( Š 1)1 2 [ j ( j Š 1)+ k ( k Š 1)]1 1... k k 1€1... j€j,(3. 1.5a) andhence ( 1... j€1...€k) ( Š 1)1 2 [ j ( j Š 1)+ k ( k Š 1)] 1... k€1...€j.(3. 1.5b) Inaddition,isospinindicesfor SU ( N )gofrom uppertolower,orviceversa,uponhermitianconjugation.Hermitianconjugationofpartialderivativesfollowsfromthereality of A B=( AzB)=[ A, zB} : ( A)= Š ( Š )AA,(3. 1.6a) where( Š )Ais Š 1forspin orindicesand+1otherwise: ( a)= Š a,( )=+ € .(3. 1.6b) Hermitianconjugationasappliedtogeneraloperatorsisde“nedby O ( O ) ,(3. 1.7)

PAGE 68

583.REPRESENTATIONSOFSUPERSYMMETRYwheretheintegrationisovertheappropriatespace(aswillbedescribedinsec.3.7)and isthehermitianconj ugateofthefunction ,asde “nedabove. Si nc ei nt eg rationde“nesnotonlyasesquilinear(hermitian)metric onthe sp aceoffunctions,asusedtode“neaHilbertspace,butalsoabilinearmetric ,we canalsode“nethetransposeofanoperator: O ( Ot ) ,(3. 1.8) where is Š for O and anticommuting,+otherwise.Whentheoperatorisexpressed asamatrix,thehermitianconjugateandtransposetaketheirfamiliarforms.Wecan alsode“necomplexconjugationofanoperator: O O ,(3. 1.9) with asin(3.1.8).Forc-numberswehave t= and *= .Forpartia lderivatives,integrationbypartsimplies( A)t= Š A.Ingen eral,wealsohav etherelation O *t= O,andtheorderin grelations( O1O2)t= O2 tO1 tand( O1O2)*= O1* O2*,as wellasth eusual( O1O2)= O2 O1 . e.Levi-Civitat enso rsandindexcontractions Thereisonlyonenontrivialinvariantmatrixin SL (2, C ),theantisymmetricsymbol C(anditscom plexconjugateandtheirinverses),duetothevolume-preserving natureofthegroup(unitdeterminant).Similarly,for SU ( N )wehavethe antisymmetricsymbol Ca1... aN(anditscom plexconjugate).Inadditionwe“nditusefultointroduce theantisymmetricsymbolof SL (2 N C ) SU ( N ) SL (2, C ), C 1... 2 N.B ecauseofanticommutativity,itappearsintheantisymmetricproductofthe2 N sof N -extended supersymmetry.Theseobjectssat isfythefollowingrelations: C= C€€, CC= [ ] ;( 3.1.10a) Ca1... aN= Ca1... aN, Ca1... aNCb1... bN= [ a1b1... aN] bN;(3. 1.10b) C 1... 2 N= C€ 2 N...€ 1, C 1... 2 NC 1... 2 N= [ 1 1... 2 N] 2 N.(3. 1.10c) The SL (2 N C )symbolcan beexpr essedintermsoftheothers:

PAGE 69

3.1.Notation59C 1... 2 N= 1 N !( N +1)! ( Ca1... aNCaN +1... a2 NC1N +1... CN2 N permutationsof i).(3.1 .11) Themagnitudesofthe C sare“xedbythe conventions C= C€€, C 1... 2 N= C€ 2 N...€ 1,(3. 1.12) whichsettheabsolutevaluesoftheircomponentsto0or1. Wehavethefo llowingrelationfort heproductofallthe s(b ecause { } =0, thesquareofanyonecomponentof vanishes): 1... 2 N= C 2 N... 1(1 (2 N )! C 2 N... 1 1... 2 N) C 2 N... 12 N,(3. 1.13) andasimilarrelationfor ,where,uptoa phasefactor, 2 Nissimplytheproductofall the s.Ourconventionsforcomplexconjugationofthe C simply 2 N = 2 N.Altho ugh seldomneeded(exceptforexpressingthe SL (2, C ) SU ( N )covariantsint ermsofcovariantsofasubgro up,as,e.g.,whenperformingdimensionalreductionorusingalight-cone formalism),wecan“xthephases(uptosigns)inthede“nitionofthe C sbythefollowingconventions: C= C€€, C 1... 2 N= C€ 1...€ 2 N Ca1... aN= Ca1... aN.(3. 1.14) Inparticular,wetake C= 0 i Š i 0 (3.1.15) For N =1wehave 2=1 2 C= i +Š. Cisthusthe SL (2, C )metri c,andcanbe usedforraisingandloweringspinorindices: = C, = C,( 3.1.16a) = 21 2 C=1 2 =1 2 = i +Š;(3. 1.16b) €= €C€€, €= C€€ €,(3. 1.16c)

PAGE 70

603.REPRESENTATIONSOFSUPERSYMMETRY € €= 21 2 C€€ € €=1 2 € €=1 2 = i €+ €Š;(3. 1.16d) V a= V bCC€€ V b b a, V a= CC€€V b a bV b,(3. 1.16e) V W V aW a= W V V21 2 b aV aV b=1 2 V aV a=1 2 V V = V+VŠ+ VTVT= Š detV€.(3. 1.16f) (Asindicatedbytheseequations,wecontractindiceswiththecontravariantindex“rst.) Ourunusualde“nitionofthesquareofavectorisusefulforspinoralgebra,butwecautionthereadernottoconfus eitwiththest a ndardde“nition.Inparticular,wede“ne 1 2 a a.(However,wh enwetransform(witha nonunimodular transformation)toa cartesianbasis,thenwehavetheusual = a a .Forthecoordin ates,wehave x2=1 4 xa xa .Ourconv entionsareconvenientforsuper“eldcalculations,butmayleadto afewun usualcomponentnormalizations.) De“ning € a b€, € b a€;(3. 1.17) wehavethei dentit ies €€= € € = .(3. 1.18) Fr om(3.1.10a)weobtainthefrequentlyusedrelation [ ]= C( C)= C( ),(3.1 .19) whichistheWeyl-spinorformoftheFierzidentities.Similarrelationsfollowfrom (3.1.10b,c). Thecomplexconjugationpropertiesof Cimplythatthecomplexconjugatesof covariant (lowerindex)spinors,includingspinorpartialderivatives(cf.(3.1.6)),havean additionalminussign: ( )= Š €.(3. 1.20) From(3.1 .11)and(3.1.18),ordirectlyfromthef acttha tantisy mmetricsymbolsde“ne

PAGE 71

3.1.Notation61determinants( detV € =( detV€)N=( Š V2)N),wehavethefollowingidentity: C€ 2 N...€ 1 1€ 1... 2 N€ 2 N= C 1... 2 N( Š )N.(3. 1.21) Finally,wede“nethe SO (3,1)Levi-Civitatensoras a b c d= i ( CCC€€C€€Š CCC€€C€€), a b c d e f g h= Š [ a e b f c g d ] h.(3. 1.22)

PAGE 72

623.REPRESENTATIONSOFSUPERSYMMETRY3.2.Thesupersymmetrygroups LiealgebrasandLiegroupsplayanimportantrolein“eldtheory;groupssuchas thePoincar egroup ISO (3,1),theLorentzgroup SO (3,1), SU (3)and SU (2) U (1)are fa miliar.ThenewfeatureneededforsupersymmetryisageneralizationofLiealgebras tosuper-Liealgebras(alsocalledgradedLie algebras;however,thistermissometimes usedinadierentway). a.Liealgebras ALieal gebraconsistsofasetofgenerators { A} ,A=1,. .., M .Theseob jects closeunderanantisymmetricbinaryoperationcalledaLiebracket;wewriteitasa commutator: [A,B]=ABŠ BA.(3. 2.1) TheLiealgebraisde“nedbyitsstructureconstants fAB C: [A,B]= ifAB CC.(3. 2.2) Thestructureconstantsarerest rictedbytheJacobiidentities fAB DfDC E+ fBC DfDA E+ fCA DfDB E=0(3.2 .3) whichfollowfrom [[A,B],C]+ [[B,C],A]+ [[C,A],B]=0.(3 .2.4) Thegeneratorsformabasisforvectorsoftheform K = AA,wherethe AarecoordinatesintheLiealgebrawhichareusually takentocommutewiththegeneratorsA.In mostphysicsapplicationstheyaretakentobereal,complex,orquaternionicnumbers. B ecausethestructureconstantssatisfytheJaco biidentities,itisalwayspossibletorepresentthegeneratorsasmatrices.WecanthenexponentiatetheLiealgebraintoaLie groupwithelements g = eiK;ingen eral,dierentrepresentationsoftheLiealgebrawill giverisetoLiegroupswithdierenttopologicalstructures.Ifasetof“elds( x )transformslinearlyundertheactionoftheLiegroup,wesay( x )isinorc arriesarepresentationofthegroup.Abstractly,wewrite ( x )= eiK( x ) eŠ iK;(3. 2.5)

PAGE 73

3.2.Thesupersymmetrygroups63togivethismeaning,wemustspecifytheactionofthegeneratorson,i.e.,[A,].For example,if K isamatrixrepresentationandisaco lumnv ector,theexpressionabove istobeinterpretedas= eiK. b.Super-Liealgebras Forsupersy mmetrywegeneralizeandconsidersuper-Liealgebras.Theessential newfeatureisthatnowtheLiebracketofso megeneratorsissymmetric.Thosegeneratorswhosebracketissymmetricarecalledfe rmionic;therestarebosonic.Wewritethe bracketasa gradedcommutator [A,B} =ABŠ ( Š )ABBA [AB).(3. 2.6) Thestructureconstantsofthesuper-Liealgebraobeysuper-Jacobiidentitiesthatfollow from: 0=1 2 ( Š )AC[[[A,B} ,C)} ( Š )AC[[A,B} ,C} +( Š )AB[[B,C} ,A} +( Š )BC[[C,A} ,B} .(3. 2.7) Again,wecande“neavectorspacewiththegeneratorsAactingasabasis;however, inthiscasethecoordinates Aassociatedwiththefermionicgeneratorsare antico mmuting numbersorGra ssmannparametersthatanticommutewitheachotherandwiththe fermionicgenerators.Gra ssmannparameterscommutewithordinarynumbersand bosonicgen erators;thesepropertiesensurethat K = AAisbosonic.Formally,we obtainsuper-LiegroupelementsbyexponentiationofthealgebraaswedoforLie groups. c.Super-Poincar ealgebra Fieldtheoriesinordinaryspacetimeareusuallysymmetricundertheactionofa spacetimesymmetrygroup:thePoincar egroupform assivetheoriesin”atspace,the conformalgroupformasslesstheories,andthedeSittergroupfortheoriesinspacesof constantcurvature.Forsupersymmetry,weconsiderextensionsofthesegroupsto supergroups.ThesewereinvestigatedbyHaag,/Lopusza nski,andSohnius,whoclassi“ed themostgeneralsymmetriespossible(actu ally,theycon sideredsymmetriesoftheSmatrixandgeneralizedtheColeman-Mandulatheoremonuni“edinternalandspacetime

PAGE 74

643.REPRESENTATIONSOFSUPERSYMMETRYsymmetriestoincludesuper-Liealgebras).Theyprovedthatthemostgeneral super-Poincar ealgebra contains,inadditionto { J, J€€, P€} (thegen eratorsofthe Poincar egro up), N fermionicsp inorialgenerators Qa (andtheirhermi tianconjugates Š Qa€),where a =1,. .., N isanisospinindex,andatmost1 2 N ( N Š 1)complexcentral char ges(calledcentralbecausetheycommutewithallgeneratorsinthetheory) Zab= Š Zba.Theal gebrais: { Qa Qb€} = a bP€,(3. 2.8a) { Qa Qb } = CZab,(3. 2.8b) [ Qa P€]=[ P€, P€]=[ J€€, Qc ]=0,(3 .2.8c) [ J, Qc ]=1 2 iC ( Qc ),(3. 2.8d) [ J, P€]=1 2 iC ( P )€,(3. 2.8e) [ J, J]= Š1 2 i ( ( J ) ),(3. 2.8f) [ J, J€€]=[ Zab, Zcd]=[ Zab, Zcd]=0.(3 .2.8g) TheessentialingredientsintheproofaretheColeman-Mandulatheorem(whichrestricts thebosonicpartsofthealgebra),andthesuper-Jacobiidentities.The N =1caseis calledsimplesupersymmetry,whereasthe N > 1caseisc alledextendeds upersymmetry. Centralchargescanariseonlyinthecaseofextended( N > 1)supersymmetry.The supersymmetrygenerators Q actassquarerootsofthemomentumgenerators P d.Positivityoftheenergy Adir ectconsequenceofthealgebraisthepositivityoftheenergyinsupersymmetrictheories.Thesimplestwaytounderstandthisresultistonotethatthetotal energycanbewri ttenas=1 2 ( P+Š PŠ)=1 2 €P€= Š1 2 €P€.(3. 2.9)

PAGE 75

3.2.Thesupersymmetrygroups65Since P€canbeobtainedfromtheanticommutatorofspinorcharges,wehave= Š1 2 N €{ Qa Qa€} =1 2 N { Qa ,( Qa )} (3.2.10) (weuse Qa€= Š ( Qa )).Therightha ndsideofeq.(3.2.10)ismanifestlynon-negative: Foranyop erator A andanystate | > < |{ A A}| > =n ( < | A | n >< n | A| > + < | A| n >< n | A | > ) =n (|< n | A| >|2+|< n | A | >|2).(3.2 .11) Hen ce,isalsononnegative.Further,ifsupersymmetryisunbroken, Q musta nnihilate thevacuum;inthiscase,(3.2.10)leadstotheconclusionthatthevacuumenergyvanishes.Althoughthisargumentisformal,itcanbemademoreprecise;indeed,itispossibletocharacterizesupersymmetrictheoriesbytheconditionthatthevacuumenergy vanish. e.Superconformalalgebra Fo rm asslesstheories,Haag,/Lopusza nski,andSohniusshowedwhatformextensionsoftheconformalgroupcantake:Th egen eratorsofthesuperconformalgroups consistofthegeneratorsoftheconformalgroup( P€, J, J€€, K€,) (thesearethe generatorsofthePoincar ealgebra,thesp ecialconformalboostgenerators,andthedilationgenerator),2 N spinorgenerators( Qa Sa )(andtheir hermitianconjugates Š Qa€, Š Sa€withatotalof8 N components),and N2furtherb osoniccharges( A Ta b) where Ta a=0.Theal gebrahasstructureconstantsde “nedbythefollowing(anti)commuta tors: { Qa Qb€} = a bP€, { Sa Sb€} = b aK€,( 3.2.12a) { Qa Sb } = Š i a b( J +1 2 ) Š1 2 a b(1 Š4 N ) A +2 Ta b(3.2.12b) [ Ta b, Sc ]=1 2 ( a cSb Š1 N a bSc ),(3.2 .12c)

PAGE 76

663.REPRESENTATIONSOFSUPERSYMMETRY[ A Sc ]=1 2 Sc ,[, Sc ]= Š i1 2 Sc ,(3. 2.12d) [ J Sc ]= Š1 2 i ( | |Sc ),[ P€, Sc ]= Š Qc€,(3. 2.12e) [ Ta b, Qc ]= Š1 2 ( c bQa Š1 N a bQc ),(3.2 .12f) [ A Qc ]= Š1 2 Qc ,[, Qc ]= i1 2 Qc ,( 3.2.12g) [ J Qc ]=1 2 i ( Qc ),[ K€, Qc ]= Sc€,(3. 2.12h) [ Ta b, Tc d]=1 2 ( a dTc bŠ c bTa d),(3.2 .12i) [, K€]= Š iK€,[, P€]= iP€,(3. 2.12j) [ J K€]= Š1 2 i ( | |K )€,[ J P€]=1 2 i ( P )€,(3. 2.12k) [ J, J]= Š1 2 i ( ( J ) ),(3. 2.12l) [ P€, K€]= i ( €€J + J€€+ €€)= i ( J a b+ a b).(3.2.12m) Allother(anti)commutatorsvanisho rarefo undbyhermitianconjugation. Thesuperconformalalgebracontainsthesuper-Poincar ealgebraasas ubalgebra; however,inthesuperconformalcase,thereare no centralcharges(thisisadirectconsequenceoftheJacobiidentities).Inthesamewaythatthesupersymmetrygenerators Q actassquarerootsofthetranslationgenerators P ,the S -supersymmetrygenerators S actassquarerootsofthespecialconformalgenerators K .The newbos oniccharges A and Ta bgeneratephaserotationsofthespinors(axialor 5rotations)and SU ( N )transformationsrespectively(allbutthe SO ( N )s ubgroupofthe SU ( N )isaxial).For N =4, theaxialcharge A dropsoutofthe { Q S } anticommutatorwhereasthe[ Q A ]and [ S A ]commuta torsare N i ndependent.Thenormalizationof A ischosen suchthat Ta b+1 N a bA generates U ( N )(e.g.,[ Ta b+1 N a bA Qc ]= Š1 2 c bQa ).

PAGE 77

3.2.Thesupersymmetrygroups67f.Super-deSitteralgebra Finally,weturntothesupersymmetricextensionofthedeSitteralgebra.The generatorsofthisalgebraarethegeneratorsofthedeSitteralgebra( P, J, J ),spinorial generators( Q, Q ),and1 2 N ( N Š 1)bosonic SO ( N )charges Tab= Š Tba.Theycanbe constructedoutofthesuperconformalalgebra(justasthesuper-Poincar ealgeb raisa subalgebraofthesuperconformalalgebra,soisthesuper-deSitteralgebra).Wecan de“nethegeneratorsofthesuper-deSitteralgebraasthefollowinglinearcombinations ofthesuperconformalgenerators: P€= P€+ | |2K€, Qa = Qa + abSb J= J, Tab= c [ bTa ] c,(3. 2.13) where,sincewebreak SU ( N )to SO ( N ),wehavelo weredtheis ospini ndicesofthe superconformalgeneratorswithakronecke rdelta.(We couldalsofo rmallymaintain SU ( N )invariancebyu singinstead absatisfying ab= baand ac bc a b,with ab= abinanappropriate SU ( N )fra me.)Thuswe“ndthefollowingalgebra: { Qa Qb } =2 ( Š i abJ+ CTab), (3.2.14a) { Qa Qb€} = a bP€,(3. 2.14b) [ Qa P€]= Š Cab Qb€,(3. 2.14c) [ J, Qc ]=1 2 iC ( Qc ),(3. 2.14d) [ J, P€]=1 2 iC ( P )€,(3. 2.14e) [ P€, P€]= Š i 2 | |2( C€€J+ C J€€),(3.2 .14f) [ J, J]= Š1 2 i ( ( J ) ).( 3.2.14g) [ Tab, Qc ]=1 2 c [ aQb ] ,(3. 2.14h)

PAGE 78

683.REPRESENTATIONSOFSUPERSYMMETRY[ Tab, Tcd]=1 2 ( b [ cTd ] aŠ a [ cTd ] b)(3. 2.14i) Thisalgebra,incontrasttothesuperconformalandsuper-Poincar ecases ,dep endsona dimensionalconstant .Physi ca lly, | |2isthecurvatureofthedeSitterspace.(Actually,thesignissuchthattherelevantspaceisthespaceofconstant negative curvature, oranti-deSitterspace.Thisisaconsequen ceofsupersymmetry:Thealgebradeterminestherelativesigninthecombination P + | |2K above.)

PAGE 79

3.3.Representationsofsupersymmetry693.3.Representationsofsupersymmetry a.Particlerepresentations Beforedisc ussing“eldrepresentationsofsupersymmetry,westudytheparticle contentofPoincar esupersy mmetrictheories.WeanalyzerepresentationsofthesupersymmetrygroupintermsofrepresentationsofitsPoincar es ubgroup.Because P2isa Casimiroperatorofsupersymmetry(itcommut eswithallthegenera tors),allelements ofagivenirreduciblerepresentationwillhavethesamemass. a.1.Masslessrepresentations We“rstco nsidermasslessrepresentations.WethencanchooseaLorentzframe wheretheonlynonvanishingcomponentofthemomentum p ais p+.Inthisf ramethe anticommutationrelationsofthesupersymmetrygeneratorsare { Qa+, Qb+} =0, { Qa+, Qb€+} = p+a b, { QaŠ, QbŠ} =0, { QaŠ, Qb€Š} =0, { Qa+, QbŠ} =0, { Qa+, Qb€Š} =0.(3. 3.1) Sincetheanticommutatorof Qa Šwithitshermitianconjugatevanishes, Qa Šmustvanishidenticallyonallphysicalstates:From(3.2.11)wehavetheresultthat 0= < |{ A A}| > =n (|< n | A| >|2+|< n | A | >|2) < n | A | > = < n | A| > =0.(3. 3.2) Ontheotherhand, Qa+anditshermitianconjugatesatisfythestandardanticommutationrelationsforannihilationandcreationo perato rs,uptonormalizationfactors(with theexceptionofthecase p+=0,whichinth isframemeans p a=0andd escribesthe physicalvacuum).Wecanthusconsiderastate,the Cliordvacuum | C > ,whichis annihilatedbyalltheannihilationoperators Qa+(orconstructsuchastatefromagiven statebyoperatingonitwithasucientnumberofannihilationoperators)andgenerate allotherstatesbyactionofthecreationoperators Qa€+.Sin ce,asusual,anannihilation operatoractingonanystateproducesanotherwithonelesscreationoperatoractingon

PAGE 80

703.REPRESENTATIONSOFSUPERSYMMETRYth eC li or dv ac uum,thissetofstatesisclosedundertheactionofthesupersymmetry generators,andthusformsarepresentatio nofthesupers ymmetryalgebra.Furthermore,iftheCliordvacuumisanirreduc iblerepresentationofthePoincar egro up,this setofsta tesisanirreduciblerepresentation ofthesupersymmetr ygro up,sinceany attempttoreducetherepresentationbyimposingaconstraintonastate(oralinear combinat ionofstates)wouldalsoconstrain theCliordvacuum(a fterapplyingan appropriatenumberofannihilationoperators;seealsosec.3.8.a).TheCliordvacuum mayalsocarryrepresentationsofisosp inandotherinternalsymmetrygroups. TheCliordvacuum,beinganirreduciblerepresentationofthePoincar egro up,is alsoaneigenstateofhelicity.Inthisframe, Qa€+hashe licity Š1 2 ,thusdet erminingthe he licitiesoftheotherstatesintermsofthat oftheCliordvacuum.(Ingeneralframes, thehelicity Š1 2 componentof Q€ isthecreationoperator,andthehelicity+1 2 component,whichisthelinearlyindependentLorentzcomponentof P€ Qa€,vanishes: { P€ Qa€, P€Qb } = b aP2P€=0,since p2=0inthema sslesscase.)Therepresentati on so ft hestatesunderisospinarealsodeterminedfromthetransformationproperties oftheCliordvacuumandthe Q s:Wetakethete nsorproductoftheCliordvacuumsrepresentationwi ththatofthecreationoperators(namely,thatformedbymultiplyingtherepresentationsoftheindividualoperatorsandantisymmetrizing). Asexampl es,weconsiderthecasesofthemasslessscalarmultiplet( N =1,2), su pe r-Yang-Mills( N =1,. ..,4),andsupe rgravity( N =1,. ..,8),de“ne dbyC liord vacuawhicharei soscalarsandhavehelicity+1 2 ,+1, and+2,respectively.(Inthe sc alarandYang-Millscases,thestatesmaycarryarepresentationofaseparateinternal symmetrygroup.)ThestatesarelistedinTab le3.3.1.Eachstateistotallyantisymmetricintheisospinindices,andthus,foragiven N ,sta teswithmorethan N isospin indicesvanish.Thescalarmultipletcontainshelicities(1 2 ,...,1 2 ŠN 2 ),superYang-Mills containshe licities(1,...,1 ŠN 2 ),andsupergravitycontainshelicities(2,...,2 ŠN 2 ).In addition,anyrepresentationofaninternalsymmetrygroupthatcommuteswithsupersy mmetry(suchasthegaugegroupofsuperYang-Mills)carriedbytheCliordvacuum iscarriedbyallstates(soinsuperYang-Millsallstatesareintheadjointrepresentation ofthegaugegroup).Thusthetotalnumberofstatesinamasslessrepresentationis 2Nk ,where k isthenumberofstatesi ntheC liordvacuum.

PAGE 81

3.3.Representationsofsupersymmetry71 he licityscalarmultipletsuper-Yang-Millssupergravity +2 = | C > + 3/2 a+1 = | C >ab+ 1/2 = | C >aabc0 aaba bcd-1/2 ababca bcde-1 a bcda bcdef-3/2 a bcdefg-2 a bcdefgh Table3.3 .1.Statesintheoriesofphysicalinterest TheCPTconjugateofastatetransformsasthecomplexconjugaterepresentation. JustasforrepresentationsofthePoincar egro up,onemayidentifyasupersymmetry representationwithitsconjugateifithasthesamequantumnumbers:i.e.,ifitisareal representation.(Intermsofclassical“elds,or“eldsinafunctionalintegral,thisselfconjugacyconditionrelates“eldstotheircom plexconjugates:see(3 .12.4c)or(3.12.11). Thus,inafunctionalintegralformalism,se lf-conjugacyiswithrespecttoatypeof char geconjugation:Achargeconjugationiscomplexconjugationtimesamatrix(see sec.3.3.b.5).)Fortheaboveexamples,thisself-conjugacyoccursfor N =4supe rYan gM illsand N =8superg ravity.(Thisisnottrueforthe N =2scal armultiplet,sincean SU (2)isospinorcannotbeidenti“edwithitsc omplexconj ugate,unlessanextraisospin i ndexoftheinternal SU (2)symmetry,i ndependentofthesupersymmetry SU (2),is a dded.Theself-conjugacythensimplycancelsthedoublingintroducedbytheextra i ndex.) a.2.Massiverepresentationsandcentralcharges Themassivecaseistreatedsimilarly,exceptthatwecannolongerchoosethe Lorentzframeabove;instead,wechoosetherestframe, p€= Š m €: { Q Q } =0, { Q Q€ } = Š m € .(3. 3.3) Nowwehavetwiceasmanycreationandannihilationoperators,the QŠsaswellasthe

PAGE 82

723.REPRESENTATIONSOFSUPERSYMMETRYQ+s.Thereforethenumberofstatesinamassiverepresentationis22 Nk .(Forexample, an N =1ma ssivevectormult iplethashelicitycontent(1,1 2 ,1 2 ,0).) Thecasewithcentralchargescanbeanalyzedbysimilarmethods,butitissimpler to understandifwerealizethatsupersymmetryalgebraswithcentralchargescanbe obtainedfromsupersymmetryalgebraswitho utcentralchargesinhigher-dimensional spacetimesbyinterpretingsomeoftheextracomponentsofthemomentumasthecentralchargegenerators(theywillcommutewithallthefour-dimensionalgenerators). Theanalysisofthestatecontentisthenthesameasforthecaseswithoutcentral char ges,sincebothcasesareobtainedfromthesamehigher-dimensionalsetofstates (ex ceptthatwedonotkeepthefullhigher-dimensionalLorentzgroup).However,the twodisting uishingcas esarenow,intermsof P2 high er Š dimens ional= P2+ Z2= 1 2 ( P aP a+ ZabZab):(1) P2+ Z2=0,which hasthesamesetofstatesasthemassless Z =0case (thoughthestatesarenowmassive,haveasmallerinternalsymmetrygroup, an dt ransformsomewhatdierentlyundersupersymmetry),and(2) P2+ Z2< 0,which hasthesamesetofstatesasthemassive Z =0case.Bythissamea nalysis,weseethat P2+ Z2> 0isnot allowed(ju stasfor Z =0weneverhave P2> 0). a.3.Casimiroperators Wecanconst ructotherCasimiroperatorsthan P2.We“rstd e“nethesupersymmetricgeneralizationofthePauli-Lubanskivector W€= i ( P€JŠ P€ J€€) Š1 2 [ Qa Qa€],(3.3 .4) wherethelasttermisabsentinthenonsupersymmetriccase.Thisvectorisnotinvarian t undersupersymmetrytransformations,butsatis“es [ W a, Q ]= Š1 2 P aQ ,[ W a, Q€ ]=1 2 P a Q€ .(3. 3.5) Asaresult, P[ aW b ]commuteswith Q ,andth usitssquare P2W2Š1 4 ( P W )2commuteswitha llthegen eratorsofthesuper-Poincar ealgebraandisaCa simiroperator. InthemassivecasethisCasimiroperatorde“nesaquantumnumber s ,the superspin. Thegeneralizationofthenonsupersymmetricrelation W2= m2s ( s +1)is

PAGE 83

3.3.Representationsofsupersymmetry73P2W2Š1 4 ( P W )2= Š m4s ( s +1).(3 .3.6) Inthemasslesscase,notonly P2=0, butalso P€Qa = P€ Qa€=0,and hence P W = P[ aW b ]=0.Howev er,usingthegenerator A ofthesuperco nformalgroup (3.2.12),wecanconstructa nobj ectthatcommuteswith Q and Q : W aŠ AP a.Thus wecand e“neaquantumnumber ,the superhelicity, thatgeneralizeshelicity 0(de “nedby W a= 0P a): W aŠ AP a= P a.(3. 3.7) Wealsocanc onstructsupersymmetryinvariantge nera lizationsoftheaxialgenerator A andofthe SU ( N )gen erators: W5 P2A +1 4 P€[ Qa Qa€], Wa b P2Ta b+1 4 P€([ Qa Qb€] Š1 N a b[ Qc Qc€]).(3.3.8) Inthemassivecase,the superchiralcharge andthe superisospin quantumnumberscan thenbede “nedastheusualCasimiroperatorsofthemodi“edgroupgenerators Š mŠ 2W5, Š mŠ 2Wa b.Inthema sslesscase,wede“netheoperators W5 € P€A +1 4 [ Qa Qa€], Wa b € P€Ta b+1 4 ([ Qa Qb€] Š1 N a b[ Qc Qc€]).(3.3.9) Thesecommutewith Q and Q whenthecondition P€Qa =0holds, whichisprecisely thema sslesscase.Since P€W5 €= P€Wa b €=0,wecan “ndmatrixrepr esentations g5, ga bsuchthat W5 c= g5P c, Wa b c= ga bP c.(3. 3.10) Thesuperchiralchargeis g5,andsuperi sospinquantumnumberscanbede“nedfromthe tra celessmatrices ga b.Allsupersy mmetricallyinvariantoperatorsthatwehaveconstructedcanbereexpressedintermsofcovariantderivativesde“nedinsec.3.4.a;seesec 3.4.d.

PAGE 84

743.REPRESENTATIONSOFSUPERSYMMETRYb.Representationsonsuper“elds Weturnnowto“eld(o -shell)representationsoft hesupersymmetryalgebras. Thesecanbedescribedinsuperspace,whichisanextensionofspacetimetoinclude extraanticommutingcoordinates.Todiscovertheactionofsupersymmetrytransformationsonsupersp ace,weuset hemethodofinducedrepresentations.Wediscussonly simple N =1supersy mmetryforthemoment. b.1.Superspace Ordinaryspacetimecanbede“nedasthecosetspace(Poincar egro up)/(Lorentz group).Similarly, global”atsuperspace canbede“nedasthecosetspace (super-P oincar egro up)/(Lorentzgroup):ItspointsaretheorbitswhichtheLorentz groupsweepsoutinthesuper-Poincar egro up.Relative tosomeorigin,thiscosetspace canbeparametrizedas: h ( x , )= ei ( x€ P€+ Q+ € Q€)(3.3.11) where x , arethecoor dinatesofsuperspace: x isthecoordinateofspacetime,and arenewfermionicspinorcoordinates.Thehaton P and Q i ndicatesthatthey areabstractgroupgenerators, not tobeconfusedwiththed ierentialoperators P and Q usedtorepresentthembelow.Thestatisticsof aredetermine dbythoseof Q Q : { } = { } = { Q } =[ x ]=[ P ]=0,(3 .3.12) etc.,thatis, areGrassmannparameters. b.2.Acti onofgeneratorsonsuperspace Wede “netheactionofthesuper-Poincar egrouponsupe rspacebyleftmultiplication: h ( x, )= gŠ 1 h ( x , ) modSO (3,1)(3.3.13) where g isagroup element,and modSO (3,1)meansthatanytermsinvolvingLorentz generatorsaretobe pushedthroughtotherightandthendropped.To“ndtheaction ofthegenerators( J P Q )onsuperspa ce,weconsider

PAGE 85

3.3.Representationsofsupersymmetry75 g = eŠ i ( J + €€ J€€), eŠ i ( € P€), eŠ i ( Q+ € Q€) ,(3. 3.14) respectively.UsingtheBaker-Hausdortheorem( eAeB= eA + B + 1 2 [ A B ]if [ A ,[ A B ]]=[ B ,[ A B ]] =0 )t or earrangetheexponents,we“nd: J & J : x €=[ e] [ e ]€€x€, =[ e] €=[ e ]€€ €, P : x a= x a+ a, = €= €, Q & Q : x a= x aŠ i1 2 ( €+ €), = + €= €+ €.(3. 3.15) Thusthegeneratorsarerealizedascoordinatetransformationsinsuperspace.The Lorentzgroupacts reduci bly: Underitsactionthe x sand sdo not transformintoeach other. b.3.Acti onofgeneratorsonsuper“elds Togetrepresent ationsofsupersymmetryonphysical“elds,weconsider super“elds ...( x , ):(generalized)multispinorfunction soversupersp ace.Unde rthesupersymmetryalgebratheyarede“nedtotransformascoordinatescalarsandLorentzmultispinors.Theymayalsobeina matrixrepresentationofaninternalsymmetrygroup. Thesimplestcaseisascalarsuper“eld,whichtransformsas:( x, )=( x , )or, i n“nitesimally, ( z ) Š ( z )= Š zMM( z ).Using(3.3.13),wewritethetransformati onas = Š i [( Q+ € Q€),]= i [( Q+ € Q€),],etc.Hen ce,justasin theordinaryPoincar ecase,theg enerators Q ,etc.,arerepre sented bydierentialoperators Q ,etc.: J= Š i1 2 ( x( € )€+ ( )) Š iM, P€= i €, Q= i ( Š1 2 €i €), Q€= i ( €Š1 2 i €);(3.3 .16) where Mgenera testhe matrix Lorentztransformationsofthesuper“eld:

PAGE 86

763.REPRESENTATIONSOFSUPERSYMMETRY[ M, ...]=1 2 C ( ) ...+ ... Forfut ureuse,w ewrite Q and Q as Q= e1 2 Ui eŠ1 2 U, Q€= eŠ1 2 Ui €e1 2 U,( 3.3.17a) where U = €i €.(3. 3.17b) Finally,fromtherelation { Q Q } = P ,weconcl udethat thedimensionof and is ( m )Š1 2 b.4.Extendedsupersymmetry Wenowgen eralizetoextendedPoincar esupersy mmetry.Inprinciple,theresults wepres entcouldbederivedbymethodssimilartotheabove,orbyusingasystematic dierentialgeometryprocedure.Inpracticethesimplestprocedureistostartwiththe N =1Poincar eresultsan dgen eralizethembydimensionalanalysisand U ( N )sy mmetry. Forgen eral N ,superspac ehasco ordi nates zA=( x€, a a€) ( x a, € ). Super“elds ... ab ...( x , )transfo rmasmultispinorsandisospinors,andascoordinate scalars.Includingcentralcharges,thesuper-Poincar egen eratorsactonsuper“eldsas thefollowingdierentialoperators: Qa = i ( a Š1 2 a€i €Š1 2 b Zba), (3.3.18a) Qa€= i ( a€Š1 2 a i €Š1 2 b€ Zba),(3.3 .18b) J= Š i1 2 ( x( € )€+ a ( a )) Š iM,(3. 3.18c) J€€= Š i1 2 ( x (€€ )+ a (€ a€ )) Š i M€€,(3. 3.18d) P€= i €.(3. 3.18e) Centralchargesarediscussedinsection4.6.

PAGE 87

3.3.Representationsofsupersymmetry77b.5.CPTinsuperspace Poincar esupersy mmetryiscompatiblewiththediscreteinvariancesCP(charge conjugation parity)andT(timereversal).WebeginbyreviewingC,P,andTin ordinaryspacetime.We describethetransformationsasactingon c -number“el ds, i.e., weuset hefunctionalinte gralformalism,ratherthanactingon q -number“eldsor Hilbertspa cestates. U ndera re”ection withrespecttoanarbitrary(butnotlightlike)axis u a,( u = u u2= 1)thecoordinatestransformas x a= R ( u ) x a= Š uŠ 2u€u€x€= x aŠ uŠ 2u au x R2= I (3.3.19) ( u x chan gessign,whilethecomponentsof x orthogonalto u are unchanged.)Tthen actsonthecoordinatesas R ( a 0)wh ileaspacere” ectioncanberepresentedby R ( a 1) R ( a 2) R ( a 3)(inte rmsofatimelikevector a 0,a nd threeorthogonalspacelike v ectors a i, i =1,2,3). Wede “netheactionofthediscretesymmetriesonarealscalar“eldby ( x)= ( x ).TheactiononaWeylspinoris ( x)= iu€ €( x ), €( x)= iu€( x ); ( x )= u2( x ).(3.3 .20) Sincethistransformationinvolves complexconjugatio n,weinterpret R asgivingCPand T.Indeed,sinceundercomplexconjugation eŠ ipx e+ ipx,wehave p a= Š ( p aŠ uŠ 2u au p ).Ther efore p0chan gessignforspacelike u ,andth isisconsistentwithourinterpretation.Thecom binedtransformationCPTissimply x Š x and the“eldstransformwithoutanyfactors(exceptforirrelevantphases).ThetransformationofanarbitraryLorentzrepresentationisobtainedbytreatingeachspinorindexas in(3.3.20). Thede“nitionofC,andthusPandCT,requirestheexistenceofanadditional, internal,discretesymmetry,e.g.,asymmetr yinvolving onlysignchanges:Forthephoton“eldC A a= Š A a;forap airofrealscalars,C 1=+ 1,C 2= Š 2gives

PAGE 88

783.REPRESENTATIONSOFSUPERSYMMETRYC( 1+ i 2)=( 1+ i 2).Forap airofspi nors,C 1 = 2 ,C 2 = 1 gives,forthe Diracspinor( 1 2€),thetransformationC( 1 2€)=( 2€, 1 ), i.e.,complexconjugationtimesamatrix.Therefore,Cgenera llyinvolvescomplexconjugationofa“eld, asdoCPandT,whereasPand CTdonot.(However,notethatthede“nitionofcomplexconjugationdependsonthede“nitionofthe“elds,e.g.,combining 1and 2as 1+ i 2.) Thegeneralizationtosuperspaceisstraightforward:Inadditiontothetransformation R ( u ) x givenabove,wehave(asforanyspinor) a = iu€ a€, a€= iu€a .(3. 3.21) Areals calarsuper“eldandaWeylspinorsuper“eldthustransformasthecorresponding component“elds,butnowwithallsuperspacecoordinatestransformingunder R ( u ).To preservethechiralityofasuperspaceorsuper“eld(seebelow),wede“ne R ( u )to always complexconjugatethesuper“elds.Wethushave,e.g., ( z)= ( z ), ( z)= iu€ ( z )= iu€ €( z ).(3.3 .22) Asforcomponents,Ccanbe de“nedasanadditional(internal)discretesymmetry whichcanbeexpressedasamatrix timeshermitianconjugation. Were markthat R ( u )transfo rmsthesupersymmetrygeneratorscovariantlyonly for u2=+1.For u2= Š 1thereisarel ativesignchangebetween and € i € .This isb ecauseCPchangesthesignof p0,whichis neededtomain tainthepositivityofthe energy(see (3.2.10)). b.6.Chiralrepresentationsofsupersymmetry Asinthe N =1case(s ee(3.3 .17)), Q Q canbewrittencompactlyforhigher N eveninthepresenceofcentralcharges: Q = e1 2 Ui ( Š1 2 b Zba) eŠ1 2 U, Q€ = eŠ1 2 Ui ( € Š1 2 b€ Zba) e1 2 U,( 3.3.23a) U €i €, €= c d€.(3. 3.23b)

PAGE 89

3.3.Representationsofsupersymmetry79Thisallowsusto“ndotherrepresentationsofthesuper-Poincar ealgebra inwhich Q (or Q )takeaverysi mpleform.Weperformnonunitarysi milaritytransformationson all generatorsA: A ( )= eŠ+1 2 UAe1 2 U,(3. 3.24) whichleadsto: Q (+)= i ( Š1 2 b Zba), Q€ (+)= eŠ Ui ( € Š1 2 b€ Zba) eU,(3. 3.25) or Q ( Š )= eUi ( Š1 2 b Zba) eŠ U, Q€ ( Š )= i ( € Š1 2 b€ Zba). (3.3.26a) Thegeneratorsactontransformedsuper“elds ( )( z )= eŠ+1 2 U( z ) e1 2 U(3.3.26b) Theserepresentationsarecalled chiral or antichiral representations,whereastheoriginal oneiscalledthe vector representation.Theycanalsobefounddirectlybythemethod ofinducedrepresentationsbyusingaslightly dierentparametrizationofthecosetspace manifold(superspace)(cf.(3.3.11)): h(+)= ei Qeix(+) Pei Q, h( Š )= ei Qeix( Š ) Pei Q,( 3.3.27a) where x( )= x i1 2 = e1 2 UxeŠ+1 2 U(3.3.27b) arecomplex(nonhermitian)coordinates.The correspondingsuperspacesarecalledchiralandantichiral,respectively.Thesimilaritytransformations(3.3.26b)canberegarded ascomplexcoordinatetransformations: ( z )= e1 2 U( )( z ) eŠ+1 2 U=( )( z( )),

PAGE 90

803.REPRESENTATIONSOFSUPERSYMMETRYz( )= e1 2 UzeŠ+1 2 U=( x( ), ).(3.3 .28) Hermitianconjugationtakesusfromachiralrepresentationtoanantichiralone: ( V(+))= V( Š ).Consequently,ah ermitianquantity V = V inthev ectorrepresentation satis“es V = eŠ U VeU(3.3.29) inthechiralrepresentation. b.7.Superconformalrepresentations Themethodofinducedrepresentationscanbeusedto“ndrepresentationsforthe superconformalgroup.However,weuseadierentprocedure.Therepresentationsof Q P ,and J areasinthesuper-Poincar ecase.Ther epresentationsoftheremaining generatorsarefoundasfollows:Inordinaryspacetime,theconformalboostgenerators K canbeconstructedby“rstperforming aninversion,thenatranslation( P transformation),and“nallyperforminganotherinversion;asimilarsequenceofoperationscanbe usedinsuperspacetoconstruct K from P and S from Q Wede “netheinversion operationasthefollowingmapbetweenchiralandantichiralsuperspace: x ( ) €=( x(Š+))Š 2x(Š+) €= ( x( ))Š 2x( ) €, a = i ( x( Š ))Š 2x( Š ) € a€= Š i ( x(+))Š 2x(+) €a a€= i ( x(+))Š 2x(+) €a = Š i ( x( Š ))Š 2x( Š ) € a€;(3. 3.30) wehave z= z .Thee ssentialpropertyofthismappingisthatitscalesasupersymetricallyinvariantextensionof thelineelement.Wewrite ds2=1 2 s€s€,where s€= dx€+i 2 ( a d a€+ a€d a ),(3.3 .31) isasupersymmetricallyinvariant1-form(invariancefollowsatoncefrom(3.3.15)). Underinversions(3.3.30),we“nd s €= Š ( x(+))Š 2( x( Š ))Š 2x(+) €x( Š ) €s€,

PAGE 91

3.3.Representationsofsupersymmetry81ds 2=( x(+))2( x( Š ))2ds2.(3. 3.32) Super“eldstransformas II ...€ ...( z )=( x(+))Š 2 d+( x( Š ))Š 2 dŠf€... f€ ...€ ...( z), (3.3.33a) f€ i ( x(+))Š 1x(+) €, f€ i ( x( Š ))Š 1x( Š ) €.(3. 3.33b) Here d d++ dŠisthecanonicaldimension(Weylweight)of,and dŠŠ d+isproportionalt othechiral U (1)weight w .Notethat chiral super“elds(“eldsdependingonly onand x(+)and ,not ;sees ec.3.5)with dŠ=0 an do nlyundottedindicesremainchiralafteraninversion. Wecancal culate S asdescribedabove:Weusetheinversionoperator II and compute Sa = II Q€ II and Sa€= IIQ II .Usingthes uperconformalcommutatoralgebra wethen compute K A T ,and. We “nd A =1 2 ( Š € € ) Š Y ,( 3.3.34a) Ta b=1 2 ( b a Š a€ b€Š1 N a b( Š € € ))+ ta b,(3. 3.34b) Sa = i ( x€Š i1 2 b b€) Qa€+ a b i ( b + i1 2 b€€) Š 2 i b [ ( tb a+1 4 b a(1 Š4 N ) Y ) Š1 2 b a( M +1 2 d d d d )],(3.3.34c) Sa€= i ( x€+ i1 2 b b€) Qa + a€ b€i ( b€+ i1 2 b €) Š 2 i b€[ Š €€( ta b+1 4 a b(1 Š4 N ) Y ) Š1 2 a b( M€€+1 2 €€d d d d )],(3.3.34d) = Š i1 2 ( { x€, €} +1 2 ([ ]+[ € € ])) Š id d d d ,(3. 3.34e) K€= Š i ( x€x€€+ x€ a€ a€+ x€a a Š1 4 a a€b b€€) +1 2 ( a a€b b Š a a€ b€ b€)

PAGE 92

823.REPRESENTATIONSOFSUPERSYMMETRYŠ i ( x€+ i1 2 a a€) M Š i ( x€Š i1 2 a a€) M€€Š x€id d d d Š 2 a b€( ta b+1 4 a b(1 Š4 N ) Y ).(3.3 .34f) Here d d d d isthematrixpieceofthegenerator;it seigenva lueistheca nonicaldimension d .Sim ilarly, Y ta barethematrixpiecesoftheaxialgenerator A andthe SU ( N )gen erators Ta b;theeige nval ueof Y is1 2 w .Thete rmsin S S prop ortionalto Y and ta bdo notfollowfromtheinversion(3.3.33),butaredeterminedbythecommutationrelations and(3.3.34a,b). b.8.Super-deSitte rrepre sentations Toconstr uctthegeneratorsofthesuper-deSi tteralgebra,weusetheexpressions fortheconformalgeneratorsandtakethelin earcombinationsprescribedin(3.2.13). Tosu mmarize,forgeneral N ,ineac hofthecaseswehavecon sideredthegeneratorsactasdierentialoperators.Inadditionthesuper“eldsmaycarryanontrivial matrixrepresentationofallthegeneratorsexceptfor P and Q inthePoincar eand deSittercases,and P Q K ,and S inthesuper conformalcase.Theymayalsocarrya representationofsomearbitraryinternalsymmetrygroup.

PAGE 93

3.4.Covariantderivatives833.4.Covariantderivatives Inordinary”atspacetime,theusualcoordinatederivative aistranslation invariant:thetranslationgenerator P a,whichis representedby €,commuteswith itself.Insupersymmetrictheories ,thesupert ranslationgenerator Qhasanontrivial an ti commutator,andhenceisnotinvariantundersupertranslations;asimplecomputationrevealsthatthefermion iccoordi natederivatives €arenotinvarianteither. Thereis,however,asimplewaytoconstructderivativesthatareinvariantundersupersymmetrytransformationsgeneratedby Q, Q€(a ndarecovariantunderLorentz,chiral, andisospinrotationsgeneratedby J, J€€, A ,and Ta b). a.Construction Intheprecedingsectionweusedthemethodofinducedrepresentationsto“nd theactionofthesuper-Poincar egen eratorsinsuperspace.Thesamemethodcanbe usedto“ndcovariantderivatives.Wede“netheoperators Dand D€bytheequation ( e D + D)( ei ( x P + Q + Q )) ( ei ( x P + Q + Q ))( ei ( Q + Q )).(3.4 .1) Theanticommutatorof Q with D canbeexamine dasfo llows: ( eŠ i ( Q + Q ))( e D + D)( ei ( Q + Q ))( ei ( x P + Q + Q )) =( eŠ i ( Q + Q )) ( ei ( Q + Q ))( ei ( x P + Q + Q ))( ei ( Q + Q )) =( ei ( x P + Q + Q ))( ei ( Q + Q )) =( e D + D)( ei ( x P + Q + Q )).(3.4 .2) Thusthe D s ar ei nv ar ia nt undersupertranslations(andalsounderordinarytranslations): { Q D } = { Q D } =[ P D ]=0.(3 .4.3) WecanusetheBak er-Hausdortheorem,(3.4.1),and(3.3.11,13)tocomputethe exp licitformsofthe D sfromthe Q s.W e “nd D= Š iQ+ €P€, D€= Š i Q€+ P€.(3. 4.4)

PAGE 94

843.REPRESENTATIONSOFSUPERSYMMETRYFor N =1,whenactingo nsuper “elds,theyhavetheform D= +1 2 €i €, D€= €+1 2 i €,(3. 4.5) andarecovariantgeneralizationsoftheordinaryspinorderivative €.For general N ,with centralcharges,thecovariantderivativeshavetheform: D = Da = +1 2 € i € +1 2 b Zba, D€ = Da€= € +1 2 i € +1 2 b€ Zba.(3. 4.6) Theycanberewrittenusing eUas: D = eŠ1 2 U( +1 2 b Zba) e1 2 U, D€ = e1 2 U( € +1 2 b€ Zba) eŠ1 2 U.(3. 4.7) Consequently,justasthegenerators Q simp lifyinthechiral(antichiral)representation, thecovariantderivativeshavethesimplebutasymmetricform: D (+)= eŠ U( +1 2 b Zab) eU, D€ (+)= € +1 2 b€ Zab, D ( Š )= +1 2 b Zab, D€ ( Š )= eU( € +1 2 b€ Zab) eŠ U.(3. 4.8) Inanyrepresentation,theyhavethefollowing(anti)commutationrelations: { D D } = CZab, { D D€ } = i € .(3. 4.9) ItisalsopossibletoderivedeSittercovariantderivativesbythesemethods.However,thereisaneasier,moreuseful,andmorephysicalwaytoderivethemwithinthe frameworkofsupergravity,sincedeSitterspaceissimplyacurvedspacewithconstant curvature.Thiswillbedescribedinsec.5.7. b.Algebraicrelations Thecovariantderivativessatisfyanumberofusefulalgebraicrelations.For N =1,theonlypo ssiblepowerof D is D2=1 2 DD.(B ecauseofantic ommutativity higherpowersvanish:( D )3=0.)Fromthea nticommuta tionrelationswealsohave

PAGE 95

3.4.Covariantderivatives85[ D, D2]= i € D€, D2 D2D2= D2, DD= D2, D22= Š 1.(3.4 .10) For N > 1wehavesim ilarrelations;forvanishingcentralcharges: Dn 1... n D 1... D n, Dn n +1... 2 N1 n C 2 N... 1Dn 1... n, Dn 1... n=1 (2 N Š n )! C 2 N... 1Dn n +1... 2 N, D2 N Š n 1... nDn 1... n= [ 1 1... n] nD2 N, ( Dn 1... n)= Dn €n... €1, ( D2 N Š n 1... n)=( Š 1)n D2 N Š n €n... €1, D2 N2 N=( Š 1)N, D2 ND2 N D2 N= N D2 N.(3. 4.11) Itisoftennecessarytoreducetheproductof D sor D swithrespectto SU ( N ),as wellaswith respectto SL (2, C ).Foreach,thereduction isdonebysymmetrizingand antisymmetrizingtheindices.Speci“cally,w e “ndtheirreduciblerepresentationsasfollows:Apr o duct D D .... D istotallyantisymmetricinitscombinedindicessincethe D santicommute;however,antisymmetryin impliesoppositesymmetriesbetween a b and ,(onep airsymmetric,theotherantisymmetric),andhenceaYoung tableauforthe SU ( N )i ndicesispairedwiththesameYoungtableau re”ected aboutthe diagonal forthe SL (2, C )i ndices.Thelatterisactuallyan SU (2)tableausinceifwe haveonly D s th enonlyundottedindicesappear,andhasatmosttworows.(Actually, for SU (2)acolumnof2isequivalen ttoacolumnof0 ,and hencethe SL (2 C )tab leau canbereducedtoasinglerow.)Therefore,theonly SU ( N )table auxtha tappearhave twocolum nsorless.The SL (2, C )representatio ncanbere addirectlyfromthe SU ( N ) tableau(ifwe k eepcolumnsofheight N ):Theg eneral SU ( N )table auconsistsofa“rst

PAGE 96

863.REPRESENTATIONSOFSUPERSYMMETRYcolumnof height p andasecondofheight q ,where p + q isthenumberof D s;thecorresponding SL (2 C )representationisa( p Š q )i ndextotallysymmetricundottedspinor. Thereforethisrepresentationof SL (2, C ) SU ( N )hasdi mensionality ( p Š q +1) p Š q +1 p +1 N p N +1 q .(3. 4.12) c.Geometryof”atsuperspace Thecovariant deriva tives de“ne thegeometryof”at superspace.Wewrite themasasup ervector: DA=( D D€ a).(3.4 .13) Ingeneral,in”atorcurvedspace,acovariantderivativecanbewrittenintermsofcoordinatederivatives M zM andconn ectionsA: DA DA MM+A( M )+A( T )+A( Z ).(3.4 .14) TheconnectionsaretheLorentzconnection A( M )=A M +A€€ M€€,( 3.4.15a) isospinco nnection A( T )=Ab cTc b,(3. 4.15b) andcentralchargeconnection A( Z )=1 2 (A bcZbc+Abc Zbc).(3.4 .15c) TheLorentzgenerators M actonlyon tangentspace indices.(Althoughthedistinction isunimportantin ”atspace,wedistinguishcurved,orcoordinateindices M N ... fromcova riantortangentspaceindices A B ... .Incurveds uperspaceweu sua llywrite thecovariantderivativesas A= EA MDM+A, DM M ADA, i.e.,weusethe ”at superspacecovariantderivativesinsteadofcoordinatederivatives:seechapter5for deta ils.) In”atsuperspace,inthevectorrepresentation,from(3.4.6)we“ndthe”atvielbein

PAGE 97

3.4.Covariantderivatives87DA M= 0 0 0 € € 01 2 i a m m€1 2 i €€m am a m ,(3. 4.16) andthe”atcentralchargeconnection A bc= Š1 2 ( C[ b a c ],0,0), Abc= Š1 2 (0, C€€ [ b€c ] a,0),(3 .4.17) allother”atconnectionsvanishing.Wecandescribethegeometryofsuperspacein termsofcovariant torsionsTAB C, curvaturesRAB( M ),and “eldstrengthsFAB( T )and FAB( Z ): [ DA, DB} = TAB CDC+ RAB( M )+ FAB( T )+ FAB( Z )(3. 4.18) From(3.4 .16-17),we“ndthat ”at superspacehasnonvanishingtorsion T € c= i a b €€(3.4.19) andnonvanishingcentralcharge“eldstrength F cd= Ca [ cb d ], F€ € cd= C€€[ c ad ] b,(3. 4.20) allothertorsions,curvatures,and“eldstr engthsvanishing.Hence”atsuperspacehasa nontrivial geometry. d.Casimiro pera tors Thecompletesetofoperatorsthatcommutewith P a, Q and Q€ (andtransformcovariantlyunder Jand J€€)is { DA, M M€€, Y ta b, d d d d } .(Ex ceptfor DA, whichisonlycovariantwithrespecttothesuper-Poincar ealgebra,a lltheseoperators arecovariantwithrespecttotheentiresuperconformalalgebra.Notethatthematrix operators M Y t d d d d actonlyontangentspaceindices.)ThustheCasimiroperators (groupinvariants)canallbeexpressedinte rmsoftheseoperators.Followingthediscussionofsubsec.3.3.a.3,itissucienttoconstruct:

PAGE 98

883.REPRESENTATIONSOFSUPERSYMMETRYP[ aW b ]= P[ af b ], f a1 2 [ Da Da€] Š i ( €M Š € M€€), W aŠ AP a= f a+ Yi a,(3. 4.21) Wa b= Š m2ta bŠi 4 €([ Da Db€] Š1 N a b[ Dc Dc€]), W5= m2Y Ši 4 €[ Da Da€](3. 4.22) Wa b a= ta bi aŠ1 4 ([ Da Db€] Š1 N a b[ Dc Dc€]); W5 a= Š Yi aŠ1 4 [ Da Da€](3. 4.23) whereweh aveused P2= Š m2for Wa b,and P€Q = €D = € =0for Wa b c(the masslesscase:seesubsec.3.3.a.3).

PAGE 99

3.5.Constrainedsuper“elds893.5.Constraine dsuper “elds Theexistenceofcovariantderivativesallowsustoconsiderconstrainedsuper“elds;thes implest(andformanyapplicationsthemostuseful)isachiralsuper“eld de“nedby D€ =0.(3 .5.1) Weobservethatt heconstraint(3.5 .1)impliesthatonachiralsuper“eld D =0and ther efore { D D } =0 Z =0. Inachiralrepresentation,theconstraintissimplythestatementthat(+)isindependentof thatis(+)( x , )=(+)( x ).Ther efore,inavectorrepresentation, ( x , )= e1 2 U(+)( x ) eŠ1 2 U=(+)( x(+), ),(3.5 .2) where x(+)isthechiralcoordinateof(3.3.27b).Alternatively,onecanwriteachiral super“eldintermsofageneralsuper“eldbyusing D2 N +1=0: = D2 N( x , )(3. 5.3) Thisformofthesolutiontotheconstraint(3.5.1)isvalidinanyrepresentation.Itis themostg eneralpossible;seesec.3.11. Similarly,wecande“neantichiralsuper“elds;theseareannihilatedby D .Note that ,thehermitianconjugateofachiralsuper“eld,isantichiral.Thesesuper“elds maycarryexternalindices. *** Thesupersymmetrygeneratorsarerepresentedmuchmoresimplywhentheyact onchiralsuper“elds,particularlyinthechiralrepresentation(3.3.25),thanwhenthey actongeneralsuper“elds.Forthesuper-Poincar ecasew ehave: Q = i ( Š1 2 b Zba), Q€ = a €, P a= i a, J= Š i1 2 ( x( € )€+ a ( a )) Š iM, J€€= Š i1 2 x (€€ )Š i M€€,(3. 5.4)

PAGE 100

903.REPRESENTATIONSOFSUPERSYMMETRYwhere Zab=0(asex plainedabove)but Zabis unrestricted. Ifwethinkof Zabasa partialder ivativewithrespecttocomplexcoordinates ab, i.e., Zab= i ab ,thenachiral super“eldisafunctionof x , andisi ndependentof .Inthesupe rconformal case, Zabmustvani sh,and,forconsistencywiththealgebra,achiralsuper“eldmust have no dottedindices(i.e., M€€=0).Onchiralsuper“elds,t heinvers ion(3.3.33)takes theform II ...( x )= xŠ 2 df€... ...( x, )= xŠ 2 df€... € ...( x, ), f€= i ( x )Š 1x€, x a= xŠ 2x a, = ixŠ 2x€a ;(3. 5.5) (notethat dŠ=0andh ence d = d+).Thegeneratorsofthesuperconformalalgebraare nowjust( 3.5.4), S€ = Š x€a S = Š a G K a= x€G ;(3. 5.6a) with G = J + [+ i (1 2 x a a+2 Š N )], = Š i ( x a a+1 2 +2 Š N + d d d d ), A =1 2 Š (4 Š N )Š 1Nd d d d Ta b=1 4 ([ b a ] Š1 N a b[ ]).(3.5.6b) Thecommutatoralgebrais,ofcourse,unchanged.Notethattheexpressionfor A containsaterm(1 Š1 4 N )Š 1d d d d ;thisimpliesthatfor N =4,either d d d d vanishes,o rtheaxial char gemustbedroppedfromthealgebra(seesec.3.2.e).Theonlyknown N =4theoriesareconsistentwiththisfact: N =4 Ya ng -M illshasnoaxialchargeand N =4conformalsuper gravityhas d d d d =0.Wefurth ernotethatconsistencyofthealgebraforbids theadditionofthematrixoperator ta bto Ta binthecas eofconfo rmalchiralsuper“elds. Thismeansthatconformalchiralsuper“eldsmustbeisosinglets,i.e.,cannotcarryexternalisospinindices.

PAGE 101

3.5.Constrainedsuper“elds91*** For N =1,acomple x“eldsatis fyingtheconstraint D2=0isc alledalinear super“eld.Areallinearsuper“eldsatis“estheconstraint D2G = D2G =0.Wh ilesuch objectsappearinsometheories,theyarelessusefulfordescribinginteractingparticle mult ipletsthanchiralsuper“elds.Acomplexlinearsuper“eldcanalwaysbewrittenas = D€€,whereasareal linearsuper“eld canbewri ttenas G = D D2+ h c ..

PAGE 102

923.REPRESENTATIONSOFSUPERSYMMETRY3.6.Componentexpansions a. -expansions B ecausethesquareofanyanticommutingnumbervanishes,anyfunctionofa “nitenumberofanticommutingvariableshasaterminatingTaylorexpansionwith respecttothem.Thisallowsustoexpandasuper“eldintermsofa“nitenumberof ordi naryspacetimedependent“elds,or components. Forgen eral N ,thereare4 N i ndepe ndentanticommutingnumbersin ,and thus4 N i =0 4 N i =24 Ncomponentsinanunconstrainedscalarsuper“eld.Forexample,for N =1,arealsc alarsuper“eldhasthe expansion V = C + + € €Š 2M Š 2 M + €A aŠ 2Š 2 € €+ 2 2D(3.6.1) with16realcomponents.Similarly,achiralscalarsuper“eldinvectorrepresentation hastheexpansion: = e1 2 U( A + Š 2F ) eŠ1 2 U= A + Š 2F + i1 2 € aA + i1 2 2 € a+1 4 2 2 A (3.6.2) with4independentcomplexcomponents. Theseexpansionsbecomecomplicatedfor N > 1super “eldsbutfortunatelyare no tn eed ed .H ow ev er ,w eg ivesomeexamplestofamiliarizethereaderwiththecomponentcontentofsuchsuper“elds.Forinstance,for N =2,ina ddition tocarrying Lorentzspinorindices,super“eldsarerepresentationsof SU (2).Arealscala r-isoscalar super“eldhastheexpansion V ( x , )= C ( x )+ + € € Š 2 M( )Š 2 abM( ab )Š 2€€ M(€€ )Š 2 ab M( ab )+ a b€( Wa b €+ a bV€)+ ...

PAGE 103

3.6.Componentexpansions93+ 4N + 4 N + ... + 2 2€€h€€+ ... + 4 4D( x )(3. 6.3) where Wa a €=0,wh ileachiralscalarisospinorsuper“eldhastheexpansion(inthechiralrepresentation) (+) a( x )= Aa+ b ( Cab+ ( ab ) ) Š 2 Fa ( )Š 2 bc( F( abc )+ CabFc) Š 3 b ( a b+ a b )+ 4D a,(3. 6.4) where a a =0.Thespinan diso spinofthecomponent“eldscanbereadfromthese expressions. Generalsuper“eldsarenotirreduciblerepresentationsofextendedsupersymmetry. As we discussinsec.3.11,chiralsuper“eldsareirreducibleundersupersymmetry(except forapossiblefurtherdecompositionintorealandimaginaryparts);wepresenttherea systematicwayofdecomposinganysu per“eldintoitsi rreducibleparts. Thesupersymmetrytransformationsofthe component“eldsfollowstraightforwardlyfromthet ransformationsofthesuper“ elds.Thus,forexample,for N =1,from V =[ i ( Q + Q ), V ]= C + + ... (witha cons tant spinorparameter )we “nd: C ( x )= Š ( + € €), ( x )= M Š €( i1 2 aC + A a), €( x )= € M Š ( i1 2 aC Š A a), M ( x )= Š €( €+ i1 2 a), A a( x )= Š ( C €+ i1 2 €)+ €( C€€+ i1 2 € €), ( x )= ( CD+ i1 2 €A€) Š i1 2 €€ M ,

PAGE 104

943.REPRESENTATIONSOFSUPERSYMMETRY D( x )= i1 2 a( €+ €), etc .,(3.6 .5) Similarly,forachiralsuper“eldwe“nd: A = Š = Š €i €A + F F = Š €i €.(3. 6.6) b.Projection Formanya pplications,the -expansionsjustconsideredareinconvenient;an alternativeistode“necomponentsbyprojectionofanexpressionasthe -independent partsofitssuccessivespinorderivatives.Weintroducethenotation X | toindicatethe i ndependentpartofanexpression X .The n,forexample,wecande“nethecomponents ofachirals uper“eldby A ( x )=( x , ) | ( x )= D( x , ) | F ( x )= D2( x , ) | .(3. 6.7) Thesupersymmetrytransformationsoftheco mponent“eldsfollowfromthealgebraof thecovariantderivatives D ;weuse( Š iQ ) | =( D ) | and { D, D€} = i €to “nd A = i ( Q + Q ) | = Š ( D + D ) | = Š D | = Š = i ( Q + Q ) D | = Š ( D + D ) D | =( D2Š €i €) | = F Š €i €A F = i ( Q + Q ) D2 | = Š DD2 | = Š €i €.(3. 6.8)

PAGE 105

3.6.Componentexpansions95Explicitcomputationofthecomponents showsthat,inthisparticularcase,the componentsinthe -expansionareidenticaltothosede“nedbyprojection.Thisisnot necessarilythecase:Forsuper“eldsthatarenotchiral,somecomponentsarede“ned withboth D sand D s;forthesecomponents,therei sanambiguityst emmingfromhow the D sand D sareorder ed.Forexample,the 2 componentofarealscalarsuper“eld V couldbede“nedas D2 DV | D DDV | ,or DD2V | .These de“nitionsdieronlyby spacetimederivativesofcomponentslowerdowninthe -expansion(de“nedwithfewer D s).Ingeneral,theywillalsodi erfromcomponentsde“nedby -expansionsbythe samederivativeterms.Thesedierencesarejust“eldrede“nitionsandhavenophysical signi“cance. Usually,o neparticularde“nitionofcomponen tsispref erable.Forexample,one modelthatwewillconsider(seesec.4.2.a)dependsonarealscalarsuper“eld V which transformsas V= V + i ( Š )underagaugetransformationthatleavesallthe physicsinvariant(hereisachiral“eld).Inthiscase,ifpossible,weselectcomponents thataregaugeinvariant;intheexampleabove, D2 DV | isthepreferredchoice. Ifthesuper“eldcarriesanexternalLorentzindex,theseparationintocomponents requiresreductionwithrespecttotheLorentzgroup.Thus,forexample,achiralspinor super“eldhastheexpansioninthechiralrepresentation(whereitonlydependson ): (+)( x )= + ( CD+ f) Š 2.(3. 6.9) Usingprojections,wewou ldde “nethecomponentsby =| D=1 2 D| f=1 2 D( )| = D2| .(3. 6.10) For N > 1asim ilarde“nitionofcomponentsby projectionispossible.Inthis case,inadditiontoreductionwithrespecttotheexternalLorentzindices,onecanfurtherre ducewithrespectto SU ( N )i ndices.

PAGE 106

963.REPRESENTATIONSOFSUPERSYMMETRYTheprojectionmethodisalsoconvenientfor“ndingcomponentsofaproductof super“elds.Forexample,theproduct=12ischiral,andha scomponents | =12| = A1A2, D | =( D1)2| +1( D2) | = 1 A2+ A12 D2 | =( D21)2| +( D1)( D2) | +1( D22) | = F1A2+ 1 2 + A1F2.(3. 6.11) Similarly,thecomponentsoftheproduct= 12canbeworkedoutinastraightforward manner,usingtheLeibnitzruleforderivatives. c.Thetransformationsuper“eld ThetransformationsofPoincar esupersy mmetry(translationsand Q -supersymmetrytransformations)areparametrizedbya4-vector aandaspinor respectively.Itis po ssibletoviewthese,alongwiththeparameter r ofR-symmetrytransformations generatedby A in(3.2.12,3.3.34a),ascomponentsofan x -independentrealsuper“eld a1 2 [ D€, D] | i D2D | r 1 2 D D2D | ,(3. 6.12) andtowritethesupersymmetrytransformationsintermsof andthecovariant deriva tives DA: = i ( aP a+ Q+ € Q€+2 rA ) = Š [( i D2D ) D+( Š iD2 D€ ) D€+(1 2 [ D€, D] ) €+ iw (1 2 D D2D )],(3.6.13) where1 2 w istheeigenvalueoftheoperator Y (thematrixpartoftheaxialgenerator A ). Thesetransformationsareinvariantundergaugetransformations = i ( Š ), chiraland x -independent.Consequently,theydependonlyon a, ,andthec omponent r TheR-transformationswithparameter r areaxialrotations ( x , )= eŠ iwr( x eir eŠ ir ).(3.6 .14)

PAGE 107

3.6.Componentexpansions973.7.Superintegration a.Berezinintegral Toconstr uctmanifestlysupersymme tricallyinvariantactions,itisusefultohave anotionof( de“nite)integrati onwithrespectto .Thee ssentialpropertieswerequire ofthe Berezin integralaretranslationinvarianceandlinearity.Considera1-dimensional anticommutingspace;thenthemostgeneralformafunctioncantakeis a + b .The mostgeneralformthattheintegralcantakehasthesameform:d ( a + b )= A + B where A B arefunctionsof a b .Imposing linearityandinvarian ceundertranslations + leads uniquelytotheconclusionthatd ( a + b ) b .The normalizat ionoftheintegralisarbitrary.Wechoose d =1(3.7 .1) a nd,aswefoundabove, d 1=0.(3 .7.2) Wecand e“nea -function:Werequire d ( Š )( a + b )= a + b (3.7.3) and “nd ( Š )= Š (3.7.4) Theseconceptsgeneralizeinanobviouswaytohigherdimensionalanticommuting spaces;for N -extendedsupersymmetry,d2 N d2 N picksoutthehighest component oftheintegrand,anda -functionhastheform 4 N( Š )=( Š )2 N( Š )2 N.(3. 7.5) Wede “ne 4+4 N( z Š z) 4( x Š x) 4 N( Š ).Wethushave d4+4 Nz 4+4 N( z Š z)( z )

PAGE 108

983.REPRESENTATIONSOFSUPERSYMMETRY= d4xd4 N4( x Š x) 4 N( Š )( x )=( z)(3. 7.6) Wenote thatallthepropertiesoftheBerezinintegralcanbecharacterizedbysayingitisidenticaltodierentiation: d f ( )= f ( ).(3.7 .7) Thishasanimportantconsequenceinthecontextofsupersymmetry:Becausesuperspaceactionsareintegratedoverspacetimeaswellasover ,anyspa cetimetotalderivati ve a ddedtotheintegrandisirrelevant(moduloboundaryterms).Consequently,inside aspa cetimeintegral,intheabsenceofcentralchargeswecanreplaced = by D Thisallowsustoexpandsuperspaceactionsdirectlyintermsofcomponentsde“nedby projection(seechap.4,whereweconsiderspeci“cmodels).Insidesuperspaceintegrals, wecanint egrate D byparts, b ecaused4 N = 2 N 2 N =0(since 2 N +1=0). Sincesupersymmetryvariationsarealsototalderivatives(insuperspace),wehaved4xd2 N Q = d4xd2 N Q€ =0,andth usforanygeneralsuper“eldthefollowingisasupersymmetryinvariant: S= d4xd4 N .(3.7 .8) Inthecaseofchiralsuper“eldswecande“neinvariantsinthechiralrepresentationby S= d4xd2 N ,(3.7 .9) sinceisafunctionofonly x aand .Infact, thisde “nitionisrepresentationindependent,sinceth eoperator U usedtochangerepresentationsisaspacetimederivative,so onlythe1partof e1 2 Ucontributesto S.Furthe rmore,ifweexpressintermsofageneralsuper“eldby= D2 N,we“nd S= d4xd2 N D2 N= d4xd4 N = S,(3. 7.10) since D€ = d € wheninsidea d4x integral.

PAGE 109

3.7.Superintegration99Similarly,thechiraldeltafunction,whichwede“neas 4( x Š x) 2 N( Š ) 4( x Š x)( Š 1)N( Š )2 Ninthechiralreprese ntation, takesthe followingfo rminarbitraryrepresentations: D2 N4+4 N( z Š z),(3.7 .11) whichisequivalentinthechiralrepresentation( D€ = € ),andi ngen eralrepresentations gives d4xd2 N [ D2 N4+4 N( z Š z)]( z ) = d4xd4 N4+4 N( z Š z)( z ) =( z )(3. 7.12) b.Dimensions SincetheBerezinintegralacts likeaderivative(3.7.7),italso scales likeaderivative;thusith asdime nsion[d ]=[ D ].However,from(3.4.9 ),weseetha tthe dimensionsof D m1 2 ,andco nsequently,ageneralintegralhasdimensiond4xd4 N m2 N Š 4andachiralintegralhasdimensiond4xd2 N mN Š 4.Inparticular,for N =1,wehaved4xd4 d8z mŠ 2andd4xd2 d6z mŠ 3. c.Superdeterminants Finally,weusesuperspaceintegralstode“nesuperdeterminants(Berezinians). Considera( k n )by( k n )dimensi onalsupermatrix M witha k by k dimensionalevenevenpart A ,a k by n dimensionaleven-oddpart B ,an n by k dimensionalodd-even part C ,andan n by n dimensionalodd-oddpart D : M = A C B D (3.7.13) wheretheentriesof A D arebosonicandthoseof B C arefermionic.Wede“nethe superdeterminantbyanalogywiththeusualdeterminant:

PAGE 110

1003.REPRESENTATIONSOFSUPERSYMMETRY( sdetM )Š 1= K dkxdkxdn dneŠ z tMz,( 3.7.14a) where z t=( x), z = x ,(3. 7.14b) and K isanormalizationfacto rchosentoe nsurethat sdet (1)=1.Theexponent xAx + xB + Cx + D canbewri tten,aftershiftsofintegrationvariableseitherin x orin ,intwoequiva lentforms: xAx + ( D Š CAŠ 1B ) or x( A Š BDŠ 1C ) x + D .Integrationov erthebosonicvariablesgivesusaninverse determinantfactor,andintegrationoverthe fermionicvariablesgivesadeterminantfactor.Weobtain sdetM intermsofordinarydeterminants: sdet ( M )= detA det ( D Š CAŠ 1B ) = det ( A Š BDŠ 1C ) detD .(3. 7.15) Thisformulahasanumberofusefulprope rties.Justaswiththeordinarydeterminant,thesuperdeterminantoftheproductofseveralsupermatricesisequaltothe productofthesuperdeterminantsofthesupermatrices.Furthermore, ln ( sdetM )= str ( lnM ), (3.7.16a) wherethesupertraceofasupermatrix M isthetraceofthee ven-ev enmatrix Aminus thetraceoftheodd-oddmatrix D : strM trA Š trD (3.7.16b) An ar bitraryin“nitesimalvariationof M inducesavariationofth esuper determinant: ( sdetM )= exp [ str ( lnM )] =( sdetM ) str ( MŠ 1 M )(3. 7.17)

PAGE 111

3.8.Superfunctionaldierentiationandintegration1013.8.Superfunctionaldierentiationandintegration a.Dierentiation Inthissectionwediscussfunctionalcalculusforsuper“elds.Webeginbyreviewingfunctionaldierentiationforcomponent“elds:Byanalogywithordinarydierentiation,functionaldierentiationofafunctional F ofa“eld A canbede“nedas F [ A ] A ( x ) = 0lim F [ A + xA ] Š F [ A ] ,(3. 8.1) where xA ( x)= 4( x Š x).(3.8 .2) Thisis not thesameasdividing F by A .The derivativecanalsobede“nedforarbitraryvariationsbyaTaylorexpansion: F [ A + A ]= F [ A ]+ A F [ A ] A + O (( A )2),(3.8 .3) wheretheproduct(,)oftwoarbitraryfunctionsisgivenby ( C B )= d4xC ( x ) B ( x ).(3.8 .4) Inparticular,from(3.8.2)we“nd ( xA B )= B ( x ).(3.8 .5) Thisde“nitionallowsaconvenientprescriptionforgeneralizeddierentiation.For example,incurvedspace,wheretheinvariantproductis( C B )= d4xg1/2CB ,the normaliz ation( A B )= B ( x )co rrespondstothefunctionalvariation A ( x)= gŠ 1/2( x ) 4( x Š x).Generally,achoiceof xisequivalenttoachoiceofthe product(,).Inparticula r,for(3.8.2,4)wehavethefunctionalderivative A ( x ) A ( x) = 4( x Š x).(3.8 .6) Incurvedspace,usingtheinvariantproduct,wewouldobtain gŠ 1/2( x ) 4( x Š x).Note thattheinnerpro ductisnot alwayssymmetric:In( C B ), C transformscontragredientlyto B .Forex ample,if A isacovariantvector,thequantityontheleft-handsideof

PAGE 112

1023.REPRESENTATIONSOFSUPERSYMMETRYtheinner-productisacovariantvector,whilethatontherightisacontravariantvector; if A isanisospinor, F A isacomplex-conjugateisospinor;etc. Insuperspace,thede“nitionsforgeneralsuper“eldsareanalogous.Theproduct (,)isd4+4 Nz ( z )( z )=d4xd4 N ( x )( x ),andthus ( z ) ( z) = 4+4 N( z Š z)= 4( x Š x) 4 N( Š ).(3.8 .7) (Appropriatemodi“cationswillbemadein curvedsuperspace.)However,forchiral super“eldswehave (,)= d4+2 Nz = d4xd2 N ,(3. 8.8) sinceandessentia llydependononly x aand ,not € .Thevariatio nisthe refore de“nedintermsofthechiraldeltafunction: z( z)= D2 N4+4 N( z Š z)(3. 8.9) sothat ( z ) ( z) = D2 N4+4 N( z Š z),(3.8 .10) andthecomplexconjugaterelation ( z ) ( z) = D2 N4+4 N( z Š z).(3.8 .11) (Again,appropriatemodi“cationswillbema deincurvedsuperspace.)Furthermore, variations ofchiralintegralsgivetheexpectedresult ( z) d4xd2 N f (( z ))= d4xd2 N f(( z )) D2 N4+4 N( z Š z) = d4xd4 N f(( z )) 4+4 N( z Š z)= f(( z)).(3.8.12) Whenthefunctionaldierent iationisonanexpressionappearinginachiralintegral with d2 N ,the D2 Ncanalwaysbeusedt oconvertittoa d4 N integral,afterwhichthe full -functioncanbeusedasin(3.8.12).

PAGE 113

3.8.Superfunctionaldierentiationandintegration103Thisresultcanalsobeobtainedbyexpre ssingintermsofageneralsuper“eld, as= D2 N:wehave ( z ) ( z) = D2 N( z ) ( z) = D2 N ( z ) ( z) = D2 N4+4 N( z Š z).(3.8 .13) Wecanthusi dentify with for= D2 N. Thesede“nitionscanbeanalyzedintermsofcomponentsandcorrespondtoordinaryfunctionaldierentiationofthecomponent“elds.Wecannotde“nefunctionaldifferentiationforconstrainedsuper“eldsotherthanchiralorantichiralones.Forexample,foralinearsuper“eld( whichcanbewrittenas= D€ €)thereis nofunctional derivati vewhichisb othlinearandascalar. b.Integration Inchapters5and6wediscussquantizationofsuper“eldtheoriesbymeansof functionalintegration.Weneedtode“neonly integralsofGaussians,asallotherfunctionalintegralsinperturbationtheoryarede“nedintermsofthesebyintroducing sourcesanddierentiatingwithrespecttothem.Thebasicintegralsare IDVed4xd4 N1 2 V2=1 ,( 3.8.14a) ID ed4xd2 N1 2 2=1,(3. 8.14b) ID ed4xd2 N 1 2 2=1,(3. 8.14c) where,e.g., IDV =i IDVi,for Vithecomponentsof V .B ecauseasuper“eldhasthe samenumberofboseandfermicomponents,manyfactorsthatappearinordinaryfunctionalintegralscancelforsuper“elds.Thuswecanmakeanychangeofvariablesthat doesnotinvolve both exp licit sand swithoutg eneratinganyJacobianfactor, b ecauseunlessthebosonsandfermionsmixnontrivially,thesuperdeterminant(3.7.14) isequaltoo ne.Forexample,achangeofvariables V f ( V X )where X isanarbitraryexternals uper“eldgeneratesnoJac obianfactor;thesameistrueforthechangeof

PAGE 114

1043.REPRESENTATIONSOFSUPERSYMMETRYvariables V V aslongas isapurelybosonicoperator.NontrivialJacobian determinantsariseforchangesofvariablessuchas V D2V or V V where is ba ck groundcovariant,e.g.,insupergravityorsuper-Yang-Millstheory,andhencecontainsspinorderivatives. To provetheprecedingassertions,weconsiderthecasewithone ;thege neralcase canbeprovenbychoosingoneparticular an dp roceedinginductively.Weexpandthe super“eldwithrespectto as V = A + ;sim ilarly,weexpandthearbitraryexternal super“eldas X = C + .Thenwecan expandthenewvariable f ( V X )as f ( V X )= f ( A C )+ [ fV( A C ) | + fX( A C ) | ](3. 8.15) where fV| fA ( f | ) A ,etc.TheJa cobianofthistransformationis sdet f V = sdet fA( A C ) 0 fAA+ fACfA( A C ) = det ( fA) det ( fA) =1.(3. 8.16) Inparticular,theexternalsuper“eld X canbeanonlocaloperatorsuchas Š 1. An i mmediateconsequenceoftheprecedingresultisthatsuper“eld -functions ( V Š V) i ( ViŠ V i)( 3.8.17a) ar ei nv ar ia nt under -nonmixingchangesofvariables: ( f ( V ))=f ( ci)=0 ( V Š ci).(3.8 .17b) Ingeneral,ifnontrivialoperatorsappearintheactions,thefunctionalintegralsare nolongerconstant.We“rstintroducethefollowingconvenientnotation: V ,t d4x d4 N Vt d2 N t d2 N t ,(3. 8.18) where V ,,and themselvescanst andforseveralsuper“eldsarrangedascolumnvectors.Wenextconsideractionsoftheform

PAGE 115

3.8.Superfunctionaldierentiationandintegration105S =1 2 tO O ,(3.8 .19) wherethe nonsingular operator O O issuchthatthecomponentsofthecolumnvector O O havethesamechiralityasthecorrespondingcomponentsof.Theseactionsgivethe “eldequations S = O O ,(3.8 .20) dueto theintegrationmeasureschosenforthede“nitionoftheintegrals(3.8.18). Wede “ne,forcommuting, ( det O O )Š1 2 ID eS,(3. 8.21) with S givenby(3.8.19).Foranticommutingweobtain( det O O )1 2 .Then(3. 8.14)can bewri ttenas det I I =1.(3. 8.22) Fromthede “nition (3.8.21)wehaveID 1ID 2e1 TO O 2=( det O O )Š 1.(3. 8.23) Wealsohave ( det O O1)( det O O2)= det ( O O1O O2).(3.8 .24) Thiscanbeprovenasfollows:Weconsidertheaction(1 tO O12+3 tO O24).(3.8 .25) Thefunctionalintegraloftheexponentialofthisactionisequaltothatof(1 tO O1O O22+3 t4),(3.8 .26) ascanbeseenfromthe“eldrede“nitions 2 O O22,4 O O2 Š 14,(3. 8.27)

PAGE 116

1063.REPRESENTATIONSOFSUPERSYMMETRYwhoseJacobianscancel. Asanimportantexampleweconsiderthe N =1casewith oneandone and no V : = O O = 0 D2 D20 .(3. 8.28) Thisoperatorsatis“estheidentity O O2= .(3.8 .29) Therefore,from(3.8.24)wehave ( det O O )2= det (3.8.30) andhencetheintegraloftheexponentialoftheaction S =1 2 [ d4xd2 (1 D2 1+2 D2 2)+ h c .] = d4xd4 ( 11+ 22)(3. 8.31) isequalt othatof S =1 2 [ d4xd2 + h c .].(3. 8.32) Inthesamemannerwehavethefollowingequivalence: d4xd4 m 2 m +1 i =1 d4xd4 ii.(3. 8.33) Asanotherexampleweconsiderthecaseofachiralspinor: = € O O = 0 i €D2i € D20 ,(3. 8.34) with O O2= 2.(3.8 .35)

PAGE 117

3.8.Superfunctionaldierentiationandintegration107Therefore d4xd4 €i € 1 2 [ d4xd2 + h c .].(3. 8.36)

PAGE 118

1083.REPRESENTATIONSOFSUPERSYMMETRY3.9.Physical,auxiliary,andgaugecomponents Insection3.6wediscussedthecomponent“eldcontentofsupersymmetrictheories.However,the“eldcontentofatheorydo esnotdetermineitsphysicalstates.Conversely,agiv ensetofphysicalstatescanbedescribedbydierentsetsof“elds. Givenasetof“eldsandtheirfreeLagrangian,wecanclassifyany component ofa “eldas oneofthreetypes:(1) physical, withapropagatingdegreeoffreedom;(2) auxiliary, withanequationofmotionthatsetsitidenticallyequaltozero;and(3) gauge, not appearingintheLagrangian.(Super)Fieldsc ancontainallthreekindsofcomponents; o-shellrepresentations(ofthePoincar eorsupersy mmetrygroup)containonlythe“rst two; andon-shellrepresentationscontai nonlyt he“rst.Wealsoclassifyany “eld asone ofthreetypes:(1)physical,containingphysicalcomponents,butperhapsalsoauxiliary and/orgaugecomponents;(2)auxiliary,containingauxiliary,butperhapsalsogauge, components;and(3) compensating, containingonlygaugecomponents. Thesimplestexampleofthisistheconventionalvectorgauge“eldofelectromagnetism.Theexplicitseparationisnecessarilynon(Poincar e)covariant,andismostconvenientlyp erformedina light-cone formalism.Inthenotatio nof(3. 1.1)wetreat xŠ= xŠ€Šasthetimecoordinate,and x+, xT, xTasspacecoordinates.Wearethus freetoconstructexpressionsthatarenonlocalin x+(i.e.containinginversepowersof +),sincet hedynamicsisdescribedbyevolutionin xŠ€Š.(Infact,t heformalis mclo sely resembles nonrelativistic “eldtheo ry,with xŠactingasthetimeand +asthe ma ss.) Thevectorgauge“eld A€transformsas A€= € .(3. 9.1) Bymakingthe“eldr ede“nitions(by A f ( A )wemean A = f ( A)andthen drop alls) A+ A+, AT AT+( +)Š 1TA+, AŠ AŠ+( +)Š 1( ŠA+Š T ATŠ TAT);(3.9 .2) weobtain thenewtransformationlaws

PAGE 119

3.9.Physical,auxiliary,andgaugecomponents109 AT= AŠ=0, A+= + .(3. 9.3) Furthe rmore,theLagrangian IL = Š1 2 FF,(3. 9.4) where F=1 2 ( €A )€(3.9.5) intermsofthe oldA ,b ecomes IL = AT ATŠ1 4 AŠ( +)2AŠ.(3. 9.6) Thus,thecomplexcomponent ATde scribesthetwophysical(propagating)polarizations, therealcomponent AŠisauxiliary(ithasnodynamics;itsequationofmotionsetsit equaltozero),andtherealcomponent A+isgauge.Inthisformalismtheobvious gaugechoiceis A+=0 (thelight-conegauge),since A+doesnotappearin IL .Howev er, gaugecomponentsareimportantforLorentzcovariantgauge“xing:Forexample, ( €A€)2 ( +AŠ+2( +)Š 1 A+)2. Wecanp erformsimilarrede“nitionstoseparat earbit rary“eldsintophysical,auxiliary,andgaugecomponents.Anyorigin alcomponentthattransformsunderagauge transformationwitha +oranonderivativetermcorrespondstoagaugecomponentof therede“ned“eld.Anycomponentthattransformswitha Štermcorrespondstoan auxiliarycomponent.Oftheremainingcomponents,somewillbeauxiliaryandsome physical(dependingontheaction),organizedinawaythatpreservesthetransverse SO (2)Lorentzcovarian ce.Fortheknown“eldsappearingininteractingtheories,the componentswithhighestspinarephysicalandtherest(whenthereareany:i.e.,for physical spin2or3 2 )a reauxiliary.Theseargumentscanbeappliedinalldimensions. Anexamplethatillustratestheseparationbetweenphysicalandauxiliary(butnot gauge)componentswithouttheuseofnonlocal,noncovariantrede“nitionsisthatofa massivespinor“eld: IL = €i €Š1 2 m ( + € €).(3.9 .7) Since and maybeconsideredasindependent“eldsinthefunctionalintegral(and,

PAGE 120

1103.REPRESENTATIONSOFSUPERSYMMETRYinfact,mustbeconsideredindependentlocallyafterWickrotationtoEuclideanspace), wecanmakethefo llowing nonunitary(butlocalandcovariant)rede“nition: € €Š1 m i €.(3. 9.8) TheLagrangianbecomes IL =1 2 m ( Š m2) Š1 2 m € €.(3. 9.9) Wethus “ndthat representstwophysicalpolarizations,while co nt ainstwoauxiliary components. Thesameanalysiscanbemadeforthesimplestsupersymmetricmultiplet:the massivescalarmultiplet,describedbyachiralscalarsuper“eld(seesection4.1).The actionis S = d4xd4 Š1 2 m ( d4xd2 2+ d4xd2 2).(3.9 .10) Wenowr ede“ne +1 m D2;(3.9 .11) a nd,usingd2 = D2,weobtaint heac tion S =1 2 m d4xd2 ( Š m2) Š1 2 m d4xd2 2.(3. 9.12) (Notetha tthere de“nitionof prese rvesitsantichirality D =0.)Nowc ontains onlyphysicaland c ontainsonlyauxiliarycomponents;eachcontainstwoBosecomponentsandtwoFermi.Ascanbecheckedusingthecomponentexpansionof,theoriginalaction(3. 9.10)containsthespinorLagrangianof(3.9.7),whereas(3.9.12)contains theLagrangian(3.9.9).Italsocontainstwoscalarsandtwopseudoscalars,oneofeach bein gaphysical“ eld(withkine ticoperator Š m2)a nd theotheranauxiliary“eld (withkineticoperator1).Formoredetailofthecomponentanalysis,seesec.4.1.

PAGE 121

3.9.Physical,auxiliary,andgaugecomponents111Aswediscussinsec.4.1,auxiliary“eldsareneededininteractingsupersymmetric theoriesforseveralreasons:(1)Theyfacilitatetheconstructionofactions,sincewithout themthekineticandvariousinteractiontermsarenotseparatelysupersymmetric;(2) b eca us eo ft hi s, actionswithoutauxiliary“eldshavesupersymmetrytransformations thatarenonlinearandcouplin gdep endent,andmakediculttheapplicationofsupersy mmetryWardidentities(e.g.,toproverenormalizability);and(3)auxiliary“eldsare necessaryformanifestlysupersymmetricquantization.Compensating“elds(seefollowingsection)arealsonecessaryforthelattertworeasons.Althoughtheydisappearfrom theclassicalaction,theyappearinsupersymmetricgauge-“xingterms.

PAGE 122

1123.REPRESENTATIONSOFSUPERSYMMETRY3.10.Compensators Inoursubsequentdiscussions,wewilloftenusecompensating“eldsorcompensators.Theseare“elds thatenteratheoryinsuchawaythattheycanbe algebraically gaugedaway.Thus,inacertainsense,theyaretrivial:Thetheorycanalwaysbewrittenwitho utthem.However,theyfrequentlysimplifythestructureofthetheory;inparticular,they canbeusedtowritenonlinearlyrealizedsymmetriesinalinearway.This isoftenimportantforquantization.Anotherapplication,whichisparticularlyrelevant tosupergravity,arisesinsituationswhere oneknows howtowriteinvariantactionsfor systemstransformingunderacertainsymmetrygroup G (e.g.,thesuperconformal gr o up):Ifonewantstowriteactionsforsystemstransformingonlyunderasubgroup H (e.g.,thesuper-Poincar egro up),onecanenlargethesymme tryofsuchsystemstothe fullgr oupbyintroducingcompensators.Afterwritingtheactionforthesystemswith theenlargedsymmetry,onesimplychooses a gauge,thusbreakingthesymmetryofthe actiondownto thes ubgroup H Asimpleexamp leinordinary“eldtheoryisfa kescal arelectrodynamics.The usualkineticaction foracomplexscalar z ( x ) S =1 2 d4x z a az (3.10.1) hasagl obal U (1)symmetry: z= ei z .Thissy mmetrycanbegaugedtriviallybyintroducingarealcomp ensatingscalar ,a ssumedtotransformunderalocal U (1)transformationas = Š .Wecanthen constructacovariantderivative a= eŠ i aei = a+ i a thatcanbeusedt ode “nealocally U (1)invariantaction S =1 2 d4x z a az (3.10.2) Fakespinorel ectrodynamicscanbeobtainedbyanobviousgeneralization. a.Stueckelbergformalism Inthepreviousexample,thecompensatorservednousefulpurpose.TheStueckel be rgformalismprovidesafamiliarexampleofacompensatorthatsimpli“esthetheory. Webeginwit htheL agrangianforamassivevector A a: IL = Š1 8 F a bF a bŠ m2( A a)2,(3. 10.3)

PAGE 123

3.10.Compensators113F a b= [ aA b ].(3. 10.4) Thepropagatorforthistheoryis: D a b= Š1 Š m2 ( a bŠ1 2 m2 a b)(3. 10.5) Wecanr ecastthetheoryinanimprovedformbyintroducinga U (1)compensator thatmakestheaction(3.10.3)gaugeinvariant.Wede“ne A a= A a+1 m a (3.10.6) where A aand tr an sformunder U (1)gaugetransformations: A a= a = m .(3. 10.7) Intermsofthese“elds,the gaugeinvariant Lagrangianis(droppingtheprime): IL = Š1 8 F a bF a bŠ m2( A a)2Š m aA aŠ ( a )2.(3. 10.8) Wenowc hoosea gaugebyaddingthegauge“xingterm ILGF= Š1 4 ( aA aŠ 2 m )2(3.10.9) an d “nd: IL + ILGF=1 2 A a( Š m2) A a+ ( Š m2) .(3. 10.10) Thepropagatorscanbetriviallyreadofrom(3.10.10):for A a, D a b= Š a b( Š m2)Š 1,andfor ,D= Š1 2 ( Š m2)Š 1.Theyhaveb e tterhighenergy behaviorth an(3.10.5).Thus,byintroducingthecompensator ,wehavesim p li“edthe structureofthetheory.Wenotetha tthecomp ensatordecoupleswhenever A aiscoupledtoaconservedsource(i.e .,inagaugeinvariantway). b.CP(1)model Anotherfamiliarexampleisthe CP (1)nonlinear -model,whichdescribesthe Goldstonebosonsofan SU (2)gaugetheoryspontaneouslybrokendownto U (1).Itconsistsofarealscalar“eld andacomplex“eld y subjectto theconstraint

PAGE 124

1143.REPRESENTATIONSOFSUPERSYMMETRY| y |2+ 2=1(3. 10.11) Thegroup SU (2)canberealized nonlinearly onthese“eldsby =1 2 ( y + y ) y = Š 2 i y Š Š1 2 Š 1( y Š y ) y .(3. 10.12) where ,and arethe(constant)parametersoftheglobal SU (2)transformations. ThesetransformationsleavetheLagrangian IL = Š [( a )2+ | ay |2+1 4 ( y ay )2](3. 10.13) invariant,butbecausethetransformationsarenonlinearthisisfarfromobvious. Wecangiveadesc riptionofthetheorywherethe SU (2)isrepresent edlinearlyby introducinga localU (1)invariancewhichisrealizedbyacompensating“eld .U nder thislocal U (1), transformsas ( x )= ( x ) Š ( x ).(3. 10.14) Wede “ne“elds ziby z1= eŠ i z2= eŠ i y .(3. 10.15) B ecauseofthisde“nitiontheytransformunderthelocal U (1)as z i= ei zi.(3. 10.16) Theconstraint(3.10.11)becomes | z1|2+ | z2|2=1.(3. 10.17) Ignoringtheconstraintthe SU (2)actslin earlyonthese“elds(seebelow): z1= i z1+ z2 z2= Š i z2Š z1.(3. 10.18) Thecomplicatednonlineartransformations(3.10.12)ariseinthefollowingmanner: whenwe“xthe U (1)gauge z1= z1 (3.10.19)

PAGE 125

3.10.Compensators115thelinear SU (2)transformations(3.10.18)donotpreservethecondition(3.10.19).Thus wemustadda  gauge-restoring U (1)transformationwithparameter i ( x )= Š1 2 Š 1( z1Š z1)= Š i Š1 2 Š 1( z2Š z2).(3. 10.20) Thecombinedlinear SU (2)transformationandgaugetransformation(3.10.16)with nonlinearparameter(3 .10.20)preservesthegaugecondition(3.10.19)andareequivalent to(3.10.12). To writeanactioninvariantunderboththeglobal SU (2)andthelocal U (1)transformationsweneedacovariantderivative forthela tter.Byanalogywithour“rst examplewecouldwrite a= eŠ i aei = a+ i a .(3. 10.21) Amanifestly SU (2)invariantchoiceintermsofthenewvariablesis a= aŠ1 2 zi azi= a+ i a Š1 2 y ay .(3. 10.22) Thisdiersfrom(3.10.21)bythe U (1) gaugeinvariant term y ay ;oneisalwaysfreeto chan geacovariantderivativeby a ddingcovarianttermstotheconnection.(ThisissimilartoaddingcontortiontotheLorentzconnectionin(super)gravity;seesec.5.3.a.3.) ThenamanifestlycovariantLagrangianis IL = Š| azi|2= Š| azi|2Š1 4 ( zi azi)2.(3. 10.23) Inthegauge(3.10.19)thisLagrangianbecomesthatof(3.10.13). Weconsid ernowanotherapplicationofcompensators:Theconstraint(3.10.17)is awkward:Itmake sthetransf ormations(3.10.18)implicitlynonlinear.Wecanavoid thisbyintroducinga s econd compensating“eld.Weobservethatneithertheconstraint northeLagrangianareinvariantunderscaletransformations.However,wecanintroduceascaleinvarianceintothetheorybywriting zi= eŠ Zi(3.10.24)

PAGE 126

1163.REPRESENTATIONSOFSUPERSYMMETRYintermsof new“elds Ziandthecompensator ( x ).Theconstraintandtheaction, writteni nte rmsof Zi, ,w ill be in va riantunderthescaletransformations Z i= eZi, = + .(3. 10.25) The SU (2)transformationsof Ziarenowthe(truly)lineartransformations (3.10.18).The U (1)andthescaletransformationscanbecombinedintoasinglecomplexscaletransforma tionwithparameter = + i (3.10.26) Z i= eZi, = + .(3. 10.27) Theconstraint(3.10.17)becomes Z Z = e2 (3.10.28) wherew ewrite Z Z | Z1|2+ | Z2|2.Inte rmsofthenewvariablestheLagrangianis IL = Š| a( eŠ Zi) |2= Š| a( eŠ Zi) |2Š1 4 eŠ 4 ( Zi aZi)2.(3. 10.29) Substitutingfor thesolutionoftheconstraint(3.10.28),amanifestly SU (2)inv ariant procedure,leadsto IL = Š| aZi Z Z |2Š1 4 ( Zi aZi)2( Z Z )2 = Š 1 Z Z ( i kŠ Zi ZkZ Z )1 2 ( a Zi)( aZk)(3. 10.30) ThislastformoftheLagrangianisexpressedintermsofunconstrained“elds Zionly.It ismanifestlyglobally SU (2 )i nv ar ia nt an da ls oi nv ar ia nt underthelocalcomplexscale transformations(3.10.27).Wecanusethisinvariancetochooseaconvenientgauge.For example,wecanchoosethegauge Z1=1;orwecanchoo seagaugeinwhichweobtain (3.10.13).Oncewechooseagauge,the SU (2)transformationsbecomenonlinearagain. Thesetwocompensatorsallowedustorealizeaglobalsymmetry( SU (2))ofthe systemlinearly.However, theyplay dierentroles: ( x ),the U (1)compensator,gauges

PAGE 127

3.10.Compensators117aglobalsy mmetryofthesystem,whereas ( x ),thescaleco mpensatorintroducesan altogethernewsymmetry.Forthe U (1)invariancewei ntrodu cedaconnection,whereas forthescaleinvarianceweintroduced ( x )dir ectly,withoutaconnection.Intheformer case,theconnectionconsistedofapuregaugepart,andacovariantpartchosentomake it manifestlycovariantunderasymmetry( SU (2))ofthesystem;hadwetriedtointroduce ( x )dir ectly,wewouldhavefounditdiculttomaintainthe SU (2)invar ian ce.In thecaseofthescaletransformationsnosuchd icultiesarise,andaconnectionisunnecessary.Asweshallsee,bothkindsofcompensa torsappearinsupers ymmetrictheories. c.Cosetspaces Compensatorsalsosimplifythedescriptionofmoregeneralnonlinear -models. Weconsid eramodelwith“elds y ( x )thatarepoints ofacosetspace G / H ;theyt ransformnonlinearlyundertheglobalactionofagroup G butlinearlywithrespecttoa subgroup H .Byintro ducinglocaltransformationsofthesubgroup H viacompensators ( x ),werealize G linearly,andthuseasily“ndaninvariantaction. Thegeneratorsof G are T S ,where S arethegeneratorsof H and T arethe remaininggenerators,with T S antihermitian.Since H isas ubgroup,thegenerators S closeundercommutation: [ S S ] S .(3. 10.31) Werequireina ddition thatthegenerators T carryarepresentationofthe H ,thatis [ T S ] T .(3. 10.32) (Thisisalwaystruewhenthestructureconstantsaretotallyantisymmetric,sincethen theabsen ceof[ S S ] T termsimpliestheabsenceof[ T S ] S terms.) Wecouldwrite y ( x )= e ( x ) TmodH butinsteadweintroducecompensating“elds ( x ),andde “ne“elds z ( x )thatare elementsofthewholegroup G : z = e ( x ) Te ( x ) S e(3.10.33) (where= ( x ) T + ( x ) S providesanequivalentparametrizationofthegroup).The new“elds z tr an sformunder globalG -transformationsand localH -transformations: z= gzhŠ 1( x ), gG hH (3.10.34)

PAGE 128

1183.REPRESENTATIONSOFSUPERSYMMETRY(whereagainwecanuseanexponentialparametrizationfor g and h ( x )ifwewish). Thelocal H transformationscanbeusedtogaugeawaythecompensators and reduce z tothecosetvariables y .Ifwechoos ethe gauge =0,thent hegl obal G -transformationswillinducelocalgauge-restoring H -transformationsneededtomaintain =0:For gH dueto (3.10.32),weuse h ( x )= g : eT= ge TgŠ 1(3.10.35) andthust he“elds y tr an sformlinearlyunder H .For gG / H ,the gaugerestoringtransformationiscomplicatedanddependsnonlinearlyon ,andth usthe“elds y transform nonlinearlyunder G / H To “ndaglobally G -andlo ca lly H -invariantLagrangian,weconsiderthefollowing quantity: zŠ 1 az a+ A aS + B aT a+ B aT .(3. 10.36) U nderglobal G -transformations,both aand B aar ei nv ar ia nt ; underlocal H -transformationswehave ( zŠ 1 az )= hzŠ 1 a( zhŠ 1) = h ahŠ 1+ hzŠ 1( az ) hŠ 1= h ahŠ 1+ h ( A aS + B aT ) hŠ 1= h ( a+ B aT ) hŠ 1(3.10.37) Becauseof(3.10.31), hShŠ 1 S and h ahŠ 1 S ;b ecauseof(3.10.32), hThŠ 1 T ; hence A atransformsasaconnectionforlocal H transformations( atransformsasa covariantderivative),and B atransformscovariantly.Therefore,aninvariantLagrangian is IL = Š1 4 tr ( B aB a)(3. 10.38) Ifwechoosethegauge ( x )=0,thisb ecomesacomplicatednonlinearLagrangianfor the“elds y ( x ).Wecanalsoc o uplethissystemtoother“eldstransforminglinearly under H byreplacin gallderiva tiveswith a. Finally,from(3.10.33)wehave zŠ 1 az = a+ eŠ S( ae S)+ eŠ S(eŠ T ae T)e S

PAGE 129

3.10.Compensators119= a+( a ) S +( a ) T + ... (3.10.39) andhence a= a+ a S + ... and B a= a T + ... = ayT + ... .Thisisw hatwe expect:Thecovariantderivativehastheusualdependenceonthecompensator,andthe Lagrangian(3.10.38)hasaterm Š1 2 tr ( ay )2,whichisa ppropri ateforaphysical“eld.

PAGE 130

1203.REPRESENTATIONSOFSUPERSYMMETRY3.11.Projectionoperators a.General Theanalysisofmanyaspectsofthesuperspaceformulationofsupersymmetric th eoriesrequiresanunderstandingoftheirreduciblerepresentationsof(o-shell)supersy mmetry(physicalandauxiliarycomponents).Weneedtoknowhowtodecomposean arbitrarysuper “eldorproductofsuper“eldsintosuchrepresentations.Inthissection wedescribeap rocedureforconstructingprojectionoperatorsontoirreduciblerepresentationsofsupersymmetryforgeneral N Thebasicideaisthatageneralsuper“eldcanbeexpandedintoasumofchiral super“elds.Achiralsuper“eldthatisirreducibleunderthePoincar eandinte rnalsymmetrygroupsisalsoirreducibleundero-shellsupersymmetry(exceptforpossibleseparationintorealandimaginaryparts,whichwecall bisection). Thus,thisexpansionperformsthedecomposition. To showthatchiralsuper“eldsareirreducibleundersupersymmetryuptobisection,wetrytoreduceachiralsuper“eld byimposing somecovariantconstraint =0.Ifwedonotconsid errealityconditions(bisection),wecannotallowconstraints relatingto .Theonlycovariantoperatorsavailableforwritingconstraintsarethe spinorderi vatives Da Da€andthespacetimederivative €.Inmomen tums pace, sinceweareo-shell,allrelationsmustbetrueforarbitrarymomentum,andhencewe canfreelydivideoutanyspacetimederivativefactors.Therefore,anyconstraintwe writedowncanbereducedtoaconstraintthatisfreeofspacetimederivatives.Ifthe constraintco ntainedany D spinorderivatives,sinceischiral, D =0,bymovingthe D stotherightwecouldconvertthemtosp acetimederivatives,whichwehavejust arguedcanberemoved.(Forexample D DD = iD .) Wethusco ncludethatanypossibleconstraintoninvolvesonlyproductsofthe spinorderi vatives Da .Howev er,byapplyingasucientnumberof D stotheconstraint,wecanconvertallofthe D stospacetimederivatives ;hen ce,anyconstrainton i ndependentof w ouldsetitselftozero(o-shell!).Thereforemustbeirreduc ible.Thisargumentisanalogoustotheproofinsection3.3thatirreduciblerepresentationsofsupersymmetrycanbeobtainedbyrepeatedlyapplyingthegenerators Qa€to theCliordvacuum | C > de“nedby Qa | C > =0:ins teadof | C > Q ,and Q with

PAGE 131

3.11.Projectionoperators121Q | C > =0,wehave, D ,and D with D =0,resp ectively. Theonlyfurtherreductionwecanperformistoimposearealityconditiononthe super“eld.Achiralsuper“eldofsuperspin s (thespincontentofitsexternalLorentz i ndices)hasa single maximumspincomponentofspin smax= s +1 2 N residingatthe N [ a1... aN]( 1... N)levelofthesuper“ eld.(Thisismosteasilyseeninthechiralrepresentation,where achirals uper“elddependsonlyon .There ductionofproductsof sinto irreduciblerepresentationsi sdonebythemeth oddescribedforther eductionofproducts ofspinorderivativesinsec.3.4.Sincethe maximumspincomponenthasthemaximum numberof sy mmetrizedSL (2 C )i ndices,itmusthavethemaximumnumberof antisymmetrizedSU ( N )i ndices,i.e .,itmusthave N i ndicesofeachtype.Termswithfewer s havefewer SL (2 C )i ndices,whereastermswithmore scannotbeantisymmetricin N SU ( N )i ndices,andhencecannotbesymmetricin NSL (2 C )i ndices.Forexamplessee (3.6.1-4)).Onlyifwecanimposearealityconditiononthehighestspincomponentcan weimposearea lityconditionontheentiresup er“eld.Thisispossiblewhen smaxisan integer.(Acomponen t“eldwithanoddnumb erofWeylindicescannotsatisfyalocal realitycondition.) a.1.Poincar eprojectors Webeginwitht hedecompositionofanarbitrarysp inorintoirreduciblerepresentationsofthePoincar egroupinor dinaryspacetime,bothbecauseitisoneofthestepsin thesuperspacedecomposition,andbecauseitillustratessomeofthesuperspacefeatures. Thisreductionismosteasilyperformedbyconvertingdottedindicesintoundottedones withtheformaloperator€= Š i € Š1 2 ,r educingunder SU (2)(bysymmetrizingand antisymmetrizing,i.e.,takingtraces),andconvertingformerlydottedindicesbackwith .(Thisinsuresthatnofractionalpowersof remain.Wegenerallyconsider Š1 2 to behe rmitian,sincewemainlyareconcernedwith = m2> 0.)Explicitly,wewritefor eachindex =€€, €=€, (  )=( ), =.(3. 11.1) Thus,forexample,avector adecomposesinthefollowingmanner:

PAGE 132

1223.REPRESENTATIONSOFSUPERSYMMETRY a=€ =€1 2 ( C + ( )) =€1 2 (€C€+( € )€) =1 2 Š 1[ €( €€) Š €( ( € )€)] =[(L+T)] a,(3. 11.2) whereLandTarethelo ngitudinalandtransverseprojectionoperatorsforafour-vector. Theprojectionscanbewrittenintermsof “eldstrengthsS and F: (L) a= Š 1€S S =1 2 €€, (T) a= Š 1€F, F=1 2 ( € )€.(3. 11.3) The“eldstrengthsarethemselvesirreduciblerepresentationsofthePoincar egro up. TheprojectionsL=LandT=Ta reinvariantundergaugetransformations =T and =L respectively.The“eldstrengthshavethesamegaugeinvarianceastheprojections: L a= Š 1 a S =0imp lies S =0,andsi milarly T a= Š 1€ F=0imp lies F=0. a.2.Super-Poincar eprojectors Projectionsofsuper“eldscanbewritteni nte rmsof“eldstrengthsinsuperspace aswell.Wewill“ndthatprojectionsofageneralsuper“eldcanbeexpressedinterms of chiral “eldstrengthswithgaugeinvariancesdeterminedbytheprojectionoperators. Thus,forasuper“eldwithdecomposition=(nn),anysingletermn=nhas a gaugeinvariance =i = nii.Eachproj ectioncanbewri tteni ntheform n= D2 N Š n( n )wherethechiral“eldstrengths( n )= D2 NDnarePoincar eand SU ( N )i rreducibleandhavethesamegaugeinvarianceasn:0= n= D2 N Š n ( n )implies ( n )=0b ecauseandhence arei rreducible.

PAGE 133

3.11.Projectionoperators123Thesameindexconversionusedin(3.11.1)canbeusedtode“netheoperationof rest-frameconjugation onacomponent“eldor genera lsuper “eld1... i€i +1...€2 sby 1... i€i +1...€2 s=1€1... i€ii +1€i +1... 2 s€2 s 2 s... i +1€i...€1, =.(3. 11.4) Forexample,wehave: = ,=€ €, H€=€€ H€.(3. 11.5) Weextend thistochiralsuper“eldsandde“nearest-frameconjugationoperator K K whichpreserveschirality,byusinganextrafactor Š1 2 N D2 Ntoconv erttheantichiral (complexconjugatedchiral)s uper“eldbacktoachiralone(andsimilarlyforantichiral super“elds).Wede“ne K K 1... i€i +1...€2 sa1... aib1... bi= D2 N Š1 2 N1... i€i +1...€2 sb1... bia1... ai, K K ...= D2 N Š1 2 N ..., K K (...)= ( K K ...), K K2=1,[1 2 (1 K K )]2=1 2 (1 K K ),(3. 11.6) where D€ ...= D ...=0.Forex ample,foran N =1chiral spinor, K K = Š D2i € €,(3. 11.7) Wecand e“ne self-conjugacy or reality under K K ifwerestrictourselvestosuper“elds thatarerealrepresentationsof SU ( N )with smax= s +N 2 integral(thelatterisrequired toinsurethatonlyintegralpowersof appear).Therealityconditionis K K ...= ...andthesplitti ngofachiralsuper“eldintorealandimaginarypartsissimply ...=1 2 (1 K K )....(3. 11.8) Inthepreviousexample,ifwei mposetherealitycondition K K =,contr actboth sideswith Dandusetheantichiralityof €,we “ndtheequivalentcondition: D= D€ €(3.11.9)

PAGE 134

1243.REPRESENTATIONSOFSUPERSYMMETRYTheserealchiralsuper“eldsappearinmanymodelsofinterest.Forexample,isoscalar realchiralsuper“eldswith2 Š N undottedspinorindicesdescribe N 2Y ang-Mills gaugemultiplets.Similarsuper“eldswith4 Š N undottedspinorindicesdescribethe conformal“el dstren gth of N 4supergravity. Tod ecomposeageneralsuper“eldintoirreduciblerepresentations,we“rstexpand itintermsofchiralsuper“elds.Inthe chiralrepresentation ( D€ = € )aTaylorseriesin gives ( x , )=2 N n =0 1 n n €1... €n( n ) €n... €1( x ), (3.11.10a) wherencanberewrittenas ( n ) €1... €n( x )= Dn €1... €n( x , ) | =0,(3. 11.10b) or,using { D€ € } = € € and { € € } =0(which implies 2 N +1=0), ( n ) €1... €n( x )=( Š 1)N D2 N 2 N Dn €1... €n( x , ).(3. 11.10c) However, isnotcovariant,andhenceneitheristheexpansion(3.11.10).Wecangeneralize(3.11.10):For anyoperator € ( )whichobeys { D€ € } = € € { € € } =0,(3. 11.11) wecanwrite ( x , )=2 N n =0 1 n n €1... €n ( n ) €n... €1( x ), (3.11.12a) where ( n ) €n... €1( x )=( Š 1)N D2 N 2 N Dn €n... €1 ( x , ).(3. 11.12b) Ifwechoose( x , ) ( x , ( )),weobtain,substituting(3.11.12b)into(3.11.12a): ( x , )=( Š 1)N 2 N n =0 1 n n €1... €n D2 N 2 N Dn €n... €1( x , ),(3. 11.13) forany satisfying(3.11.11).Amanifestlysupers ymmetricoperatorsatisfying(3.11.11)

PAGE 135

3.11.Projectionoperators125is € = Š i € D = € Š i € .(3. 11.14) Substituting(3.11.14)into(3.11.13),we“nd ( x , )= Š N 2 N n =0 1 n Dn 1... n Š i 1 €1 ... Š i n €n D2 ND2 N Dn €n... €1( x , ),(3. 11.15) whereweh aveused( Š i € D )2 N=( Š )Š ND2 N.Pus hing Dnthrough D2 Ntothe D2 N,we “nd(reorderingthesumbyreplacing n 2 N Š n ) = Š N1 (2 N )! C 1... 2 N2 N n =0 ( Š 1)n 2 N n D2 N Š n 2 N... n +1 D2 NDn n... 1( x , ) = Š N 2 N n =0 1 n ( Š 1)nD2 N Š n 1... n D2 NDn 1... n( x , ).(3. 11.16) This“nalexpressioncanbecomparedtothenoncovariant expansionin(3.11.10). Thechiral“elds D2 NDnarethecovaria ntan alogsofthe(2 N Š n )s.Wethusobtain 1=2 N n =0 1 n ( Š 1)n Š ND2 N Š n 1... n D2 NDn 1... n.(3. 11.17) Forexample,in N =1this istherelation 1= D2 D2 Š D D2D + D2D2 .(3. 11.18) Eachterminthesumisa(reducible)proj ectionoperatorwhichpicksoutthepart ofasuper “eldappearinginthechiral“eldstrength D2 NDn(whichisi rreducible under SL (2 N C ) butreducibleunder SU ( N ) Poincar e,andpossiblyalsounder K K ). Wethus havetheprojectionoperatorsn, n =0,1,. ..,2 N : n= 1 n ( Š 1)n Š ND2 N Š n 1... n D2 NDn 1... n,

PAGE 136

1263.REPRESENTATIONSOFSUPERSYMMETRY2 N n =0 n=1.(3. 11.19) Inparticular,0= Š ND2 N D2 Nand2 N= Š N D2 ND2 Nprojectoutthean tichiraland chiralpartso fresp ectively.Theprojectors(3.11.19) satisfyanumberofrelations: Orthon ormality mn= mnm(notsummed)(3.11.20) followsfrom D2 NDn D2 N=0 unless n =2 N andhencemn=0for m = n ;then m=1imp liesn=n m=2 n.Therearerel ationsbetweenthes:nis equaltothetransposeandtothecomplexconjugateof2 N Š nn=t 2 N Š n=1 (2 N Š n )! Š ND2 N Š n 1... 2 N Š n D2 NDn 1... 2 N Š n,(3. 11.21) n=* 2 N Š n=1 (2 N Š n )! ( Š 1)n Š N Dn €1... €2 N Š nD2 N D2 N Š n €1... €2 N Š n.(3. 11.22) Combining(3.11.21)and(3.11.22),we“ndanotherformofn: n= n= 1 n Š N Dn €1... €nD2 N D2 N Š n €1... €n.(3. 11.23) Thecomplexconjugationrelation(3.11.22)impliesthathalfofthesareredundantfor realsuper“elds: V = V 2 N Š nV =* n V = (nV ). Reduction ofthesintoirreducibleprojectionoperatorsisnoweasy: (1)Algebraicallyreduce D2 NDn under SU ( N ) Poincar e(wheremayhavefurther isospi norandWeylspinorindices); (2)Whent hereducedchiral“eldstrength D2 NDnisinareal representationof SU ( N ) andhas s +1 2 N integral,furtherreducebybisection,i.e.multiplicationby1 2 (1 K K ). Toperform( 1)itisconvenientto“rstreduce D2 NDnbyusingt heto talantisymmetryof the D s(sees ec.3.4),andthenreducethetensorproductoftheirreducible

PAGE 137

3.11.Projectionoperators127representationsof D2 NDnwiththerepresentationofthesuper“eldasusual.Ifwe onlywanttopreserve SO ( N ),furtherre ductionisperformedi nstep1;f orstep2, D2 NDnisalway sinareal representationof SO ( N ). Althoughncontainstheproductof2 ND sand2 N D sandisthu sinitssimplestform,n ,obtai nedbydir ectly introducing 1 2 (1 K K )infrontofthe D2 N,contains 2 ND sand4 N D sinthe K K term,andcanbefurthersimpli“ed.Aftersomealgebra we “nd: For n N : K K D2 NDn 1... nb1... bn=( Š 1)2 sn 1 2 ( n Š N ) D2 ND2 N Š n 1... nC11... Cnna1... an,(3. 11.24) or K K D2 NDn 1... 2 N Š nb1... b2 N Š n=( Š 1)2 sn 1 2 ( n Š N ) D2 NC11... C2 N Š n2 N Š nD2 N Š n 1... 2 N Š na1... a2 N Š n,(3. 11.25) where2 s extraWeylspinorindices,andextraisospinorindices,reducedasinstep1,are impliciton. For n N : K K D2 N D2 N Š n €1...b1...=( Š 1)2 sn 1 2 ( N Š n )D2 N Dn€1...C€1€1...a1...,(3. 11.26) or K K D2 N D2 N Š n €1...b1...=( Š 1)2 sn 1 2 ( N Š n )D2 NC€1€1... Dn €1...a1....(3. 11.27) Asanexampleofthi ssimp li“cation,weconsiderthe N =1chiral“e ldabove(3.11.7)for thesp ecialcasewhenitisa“eldstrengthofarealsuper“eld V := D2DV K K = Š D2i € D2 D€V = D2DV =.(3. 11.28) Wenowcoll ectourresults:Thesuperprojectorstakethe“nalform Ifbisectionispossible: n N :n i =1 n Š N Dn €1... €n1 2 (1 K K ) IPiD2 N D2 N Š n €1... €n,

PAGE 138

1283.REPRESENTATIONSOFSUPERSYMMETRYn N :n i =1 (2 N Š n )! Š ND2 N Š n 1... 2 N Š n1 2 (1 K K ) IPi D2 NDn 1... 2 N Š n,(3. 11.29) Ifbisectionisnotpossible: n i=eithero ftheabovewith1 2 (1 K K )dro pped,(3 .11.30) wherethe IPiare SU ( N ) Poincar eproj ectorsactingontheexplicitindices(including thoseofthesuper“el d).Wehavechosentheparticularformsofnfrom(3. 11.19,21-23) thatminimizethenumberofindicesthatthe IPiacton.Thechiralexpansion,besides itssimplicity,hastheadvantagethatthechiral“eldstrengthsappearexplicitly,andthe superspinandsuperisospinoftherepresentationontowhichprojectsarethoseofthe chiral“e ldstrength. b.Examples b.1.N=0 Webeginbyg ivingafewPoincar ep rojectionoperators.Theprocedurefor“ndingthemwas discussedinconsiderabledetailins ubsec.3. 11.a.1,soherewesimplylist results.Scalarsandspinorsareirreducible(nobisectionispossibleforaspinor: s +N 2 =1 2 isnotaninteg er).A(real)vectordecomposesintoaspin1andaspin0projection(see(3. 11.2,3)).Foraspinor-vector € =€ wehave: =1 3! ( )+1 3! ( ( ) C+ ( ) C)+1 2 C andhence € =(3 2 +T1 2 +L1 2 ) € , where 3 2 € = Š Š 1€w, w=1 6 ( €€ ); T1 2 € = Š Š 1€[ Cr+ Cr], r=1 6 ( € )€ ; L1 2 € = Š 1€s, s=1 2 a a .(3. 11.31)

PAGE 139

3.11.Projectionoperators129Forareal two-i ndextensor h a c=€€h wehave h =1 4! h( )+( C ( q ) + C ( q ) )+ CCq + C ( C ) r +1 4 ( Ch ( | | )+ Ch( | | ) ),(3. 11.32) where q= Š1 32 ( h( | | ) + h( | | ) + h( | | ) + h( | | ) ), r = Š1 24 h( ( ) ), q =1 4 h .(3. 11.33) Thereforethecompletedecompositionofthethetwo-indextensorisgivenby h€ €=(2, S+1, S+L 0, S+T 0, S+1, A ++1, A Š) h€ €, wheretheprojectorsarelabeledbythespin(2,1,0),thesymmetricandantisymmetric partof h a b( S and A ),longit udinalandtransverseparts( L and T ),andsel f-dualand anti-self-dualparts(+and Š ).Theexplicitformoftheprojectionoperatorsis 2, Sh€ €= Š 2€€w, w=1 4! ( €€h€ )€; 1, Sh€ €= Š 2€€[ C ( w ) + C ( w ) ], w= Š1 32 (€[ € )h( € )€+ € )h( € )€]; L 0, Sh€ €= Š 2€€S S =1 4 €€h€ €; T 0, Sh€ €= Š 2( CC€€ + €€) T T = Š1 12 ( € )€h€ €; 1, A +h€ €= Š 1€€l+ ( ), l+ ( )= Š1 4 h( € )€; 1, A Šh€ €= Š 2€€lŠ ( ), lŠ ( )= Š1 4 ( € )€h€ €.(3. 11.34) (The “eldstrengths wand T areproportionaltothelinearizedWeyltensorand scalarcurvaturesrespectively.)Fromthisdecomposition,weseethatthetwo-index

PAGE 140

1303.REPRESENTATIONSOFSUPERSYMMETRYtensor“eldconsistsofirreduciblespins2+1+1+1+0+0. b.2.N=1 Weconstr ucttheirreducibleprojectionoperatorsforacomplexscalarsuper“eld .From(3.9 .26-32)wehave,forthecaseswithoutbisection( s +1 2 N = s +1 2 ishalfintegral,sothat s isintegral) 0= Š 1D2 D2,1= Š Š 1D D2D,2= Š 1 D2D2.(3. 11.35) Sincehasnoexternalindiceswecangodirectlytostep2.Thechiral“eldstrengths D2and D2D2donotsati sfytheconditionthat s +1 2 isintegral,whereas D2D does.For N =1,theco ndition ofbeinginarealisospinrepresentationistriviallysatis“ed,andthatmeansthat1needstobebisected: 1=1++1 Š, 1 = Š Š 1D1 2 (1 K K ) D2D.(3. 11.36) Therefore,from(3.11.26-7), 1 = Š Š 1D D2D1 2 ( ),(3. 11.37) andthus0,1 and2completelyreduce.Theseirreduciblerepresentationsturn outtodescribetwoscalarandtwovectormultiplets,respectively. Wegivenextth ed ecomposit ionofthespinorsuper“eld.To “ndtheirreduciblepartsof1wePoincar ere ducethechiral“eldstrength D2D=1 2 [ C D2D+ D2D( )].Thisgivestheprojectionsn sforsuperspin s ofthischiral“eldstrength: 1,0=1 2 Š 1D D2D,1,1= Š1 2 Š 1D D2D( ).(3. 11.38) Thelatterirreduciblerepresentationisaconformalsubmultipletofthe(3 2 ,1)multiplet(seesection4.5).For0and2wemustbi sect: 01 2 = Š 11 2 (1 K K ) D2 D2= Š 1D21 2 ( D2 i € €),

PAGE 141

3.11.Projectionoperators13121 2 = Š 11 2 (1 K K ) D2D2= Š 1 D21 2 ( D2 i € €).(3. 11.39) Equivalentformsare: 01 2 = Š Š 1D1 2 ( D D2 D€D2 €), 21 2 = Š 1 D2D1 2 ( D D€ €).(3. 11.40) Finallywedecomposetherealvectorsuper“eld H€.B ecauseofitsrealitybisectionisunnecessary.Poincar eproj ectionisperformedbywriting H€=€Hand (anti)symmetrizingintheindicesofthechira l“eldstr engths.Toensurethattheprojectionoperatorsmaintaintherealityof H€,wecombi nethe2swiththe0s,since from(3.9 .24)2H€=(0H€).Weobtain T 0,1H€=1 2 Š 1€{ D2, D2} H( ), L 0,0H€=1 2 Š 1€{ D2, D2} H T 1,3 2 H€= Š1 6 Š 1€D D2D( H ), T 1,1 2 H€=1 6 Š 1€( D D2DH( )+ D D2DH( )), L 1,1 2 H€= Š1 2 Š 1€D D2DH ,(3. 11.41) where T and L denotetransverseandlongitudinal.Reexpressing Hintermsof H€, we “nd T 0,1H€=1 2 Š 2€{ D2, D2} ( €H )€, L 0,0H€=1 2 Š 2€{ D2, D2} cH c, T 1,3 2 H€=1 6 Š 2€D D2D( €H )€,

PAGE 142

1323.REPRESENTATIONSOFSUPERSYMMETRYT 1,1 2 H€=1 6 Š 2€( D D2D( €H )€+ D D2D( €H )€), L 1,1 2 H€= Š1 2 Š 2€D D2D dH d.(3. 11.42) b.3.N=2 Webeginbygiv ingtheex pressi onsfor SL (4, C ) C sintermsofthoseof SU (2)and SL (2, C ): C = CabCcdCCŠ CadCcbCC, C € € € €= CabCcdC€€C€€Š CadCcbC€€C€€, C = CabCcdCCŠ CadCcbCC, C € € € €= CabCcdC€€C€€Š CadCcbC€€C€€.(3. 11.43) Wede “nethe SU (2) Poincar ere ducti onof D2 asfollows: D2 = CD2 ab+ CbaD2 D2 € €= C € € D2 ab+ Cba D2€€, D2 = CCacCdbD2 cd+ CbaD2 D2 ab=1 2 Da Db = D2 ba=( D2 ab), D2 =1 2 CbaDa Db = D2 = Š ( D2€€).(3. 11.44) Thesetof(possibly)reducibleprojectionoperatorsis: 0,0= Š 2D4 D4,4,0= Š 2 D4D4, 3,1 2 = Š 2D D4D3 ,1,1 2 = Š Š 2D3 D4D 2,0= Š 2CcaCbdD2 ab D4D2 cd,

PAGE 143

3.11.Projectionoperators1332,1= Š 2D2 D4D2 .(3. 11.45) Inwriting2,0and2,1wehavetaken2de“nedby(3.11. 19)andused(3.11.44)tofurtherreduceit.Weca nnowd ecompose N =2super“elds. Westartwit hacomplex N =2scal arsuper“eld.Weneednotbisecttheterms obtainedfrom1,1 2 and3,1 2 .Bis ectingtherest,we“ndeightmoreirreducibleprojections. 0,0 = Š 21 2 (1 K K ) D4 D4= Š 2D41 2 ( D4 ), 4,0 = Š 21 2 (1 K K ) D4D4= Š 2 D41 2 ( D4 ), 2,0 = Š 2CcaD2 ab1 2 (1 K K ) D4CbdD2 cd= Š 2CcaCbdD2 ab D4D2 cd1 2 ( ), 2,1 = Š 2D2 1 2 (1 K K ) D4D2 = Š 2D2 D4D2 1 2 ( ).(3. 11.46) Wegivetwom oreresultswithoutdetails:Forthe N =2v ectormult iplet,describedby areals calar-isovectorsuper“eld Va bwe “nd 0,0,1 Va b= Š 2D4( D4 ) Va b, 1,1 2 ,3 2 Va b=1 3! Š 2CdbD3 c D4Ce ( aDc Vd ) e, 1,1 2 ,1 2 Va b=1 3 Š 2CdbD3 c D4De Cc ( aVd ) e, 2,1,1Va b= Š 2D2 D4D2 Va b, 2,0,2Va b= Š1 4! Š 2CbgCceCfdD2 ef D4D2 ( cdVa hCg ) h, 2,0,1Va b= Š1 4 Š 2CcdCbeD2 d ( a | D4( D2 | e ) fVc f+ D2 cfV| e ) f), 2,0,0Va b=1 3 Š 2CbcD2 ac D4CfeD2 deVf d,(3. 11.47) wheretheprojectionoperatorsarelabeledbyprojectornumber,superspin,superisospin, and K K conjugation .A gain,toconstructrealprojectionoperators,thecomplex

PAGE 144

1343.REPRESENTATIONSOFSUPERSYMMETRYconjugatemustbeaddedforthe0sand1s(0 0+4,1 1+3).Finally, forthespinor-isospinorsuper“elda (theunconstrainedprepotentialof N =2supergravit y)we “nd 0,1 2 ,1 2 a = Š 2D4 D4a 4,1 2 ,1 2 a = Š 2 D4D4a 2,1 2 ,3 2 a =1 3! Š 2D2 de D4Cb ( dCe | cD2 bc| a ) 2,1 2 ,1 2 a =2 3 Š 2CadCecD2 de D4D2 cbb 2,3 2 ,1 2 a =1 3! Š 2D2 D4D2 ( a ), 2,1 2 ,1 2 a =2 3 Š 2D2 D4D2 a 1,1,1 a =1 8 Š 2 Db€D4€( Š 1 D3 ( b (€€ )a ) Š+ i D( a (€ b )€ )), 1,1,0 a =1 8 Š 2 Da€D4€( Š 1 D3 b (€€ )b Š+ i Db (€ b€ )), 1,0,1 a = Š1 8 Š 2 Db€D4€( Š 1€ D3 ( b€a ) Š+ i D( a€ b )€), 1,0,0 a = Š1 8 Š 2 Da€D4€( Š 1€ D3 b€b Š+ i Db€ b€), 3,1,1 a =1 4 Š 2Db D4D2 Cc ( a( Dc b ) Db )€ c€), 3,1,0 a =1 4 Š 2CabDb D4D2 ( De e De€ e€), 3,0,1 a =1 4 Š 2CbdDb D4CacD2 dc( De e De€ e€), 3,0,0 a =1 12 Š 2CabDb D4CcdD2 ce( Dd e De€ d€). (3.11.48a)

PAGE 145

3.11.Projectionoperators135Thereare22irreduciblerepresentations.Onesimpli“cationispossible:Using(3.9.21) insteadof(3.9.25)forjustthe“rsttermin1 K K ,we “nd 1,1,1 a =1 8 Š 2( Š D3 b D4D( b ( a ) )Š+ i € Db€D4 D( a (€ b )€ )), 1,1,0 a =1 8 Š 2( Š D3 a D4De ( e )Š+ i € Da€D4 De (€ e€ )), 1,0,1 a = Š1 8 Š 2( D3 b D4D( b a ) Š+ i € Db€D4 D( a€b )€), 1,0,0 a = Š1 8 Š 2( D3 a D4De e Š+ i € Da€D4 De€e€).(3. 11.48b) b.4.N=4 Webeginbyde “ningasetofirreducible D -operators: D2 = CD2 ab+ D2 [ ab ] D3 = Cd cbaD3 d +( CD3 [ ac ] b Š CD3 [ ab ] c ) D4 = Cd cbaD4 +1 2 ( CCCcdefD4 [ ab ] [ ef ]Š CCCbcefD4 [ ad ] [ ef ]) +( CCeacdD4 b e Š CCeabdD4 c e + CCeabcD4 d e ) D5 = Cd cbaD5 d +( CD5[ ac ] b Š CD5[ ab ] c ) D6 = CD6 ab+ D6[ ab ] .(3. 11.49) Theysatisfythefollowingalgebraicrelations D2 ab= D2 ba, D6 ab= D6 ba, D4 a a = D4 [ ab ] [ cb ]= Ca bcdD3 [ ab ] c = Ca bcdD5[ ab ] c =0.(3. 11.50) All SL (2, C )i ndicesonthe Dnsaretotallysymmetric.Wealsohave D4 1... 4=1 4! C 1... 8D4 5... 8,

PAGE 146

1363.REPRESENTATIONSOFSUPERSYMMETRYD4 1... 4=1 4! C 1... 8D4 5... 8,(3. 11.51) andthese imply D4 = Cd cbaD4 +1 2 ( CCCcdefD4 [ ef ] [ ab ]Š CCCbcefD4 [ ef ] [ ad ]) +( CCeacdD4 e b Š CCeabdD4 e c + CCeabcD4 e d ),(3. 11.52) ascanbeveri“edbysubstitutingexplicitvaluesfortheindices. Weconsid ernowacomplexscalar N =4super“eld and “nd“rst 0,0,1= Š 4D8 D8,8,0,1= Š 4 D8D8, 1,1 2 ,4= Š 4 D€ D8 D7 €,7,1 2 4= Š 4D D8D7 2,0,10= Š 4 D2 abD8 D6 ab, 2,1,6=1 2 Š 4 D2[ ab ]€€D8 D6 [ ab ]€€, 3,3 2 4= Š 4 D3 a€€€D8 D5 a€€€, 3,1 2 ,20=1 3! Š 4 D3[ ab ] c€D8 D5 [ ab ] c€, 4,2,1= Š 4 D4€€€€D8 D4€€€€, 4,1,15= Š 4 D4 a b€€D8 D4 b a€€, 4,0,20= Š 4 D4 [ cd ] [ ab ]D8 D4[ cd ] [ ab ], 5,3 2 ,4= Š 4D3 a D8D5 a 5,1 2 20=1 3! Š 4D3 [ ab ] c D8D5[ ab ] c 6,0, 10= Š 4D2 ab D8D6 ab,

PAGE 147

3.11.Projectionoperators1376,1,6=1 2 Š 4D2 [ ab ] D8D6[ ab ] ,(3. 11.53) wherethesuperisospinquantumnumberherereferstothedimensionalityofthe SU (4) representation.Theonlyprojectorsthatneedbisectionaretherealrepresentationsof SU (4 ): th e1 ,6 15, an d 20 .W e “nd: 0,0,1 = Š 4D81 2 ( D8 2 ), 8,0,1 = Š 4 D81 2 ( D8 2 ), 4,2,1 = Š 4 D4€€€€D8 D4€€€€1 2 ( ), 2,1,6 =1 2 Š 4 D2[ ab ]€€D81 2 ( D6 [ ab ]€€ 1 2 Ca bcd D2[ cd ]€€ ), 6,1,6 =1 2 Š 4D2 [ ab ] D81 2 ( D6[ ab ] 1 2 Ca bcd D2 [ cd ] ), 4,1,15 = Š 4 D4 a b€€D8 b a€€1 2 ( ), 4,0,20= Š 4 D4 [ cd ] [ ab ]D8 D4 [ ab ] [ cd ]1 2 ( ),(3. 11.54) andatotalof22irreduciblerepresentations.The6isarealrepresentationonlyifwe useadualitytransformationinth erest-fram econju gation(3.11.4): X[ ab ]=1 2 Ca bcd X[ cd ].Thiso ccursforrank1 2 N antisymmetrictensorsof SU ( N )when N isamulti pleof4.

PAGE 148

1383.REPRESENTATIONSOFSUPERSYMMETRY3.12.On-shellrepresentationsandsuper“elds Insection3.9wediscussedirreduciblerepresentationsofo-shellsupersymmetry intermsofsuper“elds;herewegivethecorresp ondinganalysisofon-shellrepresentations.We“r stdiscussthedescriptionofon-shellphysicalcomponentsbymeansof“eld strengths.Wethendescribea(non-Lorent z-covariant)subgroupofsupersymmetry, whichwecall on-shell supersymmetry,underwhich(reducibleorirreducible)o-shell representationsofordinary(oro-shell)su persy mmetrydecomposeintomultipletsthat containonlyoneofthethreetypesofcomponentsdiscussedinsec.3.9.Byconsidering representationsofthissmallergroupintermsof on-shell super“elds(de“nedinasuperspacewhichisanon-Lorentz-covariantsubspaceoftheoriginalsuperspace),wecanconcentrateonjustthephysicalcomponents,andthusonthephysicalcontentofthetheory. a.Fieldstrengths Forsimp licitywerestrictourselvestomassless“elds.(Massive“eldsmaybe treatedsimilarly.)Itismoreconvenienttod escribethephysicalcomponentsintermsof “eldstrengthsratherthangauge“elds:EveryirreduciblerepresentationoftheLorentz group,whenconsideredasa “eldstrength, satis“escertainuniquec onstraints(Bianchi identities)plus“ eldequations,andcorrespondstoauniquenontrivialirreduciblerepresentationoft hePoincar egroup(a zeromasssinglehe licitystate).Ontheotherhand,a givenirreduciblerepresentationoftheLorentzgroup,whenconsideredasagauge“eld, maycorrespondtoseveralrepresentationsofthePoincar egro up,dependingontheform ofitsgaugetransformation. Speci“cally,any“eldstrength 1... 2 A€1...€2 B, totallysymmetric inits2 A undotted i ndicesandinits2 B dottedindices,hasmassdimension A + B +1ands atis“est heconstraintsplus“eldequations 1€1... 2 A€1...€2 B= €11... 2 A€1...€2 B=0,(3. 12.1a) 1... 2 A€1...€2 B=0.(3. 12.1b) TheKlein-Gordonequation(3.12.1b)projectsontothemasszerorepresentation,while (3.12.1a)projectontothehelicity A Š B state.TheKlein-Gordonequationisaconsequen ceoftheothersexceptwhen A = B =0.Tosolvethesee quationswegoto

PAGE 149

3.12.On-shellrepresentationsandsuper“elds139momentumspace:Then(3.12.1b)sets p2to0(i.e., ( p2)),andwemaychoosethe Lorentzframe p+€+= p+€Š=0, pŠ€Š =0.Inthisfra me(3.12.1a)statesthatonlyonecomponentof isnonvan ishi ng: + ... +€+ ...€+.Since each+i ndexhasah e licity1 2 andeach €+hashe licity Š1 2 ,theto talhelicityof is A Š B ,andofitsc omplexco njugate B Š A .Intheca seswhere A = B wemaychoose real(sinceithasanequalnumber of dottedandundottedindices),sothatitdescribesasinglestateofhelicity0. Themostfamiliarexamplesof“eldstrengthshave B =0: A =0istheusual descriptionofascalar, A =1 2 aWeyls pinor, A =1describesa v ector(e.g .,the photon), A =3 2 thegravitino,andthecase A =2istheWeylt ensorofthegraviton.Sinceweare describingonlytheon-shellcomponents,wedonotsee“eldstrengthsthatvanishon shell:e.g.,ingravitytheRiccitensorvanishesbytheequationsofmotion,leavingthe Weyltensor astheonlynonvanishingpartoftheRi emanncurvaturetensor.(Thishappensbeca use,alth oughthesetheoriesareirreducibleonshell,theymaybereducibleo she ll;i.e.,the“eldequationsmayeliminatePoincar erepresentations note liminatedby (o-shell)constraints.)Themostfamiliarexampleof A B =0isthe“elds trengthof thesecond-rankantisymmetrictensorgauge“eld:( A B )=(1 2 ,1 2 )(sees ec.4.4.c).Some lessfamiliarexamplesarethespin-3 2 representationofspin1 2 ,( A B )=(1,1 2 ),thespin-2 representationofspin0,( A B )=(1,1 ),andthehigher-derivativerepresentationofspin 1,( A B )=(3 2 ,1 2 ).Generally,t heo-shelltheorycontainsmaximumspinindicatedby thei ndicesof : A + B Althoughtheanalogousanalysisforsupersymmetricmultipletsisnotyetcompletelyunderstood,theon-shellcontentofsuper“eldscanbeanalyzedbycomponent projection.Inparticular,acompletesuper“eldanalysishasbeenmadeofon-shellmultipletsthatcontainonlycomponent“eldstrengthsoftype( A ,0).This issucientto describeallon-shellmultiplets:Theorieswith“eldstrengths( A B )des cribethesame on-shellhelicitystatesastheorieswith( A Š B ,0),andareph ysicallyequivalent.They onlydierbytheirauxiliary“eldcontent.Furthermore,type( A ,0)theorie sa llowthe mostgeneralinteractions,whereastheorieswith B =0“eldsarege nerallymore restrictedintheformoftheirself-interactionsandinteractionswithexternal“elds.(In somecases,theycannotevencoupletogravity.)

PAGE 150

1403.REPRESENTATIONSOFSUPERSYMMETRYBeforedisc ussingthegeneralcase,weconsideraspeci“cexampleindetail.The mult ipletof N =2superg ravity(seesec.3.3.a.1)withhelicities2,3 2 ,1,isdes cribedby component “eldstrengths ( x ), a ( x ), ab ( x ).Theyha vedime nsion3,5 2 ,2 respectively,andsatisfythe component Bianchiidentitiesand“eldequations(3.12.1). Weintr o ducea super“eld strength F(0) ab ( x )thatcontain sthelowe stdimensioncomponent “eldstrengthatthe =0le vel: F(0) ab ( x ) | = CabF(0) ( x ) | = ab ( x ).(3. 12.2) Werequirethat all thehighercom ponentsof F(0)arecomponent“eld strengthsofthe theory(ortheirsp acetimederivatives;super“eldstrengthscontainnogaugecomponents and,onshell,noauxiliary“elds).Thus,forexample,wemusthave Da ( F(0) ab )| =0, whereas Cc ( dDc F(0) ab ) | = D €F(0) ab | =0.Sinceasup er“eldthatvanishesat =0 vanishesid entically(asfo llowsfromthesupersymmetrytransformations,e.g.,(3.6.5-6)) Cc ( dDc F(0) ab ) = D €F(0) ab =0.Fromthese argumentsitfollowsthatthesuper“eld equationsandBianchiidentitiesare: D €F(0) ab =0, D F(0) ab = c [ aF(1) b ] D D F(0) ab = c [ ad b ]F(2) D D D F(0) ab =0;(3. 12.3) where F(1)( x )and F(2)( x )aresupe r“eldscontainingthe“eldstrengths b ( x )and ( x )atthe =0leve l.Bya pplyingpowersof D and D€ totheseequationswe recoverthecomponent“eldequationsandBianchiidentities,andverifythat F(0) ab containsnoextracomponents. Genera lizationtotherestofthesupermultipletsinTable3.12.1isstraightforward: Weintro duceasetofsuper“eldswhichat =0arethecomp onent“eldstrengths(asin (3.12.1))thatdescribethestatesappearinginTable3.3.1:Thesesuper“eldssatisfyaset ofBianchiidentitiesplus“eldequations(asintheexample(3.12.3))thatareuniquely determinedbydimension alanalysisandLorentz SU ( N )covar ian ce.

PAGE 151

3.12.On-shellrepresentationsandsuper“elds141 he licityscalarmultipletsuper-Yang-Millssupergravity +2 F+ 3/2 Fa +1 FFab + 1/2 FFa Fabc 0 FaFabFa bcd-1/2 Fab€Fabc€Fa bcde€-1 Fa bcd€€Fa bcdef€€-3/2 Fa bcdefg€€€-2 Fa bcdefgh€€€€ Table3. 12.1.Fieldstrengthsintheoriesofphysicalinterest Wenowconsi derarbitrarysupermultipletsoftype( A ,0).Thereare twocases: Foranon-shel lmulti pletwithlowestspin s =0,thesuper“eldstr engthhastheform F(0) a1... am,N 2 m N ,andis totallyantisymmetric inits mSU ( N ) isospin i ndices.If thelowestspin s > 0,thesup er“eldstrengthhastheform F(0) 1... 2sandis totallysymmetric inits2 sWeylspinor i ndices.Totreatbothcasestogether,for s > 0wewrite F(0) a1... aN1... 2s= Ca1... aNF(0) 1... 2s.Thent hesuper“eldstren gthhastheform F(0) a1... am1... 2sandistotallyantisymmetricinitsisospinorindicesandtotallysymmetric initsspino ri ndices.Ithas(mass)dimension s +1. Thissuper“eldcontainsalltheon-shellcomponent“eldstrengths;inparticular,at =0,itcontainsthe “eldstrengthoflowestdimension(andthereforeoflowestspin). For s =0,thesuper“el dstrengthd escribeshe licitiesm Š N 2 ,m Š N +1 2 ... ,m 2 ,andits hermitianconjugated escribeshe licitiesŠ m 2 ,Š m +1 2 ... ,N Š m 2 .Since m N ,some he licitiesappearinboth F(0)and F(0).For s 0,thesuper “eldstrengthdescribeshelicities s s +1 2 ... s +N 2 ,anditshe rmitianconjugatedescribeshelicities Š ( s +N 2 ), ... Š s .Inthiscase,po sitivehe licitiesappearonlyin F(0)andnegative he licitiesonlyin F(0).Forbothcasest hesuper“eldstr engthtogetherwithitsconjugate describe(perhapsmu ltiple)helicities s ( s +1 2 ), ... ( s +m 2 ).

PAGE 152

1423.REPRESENTATIONSOFSUPERSYMMETRYThehigher-spincomponent“eldstrengthsoccurat =0inthesup er“elds F( n )obtainedby a pplying nD s(for n > 0)or Š n D s(for n < 0)to F(0).Theyare totally antisymmetricintheir m Š n isospinorindicesandtotallysymmetricintheir2 s + n spin i ndices,andsatisfythefollowingBianchiidentitiesand“eldequations: n > 0: Dn n... 1F(0) a1... am1... 2s=1 ( m Š n )! b1[ a1... bnanF( n ) an +1... am] 1... 2s 1... n,(3. 12.4a) n < 0: Dn€ Š n...€ 1F(0) a1... am= F( n ) a1... amb1... bŠ n€1...€Š n,(3. 12.4b) with m Š N n m ;inparticular,for s > 0, DF(0)=0.Theseeq uation sfo llowfrom therequirementthat all componentsoftheon-shell super“eld strength(de“nedbyprojection)areon-shell component “eldstrengths.The =0compon entofthesuper“eld F(0)isthe lowestdimension component“eldstrength;thisdeterminesthedimensionand indexstructureofth esuper “eld.Thehighercomponentsofthesuper“eldareeither higherdimensioncomponent“eldstrengths,orvanish;thisdeterminesthesuper“eld equationsandBianchiidentities.Notethatthedierencebetweenmaximumandminimumhe licitiesinthe F( n )isalways1 2 N Inthespecialcase s =0, m even,and m =1 2 N wehaveina ddition to(3.12.4a,b) theselfconjugacyrelation F(0) a1... a1 2 N= 1 (1 2 N )! Ca1... aN F(0) a1 2 N... aN.(3. 12.4c) Forthisc aseonly, F(+ n )isrelatedto F( Š n );thisrelatio nfo llowsfrom(3.12.4c)for n =0, andfromspinorderivativesof(3.12.4c),using(3.12.4a,b),for n > 0.Eqs.(3.12.4a,b) are U ( N )covariant,where as,becausetheantisymmetrictensor Ca1... aNisnot phase invariant,(3.12.4c)isonly SU ( N )covaria nt;thus, self-conjugatemultipletshavea smallers ymmetry. b.Light-coneformalism Whenstudyingonlytheon-shellpropertiesofafree,masslesstheoryitissimpler torepres entthe“eldsinaformwherejustthephysicalcomponentsappear.As describedinsec.3. 9,weusealight-coneformalism,inwhichanirreduciblerepresentationofthePoincar egroupisgivenbyasi nglecomponent(complexexceptforzero

PAGE 153

3.12.On-shellrepresentationsandsuper“elds143he licity).Forsuper“eldswemakealight-conedecompositionof aswellas x .Weuse thenotation(see(3.1.1)): ( x+€+, x+€Š, xŠ€+, xŠ€Š) ( x+, xT, xT, Š xŠ),( a+, aŠ) ( a, a),(3. 12.5a) ( +€+, +€Š, Š€+, Š€Š) ( +, T, T, Š Š),( a+, aŠ) ( a, a).(3. 12.5b) (The spinor deriva tive ashouldnotbeconfusedwiththe spacetime deriva tive a). U nderthetransverse SO (2)partoftheLorentzgroupthecoordinatestransformas x = x, xT = e2 i xT, a = ei a, a = eŠ i a,andtheco rrespondingderivativestransformintheopp ositeway. Insec.3.11wedescribedthedecompositionofgeneralsuper“eldsintermsofchiral “eldstrengths,whichareirre ducib legaugeinvariantrepresentationsofsupersymmetry oshell.Althoughtheycontainnogaugecomponents,theymaycontainauxiliary“elds thatonlydropoutonshell.Toanalyzethedecompositionofanirreducible o-shell representationofsupersymmetryintoirreducible on-shell representations,weperforma nonlocal,nonlinear,nonunitarysimilaritytr ansformationonthe“eldstrengthsandall operators X : = eiH, X= eiHXeŠ iH; H =( ai a) T+ .(3. 12.6a) Weuset histransformationbecauseitmakessomeofthesupersymmetrygenerators independent of ainthechiralrepresentatio n.Droppingprimes,wehave Qa+= i a, Qa€+= i ( aŠ ai +), QaŠ= i ( a+ a T+ ), Qa€Š= i ( aŠ ai T+ i a + ).(3. 12.6b) Thus Q+and Q€+arelocalanddependonlyon a, a,and +, but not on a, a, Š,and T,whereas QŠand Q€Šarenonlocalanddependonall and a.Weexpandthe transformedsup er“eldstrengthinpowersof a(theexterna li ndicesofaresuppressed): ( x€, a, a)=N m =0 1 m ma1... am( m ) am... a1( x€, a),(3. 12.7)

PAGE 154

1443.REPRESENTATIONSOFSUPERSYMMETRYwherethe mthpower m,and thus ( m ),istot allyantisymmetricinisospinorindices. Each ( m )isarepresentationofa subgroupofsupersymmetrythatwecallonshellsupersymmetry,andthatincludesthe Q+transformations,thetransverse SO (2) partofLorentztransformations(an daco rrespondingconformalboost), SU ( N )(or U ( N )),andallfourtranslations(aswellasscaletransformationsinthemasslesscase). Althougheach ( m )isarealizationofthefullsupersymmetrygroupo-shellaswellas on-shell,on-shellsupersymmetryisthemaximalsubgroupthatcanberealizedlocally (andintheinteractingcase,linearly). Theremaininggeneratorsofthefullsupersymmetrygroup(includingtheother Lorentzgenerators,thatmix awith a)mixthev arious ( m )s.Inparticular, QŠand Q€Šallowustodistinguishphysicalandauxiliaryon-shellsuper“elds:Auxiliary“elds vanishon-s hell,andhencemusthavetr ansformationsproportionalto“eldequations. WegotoaL orentzframewhere T=0.Inthi sfra me, QaŠ= i aand Q€Š= i ( a+ i a + ).The QŠand Q€Šsupersymmetryvariationofthehighest acomponentof, ( N )Š ( N ) i + ( N Š 1)(3.12.8a) isproportionalto ,w hi ch identi“esitasanauxiliary“eld.Setting ( N )to zeroonshell,weiteratetheargument:thevariationofthenextcomponentof, ( N Š 1)Š ( N Š 1) i + ( N Š 2)(3.12.8b) isagainproportionalto ,etc.W e “ndthat only (0)hasavariation not prop ortional to .Thisi denti“esitasthephysicalon-shellsuper“eld. Thus,on-shell,reducesto (0).(Allother ( m )vanish.)IntheL orentzframe chosen above( T=0), QŠand Q€Švanishwhen actingon (0),andth usthissuper“eldis alocalrepr esentati onofthe full supersymmetryalgebraonsh e ll,namely,itdescribes themultipletof physicalpolarizations.Byexpandingactionsin ,itcanbeshownthat (0)representsthemultipletofphysicalcomponentswhiletheother ( m )srepres entmultipletsofauxiliarycomponents.

PAGE 155

3.12.On-shellrepresentationsandsuper“elds145Wecanalsod e“ne(chiralrepresentation)spinorderivatives Da, Dathatare covariantundertheon-shellsupersymmetry: Da= a+ ai +, Da= a; { Da, Db} = a bi +.(3. 12.9) Whenabisectioncondi tionisimposedonthechiral“eldstrength(i.e.,isreal, asdiscussedinsec.3.11),wecanexpressthe conditionintermsoftheon-shellsuper“elds.Forsuperspin s =0,theco ndition D2 N = 1 2 N(3. 12.10) b ecomes DN ( m ) a1... am= iN m Š1 2 N( i +)N Š m1 ( N Š m )! CaN... a1( N Š m ) aN... am +1(3.12.11) (where DN1 N CaN... a1DN a1... aN)andsim ilarlyforsuperspin s > 0.Ingeneral,anonshellrepresentationcanbereducedbyarealityconditionoftheform DN ( i +)1 2 N whenthemiddle( 1 2 N)compo nentof hashelicity0(i.e.,isinvariantundertransverse SO (2)Lorentztransformations).(Comparethediscussionofrealityofo-shell representationsinsec.3.11.) Puttingtogethertheresultsofsec.3.11andthissection,wehavethefollowing reductions:generalsuper“elds(4 N s;physical+auxiliary+gauge) chiral “eld strengths(2 N s;physical+auxiliary=irreducibleo-shellrepresentations) chiral on-shellsu per“elds( N s;physical=irreducibleon-shellrepresentations).Allthree typesofsuper “eldscansatisfyrealityconditions;th erefore,thesmallesttypeofeachhas 24 N,22 N,and2Ncomponents,respectively(whentherealityconditionisallowed),andis area lscalarsuper“eld.Allotherrepresentationsare(realorcomplex)super“elds with(Lorentzorinternal)indices,andthushaveanintegralmultipleofthisnumberof components.Thesecountingargumentsforo-shellandon-shellcomponentscanalso beobtainedby theusualoperatorarguments(o-shell,thecountingisthesameasfor on-shellmassivetheories,since p2 =0), butsuper“eldsallowanexplicitconstruction, andarethusmoreusef ulforapplications. Similarargumentsapplytohigherdimensions:Wecanusethesamenumbers th ere(buttakingintoaccountthedierenceinexternalindices),ifweunderstand4 N  tomeanthenumberofanticommu tingcoor dinatesinthehigherdimensionalsuperspace.

PAGE 156

1463.REPRESENTATIONSOFSUPERSYMMETRYForsimpl esupersy mmetryin D < 4,becausec hiralitycannotbede“ned,thecountingof statesisdierent.

PAGE 157

3.13.O-shell“eldstrengthsandprepotentials1473.13.O-shell“eldstrengthsandprepotentials Wehaveshownh owsuper“elds canbereducedtoirreducibleo-shellrepresentations(sec.3.11),whichcanb eredu cedfurthertoon-shellsuper“eldstrengths(sec. 3.12).To“ndasuper “elddescriptionofagivenmultipletofphysicalstates,weneedto reversetheprocedure:Startingwithan onshellsuper“eldstrength F(0)thatdescribes themultiplet,weneedto“ndthe oshellsuper“eldstrength W thatreducesto F(0)on shell,andthen“ndasuper“eld prepotential inte rmsofwhich W canbeexpressed. Thereisnounambiguouswaytodothis:Thesame F(0)isdescribedbydierent W s, andthesame W isdescribedbydierents.How ever,foraclassoftheoriesthat in cludesmanyofthemodelsthatareunderstood,weimposeadditionalrequirementsto reducetheambiguityand“ndauniquechiral“eldstrengthandafamilyofprepotentials foragivenmultiplet. Themultipletsweconsiderhaveon-shellsuper“eldstrengthsofLorentzrepresentationtype( A ,0)(superspin s = A )andare isoscalars:F(0) 1... 2 s.From( 3.12.4b),this impliesthatthe F(0)sare chiral andthereforecanbegeneralizedtoo-shellirreducible (uptobisection)“eldstrengths W1... 2 s, D€ W1... 2 s=0.Physi ca lly,the W scorrespond to“eldstrengthsofconformallyinvariantm odels.(Theyt ransforminthesamewayas Ca1... aN:as SU ( N )scalars butnot U ( N )scalars). Inthephysicalmodelswherethese super“eldsarise,thechiralityandbisectionconditionson W arelinearizedBianchi identities.Wecanuset heprojectionoperatoranalysisofsec.3.11tosolvetheidentities byexpr essingthe W sintermsofappropriateprepotentials. Whenthereisnobisection( s +1 2 N notaninteger),the W saregeneralchiral super“elds: W1... 2 s= D2 N1... 2 s.The1... 2 ssmaybeexpressedintermsofmorefundamentalsuper“elds.Aninterestingclassofprepotentialsarethosethatcontainthe lowest superspins:Inthatcase,the W shavetheform N 2 s Š 1: W1... 2 s=1 (2 s )! D2 NDN ( 1... NN +1€1... N + M€MN + M +1... 2 s)€1...€M, N 2 s Š 1: W1... 2 s=1 ( Š M )!(2 s )! D2 NDN [ a1... aŠ M] [ b1... bŠ M] ( 1... 2 s Š 12 s) b1... bŠ Ma1... aŠ M(3.13.1) whereisanarbitrary(complex)super“eldand M = s Š1 2 ( N +1).

PAGE 158

1483.REPRESENTATIONSOFSUPERSYMMETRYIf W isbis ected( s +1 2 N integer,(1 Š K K ) W =0), thenmust beexpr essedin termsofa real prepotential V thathasmaxim umsuperspin s W hasaformsimilarto (3.13.1): N 2 s : W1... 2 s=1 (2 s )! D2 NDN ( 1... NN +1€1... N + M€MVN + M +1... 2 s)€1...€M, N 2 s : W1... 2 s=1 ( Š M )! D2 NDN [ a1... aŠ M] [ b1... bŠ M] 1... 2 sVb1... bŠ Ma1... aŠ M(3.13.2) where M = s Š1 2 N Whetherornot W isbis ected,ambiguityremainsintheprepotentials, V ,since th eymaystillbeexpressedasderivativesofmorefundamentalsuper“elds:Thisleadsto varianto-shellm ultiplets(seesec.4.5.c).O urexpression (3.13.1)forintermsof isanexampleofsuchana mbiguity:T hereisnoapriorireasonwhymustta kethe specialform, unless itisobtainedasasubmulti pletofabisectedhigherN mult iplet(as, forexample,inthecaseofthe N =1spin3 2 ,1mult iplet(sec.4. 5.e),which isas ubmultipletofthe N =2superg ravitymultiplet).Modulosuchambiguities,theexpressions for W intermsofand V arethemostgenerallocalsolutionstotheBianchiidentities constraining W (i.e.,chirality,andifpossible,bisection).

PAGE 159

Contentsof 4.CLA SSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS 4.1.Thescalarmultiplet149 a.Renormalizablemodels149 a.1.Actions149 a.2.Auxiliary“elds151 a.3.R-invariance153 a.4.Super“eldequations153 b.Nonlinear -models154 4.2.Yang-Millsgaugetheories159 a.Prepotentials159 a.1.Linearcase159 a.2.Nonlinearcase162 a.3.Covariantderivatives165 a.4.Fieldstrengths167 a.5.Covariantvariations168 b.Covarian ta pproac h 170 b.1.Conven tionalconstraints171 b.2.Representation-pre servingconstraints172 b.3.Gaugechiralr epresentation174 c.Bianchiidentities174 4.3.Gauge-invariantmodels178 a.Renormalizablemodels178 b.CP(n)m odel s 179 4.4.Superforms181 a.General181 b.Vectormu ltiplet185 c.Tensormu ltiplet186 c.1.Geometricformulation186 c.2.Dualitytransformati ontochiralmultiplet190 d.Gauge3-formmultiplets193 d.1.Re al3-form193 d.2.Comple x3-for m 195 e.4-formmu ltiplet197 4.5.Othergaugemultiplets198 a.GaugeWess-Zu minomodel198

PAGE 160

b.Thenonminimalscalarmultiplet199 c.Morevariantmultiplets201 c.1.Vectormultiplet201 c.2.Tensormultiplet203 d.Super“eldLagran gemultipliers203 e.Thegravitinoma ttermultiplet206 e.1.O-shell“eldstrengthandprepotential206 e.2.Compensators208 e.3.Duality211 e.4.Geometricformulations212 4.6.N-extendedm ultiplets216 a.N=2multiplets216 a.1.Vectormultiplet216 a.2.Hypermultiplet218 a.2.i.Freetheory218 a.2.ii.Interactions219 a.3.Tensormultiplet223 a.4.Duality224 a.5.N=2super“eldLagrangemultiplier227 b. N=4Yang-Mills228 b.1.Minimalf ormulation228 b.2.L agrangemultiplierformulation229

PAGE 161

4.CLA SSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS Inthischapterwediscussinteracting“eldtheoriesthatcanbebuiltoutofthe super“eldsofglobal N =1Poincar esupersy mmetry.Thisrestrictsustotheories describingparticleswithspinsnohigherthan1.Thesimplestdescriptionofsuchtheoriesisintermsofchiralscalarsuper“eldsforparticlesofthescalarmultiplet(spins0 and1 2 ),andrealscalargaugesuper“eldsforp articlesofthevectormultiplet(spins1 2 and1).However,otherdescriptionsarepossible;wetreatsomeoftheseinageneral frameworkprovidedbysuperforms.Wedescribe N =1theori esandalsoextended N 4theoriesi nte rmsof N =1 super“elds.Ourprimarygoalistoexplainthestructureofthesetheoriesinsup erspace.Wedonotdiscussphenomenologicalmodels. 4.1.Thescalarmultiplet a.Renormalizablemodels Thelowestsuperspinrepresentationofthe N =1supersy mmetryalgebraiscarriedbyachiralscalarsuper“eld.Insec.3.6wedescribeditscomponentsandtransformations.Inthechiralrepresentationwehave(+)= A + Š 2F ,with complex scalarcomponent“elds A =2Š1 2 (A+iB), F =2Š1 2 (F+iG),andthetransformationsof (3.6.6). a.1.Ac tions To “ndsuperspaceactionsforthechiralsup er“eldweusedimensionalanalysis: Thesuper“eldcontainstwocomplexscalarsdieringbyoneunitofdimension(recall that hasdimension Š1 2 );however,itcontainsonlyon espi nor,andwerequirethis spinortohavetheusualphysicaldimension3 2 .Therefo re,weshouldassignthesuper“elddimension1.Thisleadsustoauniquechoiceforafree(quadratic)masslessaction withnodimensionalparameters: Skin= d4xd4 (4. 1.1) (seesec.3.7.afor ades criptionoftheBerezinintegral).Uptoanirrelevantphasethere

PAGE 162

1504.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSisa uniquemasstermandauniqueinteractiontermwithdimensionlesscouplingconstant: S( m )= d4xd2 (1 2 m 2+ 3! 3)+ d4xd2 (1 2 m 2+ 3! 3).(4.1 .2) TheresultingactiondescribestheWess-Zuminomodel. Alloftheintegralsare i ndependentoftherepresentation(vector,chiralorantichiral)inwhichthe“eldsaregiven;theintegrandsindierentrepresentationsdierby total x -derivat ives(fromthe eUfactors,see( 3.3.26)),thatvanishupon x -integration. Wecanexpressthea ctioninitscomponentformeitherbystraightforward -expansionandintegration,orby D -projection.Intheformerapproach,wewritefor example,intheantichiralrepresentationfor = ( Š ),and( Š )= eU(+): Skin= d4xd4 ( Š )eU(+)= d4xd4 [ A + € €Š 2 F ] e €i €[ A + Š 2F ],(4.1 .3a) andaftersomealgebraobtain Skin= d4x [ A A + €i €+ FF ].(4.1 .3b) Itissimplertousetheprojectiontechnique;wewrited4xd4 = d4x D2D2and Skin= d4xd4 = d4x D2[ D2] | = d4x [ D2D2+( D2 )( D2)+( D€ )( D€D2)] | .(4. 1.4) Usingtheidentities DD2= Di and D2D2= ,whichfollowfromthechiralityof ,andthede“nitionofthecomponents(3.6.7),weobtain(4.1.3b). Toevaluatech iralintegralsbyprojectionwewrite,foranyfunction f () d4xd2 f ()= d4xD2f ()

PAGE 163

4.1.Thescalarmultiplet151= d4x [ f()( D )2+ f() D2] | = d4x [ f( A ) 2+ f( A ) F ].(4.1 .5) Inparticularweobtainforthemassandinteractionterms S( m )= d4x { m [ 2+ AF ]+ [ A 2+1 2 FA2]+ h c } .(4. 1.6) (Witho utlossofgenerality,wecanchoose m and real.) Wecoulda ddalinearterm anditshe rmitianconjugatetotheaction(4.1.2). Suchatermwouldaddtothecomponentactionalinear F + F term.However,in theWess-Zuminomodelsuchatermcanalwaysbeeliminatedfromtheactionbyaconstantshift + c .Lineart ermsdohoweverplayanimportantroleinconstructing modelswithspontaneoussupersymmetrybreaking(seesec.8.3). a.2.Auxiliary“elds Thecomponent“eld F doesnotdescribeanindependentdegreeoffreedom;its equationofmotionisalgebraic: F = Š m A Š1 2 A2.(4. 1.7) Ifweeliminatethe auxiliary “eld F fromtheactionandthetransformationlaws,we“nd S = d4x [ A ( Š m2) A + €i €+ m ( 2+ 2) Š1 2 m ( A A2+ AA2) Š1 4 2A2 A2+ ( A 2+ A 2)],(4.1.8) and A = Š = Š €i €A Š ( m A +1 2 A2).(4.1 .9) ThereforetheWess-Zuminoactiongivesequalmassestothescalarsandthespinor, cubicandquarticself-interactionsforthescalars,andYukawacouplingsbetweenthe

PAGE 164

1524.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSscalarsandthespinor,allgovernedbyacommoncouplingconstant. Aftereliminating F ,thesupersy mmetrytransformationso fthespinor arenonlinear;thismakesananalysisofthesupersymmetryWardidentities without theauxiliary “e ldsdicult.Thisisnottheonlyproblemcausedbyeliminatingauxiliarycomponent “elds:Thetransformationsarenotonlynonlinear,butalsodependentonparametersin theLagrangian,anditisdiculttodiscoverfurthersupersymmetrictermsthatcouldbe addedtothecomponentLagrangian(e.g.,gaugecouplings).Furthermore,equation (4.1.7)isnotitselfsupersymmetric unless theequationofmotionofthespinorissatis“ed;onlythenis F = Š €i €(4.1.10a) thesameas F ( A )= ( Š m A Š1 2 A2)=( m + A ) € €.(4. 1.10b) Forthisreaso n,formulationsofsupersymmetricth eoriesthatlackthecomponentauxiliary“eldsareoftencalled on-s hell supersymmetric.Indeed ,ifwecalcula tethecommuta toroftwosupersymmetrytransformation sactingonthe spinor,we“ndthatthe “elds A ,formarepres entationofthealgebra(i.e.,thealgebracloses)onlyifthe spinorequationofmotionissatis“ed. TheWess-Zuminomodelcanbegeneralizedtoincludeseveralchiralsuper“elds. Themostgeneralactionthatleadstoaconventionalrenormalizabletheoryis S = d4xd4 ii+ d4xd2 P(i)+ h c .,(4.1 .11) wherePisapolynomialofmaximumdegree3inthe“elds.Thecomponentactionhas theform S = d4x [ Ai Ai+ i€i € i+ FiFi] + d4x [Pi( A ) Fi+Pij( A )1 2 i j+ h c .],(4. 1.12) where

PAGE 165

4.1.Thescalarmultiplet153Pi= P Ai ,Pij= 2P Ai Aj .(4. 1.13) Inparticular,eliminationoftheauxiliary“eldsgivesthescalarinteractionterms(the scalarpotential U ): Š U ( Ai)= Ši | Pi|2.(4. 1.14) Asaconsequenceofsupersymmetry(see(3.2. 10))thepotentialispositivesemide“nite. Theaction(4.1.11)canalsobeinvariantunderaglobalinternalsymmetrygroupcarried bythes. a.3.R-invariance AnadditionaltoolusedtostudythesemodelsisR-symmetry(3.6.14).Thisisthe chiralsy mmetrygeneratedbyrotating and byopposi te phases(sothatd4 is invari ant butd2 isnot)andbyrotatingdierentchi ralsuper“eldsbyrelatedphases: ( x , ) eŠ iwr( x eir eŠ ir ).(4.1 .15) Itmaybe,butisnotalways,possibletoassignappropriateweights w tothevarious super“eldstomakethetotalactionR-invariant.Forexample,withonlyonechiralmultiplet,R-invaria nceholdsifeitheramassoradimensionlessself-couplingispresent,but notboth:Theappropriatetra nsformationsw eightsare w =1and w =2 3 respectively. Withmorethanonechiralmultiplet,itisp ossibletowriteR-symmetricLagrangians havingbothmassandinteracti onterms:Achiralself-interac tiontermisR-invariantif itstotalR-weight w =2( i.e.,thesumoftheR-weightsofeachsuper“eldfactoris2). a.4.Super“eldequations Fromtheactionf orachirals uper“eld,weobtaintheequationsofmotionbyfunctionaldierentiation(see(3.8.10,11)).Forexample,includingsources,wehave S = d4xd4 + { d4xd2 [P()+ J ]+ h c } ,(4. 1.16) fromwhich,using( 3.8.9-12),wederivetheequations D2 +P()+ J =0 ,( 4.1.17a)

PAGE 166

1544.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSD2+ P( )+ J =0.(4. 1.17b) Weconsid er“rstthemassivenoninteractingcaseP()=1 2 m 2.M ul tiplying(4.1.17a) by D2,we “nd D2 D2 + mD2+ D2J =0.(4. 1.18) Substituting(4.1.17b)into(4.1.18)andusingthechiralityof( D2 D2 = ),we obtain ( Š m2) = m J Š D2J .(4. 1.19) Similarly,we“nd ( Š m2)= mJ Š D2 J .(4. 1.20) andtheseequationscanbereadilysolved. Forarbit raryP(),wederivetheequationsofmotionforthecomponent“eldsby projectionfromthesuper“eldequations.Successivelyapplying D sto(4.1.17a)we“nd F +P( A )+ JA=0 i € €+P( A ) + J =0 A +P( A ) 2+PF + JF=0(4.1 .21) aswouldbeobtainedfromthecomponentLagrangian. b.Nonlinear -models Ifrenormalizabilityisnotanissue,wecanconstructgeneralsupersymmetric actionsbytakingarbitraryfunctionsof, ,andtheirderivatives,andintegratingover superspace.Aninterestingclassofsupersy mmetricmodelsthatcanbeconstructedout ofchiralsuper“eldsisthegeneralizednonlinear -model.Inordinaryspacetime,ageneralizednonlinear -modelisdescribedby“elds ithatarethecoordinatesofanarbitrarymanifold.Theactionofsuchamodelis S= Š1 4 d4xgij( ai)( aj),(4.1 .22)

PAGE 167

4.1.Thescalarmultiplet155where gij( i)isthemetric tensor de“nedonthemanifold.Thesupersymmetricgeneralizationofthesemodelsisdesc ribedbychiralsuper“eldsiandtheirco njugates iwhicharethecomplexcoordinatesofanarbitraryK¨ ahlermanifold(seebelow).(Weuse agrouptheo reticconvention:Upperandlowerindicesarerelatedbycomplexconjugation,andallfactorsofthemetricarekeptexplicit.)Theactiondependsonasinglereal function IK (, )de“neduptoarbitraryadditivechiralandantichiraltermsthatdo notcontribute: S= d4xd4 IK (i, j).(4.1 .23) Thecomponentcontentofthisactioncanbeworkedoutstraightforwardlyusingthe projectiontechnique;we“nd S= Š1 2 d4x 2IK Ai Aj ( aAi)( a Aj)+ ... .(4. 1.24) Thishastheform(4.1.22)ifweidentify2IK Ai Aj asthemetric gij.Acomplexm anifold whosemetriccanbewritten(locally)intermsofapotential IK iscalledK¨ ahler;thusall four-dimension alsupersymmetricnonlinear -modelsarede“nedonK¨ ahlermanifolds. Conversely,anybosonicnonlinear -modelwhose“eldsresideonaK¨ ahlermanifoldcan beextendedtoasupers ymmetricmodel.Theremainingtermsin(4.1.24)providecouplingsbetweenthescalar“ eldsandthespinor“elds. K¨ ahlergeometryisaninterestingbranchofcomplexmanifoldtheorythatmathematicianshaveinvestigatedextensively.Herewediscussonlythoseaspectsrelevantto subsequenttopics(e.g.,sec.8.3.b).Wede“ne IKj1... jni1... im= i1 ... im j1 ... jn IK .( 4.1.25a) Inparticular,themetricis IKi j= 2IK i j .(4. 1.25b) Equivalently,wecanwritethelineelementas ds2= IKi jd id j.(4. 1.25c)

PAGE 168

1564.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSThemetric,liketheaction(4.1.23),isinvariantunderK¨ ahlergaugetransformations IK IK +()+ ( )(4.1.26) oftheK¨ ahlerpotential IK .Fie ldrede “nitions= f ()d e“ne holomorphic coordinate tr an sformationsonthemanifold;underthese,theformofthemetric(4.1.25b,c)ispreserved,whereasunderarbitrary nonholomorphic coordinatetransformations,ingeneral termsoftheform gijd id jand gijd id jaregeneratedinthelineelement.The nonhermitianmetriccoecients gij, gijare not relatedto IKijand IKij.Whenw orking withsuper“elds,sinceiischiral,onlyholomorphiccoordinatetransformationsmake obvioussense;however,wecanperformarbitrarycoordinatetransformationsonthe scalar“elds Ai. Usingthegaugetransformations(4.1.26)andholomorphiccoordinatetransformations,itispo ssiblet ogot oa normalgauge where,atanygivenpoint0, 0,evaluated at = =0, IKi1... im= IKj1... jn=0 foralln m ,( 4.1.27a) IKj i1... im= IKj1... jni=0 foralln m > 1,(4.1 .27b) IKi j= i j,(4. 1.27c) with i j=(1, 1,... Š 1, Š 1,. ..)depe ndingonthesignatureofthemanifold.Iftheidescribephysicalmattermultiplets, i j= i j.Inano rmalgauge,alltheconnections vanishatth epoint0,theRiemannc urvaturetensorhastheform: Ri j k l= IKik jl,( 4.1.28a) withallothercomponentsrelatedbytheusualsymmetriesoftheRiemanntensoror zero,andhencetheRiccitensorissimply: Rk j= IKik ji.(4. 1.28b) Inageneralgauge,theconnectionis ij k= IKij l( IKŠ 1)l k(4.1.29a) where( IKŠ 1)l kistheinverseofthemetric IKk l;allothe rcomponentsa rerelatedby complexconjugationorarezero.Thecontractedconnectionis,asalways,

PAGE 169

4.1.Thescalarmultiplet157i ij j=[ lndetIKk l]i.(4. 1.29b) TheRiemanntensorinageneralgaugeis Ri j k l= IKik jlŠ ( IKŠ 1)m nIKik mIKn jl(4.1.30a) andtheRiccite nsorhasthesimpleform Rk j Ri j k lIKl i=[ lndetIKi l]k j.(4. 1.30b) Manifoldscanhavesymmetries,or isometries. OnaK¨ ahlermanifold,anisometry ofthemetricis,ingeneral,aninvarianceoftheK¨ ahlerpotential IK uptoaK¨ ahler gaugetransformation(4.1.26).Onecanrequiretheisometrytobean invariance ofthe potentia l.(Actually,thisiso nlytrueifthereisapointon themanifoldwheretheisometrygroupisunbroken,i.e.,thetransformati onsdonotshiftthepoint.)This(partially) “xestheK¨ ahlergaugeinvariance:Itisnolongerpossibletogotoanormalgauge (4.1.27).However,holomorphiccoordinatetransformationsstillmakeitpossibleto choose normal coor dinates, wherethemetric IKi jsatis“es(4.1 .27c),anditsholomorphic deriva tives( IKi1j)i2... im IKj i1... imsatisfy(4.1.27b)(likewisefortheantiholomorphic de rivatives)buttheconditions(4.1.27a)are not satis“ed. Inarbitrarycoordinatesystems,the isometri esactonthecoordinatesas i=AkAi, i= AkAi(4.1.31) wherethesarein“nitesimalparameters(= areconsta nt unlessweintroduce gauge“eldsandgaugetheisometrygroup;supersymmetricgaugetheoriesarediscussed intheremai nderofthischapter),andthe k(, )sare Killingvectors. Thesesatisfy K illingsequations: kAi ; j+ kAj ; i= kAi ; j+ kAj ; i=0 (4.1.32a) kAi ; j+( IKŠ 1)i kkAl ; kIKl j=0.(4. 1.32b) where ki ; j= ki j= ki j (4.1.32c) and

PAGE 170

1584.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSki ; j= ki j+ kkjk i= ki j + kkIKjk l( IKŠ 1)l i(4.1.32d) For holomorphic K illingvectors ki= ki(), ki= ki( ),( 4.1.32a)isatrivialityand (4.1.32b)followsdirectlyfrom IKikAi+ IKikAi=0,(4. 1.33) whichisjustthestatementthattheK¨ ah le rp ot entialisinvariantundertheisometries. (Actually,invarianceuptogaugetransformations(4.1.26)sucestoimply(4.1.32b).) Wecanalsowrite thetransformations(4.1.31)as i=AkAj j i, i= AkAj j i;( 4.1.34a) Thisformexponentiatestogivethe“nitetransformation: i= exp (AkAj j )i, i= exp ( AkAj j ) i.(4. 1.34b) Forthecasesw henthereexistsa“xedpointonth emanif old,wecanchooseaspecial coordinatesystem(thatingeneralis not compatiblewithnormalc oordinates )wherethe transformations(4.1.31,34)takethefamiliarform i= i A( TA)i jj, i= Š i j A( TA)j i(4.1.35a) or,for“nitetransformations, i=( ei ATA)i jj, i= j( eŠ i ATA)j i.(4. 1.35b) Inarbitrarycoordinates,thenotionofmultiplyingvectorsby i isrepres entedby mult iplicationbyatwoindextensorcalledthe complexstr ucture. Ithastheproperty thatitssquareis Š 1 aKroneck erdelta.ForaK¨ ahlermanifold,thecomplexstructure iscovariantlyconstantandpreservesthemetric. Itmayhappenthatthereexistnontrivial nonholomorphic coordinatetransformationsthat do preservetheformofthe metric(4.1.25b,c);thenonecanshowthatthe manifoldis hyperK¨ ahler. Suchmanifoldshavethreelinearlyindependentcomplexstructuresandarelocallyquaternionic.Theyareeven(complex)dimensional;all hyperK¨ ahlermanifoldsareRicci”at,thoughtheconverseistrueonlyinfour(real) dimensions(twocomplexdimensions).

PAGE 171

4.2.Yang-Millsgaugetheories1594.2.Yang-Millsgaugetheories a.Prepotentials Ingeneral,wecan“ndaformulationofanysupersymmetricgaugetheoryeither byst udyingo-shellrepresentationstode rivethefree(linear)theoryintermsof unconstrained gaugesuper“eldsor prepotentials, orbypostulating covariant derivati vesand imposing covariant constraintsonthemuntilallquantitiescanbeexpressedintermsof asinglei rreduciblerepresentationofsupersy mmetry.Intheformercase,wemustconstructcovariantlytransformingderivativesoutoftheunconstrained“eldsandgeneralize tothenonlinearcase,whereasinthelattercasewemustsolvethecovariantconstraints intermsofprepotentials.Westudybotha pproachesandexhibittherelationbetween them. a.1.Linearcase Fromtheanalysiso fs ec.3.3.a.1,the N =1v ectormultipletconsistsofmassless spin1 2 andspin1physicalstates.Wedenotethecorrespondingcomponent“eld strengthsby f.A ccordingtothediscussionofsec.3.12.a,theselieinanirreducib leon-shellchiralsuper“eldstrength(0) ,whichsati s“esthe“eldequationsand Bianchiidentities D(0) =0.Theco rrespondingirreducibleo-shell“eldstrengthisa chiralsuper“eld W, D€W=0,satisfy ingthe bisectioncondition( s +1 2 N =1 2 +1 2 is aninteger) K K W= Š W,whichcanb ewri tten(see(3.11.9)) DW= Š D€ W€.(4. 2.1) (Wehavea Š signinthebisectionconditiontoobtainusualparityassignmentsforthe components.)Therefore,by(3.13.2),itcanbeexpressedintermsofanunconstrained realscalarsuper“eldby W= i D2DV W€= Š iD2 D€V V = V ,(4. 2.2) andthist urnsouttobethesimplestdescriptionofthecorrespondingmultiplet. Thede“nitionof Wis in va riantunder gaugetransformations withachiralparameter

PAGE 172

1604.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSV= V + i ( Š ), D€= D =0.(4 .2.3) Laterwegeneralizethistoanonabeliangaugeinvariance,butforthemomentweanalyzethesimplestcas e.Theprepotential V canbeexpandedincomponentsbyprojection: C = V | = iDV | €= Š i D€V | M = D2V | M = D2V | A€=1 2 [ D€, D] V | = i D2DV | €= Š iD2 D€V | ,D=1 2 D D2DV | .(4. 2.4a) (Toavoidconfusionwith D,w ed en otetheDauxiliary“eldbyD.)Asdisc ussedin sec.3.6.b,thereissomechoiceintheorderofthe D swhichsimplyamountsto“eld rede“nitions.Theparticularformwechosein(4.2.4a)issuchthatthephysicalcomponentsareinvariantunderthegaugetransformations(exceptforanordinarygauge transformationofthevectorcomponent“eld).Bymakingasimilarcomponentexpansion 1= | ,= D | ,2= D2 | ,(4. 2.4b) we “nd C = i ( 1Š 1), =, M = Š i 2, A€=1 2 €(1+ 1), =0, D=0.(4. 2.5) Thus,allthecomponentsof V canbegaugedawayby nonderivative gaugetransformationsexceptfor A a, andD.Thev ectorandspinorarethephysicalcomponent“elds ofthemultiplet;Disanauxiliary“eld.They(andtheirderivatives)aretheonly

PAGE 173

4.2.Yang-Millsgaugetheories161componentsappearingin W: = W| f=1 2 D( W )| ,D= Š1 2 iDW| i € €= D2W| .(4. 2.6) Thesymmetricbispinor fanditsconjugate f€€aretheselfdualandantiself-dual partsofthecomponent“el dstrengthoft he gauge“eld A a.The gaugeinwhich A ,Daretheonly nonzerocomponentsof V iscalledthe Wess-Zuminogauge. Theremaining gaugefreedomistheusualabeliangaugetransformationofthevectorcomponent“eld. TheWess-Zumino(WZ)gauge breaks supersymmetry:Thesupersymmetryvariationsof and M violatethegaugecondition C = = M =0,e .g., = i M + i €( i1 2 €C Š A€),(4.2 .7) does not vanishintheW Z gauge.Wecande“netransformationsthatpreservetheWZ gaugebyaugmentingtheusualsupersymmetrytransformationswithgauge-restoring gaugetransformations.Thus,insteadof V = i ( € Q€+ Q) V ,(4. 2.8) wetake WZV = i ( € Q€+ Q) V + i ( Š )WZ= i ( € Q€ WZ+ Q WZ) V ,(4. 2.9) whereWZischosentorestoretheWZgaugeconditionbycancelingthetermsin V thatviolateit.Speci“cally, WZ=0requires WZ( DV ) | =0.(4. 2.10) Using D2V | =0and { D€, D} V | = aV | =0(inthe WZ gauge),wehave WZ= DWZ| = i € D€DV | = i €A€.(4. 2.11) Sim ilarly,from WZM =0we “nd

PAGE 174

1624.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS2 WZ= D2WZ| = Š € €.(4. 2.12) Finally,from WZC =0,we “ndthat1 WZ= 1 WZ.There mainingrealscalarinWZis theusualcomponentgaugeparameterforthevectorgauge“eld(see(4.2.5)). TheWZgaugepreservingsupers ymmetrytransformationsare A a= Š i ( €+ €), = Š f+ i D, D=1 2 €( €Š €).(4.2 .13) Thecommutatoralgebraofthesetransformationsclosesonlyuptogaugetransformationsofthevector“eld. Theneedforgauge-restoring transformationsmakessupersymmetricquantizationintheWZgaugeimpossible. The(vector)gauge-“xingprocedure,bybreakinggaugeinvarian ce,alsobreakssupersymmetry. Fromtherequi rementthatthephy sicalcomponents A aand havecanonical dimension,weconcludethat V hasdimensionzero.Bydimensionalanalysisandgauge in va rianceunderthetransformationswe“ndtheaction S = d4xd2 W2=1 2 d4xd4 VD D2DV .(4. 2.14) Replacing d2 by D2andusing(4.2.6),weobtainthecomponentaction S = d4x [ Š1 2 ff+ €i €+D 2].(4.2 .15) Wehave notaddedthehermitianconjugateto S ; ImS isatotalderivativeandcontri butesonlyasurfaceterm( d4x a b c df a bf c d+spi norialterms).The“eldDis cl ea rlyauxiliary. a.2.Nonlinearcase Thenonabeliangeneralizationcanbemotivatedbystartingwithaglobalinternal symmetryandmakingitlocal.Forthispurposeweconsideramultipletofchiralscalar “eldstransformingaccordingtosomerepre sentationofaglobalgroupwithgenerators TAandconstantparameters A:

PAGE 175

4.2.Yang-Millsgaugetheories163= ei = ATA, TA= TA.(4. 2.16) Weextend thistoalocaltransformationinsuperspace.Clearly,tomaintainthechiralityofthelocalparametersshouldbechiral .Wetherefo reconsidertransformationsof theform = ei ,=ATA, D€=0,(4 .2.17) andcorrespo ndingly,fortheantichiral ,transformingwiththecomplexconjugaterepresentation, = eŠ i = ATA, D =0.(4 .2.18) TheLagrangian is invariantiftheparameters Aarereal.Forlocaltransformations = andwemustintro duceagau ge“eldtocovariantizetheaction.Thesimplestprocedureistointroduceamultipletofrealscalarsuper“elds VAtransformingin thefollowingfashion: eV= ei eVeŠ i V = VATA.(4. 2.19) Intheabeliancase,thistransformationisjust(4.2.3).Wecovariantizetheactionby d4xd4 eV.(4.2 .20) Thegauge“eld V actsasa converter ,cha ngingarepre sentationtoa representationofthegroup.Thus, ( eV)= ei ( eV),(4.2.21a) andsimilarly ( eV)=( eV) eŠ i .(4. 2.21b) Inthenonabeliancase,eventhein“nitesimalgaugetransformationsof V are highlynonlinear.Nonetheless,asinthea be liancasetheycanbeusedtoalgebraically gaugeawayallbutthephysicalcomponentsof VAandtakeustotheWess-Zumino gauge:Startingwithanarbitrary V ,weperform successivegaugetransformationsto gaugeaway C ,and M .Requiri ngthat the“rsttransformationgaugeaway C we “nd,byevaluating(4.2.19)at =0:

PAGE 176

1644.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS1= eV| = ei (1)eVeŠ i (1)| =( ei (1)| ) eC( eŠ i (1)| ),(4.2 .22) andhencewemustchoose(1) 1=(1)| = Š i1 2 C .The gauge C=0ispr eservedbyall furthertransformationswith Im 1=0.To gaugeaway wechooseas econdgauge transformation(2)(with(2) 1=0)byreq uiring 0= DeV| = D( ei (2)eVeŠ i (2)) | = DV|Š iD(2)| = Š i Š i (2) ,(4. 2.23) andhence(2) = D(2)| = Š .Fina lly,wecan“ndathirdtransformation(3)to gaugeaway M .IntheWZ gauge,theonlygaugefreedomleftcorrespondstoordinary gaugetransformationsofthevector“eld A a,withpar ameter= = ( x ). Asintheabeliancase,theWZgaugeisno tsupersy mmetric,andgauge-restoring transformationsarerequiredtode“netheWZ gaugesupersymmetrytransformations. Theparameterofthetransformationsisstill(4.2.11-12),butthetransformationsnow b ecomenonabelianandhencenonlinear.To“ndthem,wecomputethein“nitesimal gaugetransformationsof V :Webeginby de“ningthesymbol LVX =[ V X ],(4.2 .24) sothat eVXeŠ V= eLVX .(4. 2.25) From[ V eV]=0weobtain ( V ) eV+ V ( eV) Š eV( V ) Š ( eV) V =0 ,( 4.2.26a) or eŠ1 2 V( V ) e1 2 VŠ e1 2 V( V ) eŠ1 2 V+ eŠ1 2 V[ V eV] eŠ1 2 V=0,(4. 2.26b) andhence 2 sinh(1 2 LV)( V )= eŠ1 2 VLV( eV) eŠ1 2 V= LV[ eŠ1 2 Vi e1 2 VŠ e1 2 Vi eŠ1 2 V]

PAGE 177

4.2.Yang-Millsgaugetheories165= iLV[ cosh(1 2 LV)( Š ) Š sinh(1 2 LV)( +)],(4 .2.27) fromwhichi tfo llows V = Š1 2 iLV[ ++ coth(1 2 LV)( Š )] = i ( Š ) Š1 2 i [ V +]+ O ( V2).(4.2 .28) Fromthetransfo rmations(4.2.28)andtheparameter(4.2.11-12)we“ndthenonabelianWZgauge-preservings upersymmetrytransformations: A a= Š i ( €+ €), = Š f+ i D, D=1 2 €( €Š €),(4.2 .29) wherenow fisthesel f-dualpartofthe nonabelian “eldstrengthand €= €Š iA€.The nonlinearitycomesfromthegauge -covariant izationofthelinear transformations(4.2.13).Thecomponentsofthenonabelianvectormultipletarecovariantgeneralizationsoftheabeliancomponents;intheWZgauge,theyarethesameas (4.2.4a)(seealso(4.3.5)). a.3.Covariantderivatives Thegauge“eld V canbeusedtoconstructderivatives,gaugecovariant with respectto transformations A= DAŠ i A=( €, €),(4.2 .30) de“nedbytherequirement ( A)= ei ( A),(4.2.31) i.e., A= ei AeŠ i ,( 4.2.32a) or

PAGE 178

1664.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS A= i [, A].(4.2 .32b) Sinceischiral, € D€iscovariantwithoutfu rthermodi“cation: €= ei D€eŠ i = €.(4. 2.33) Theundottedspinorderivative Discovariantw ithres p ectto trans formations.We canuse eVtoconv ertitintoaderivativecovariantwithrespectto(see(4.2.21)); eŠ VDeVtransformscorrectly: =( ei eŠ VeŠ i ) D( ei eVeŠ i ) = ei eŠ VDeVeŠ i = ei eŠ i .(4. 2.34) Finally,weconstruct abyan alogywith(3.4.9): a= €Š i {, €} .Itscovariancefollowsfromthatof and €. Wesu mmarize: A=( eŠ VDeV, D€, Š i {, €} ).(4.2 .35) Thesederivativesarenothermitian.Theirconjugates Aarecovariantwith respectto transfo rmations: A=( D, eV D€eŠ V, Š i { €} ), A= ei AeŠ i .(4. 2.36) Thederivatives A( A)areca lled gaugechiral(antichiral)representation covariant derivatives.Theyarerelatedbyanon unitarysimilaritytransformation A= eVAeŠ V.(4. 2.37) Thisisanalogoustotherelationbetweenglobalsupersymmetrychiralandantichiral representations DA ( Š )= eUDA (+)eŠ U(4.2.38) of(3.4.8).

PAGE 179

4.2.Yang-Millsgaugetheories167Thegaugecovariantderivativesareusuallyde“nedintermsofvectorrepresentation DAs;ifweexpresstheseintermsofordi naryderivatives,(4.2.35)becomes A=( eŠ VeŠ1 2 Ue1 2 UeV, e1 2 U €eŠ1 2 U, Š i {, €} ).(4.2 .39) Byafurthersimilaritytransformation A eŠ1 2 UAe1 2 U,wegotoanewrepresentation thatischiralwithrespecttobothglobalsupersymmetryandgaugetransformations: A=( eŠ1 2 UeŠ VeŠ1 2 Ue1 2 UeVe1 2 U, €, Š i {, €} ).(4.2 .40) Wede “ne Vby e1 2 UeVe1 2 U= eU + V .(4. 2.41) Inthisform,itisclearthat Vgaugecovariantizes U : i €€ ... + i €( €Š iA€)+ ... .Thiscom binationtr ansformsas ( eU + V )= ei ( eU + V ) eŠ i = €=0.(4 .2.42) Therealsoexistsasymmetric gaugevectorrepresentation thattreatschiralandantichiral“eldsonthesamefooting.Sucharepresentationusesacomplexscalargauge“eld ,andrequiresalargergaugegroup.Wediscussthevectorrepresentationinsubsec. 4.2.b,wherethecovariantderivativesarede“nedabstractly,andwhereitentersnaturally. a.4.Fieldstrengths Thecovariantderivativesde“ne“eldstrengthsbycommutation: [ A, B} = TAB CCŠ iFAB,(4. 2.43) with V = VATA,and TAintheadjointrepresentation.Fromtheexplicitformofthe covariantderivatives(4.2.35)we“ndthatthetorsion TAB Cisthesam eoneas in”at globalsuperspace(3.4.19),andsome“eldstrengthsvanish: F= F€€= F€=0.(4. 2.44) Theremaining“eldstrengthsare F€ €= C€€ D2( eŠ VDeV)= iC€€W,

PAGE 180

1684.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSF€ €=1 2 ( C€€( W )+ C(€W€ )), W i D2( eŠ VDeV), W€ eŠ V W€eV eŠ V( Š W)eV.(4. 2.45) (Recallthat W€ ( W)implies W€=( Š W)(3.1.20).)Thusallt he“eldstrengthsof thetheoryare expressedintermsofasinglespinor Wthatisthenonlinearversionof (4.2.2).Itsatis“esBianchiidentitiesanalogousto(4.2.1): W= Š€W€.(4. 2.46) Itischiral,hasdimension3 2 ,andcanbeusedtoc onstructagaugeinvariantaction S =1 g2 tr d4xd2 W2= Š1 2 g2 tr d4xd4 ( eŠ VDeV) D2( eŠ VDeV), V = VATA, trTATB= AB.(4. 2.47) Asintheabeliancase,thisactionishermitianuptoasurfaceterm(seediscussionfollowing(4.2.15)). a.5.Covariantvariations Toderivethe“eldequatio nsfromtheaction( 4.2.47),weneedtovarytheaction withrespectto V .Howev er,since V isnotacovarianto bj ect,thisresultsinnoncovariant“eldequations(althoughmultiplicationbyasuitable(butcomplicated)invertible operatorcovariantizesthem).Inaddition,variationwithrespectto V iscomplicated b ecause V appearsin eVfactors.Wethereforede“ne acovariantva riationof V by V eŠ V eV= 1 Š eŠ LVLV V = V +....( 4.2.48) V satis“esthechiralrepresentationhermiticityconditionasin(4.2.37).Inpractice, wealwaysvary anactionwithrespectto V byexpr essingitsv ariationintermsof eV, andthenrewritingthatintermsof V .Wethusde “neacovariantfunctionalderivative F [ V ] V by(cf. (3.8.3))

PAGE 181

4.2.Yang-Millsgaugetheories169F [ V + V ] Š F [ V ] ( V F [ V ] V )+ O (( V )2).(4.2 .49) Wenowo btaintheequationsofmotionfrom: g2 S = itr d4xd4 ( eŠ VDeV) W= itr d4xd4 [ eŠ VDeV, V ] W= Š itr d4xd4 V W,(4. 2.50) whichgives g2 S V = Š i W=0.(4. 2.51) *** Attheendofs ec.3.6weexpressedsupersymmetryt ransformationsintermsofthe spinorderi vatives D.Usingthecovar iantderivati vesthatweh aveconstru cted,wecan writemanifestlygaugecovariantsupersymmetrytransformationsbyusingtheform (3.6.13)(for w =0)anda ddingthegaugetr ansformation = i D2(D ), (4.2.52a) whereAisde “nedin(4.2.30).Wethen“nd eŠ VeV=( W+ W€ €) =( WeŠ VDeV+ eŠ V W€eV D€) (4.2.52b) (where isareal x -independentsuper“eldthatcommuteswiththegroupgenerators, e.g., = D ).Since(4.2.52b)ismanifestlygaugecovariant,itpreservestheWessZuminogauge(butitisnotasymmetryofthe actionaftergauge-“xing).Thecorrespondingsupersymmetr ytrans formati onsfor covariantly chiralsuper“elds, €=0 witharbitraryR-weight w are = Š i 2[( ) + w ( 2 )].(4.2.52c)

PAGE 182

1704.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSb.Covariantapproach InthissubsectionwediscussanotherapproachtosupersymmetricYang-Mills theorythatreversesthedir ectionoftheprevioussection.Wepostulatederivatives tr an sformingcovariantlyunderagaugegroup,imposeconstraintsonthem,anddiscover thattheycanbeexpressedint ermsofprepotentials.Thisprocedurewillproveespeci allyusefulinstudyingsupergravityandextendedsuper-Yang-Mills,sowegivea detailedana lysisforthesimplercaseof N =1 su pe r-Yang-Mills. Westartwitht heordinarysuperspacederivatives DAsatisfying [ DA, DB} = TAB CDC,where TAB Cisthetorsionandhasonlyonenonzerocomponent T€ c(see(3.4.19)).ForaLiealgebrawithgenerators TAwecovarian tizetheder ivatives byintro ducingconnection“elds A= DAŠ i A,(4. 2.53) whereA=ABTBishermitianand A= Š ( Š )AA.Atthecomp onentlevelwehave = v+i 2 €v€+ ... a= w a+ ... ,( 4.2.54a) andhence = Š iv+i 2 €( €Š iv€)+ ... €= €Š i v€+i 2 ( €Š i v€)+ ... a= aŠ iw a+ ... ,(4. 2.54b) sothat thecompone ntderivati vesarecova riantized. Undergaugetransformationsthecovariantderivativesarepostulatedtotransform as A= eiKAeŠ iK,(4. 2.55) wheretheparameter K = KATAisarealsup er“eld. K = ( x )+ K(1) ( x )+ € K(1)€( x )+ ... .(4. 2.56) Thisisverydierentfromwhatemergedintheprevioussection:Insteadofchiral

PAGE 183

4.2.Yang-Millsgaugetheories171representationderivativestransformingwiththechiralparameter,wehave vectorrepresentation hermitianderivatives,transformi ngwith thehermitianparameter K .The asymmetricformoftheprevioussectionwillemergewhenwemakeasimilaritytransformationtogotothechiralrepresentation. Fo ri n“nitesimal K ,we “ndthecomponenttr ansformations: v€=[ €Š iv€, ] Š i €, w€=[ €Š iw€, ],(4.2 .57) where K | € [ D€, D] K | = €.Thecompon entgaugeparameter €canbe usedto gaugeaway Imv€algebraically;however,thecomponent“elds Rev€and w€bothre mainastwoaprioriindependentgauge“eldsforthe same componentgauge transformation.Toavoidthisweimpose constr aints onthecovariantderivatives. b.1.Conven tionalconstraints Fieldstrengths FABar ed e“nedby(4.2.43).Substituting(4.2.53)we“nd FAB= D[ AB )Š i [A,B}Š TAB CC.(4. 2.58) Inparticular, F€= D €+ D€Š i { €}Š i €.(4. 2.59) Ifweimposetheconstraint F€=0,(4. 2.60) (4.2.59) de“nes thev ectorconnection€intermsofthespinorc onnections.(Incomponents,this expresses w€intermsof v€and v.) Inanytheoryonecanaddcovarianttermstotheconnections(e.g.,(3.10.22)) withoutchangingthetransformationofthecovariantderivatives.Ifwedidnotimpose theconstraint(4.2.60)ontheconnectionsA,wecouldde “neequallysatisfactorynew connections A=(, €,€Š iF€)thati denticallysatisfytheconstraints.Forthis reason(4.2.60)iscalleda con ventional constraint .Itimp lies A=( €, Š i {, €} ).(4.2 .61)

PAGE 184

1724.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSThetheorynowisexpressedentirelyintermsoftheconnection.Howev er,it containsspin s > 1 gaugecovariantcomponent“elds,forexample ( )€ F( )€| = i [ D€, D( ] )| + ... .(4. 2.62) Italsocontainsasuper“eldstrength Fwhose -independentcomponent f= F| = D( )| + ... ,(4. 2.63) isadimensiononesymmetricspinor(equivalenttoanantisymmetricsecondranktenso r).Becauseofitsdimension,itcannotbetheYang-Mills“eldstrength.Althoughin pr inciplethetheorymightcontainsuch“elds(asauxiliary,notphysical,components),in thecovariantapproachtherearegenerallyfurthertypesofconstraintsthateliminate (many)suchcomponents. b.2.Representation-p reservingconstraints Toco uplescalarmultipletsdescribedbychiralscalarsuper“eldstosuper-YangM illstheory,wemustde“ne covariantl ychiral super“elds:Thecovariantderivatives transformwiththehermitianparameter K ,andall“el dsmusteitherbeneutralor transformwiththesameparameter.However, K isnotchiral,andgaugetransformationswillnotpreservechiralityde“nedwith D€.Instead wede “neacovaria ntlych iral super“eldby €=0,= eiK, =0, = eŠ iK.(4. 2.64) Thisimplies 0= { €, €} = Š iF€€.(4.2 .65) Consistencyrequiresthatweimposethe repres entation-preserving constraint F= F€€=0.(4. 2.66) Thiscanbewrittenas {, } =0.(4. 2.67) Themostgeneralsolutionis

PAGE 185

4.2.Yang-Millsgaugetheories173= eŠ De,=ATA,(4. 2.68) whereAisanarbitrary complex super“eld.Eq.(4.2.67)statesthat satis“esthe samealgebraas D,andthesol utionexpressesthefactthat theyareequivalentuptoa complex gaugetransformation.Hermitianconjugationyields €= e D€eŠ .(4. 2.69) Thus Aiscompletelyexpressedintermsoftheunconstrained prepotential bythe solutions(4.2.61,68,69)totheconstraints(4.2.60,66). The K gaugetransformationsarerealizedby ( e)= eeŠ iK.(4. 2.70) However,thesolutiontotheconstraint(4.2.67)hasintroducedanadditionalgauge invariance:Thecovariantderivatives(4.2.68)are invariant underthetransformation ( e)= ei e, D€=0.(4 .2.71) Therefore,thegaugegroupofislargerthanthatofA. Wede “nethe K -invarianthermitianpartofby eV= ee .(4. 2.72) The K gaugetransformationscanbeusedtogaugeawaytheantihermitianpartof. Inthisgauge,= =1 2 V ,andtra nsformationsmustbea ccompaniedbygaugerestoring K transformations: ( e)= ei eeŠ iK (), eŠ iK ()= eŠ eŠ i ( ei e2eŠ i )1 2 .(4. 2.73) In any gauge,thetransformationof V is ( eV)= ei eVeŠ i .(4. 2.74) Wehave de“nedcovariant lychiralsuper“eldsby(4.2.64).Wecanuse(see (4.2.69))toexpressthemintermsof ordinary chiralsuper“elds0(whichwecalledin sect.4 .2.a):

PAGE 186

1744.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS= e 0, D€0=0.(4. 2.75) Thefactor e converts K -transforming“eldsinto-transforming“elds: (0)=( eŠ )= ei 0.(4. 2.76) *** Ausefulidentityt hatfo llowsfromtheexplicitform(4.2.68)expresses in termsofanarbitraryvariation : =( eŠ ) e+ eŠ e=[ eŠ e].(4.2 .77) b.3.Gaugechiralrepresentation Wecanalsouset ogotogau gechiralrepresentationinwhich all quantitiesare K -inertandtransformonlyunder.Thisisa nalogoustoandnottobeconfusedwith thesupersymmetrychiralrepresentation( 3.3.24-27),(3.4.8).Wemakeasimilarity transformation 0 A= eŠ Ae =( eŠ VDeV, D€, Š i {0 0€} ), 0= eŠ 0= e = (0) eV.(4. 2.78) Thequantities 0 Aand0arethechiralrepresentation Aandoftheprevioussubsection.Wesometimeswritethechiralr epresentationhermitianconjugateof0as0toavoidconfusionwiththeordinaryhermitianconjugate 0 (0). Inthechiralrepresentationweseenotraceofor K :Only V andappear. However,wenecessarilyhaveanasymmetr ybetw eenchiralandantichiralobjects. c.Bianchiidentities Insubsection4.2.aweanalyzedthephysicalcontentofthetheoryusingcomponentexpansionsandtheWess-Zuminogauge.Alternatively,wecan“ndthe“eldcontentofthetheorybysolvingtheBianc hiidentities.ThesefollowfromtheJacobi

PAGE 187

4.2.Yang-Millsgaugetheories175identities: [ [ A[ B, C )}} =0 ,( 4.2.79a) whichimply [ AFBC )Š T[ AB | DFD | C )=0.(4. 2.79b) Normallytheseequationsaretrivialidenti ties.How ever,onceconstraintshavebeen imposedonsome“eldstrengths,theygiveinformationabouttheremainingones,andin particularallowonetoexpressallthe“eldsstrengthsintermsofabasicset.Wenow describetheprocedure. Wesolvetheeq uations( 4.2.79)subjecttotheconstraints(4.2.60,66)startingwith th eo nesoflowestdimension.Foreachequation,weconsidervariouspiecesirreducible undertheLorentzgroup,andseewhatrelationsareimpliedamongthe“eldstrengths. Thus,forexample,therelation[ {( } )]=0isidentica llysati s“edwhen F =0.From[ {, } €]+[ { €, ( } )]=0, we “nd F( )€=0,(4. 2.80) whichimplies,forsomespinorsuper“eld W, F €= Š iC W€.(4. 2.81) From[ {, } c]+ { [ c, ( ], )} =0we “nd C ( ) W€=0,(4. 2.82) whichimplies W€=0.(4. 2.83) From[ {, €} c]+ { [ c, ], €} + { [ c, €], } =0weobtain F€ €+ C € W€+ C€€W=0,(4. 2.84) whichseparatesintotwoequations: F€ €=1 2 ( C (€ W€ )+ C€€( W )) C f€€+ C€€f,(4. 2.85) and

PAGE 188

1764.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSW+ € W€=0.(4. 2.86) Thesecanbereexpressedas W= iCD+ f,D= D= Ši 2 W.(4. 2.87) Finally,[[ [ b], c ]]+[[ b, c], ]=0and[[ [ a, b], c ]]=0areau tomaticallysatis“edasaco nsequenceofthepreviousidentities.From(4.2.87)wealsoobtain €D=1 2 €W, €f= i1 2 ( €W ),(4. 2.88) and f=1 2 C ( i )€ W€.(4. 2.89) Therefore,allthe“eldstrengthsareexpressedintermsofthechiral“eldstrength W.Inparticular, thecommutatorsofthecovariantderivativescanbewrittenas: {, } =0, {, €} = i €, [ €, i €]= Š iC€€W, [ i a, i b]= i ( C€€f+ C f€€).(4.2 .90) Furthe rmore,theset F = { W,D, f} ,(4. 2.91) is closedundertheoperationofapplying and €:Onlys pacetimederivatives aof F aregenerated.Thesesuper“eldsarethenonlinearo-shellextensionofthesuper“eld strengths( n )ofsec.3.12.Thecovariantcomponentsarethe =0proj ectionsofthese super“elds.ThustheconstraintsandtheBianchiidentitiesdirectlydeterminethe“eld contentofthetheory. ***

PAGE 189

4.2.Yang-Millsgaugetheories177Theexistenceofageometricsuperspaceformulationintermsofa(constrained) connectionAisimportant.ForquantizedsuperYang-Millstheories,thegeometric(or covariant)formulationcanbecombined withthebackground“eldmethodtoderive improvedsuper“ eldpower-countinglaws.WecanalsouseAtogeneralizet heconcept ofthepath-orde redphasefactortosuperspace: IP [ e( i dzAA)],(4.2 .92) wherethedierentialsuperspaceelement dzAistobeinterpretedas d dzA for some parametrizationofthepath.(Inparticular,d is not aBer ezinintegral.)Ifwe chooseaclo sedpath,thisquantityde“nesasupersymmetricWilsonloop.ThusnonperturbativestudiesofordinaryYang-MillstheoriesbasedonthepropertiesoftheWilson loopshouldbeextendibleintosuperspace.(Thereisalsoamanifestlycovariantformof pathordering,expresseddirectlyintermsofcovariantderivatives:seesec.6.6.)

PAGE 190

1784.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS4.3.Gauge-invariantmodels a.Renormalizablemodels Inthissubsectionweconsiderpropertieso fsystemsofin teractingchiralandreal gaugesuper“eldswithactionsoftheform S = d4xd4 j(eV)j ii+ tr d4xd2 W2+[ d4xd2 P(i)+ h c .](4.3.1) (i nt hegauge-chiralrepresentation),invariantunderagroup G .Here Vi j= VA( TA)i jand( TA)i jisa(ingeneralreducible)matrixrepresentationofthegenerators TAof G Inthevectorrepresentation,(4.3.1)takestheform S = d4xd4 ii+ tr d4xd2 W2+[ d4xd2 P(i)+ h c .](4.3.2) whereweh aveused treŠ fe = trf inthechiralintegral,andrewrittentheactionin termsofcovariantlychiralsuper“elds.Thegaugecouplinghasbeensetto1,butcanbe restoredbytherescalings W gŠ 1W. S maybeR-symmetric,withthegaugesuper“eldtransf ormingas V( x , )= V ( x eŠ ir eir ). Anothertermcanbeaddedtotheaction:If G isabelian,orhasanabeliansubgroup,the Fayet-Iliopoulos term SFI= tr d4xd4 V = tr d4x D,(4. 3.3) isgaugeinvariant. Componentactionscanbeobtainedbytheprojectiontechniqueswehavediscussedbefore.Amoreecientand,upto“eldrede“nitions,totallyequivalentprocedureistode“ne covariantc omponents byproj ectingw ithcovariantderivatives.Thus, foracovariantlychiralsuper“eldwede“ne A = | = | F = 2 | .(4. 3.4) Sim ilarly,thecovariantcomponentsofthegaugemultipletcanbeobtainedbyprojection from W(here fdenotesthecomponent“eldstrength): = W| f=1 2 {( W )}| ,

PAGE 191

4.3.Gauge-invariantmodels179i € €=1 2 [ {, W} ] | ,D= Ši 2 {, W}| .(4. 3.5) Thecovariant deriva tive €| isthecovariantspace-timederivative.Toobtaincomponentacti onsbycovariantprojection,weusethefactthatonagauge invariant quantity D2D2= 22. Thecomponentactionthatresultsfrom(4.3.1)plus(4.3.3)takestheform S = d4x[Ai Ai+ ii € € i+ i Ai( )i j jŠ i € i( €)i jAj+ Ai(D)i jAj+ Fi Fi+ tr ( [ i €, €] Š1 2 ff+D 2) + tr D+(PiFi+1 2 Pij i j+ h c .)](4.3.6) where 1 2 a a,Pi,Pijarede“nedin(4.1.13),( )i j= ATA,e tc.Theauxiliary “eldDcanbeeliminatedalgebraica llyusingits“eldequations.Thisleadstointeractiontermsforthespin -zero“eldsofthechiralmultiplets: Š UD= Š1 4 [ Ai( TA)i jAj+ trTA]2(4.3.7) inadditiontothoseobtainedbyeliminating F (see(4.1.14)). b.CP(n)models Insec.4.1.bwediscussedsupersymmetricnonlinear -modelswrittenintermsof chiral andantichirals uper“eldsthatarethecomplexcoordinatesofaK¨ ahlermanifold. Somenonlinear -modelscanbewritten linearlyifweintroducea(classically)non-propagatinggauge“eld.Weconsiderheresupersymmetricextensionsofthebosonic CP ( n ) models.Thebosonicmodelsarestraightforwardgeneralizationsofthe CP (1)modelof sec.3.10.Theyarewrittenintermsof( n +1)complexsc alar“elds ziconstrainedby zi zi= c ;theacti oniswrittenbyintroducinganabeliangauge“eldwithnokinetic term: S = d4x [ | ( €Š iA€) zi|2+D( | zi|2Š c )],(4.3.8) whereDisaLagrangemultiplier“eld.Eliminating A€byitsc la ssical“eldequation,

PAGE 192

1804.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSwe “ndtheactiongivenin(3.10.23).ThisactionisstillinvariantunderthelocalU(1) gaugetransformation z ei z z eŠ i where ( x )isarealpa rameter.Itcanbe rewrittenintermsof n +1 unconstrained“elds Ziintheform(3.10.30). Inthesupersymmetriccase,themodelismostconvenientlydescribedintermsof ( n +1)chiral“ elds(andtheircomplexconjugates ),andasingleabeliangauge“eld V .Theacti on,whichisgloballysupersymmetric, SU ( n +1)invar iant,andlocally gaugeinvariant,is: S = d4xd4 (i ieVŠ cV ).(4.3 .9) NotethepresenceoftheFayet-Iliopoulosterm.Uponeliminatingthegauge“eld V by its“eldequationwe“nd S = d4xd4 cln (i i).(4.3 .10) Thisactionisstillinvariantunderthe(local)abeliangaugetransformation ei Wecanusethis invariancetochooseagauge,e.g.,i=( c ua).Incomponents,(4.3.10) givestheactiongeneralizing(3.10.30)forthe CP ( n )non linear -modelcoupledtoa spinor “eld. Theaction(4.3.9)hasastraightforwardgeneralization: S = d4xd4 ( i( eV)i jjŠ ctrV ),(4.3 .11) where,asin(4.3.1), V = VATAand( TA)i jisa(ingeneralreducible)matrixrepresentationofthegenerators TAofsomegroup.However,incontrastto(4.3.9),whenwevary (4.3.11)withrespectto V ,wegetan equationthatingeneraldoesnothaveanexplicit solution: eVTA Š ctrTA=0.(4. 3.12) (Toderive(4.3.12),weusethecovariantvariation(4.2.48) V = VATA eŠ V eV,and tr V = tr V .)

PAGE 193

4.4.Superforms1814.4.Superforms a.General Inordinaryspacetime,thereisafamilyofgaugetheoriesthatcanbeconstructed systematically;thesetheoriesareexpressedintermsof p -forms p=1 p dx m1/ / \ \dx m2/ / \ \.../ / \ \dx mp m1 m2... mpwherethedierentialssatisfy dx m/ / \ \dx n= Š dx n/ / \ \dx m.Thetowero ftheorie sbased onformsis:0=scalar,1= v ectorgauge“eld,2=tenso r gauge“eld,3=a ux iliary“eld,and4=not hing “eld.Theirgau getransformations,“eldstrengths,andBianchiidentitiesaregivenby gaugetransformation: p= dKp Š 1, fieldstrength: Fp +1= d p, Bianchii dentity: dFp +1=0.(4. 4.1) Here Kp,p, Fpare p -formgaugeparameters,gauge“elds,and“eldstrengthsrespectively,and d = dx m m.Byde “nition, Š 1-formsvanish,and5-forms(or(D+1)-formsin Ddimension s)vanishbyantisymmetry.TheBianchiidentitiesandthegaugeinvariance ofthe“eldstrengthsareautomaticconsequencesofthePoincar ele mma dd =0. Insuperspacethesameconstructionispossible,usingsuper p -forms: p=( Š 1)1 2 p ( p Š 1)1 p dzM1/ / \ \.../ / \ \dzMpMp... M1(4.4.2a) (notetheorde ringoftheindices),wherenow dzM/ / \ \dzN= Š ( Š )MNdzN/ / \ \dzM,(4. 4.2b) thecoecientsofthefo rmaresuper“elds,and d = dzMM.Thesameto werof gauge parameters,gauge“elds,“eldstrengths,andBianchiidentitiescanbebuiltup(now usingthe super Poincar ele mma dd =0).Anadvant ageofthisdescriptionof”atsuperspacetheoriesisthatitgeneralizesimmediatelytocurvedsuperspaceanddetermines thecouplingoftheseglobalmultipletstosupergravity. However,superformsdonotdescribeirreduc iblerepresentation sofsuper symmetry unlessweimposec onstraints.Tomaintaingaugeinvariance,theseconstraintsshouldbe

PAGE 194

1824.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSimposedo ntheco ecientsofthe“eldstrengthform;whentheconstraintsaresolved, thecoecientsofthe(gauge)potentialformareexpressedintermsofprepotentials.In table4.4.1theprepotentials Apcorrespondtothe constr ained super p -form Apandthe expressions dA pcorrespondto dAp. p ApdA p 0 i ( Š ) 1 Vi D2DV 21 2 ( D+ D€ €) 3 V D2V 40 Table4.4 .1.Simplesuper“elds(prepotentials)correspondingtosuperforms InthisTableandarechi raland V isreal.Therelation Ap= A4 Š pcorrespondsto Hodge dualityofthecomponentforms. Theconstrainedsuper p -formscorrespondtoparticularprepotentials Apwhether Apisagaugeparameter Kp,apotentialp,a“eldst rength Fp,oraBianchiidentity ( dF )p.Theex p licitexpressionsfor Apintermsof Aptakethesameformwhether A is K ,, F ,or dF .Thusthe prepotentialsgiverisetoatowero ftheoriest hatmim ics(4.4.1): Thegauge“eldstrengthandBianchiidentitiesatonelevelarethegaugeparameterand “eldstrengthatthenextlevel.If Ap Š 1, Ap,and Ap +1arethegaugeparameter Kp Š 1,the gauge“eldp,andthe“el dstrength Fp +1superforms,respectively,thenthegauge transformation,“eldstrength,andBianchiidentitiesofthe prepotentials are gaugetransformation: p= dK p Š 1, fieldstrength: Fp +1= d p, Bianchii dentity: dF p +1=0.(4. 4.3) TheLagrangiansforall p -formtheoriesarequadraticinthe“eldstrengths,without extraderivatives.Wediscussde tailsinthesubsectionsthatfollow.

PAGE 195

4.4.Superforms183Underasupersymmetrytransformationth esuperfo rmsarede“nedtotransformas ( z, dz)=( z dz ),(4.4 .4) where(cf.(3.3.15)) dz=( d , d €, dx €)=( d , d €, dx€Ši 2 ( €d + d €)).(4.4.5) Consequently,thecoecientsMN ...mixundersupersymmetrytransformationsandthis makesitdiculttoimposesupersymmetric constraintsonthem.Tomaintainmanifest supersymmetry,wethereforegotoatangentspacebasis,parametrizedbythedualsof thecovariantderivatives DAratherthanthedualsof M.Weusethe”at superspace vielbeins DA M(3.4.16): DA= DA MM=( D, D€, €),(4.4 .6) andthe dualforms A dzM( DŠ 1)M A.(4. 4.7) From(3.4 .18),the D ssati sfy D[ ADB ) M= TAB CDC M,(4. 4.8) andhence d A=1 2 C/ / \ \BTBC A.(4. 4.9) Inthis -basiswewr iteasuperformas p=( Š 1)1 2 p ( p Š 1)1 p A1/ / \ \.../ / \ \ApAp... A1.(4. 4.10) Wealsohave d dzMM= ADA.Theta ngentspacecoecientsAp... A1ofthe p -form donotmixundersupersymmetrytransformationsbecause Aisinvariant.Wecannow imposesupersymmetricconstraint soni ndividualcoecientsofaform. Inthisbasis,thecoecientsofthe“eldstrengthform(onwhichweimposethe constraints) Fp +1= d phavethefollowingexpressi onintermsofthegauge“elds: FA1... Ap +1=1 p D[ A1A2... Ap +1)Š1 2( p Š 1)! T[ A1A2| BB | A3... Ap +1),(4. 4.11) wherethetorsiontermscomefrom(4.4.9).TheBianchiidentityon F takesasimilar

PAGE 196

1844.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSappearance.Equation(4.4.11)istheessentialresultweneedforthediscussionofsubsecs.4.4 .b-e. Wenows ummarizesomeoftheresultsofsubsecs.4.4.b-e.Inparticular,wegive theexplicitexpressionsforthecoecientsofthesuperforms Apintermsoftheprepotentials Ap(oftable4.4.1)forall p .Inthecaseofp,theseex pressionsarefoundby solvingtheconstraintsoncertaincoecientsof Fp +1andchoosingasuitable K -gauge ( = dK ).Theexpressionsfor K followfromthenewinvariancefoundwhensolving theseconstraints.Theexpressionsfor F fo llowfromsolvingthoseBianchiidentities dF that explicitly expressonepartof F intermsofanotherinthepresenceoftheconstraints.F inally,for dF ,theex p licitexpressionscorrespondtotheremainingpartofthe Bianchiidentitiesthatarenotalgebraicallysoluble.(Forclari“cation,seesubsecs.4.4.be,wheretheexpressionsareworkedoutindetail.)We“nd: p =0: A =1 2 ( A + A ); p =1: A= i1 2 D A A a=1 2 [ D€, D] A ; p =2: A= A€=0, A b= iC A€, A a b=1 2 ( C€€D( A )+ C D(€ A€ )); p =3: A= A€= A c=0, A€ c= T€ c A A b c= Š C€€C ( D ) A A a b c= d a b c[ D€, D] A ; p =4: A= A€= A€€= A d= A€ d= A€ c d=0, A c d=2 C€€C ( C ) A A b c d=2 a b c d D€ A A a b c d=2 i a b c d( D2 A Š D2 A ),(4.4 .12) whereforeven p D€ A =0,and forodd p A = A .

PAGE 197

4.4.Superforms185Forexample,inthecaseof thevectormultipletofs ec.4.2,wefoundthevector representationpotentialsAgivenbythecase p =1above,with A = V (inthe v ector representation,andinthegaugewhere= =1 2 V ;then K =1 2 (+ ),asgiven aboveby p =0); the“eldstrengths FABby p =2above,with A= W= i D2DV (by table4.4.1);andtheremainingBianchiidentity dF on Wby p =3above,with A =1 2 ( DW+ D€ W€)( againbytable4.4.1; dF =0thusre ducesto A ( W)=0).Further exampleswillbederivedintheremainderofthissection.(Notethatanaction writtenintermsofasuper0-formdoes not describethe most general chiralmu ltiplet theory:T he“eldstrength FA= DAalwayshasth einvariance = k ,where k isa realconstant.Here F = [ i ( Š )]=0for = k .Thisinvar ianceexcl udesmass terms,and hasconsequencesevenforthefreemasslessmultipletwhenitiscoupledto supergravity.) b.Vectormultiplet Asanintroduction,wedescribetheabelianvectormultipletinthelanguageof superforms.Webeginwitharealsuper1 Š form 1= + € €+ a a,(4. 4.13) withgaugetransformation 1= dK0,where K0isa0 Š form(scalar).The“eld strengthisasuper2 Š form F2= d 1,withsuper “eldcoecientsthatfollowfrom (4.4.11): F = D( ), F ,€= D €+ D€Š i €, F b= D bŠ b, F a b= Š1 2 C€€( € )€+ h c ..(4.4 .14) Weimposeaconvent ionalconstraint F ,€=0whicha lgebraicallydetermines€.We furtherrestricttheformbyim posingthec onstraint(4.2.44) F =0.Theso lutionto theconstraintsis

PAGE 198

1864.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS= iD, €= Š i D€ a= Š i ( D €+ D€),(4.4 .15) andtheprepotentialtransformsas = Š iK0+ .(4.4 .16) Thetransformationsareaninvarianceof1introducedbysolvingtheconstraints. Itisalwaysobvious,byexaminingequationssuchas(4.4.14),whatconventional constraintscanbeimposed.Findingadditionalconstraintsismoredicult.Ingeneral, ifwewishtode scribeamultipletthatcontainsa componentp -form,werequirethatit bethe =0compon entofasuper p -formcoecientwithonly v ectorindices (e.g.,in (4.4.14) F a b| istheYang-Mills“eldstrength),andthereforewewillnotconstrainthis coecient.Forthesamereasonweassigndimension2tothiscoecient,andthisdeterminesthedimensionofthesuperform.Asaconsequence,coecientswithmorethan twospi norindiceshavetoolowdimensiontocontaincomponent“eldstrengths(orauxiliary“elds),andmustbeconstrainedtozero.Wealsoconstraintozerocoecients thatcontainatthe =0levelcomponentfo rmsthatarenotpresentinthemultiplet. c.Tensormultiplet c.1.Geometricformulation Theantisymmetric-tensorgaugemultipletcontainsamongitscomponent“eldsa second-rankantisymmetrictensor(2-form).Todescribeitinsuperspaceweconsidera super2-form2: Š 2=1 2 / / \ \ + €/ / \ \ ,€+ b/ / \ \ b+1 2 b/ / \ \ aC€€( )+ h c ., (4.4.17) wherew ehaveusedthesymmetriesoftowrite a b= C€€( )+ h c ..Thegauge variations 2= dK1are = D( K ), ,€= D K€+ D€KŠ iK€,

PAGE 199

4.4.Superforms187 b= DK bŠ bK, ( )= Š1 2 ( €K )€.(4. 4.18) The“eldstrengthsfollowfromthede“nition(4.4.11): F , =1 2 D( ), F ,€= D( )€+ D€ + i ( )€, F , c= D( ) c+ c , F ,€ c= D € c+ D€ c+ c ,€Š iC€€( )Š iC(€€ ), F b c= C€€( D( )Š1 2 ( € )€ )+ C( D(€€ )+1 2 (€€ ), ), F a b cŠ a b c dF d= Š i ( C€€CF€Š CC€€F€), F€= Š i ( € (€€ )Š €( )).(4.4 .19) wherewehaveused(3.1.22). Wecanimposet woconv entionalconstraints.The“rst, F ,€=0,(4. 4.20) gives ( )€= i [ D( )€+ D€ ],(4.4 .21) whichimplies €= iC €+ i [ D ,€+1 2 D€ ],(4.4 .22) foranarbitraryspinor €.Thes econdconventionalconstraint, F( ,€ )€=0,(4. 4.23) gives

PAGE 200

1884.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS( )= Š i1 4 [ D( € )€+ D€( )€Š ( € )€],(4.4 .24) whichimplies ( )=1 2 D( )Š1 2 D2 Š1 2 i ( € ),€.(4. 4.25) Thepotential ,€ispuregauge:Itcanbegaugedtozerousing(4.4.18).Toeliminate theremaini ngunwantedphysicalstateswechoosetwoadditionalconstraints F , = F , c=0.(4. 4.26) The“rstimplies ispuregauge,andth es econdimposes D €=0, D€=0.(4. 4.27) Inthegauge = ,€=0,allofABisexpr essedintermsof;thusthesuper“eld isthechiralspinorprepotentialthatdescribesthetensorgaugemultiplet. Theconstraintsalsoimplythatallthenonvanishing“eldstrengthscanbe expressedintermsofasingleindependent“eldstrength G = Š1 2 ( D+ D€ €).(4.4 .28) Forexample, F ,€ c= i €€G = T€ cG .(4. 4.29) G isalinearsuper“eld: D2G =0 .I ti si nv ar ia nt undergaugetransformationsoftheprepotential = i D2DL L = L .(4. 4.30) Projectingthecomponentsofwehave: =| t=1 2 D( )| =( )| A+ i B= Š D| = D2| (4.4.31) Thecomponentsofthegaugeparameterthatenter are: L= i D2DL | ,

PAGE 201

4.4.Superforms189L(1)= D D2DL | = L(1), L( )= i1 2 D( D2D )L | =1 2 ( €[ D ), D€] L | Š1 2 ( €L )€, L€= L€.(4. 4.32) Thecomponents andBcanbealgebraicallygaugedawayby Land L(1)respectively, whereas L€istheparameteroftheusualgaugetransformationforthetensorgauge “eld t.Thespinor isthephysicalspinorofthetheory(uptotermsthatvanishin theWZgauge).Thegaugeinvariantcomponentsarefoundbyprojectingfromthe“eld strength G : A= G | = DG | =1 2 ( Š i € €), f a= F a| =[ D€, D] G | = i ( €tŠ € t€€), D2G = D2G =0.(4. 4.33) Sincethereisonlyonephysicalspinorinthemultiplet, G hasdimensi onone.This determinesthekineticactionuniquely: Sk= Š1 2 d4xd4 G2.(4. 4.34) Thecorrespondingcomponentactionis Sk= d4x [1 4 A A+1 4 ( f a)2+ €i €].(4.4 .35) Notethatnoneofthe“eldsisauxiliary.Thephysicaldegreesoffreedomarethoseof thescalarmultiplet .Onshe ll,theonlydierenceisther eplacementofthephysical pseudoscalarbythe“ eldstrengthofthean tisymmetrictensor.

PAGE 202

1904.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSc.2.Dualitytransformationtochiralmultiplet Wecanwritetw o“rsto rderactionsthatareequivalentto Sk.Intro ducingan auxiliarysuper“eld X ,wed e“ne S k= d4xd4 [1 2 X2Š GX ]. (4.4.36a) Varying X andsubstitutingtheresultbackinto S k,wereobtain Sk.Wealsoseethat thetensormultipletis classicallyequivalenttoachiralscalarmultiplet:Varying,we obtain D2DX =0,whichis solvedby X = + D€ =0 Substitutionbackinto S kyieldstheusualkinetica ctionforachiralscalar (b ecause ischiraland G islinear, d4xd4 G =0).B ecausethesame“rstorderactioncanbeusedtodescribethetensormultipletandthechiralscalarmultiplet,wesaythattheyare dual toeachother. Alternativel y,wecanwrite S k= d4xd4 [ Š1 2 X2+( + ) X ].(4.4 .36b) Varying X andsubstitutingtheresultbackinto S k,weobtaint heusualkineticaction forthechir alscalar ;varying ,we “nd D2X = D2X =0,whichis solvedby X = G Substitutionbackinto Syields Sk(4.4.34). Thetensormultipletadmitsarbitrary(nonrenormalizable)self-interactionswitha dimensionalcouplingconstant : S = 2 d4xd4 f ( Š 1G ).(4.4 .37) Thecomponentactioncontainsquarticfermi onself-interactionsandYukawaterms €F€,multi p liedbyderivativesof f ( Š 1A ).Remarkably,wecanperformthedualitytransformationtoachirals calarmultipletevenintheinteractingtheory.The“rst orderactionequivalentto S is: S= 2 d4xd4 [ f ( X ) Š Š 1( + ) X ].(4.4 .38) Varying ,we “nd X = Š 1G (thenormalizationcanbechosenarbitrarily),andreobtaintheinteractingaction(4.4.37).Varying X ,we “ndthedualactionintermsof :

PAGE 203

4.4.Superforms191S= 2 d4xd4 IK (Š 1( + ))(4. 4.39) where IK isthe Le gendretransform of f : IK (Š 1( + ))= f ( X( Š 1( + ))) Š Š 1( + ) X (Š 1( + )), f ( X ) X Š 1( + ).(4.4 .40) Thedualaction(4.4.39)isrecognizableastheactionforanonlinear -model(seesec. 4.1.b,e.g.(4.1.23)). Wecanalsop erformthereversedualitytransfo rmation,thatis,startwithatheorydescribedbyachiralscalarsuper“eldand“ndanequivalenttheorydescribedbya tensormultiplet.Althoughwecan“ndthemodeldualtoanarbitrarytensormultiplet model,thereverseisnottrue:Forachiralscalarmodel,possiblywithinteractionsto otherchiraland/orgaugemultiplets,wecan“ndthedualtensormodelonlyiftheoriginalactiondependsonlyon + ,orequivalen tly,de “ning eŠ 1,on .Thus, startingwithanaction S= 2 d4xd4 IK (Š 1( + ))(4. 4.41) wecanwrite the“rstorderaction S= 2 d4xd4 [ IK ( X )+ Š 1GX ](4. 4.42) Varying G yields X = Š 1( + )and(4.4 .41),whereasvarying X leadsto(4.4.37), wherenow f isthe(inverse)Legendretransformof IK : f (Š 1G)= IK ( X( Š 1G ))+ Š 1GX (Š 1G), IK ( X ) X = Š Š 1G .(4. 4.43) Wecannow “ndasecondtensormultipletmodeldualtothe free chiralsc alarmultiplet.Web eginwith S= 2 d4xd4 = d4xd4 eŠ 1( + )(4.4.44)

PAGE 204

1924.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSWewritethisin“ rstorderformas S imp= 2 d4xd4 [ eXŠ GX ],(4.4 .45) and “ndthedualaction Simp= Š 2 d4xd4 GlnG ,(4. 4.46) wherenow G hasanonvanishingclassicalvacuumexpectationvalue.Thisdualityholds eveninthepresenceofsupergravity,wheretheequivalenceistothesuperconformal formofthescala rmulti plet( ),asopposedtothe( + )2formobtainedfrom (4.4.36);ingeneralcurvedsuperspace,thesetwoLagrangiansaredierent.Themodel describedbytheaction Simp(4.4.46)iscalledtheimprovedtensormultiplet,because, unliketheunimprovedaction(4.4.34), Simpisconformallyinvariant.(Bothareglobally scaleinvariant,buttheactionforananti symmetric tensorbyitselfisnotinvariant underconformalboosts.) Itisinterestingtostudywhathappenstotheinteractionsofachiralmultiplet afteradualitytransformation.Hereweconsiderinteractionswithagaugevectormultiplet(forotherexamples,seesecs.4.5 e,4.6,5.5).Foranactionoftheform Sgauge= d4xd4 IK ( + + V)+ d4xd2 W2(4.4.47) where V isanabeliangaugesuper“eld, W isits“eldstrength,and IK ( + + V) IK ( ln( eV )),wecanwritethe“rstorderaction S gauge= d4xd4 [ IK ( X + V )+ GX ]+ d4xd2 W2.(4. 4.48) Varying G gives(4.4.47);varying X gives S G= d4xd4 [ f (G) Š GV ]+ d4xd2 W2.(4. 4.49) Thusthegaugeinteractionsoftheoriginaltheoryaredescribedbythesingleterm GV inthedualtheory(thiscouplingisgaugeinvariantbecause G islinear).Observethat fortheusualkineticterm IK = eV ,the dualtheoryhasthei mprovedLagrangian Š GlnG Š GV (4.4.46)ratherthan Š1 2 G2Š GV (4.4.34).Itisstraightforwardtoverify

PAGE 205

4.4.Superforms193thatthelattertheorydescribesamassivevectormultipletratherthanascalarcoupled t oavector.Anoth erwaytodescribeamassivevectormultiplet,butwithoutvector “elds,isintermsofthechiralspinoralonebyaddingamassterm(whichbreaksthe gaugeinvariance(4.4.30))to Sk(4.4.34): Sm= Š1 2 m2 d4xd2 ()2+ h c ..(4.4 .50) Sk+ Smdescribesamassivevector mult iplet.Thecomponentantisymmetrictensor describesamassivespin1“eld, and describeamassiveDi racspinor,Aisamassi ve sc alar,andBisauxiliary. d.Gauge3-formmultiplets d.1.Real 3-form Webeginb yconside ringa real 3-form.Itha sthefo llowingi ndependentcoecientsuper“elds , , ,€, , c, ,€ c, ,( ), ,(€€ ), a,(4. 4.51) wherew ehaveusedthesymmetriesoftowritei tinte rmsofLorentzirreduciblecoecients. b c= C€€ ,( )+ C ,(€€ ), a b c= Š a b c d d= Š i ( C€€C€Š CC€€€),(4.4 .52) Theindependent“eldstrengthsare F , , F , ,€, F ,€ ,€, F , d, F ,€ d, F ,( ), F ,(€€ ), F ,€ ,( ),

PAGE 206

1944.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSF b, F .(4. 4.53) Thelast“ve“eldstrengthsareLorentzirreduciblecoecients,e.g.(see3.1.22)), F a b c d= F a b c d= i ( CCC€€C€€Š CCC€€C€€) F .(4. 4.54) Weimposethefo llowingco nstraintsonthe“eldstrengths: F , = F , ,€= F ,€ ,€=0, F , d= F ,€ d= F ,(€€ )= F ,€ ,( )=0, F ,( )=2 C ( C ) ,(4.4 .55) where i sa n undeterminedgaugeinvariantsuper“eld.Solvingtheconstraintsgives , = ,€= , c= ,(€€ )=0, ,€ c= iCC€€V ,( )= Š C ( D )V €=[ D€, D] V V = V ,(4. 4.56) uptoapuregaugetr ansformationofABC.Giventhesol ution,we“nd = D2V .(4. 4.57) Theprepotential V has gaugetransformations V = Š1 2 ( D+ D€ €), D €=0.(4. 4.58) Thephysicalcomponent“eldsofthismultipletare = | = D2V | = D | = D D2V h =( D2+ D2 ) | = { D2, D2} V | f = Š i ( D2 Š D2 ) | =1 2 a a| =1 2 €[ D€, D] V | .(4. 4.59)

PAGE 207

4.4.Superforms195Thequantity f isthe“eldstrengt hofthecomp onentgaugethree-form l€=€| .The componentthree-formtransformsas(cf.(4.4.33)for f a) l€= i1 2 ( €D( )Š € D(€ € )) | ,(4. 4.60) sothatits“el dstrength f isinvariant. The“eldstrengthisachiral“eldofdimensionone(determinedby ),and hencethekineticactionis S = d4xd4 .(4.4.61) Itgivesconventionalkinetictermsforthecomponents and ;thescal ar“eld h isan auxiliary“eldandthegauge“eld l€enters theactionthroughthesquareofits“eld strength f Suc ha “e lddoesnotpropagatephysicalstatesinfourdimensions. Theonlydierencebetweenthismultiplet,describedby,andtheusualchiral scalarmultip letisthereplacementoftheimagina rypart(thepseudoscalar“eld)of the F auxiliary“eldbythe“eldstrengthofthecomponentgaugethree-form.Massand interactiontermsforcanalsobeusedfor.However,atthecomponentlevel,after e liminationoftheauxiliary“eldsthetheoriesdier:Wenolongerobtainalgebraic equations,since f isthederivativeofanother“eld l€.Another dierenceisthatthe superthree-form gaugemultiplet ca nnot be co upl ed toYang-Millsmultiplets. d.2.Comple x3-form Acomplexsupert hree-formmult ipletcanbetreatedinthesameway.Ithasmore i ndependentcoecientsuper“elds: , , ,€, ,€ ,€,€ ,€ ,€, , c, ,€ c,€ ,€ c, ,( ), ,(€€ ),€ ,( ),€ ,(€€ ), a.(4. 4.62) (Forexample, (€€ ) = € ( ).)Correspondi ngly,therearemoreindependent“eld

PAGE 208

1964.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSstrengths.Theseare F , , F , ,€, F ,€ ,€, F ,€ ,€ ,€, F€ ,€ ,€ ,€, F , d, F ,€ d, F ,€ ,€ d, F€ ,€ ,€ d, F ,( ), F ,(€€ ), F ,€ ,( ), F ,€ ,(€€ ), F€ ,€ ,( ), F€ ,€ ,(€€ ), F b, F€ b, F ,(4. 4.63) Theconstraintshowever,setmore“eldstrengthstozero.Thenonzeroonesare F ,( ), F b, F ,(4. 4.64) andwestillimposetheconstraint: F ,( )=2 C ( C ) .(4.4 .65) Theonlyformcoecientsthatarenotpuregaugearegivenby ,( )= Š C ( D ) D€ €, € ,(€€ )= Š C€ (€ D€ ) D€ €= C€ (€ D2 € ), ,€ c= iCC€€ D€ €, a=[ D€, D] D€ €,(4. 4.66) (uptoarbitrarygaugetransf ormationterms).Theseexpre ssionsallowustocompute; we “ndthatitisexpressedint ermsoftheprepotentialasfollows: = D2D, D€=0.(4 .4.67) Thegaugetransformationsoftheprepotentialare =+ DL( ), D€=0.(4. 4.68) Thecomponentscontainedinthe“eldstrengthare A = | = D2D| ,

PAGE 209

4.4.Superforms197= D | = D D2D| f = D2 | =i 2 €[ D€, D] D| ,(4. 4.69) where f isthe“eldstrengthofa complex 3-form. Thismultipletcanbedescribedintermsoftworealsuper3-formmultiplets: =1+ i 2.Theco nstraintsimp osedabovearetheonesgiveninsec.4.4.d.1for1and2,plusthea ddition alconstraint F€ ,€ ,(€€ )=0.Thisi ssimplyt heconstraint 1+ i 2= D2( V1+ iV2)=0,whic himp lies V1+ iV2= D€ €. The“eldstrengthischiralandofdimensionone.Thereforealloftheactionformulaefortheusu alchiralscalarcanbeusedfor.Asfortherealgaugethree-form mu ltiplet,theequationsofmotionfortheauxiliary“eldsarenolongerpurelyalgebraic. Again,thismultipletcannotbecoupledtoYang-Millsmultiplets. e.4-formmultiplet The“nalsuperformweconsiderhas no physicaldegrees offreedom.Itis describedbyarealsuper4-formABCD.The “eldstrengthsupertensorisasuper5-form FABCDE.Therefo rethe“eldstrengthwithall“vevectorindicesvanishesbyantisymmetry. Asconstraintsweimposetheequations FABCDE=0.Thisi mpliesthatallof ABCDispuregauge.Sinceall“eldstrengthsvanish,nogaugeinvariantactionispossibleattheclassicallevel.However,thismul tiplet(andthecorre spondingcomponent form)hassomeunusu alpropertiesatthequantumlevel,becauseitsgauge“xingtermis notzero.

PAGE 210

1984.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS4.5.Othergaugemultiplets a.GaugeWess-Zuminomodel In(3.5.3)wenotedthatachiralsuper“eldcanbeexpressedintermsofan unconstrainedsuper“eld = D2.(4.5 .1) The“eldprovidesanalternatedescriptionofthescalarmultiplet.Theactionswe consideredinsecs.4.1-2canbeexpressedintermsof.Forexample,theWess-Zumino action(4.1.1-2)becomes S = d4xd4 [( D2 )( D2)+1 2 m ( D2+ D2 ) + 3! (( D2)2+ ( D2 )2)],(4.5.2a) wherew ehaveused d4xd2 ( D2)2= d4xd4 D2,(4.5 .2b) etc. Thesolution(4.5.1)ofthechiralityconstraintintroducestheabeliangaugeinvariance = D€ €(4.5.3) where isanunconstrainedsuper“eld.Thegaugeinvariantsuper“eldisthechiral “eldstrengthofthe gaugesuper“eld,andtheactionisobviouslyinvariant.The gaugetransformationcanbeusedtogotoaWZgauge,byalgebraicallyremovingallthe componentsofexceptthosethatappearin.Inthisformulationthecouplingto su pe rY ang-Millscanbeachievedbycovariantizingthederivatives:Ifiscovariantly chiral,then= 2, = €€.U nde rY ang-Millsgaugetransformationstransformsinthes amewayas.

PAGE 211

4.5.Othergaugemultiplets199b.Thenonminimalscalarmultiplet Thismultiplethasanumberofinterestingfeatures:(a)Itisamultipletwhere thespinofauxiliary“elds ex ceeds thatofthephysical“elds ;(b) noneofthecomponent “elds(inaWess-Zuminogauge)ofthismultipl etaregauge“elds,eventhoughthemultipletisdescribedbyagaugesuper“eld;(c)thismultiplet,unlikeotherscalarmultiplets, formsa reduci ble representationofsupersymmetry. Weintro duceageneralspinorsuper“eldwiththegaugetransformation = DL( ), L( )ar bitrary.Anactionthatisinvariantunderthisgaugetransformationis S = Š d4xd4 ,= D€ €,(4. 5.4) The“eldstrengthsatis“es D2=0,soth atitisa complex linearsuper“eld;incontrast,the“eldstrengthofthetensorgau gemultipletisareallinearsuper“eld. Thecomponent“eldsofthemultipletare A = | €= D€ | = D | P€= D€D | F = D2 | €=1 2 D D€D | .(4. 5.5) Thecomponentactionis S = d4x [ A A + €i €Š| F |2+2 | P€|2+ + € €],(4.5 .6) withpropagatingcomplex A and .A lltheother“eldsareauxiliary. Intermsofsuper“eldswecanseethattheaction(4.5.4)describesascalarmultiplet.Theconstraintand“eldequationsfor are: D2 =0, D€ =0.(4 .5.7) Thesearethesameasthosefortheon-shellchiralscalarmultiplet,butwithconstraint and“eldequationinterchanged.

PAGE 212

2004.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSTo se et he reducibilityofthismultipletweusethesuperprojectorsofsec.3.11. Theactioncanbewritten S = d4xd4 €i €[21,0+(2,1 2 ++2,1 2 Š)], 1,0= Š1 2 Š 1D D2D, 2,1 2 = Š 1 D2D1 2 ( D D€ €).(4.5 .8) Thusthemultipletconsistsofthreeirreduciblesubmultiplets:oneofsuperspin0,and twoofsuperspin1 2 Incontrasttothechiralscalarmultiplet ,itisnotpo ssibletointroducearbitrary massandnonderivativeself-interactionterms.However,wecanwritedowntheaction S = d4xd4 f (, ),(4.5.9) where f ( z z )= f ( z z ).Thus,forexample,itispossibletoformulatesupersymmetric nonlinear -modelsintermsofthenonminimalscalarmultiplet.Furthermore,thenonminimalmultipletcanbecoupledtoYang-Millsmultipletsbycovariantizingthederivatives:= *** Justasforthetensormultiplet(sec.4.4.c),wecanexhibitthedualityofthenonminimalscalarandchiralmultipletsbywritinga“rstorderaction.Mostofthediscussionofsec.4.4.c.2hasananalogforthenonminimalscalarmultiplet,except,sincethe mult ipletisdescribedbyalinearsuper“eld,theLegendretransformistwodimensional andhencethereisnorestrictionontheformofthenonlinear -modelthatcanbe described.Thetwo“rstorderactionsequivalentto(4.5.9)are(see(4.4.38,42)): S= d8z [ f ( X X ) Š X Š X ], D€= 0, (4.5.10a) and

PAGE 213

4.5.Othergaugemultiplets201S= d8z [ IK ( X X )+ X + X ], D2=0,(4.5 .10b) where X isacomplex unconstrainedsuper“eldand IK istheLeg endretransformof f Justasforthetensormultip let,thisdualitytransformationcanbeperformedevenin thepresenceofinteractionswithothermultiplets(e.g.,supergravity). c.Morevariantmultiplets Aswehaves een,severalinequivalentsuper“eldformulationscandescribethe samesetofphysicalstates.The(0,1 2 )multi pletcanbedescribedbyachiralscalar,a gaugetwo-form,real(orcomplex)gaugethree-forms,oragaugespinor.Thechiral scalarprovidesthesimplestrepresentation.Allbutoneoftheotherrepresentationsare obtainedbyreplacingeitherthephysicalorauxiliary“eldbycomponent2-formsor 3-formsrespectively.Wecallthesevariantrepresentationsofthescalarmultiplet.In general,variantrepresentationsareveryrestr ictedineithertheirself-interactionsorcouplingstoothermultiplets.Inthissubsectionwediscussvariantvectorandtensormultiplets. c.1.Vectormultiplet We ha ve de sc ribedsuperYang-Millstheoriesintermsofahermitiangaugeprepotential V .Itcontainsaco mponentvectorasitshighestspincomponent: A a=1 2 [ D€, D] V | .Thereis,howev er,asmallersuper“eldthatcontainsacomponent v ector:Achiral dotted spinor€( D€€=0),hasa sitshighests pincomponent A aŠ i ( D€ + D€) | .(4. 5.11) Thesuper“eld€isreducible;itcanbebi sected(see(3.11.7)1 2 (1 K K )€=1 2 (€Š+ Š 1 D2i € ).(4.5 .12) Sincewewanttodescr ibeagaugetheory,wegaugeawayoneoftherepresentations insteadofconstrainin git.Thetra nsformation

PAGE 214

2024.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS €= D€( D+ D€ €)= D2 €Š i €=(1+ K K ) D2 €, D €=0,(4. 5.13) canbeusedtogaugeaway(1+ K K )€,andleaves(1 Š K K )€inert.Thegaugeparameter D+ D€€describesthetensormultipletofsec.4.4. The“eldstrengthfor isthelowestdimensionlocalgaugeinvariantsuper“eld: W= D2(1 Š K K ) = D2 + i €€.(4. 5.14) The“eldstrength Wisthefamiliarchiral“eldstrengthofthegaugemultiplet describedby V butnowwith= ,€=€, a= Š i ( D€ + D€),and A a= a| .I tscomponentsarethesame,exceptfortheauxiliary“eldD: W| f1 2 D( W )| =1 2 ( €A )€, D 1 2 iDW=1 2 €( D€ Š D€) | =1 2 €B€.(4. 5.15) We thusseetheauxiliarypseudoscalarhasbeenreplacedbythe“eldstrengthofagauge three-form.Theactionisstill(4.2.14),and incomponentsdiersfromtheusualvector mult ipletonlybythe replacementD1 2 €B€. Thisvariantformofthevectormultipletcanalsobeobtainedfromthecovariant approachofsec.4.2:Intheabeliancase,wecansolvetheconstraint F= D( )=0 by= .Justast heusualsolution= Š i1 2 DV directlyintermsofthereal scalarprepotential“xedsomeofthe K invariance(correspondingtoagaugecondition D D2= Š D€D2 €,whichim p lies K =1 2 (+ )),thevariantsolution“xessomeof the K invariancewiththegaugecondition D=0(which,t ogetherwiththeconstraint, impliesthatisantichiral),reducingitto K = D+ D€ €. Thecovariantderivativescanbeusedtocouplethisabelianmultiplettomatter. However,= isnotasolutiontothenonabelianconstraints,nortotheabelian onesingeneralcurvedsuperspace.Thus,likeothervariantmultiplets,itislimitedin thetypesofintera ctionsitcanhave.

PAGE 215

4.5.Othergaugemultiplets203c.2.Tensormultiplet Thevariantrepresentationforthetensormultipletisdescribedbythesamechiral spinorsuper“eldastheusualone(4.4.27),butthegaugetransformationischanged. Inplaceoftherealscalarparameter L (4.4.30),weusethechiraldottedspinor€. Explicitly,themodi“edgaugevariationis(cf.(4.5.14)) = D2 + i €€.(4. 5.16) Thisleadstotheusualtransformationsfor t( )andleaves A and invariant(see (4.4.31)).Butthevariationofthecomponent“eld B = i1 2 ( D€ €Š D) | is B = Š1 2 €v€.(4. 5.17) Thereforethiscomponent“eldisagaugefour-form. Theactionforistheusualoneproportionalto G2,andthefourformdoesnot appearin G2andintheaction.However,atthequantumlevel,thefour-formwould reappearingauge“xingterms,andchiraldottedspinorswouldappearasghosts. d.Super“eldLagrangemultipliers Wehavegi venanumberof examplesofsupersymmetr ictheori esthatdescribe thescalarmultipleton shell(samephysicalstates)butareinequivalentoshell.They dierprimarilyinthetypesofinteractionstheycanhave.Sofar,wehavefoundthat thesimplestformulationofthescalarmultiplet,achiralscalarsuper“eld(or,equivalentlyevenoshell, D2onageneralscalar),hasthemostgeneralinteractions.However,inextendedsupersymmetrynoneoftheknown N =2theori esequivalenton-shell tothe N =2scal armultipletcanhavealltheinteractionsknownfromon-shellformulations.Wenowintro duceaformofthe N =1scal armultipletthati sas ubmultipletof ano-shellformulationofthe N =2scal armultiplet.Itsmostdistinctivefeatureisa super“eldthatappearsonlyasaLagrangemultiplier.Thisformulationhassomedrawbacksincommo nwiththetensormu ltiplet(anothertheoryequivalenttothescalarmultipletonshell),towhichitis closelyrelated:(1)Itdoesnothaverenormalizableselfinteractions(i.e,tho secorrespondingtotermsd4xd2 P()),(2)itisrestrictedinits couplingstosupergravity,and(3)itisnotano-shellrepresentationofthe(chiral) U (1) symmetrywhichthescalarmultiplethasonshell(correspondingto= ei ).Onthe

PAGE 216

2044.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSotherha nd,unlikethetensormultiplet(andmostv ariantformsofthescalarmultiplet), itdoescoupletoYang-Mills.However,becauseof(3),itcanonlybea real representationofanyinternalsymmetrygroup,andcouplestoYang-Millsaccordingly(e.g.,itcan coupletoa U (1)vectormultipletonlyasadoubletofoppositecharges). Theformulationisdescribedbyageneralspinorgaugesuper“eldwithatermin theactionlikethatofthechiralspinorgaugesuper“eldofthetensormultiplet,anda realscalarsuper“eldLagrangemultiplierwithatermintheactionthatconstrainsto zerothesubmultipletsintheformertermthatdontoccurinthetensormultiplet. Explicitly,theactionis S = Š d4xd4 (1 2 F2+ YG ), F =1 2 ( D+ D€ €), G = i1 2 ( DŠ D€ €);(4.5 .18) withgaugeinvariance = DL( )(4.5.19) intermsofageneralsuper“eldgaugeparameter.TheBianchiidentitiesand“eldequationsare: Bianchiidentities : D2( F Š iG )= 0, (4.5.20a) fieldequations : D€( F + iY )= G =0.(4. 5.20b) IfwemakeadualitytransformationbyswitchingtheBianchiidentitieswiththe“eld equations,weobtaintheusualformulationofthescalarmultiplet,withtheidenti“cations F =1 2 (+ ), Y =1 2 i ( Š ), G =0.(4. 5.21) Intermsofirreduciblerepresentationso fsupersy mmetry,thistheorycontains superspins1 2 +1 2 +0inand1 2 +0in Y .There presentationsin Y setthecorrespondingonesintozeroonshell,leavingtheremainingoneasatensormultiplet. However,unlikethetensormultiplet,thephysicalspinzerostatesareallrepresentedby scalars:Thevectorobtainedbyprojectionfrom[ D€, D] F isanunconstrained

PAGE 217

4.5.Othergaugemultiplets205auxiliary“eld,appearingat 2 orderin,whereasthecorre spondingvectorinthetensormultipletisthetransverse“eldstrengthofthetensorappearingat orderi nthe chiralsp inor.Thist heoryh asthesamecomponent-“eldcontentasthenonminimal sc alarmultipletplusanauxiliaryrealscalarsuper“eld. CouplingtoYang-Millsisstraightforward;however,sincebothand appearin F andin G mu sttransformundera real representationoftheYang-Millsgroup.We covariantizebyreplacingthespinorderivativesinthede“nitionsof F and G byYangMillscovariantspinorderivatives.Invaria nceoftheactionundertheYang-Millscovariantizationof(4.5.19)thenrequires 0= = L( )= Š i1 2 FL( ),(4. 5.22) implyi ngthesamerepresentation-preservingconstraint F=0asforthech iralscalar formulation.ThetotalsetofBianchiidenti tiesand“eldequationsisthesameasforthe chiralsc alar. Justasthereisanimprovedformofthete nsormult iplet,withsuperconformal invariance,thereisanimprovedformofthiss calarmultiplet.Inanalogytothetensor mult iplet,itisobtainedbyreplacing1 2 F2in(4.5.18)with FlnF (cf.(4.4.46)).Furthermore,the“rst-orderformulationofthismultipletturnsouttobeequivalenttothe“rstorderformulationofthe nonminimal scalarmultiplet.Westartwith(cf.(4.4.45)) S = d4xd4 [ eXŠ XF Š YG ].(4.5 .23) The“rstorderformofthenonminimalscalarmultipletisusuallywrittenas(4.5.10b): S = d4xd4 [ X XŠ ( XD + h c .)].(4 .5.24) Uponeliminationofthecomplexscalar X,thisgivesth es econd-orderformofthenonminimalscalarmultiplet(4.5.4).Uponeliminationofthespinor ,weobtai nthe constraint D€X=0,whose solution X=inte rmsofachiralscalargivestheminimal scalarmultiplet(provingtheirduality).Theequivalenceoftheactions(4.5.23)and (4.5.24)followsfromthechangeofvariables = X Š 1, X= e1 2 ( X + iY ),(4. 5.25)

PAGE 218

2064.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS(someintegrationby partsisnecessarytoshowtheequivalence). e.Thegravitinomattermultiplet Thusfarwehaveconsideredsupermultipletswithphysical“eldsofspinoneor less.Weconcludeourdiscussionofglobal N =1mult ipletsbyconsideringonewitha spin1and aspin3 2 (gravitino)component “eld.Itispossiblet odiscu ssitwithout introducingsupergravityonlyifthemultipletdescribesafreetheory.Thegravitino mult ipletisofinterestb ecausemanyofthefeaturesencounteredinthesuper“eldformulationofsupergravity,sucha si rreduciblesubmultiplets,compensators,andinequivalent o-shellformulations,arealreadypresent. e.1.O-shell“eldstrengthandprepotential Fo llowingthediscussionofsec.3.12.awedes cribethismultipleton-shellwithcomponent “eldstrengths (v ector“eldstrength)and (theRarita-Schwinger“eld strength ),totallysymmetricintheirindices.Wedenotetheo-shellsuper“eldstrength correspondingto by W.Itisachiral“e ldstrengthofsuperspin1,nobisectionis po ssible( s +1 2 N =3 2 isnotaninteger),andthereforewewrite(see(3.13.1)) W=1 2 D2D( ),(4. 5.26) intermsofageneralspinorsuper“eld.From dimensionalanalysis(thegravitino“eld hascan onicaldimension3 2 ),thedimensionof W is2. Thegaugetransformationsthatleave W invariantare =+ D, D€=0, = ,(4.5 .27) Toan alyzethetransformationsofthecomponents,wede“ne t= W| =1 2 D2D( )| = D( W )| =1 2 D( D2D )| = i1 2 ( €1 2 [ D€, D] )| = DW| =1 2 D D2D( )| ,

PAGE 219

4.5.Othergaugemultiplets207f= D2W| =1 2 D2 D2D( )| = i1 2 ( €D2 D€ )| .(4. 5.28) Weidentify thegravitinoand(comple x)v ector“eldstrengths, and f.Theco rrespondingcomponentgauge“eldsappearinandaregivenby a = i1 2 [ D€, D]| A a= iD2 D€| ,(4. 5.29) Theirgaugetransformationsare a =[1 2 a( D Š )+ iC D€D2] | A a= Š aD2 | .(4. 5.30) Thegravitino“eld,inadditiontoundergoingaRarita-Schwingergaugetransformation describedbythe“rstterm,alsoistranslatedby iC €, €= D€D2 | .(When coupled to N =1superg ravitytogive N =2superg ravity,thetransformation(4.5.30)ispartof the N =2superconfo rmalgroup:The“rsttermbecomesthesecondlocal Q -supersymmetrytransformation,thesecondterm,thesecond S -supersymmetrytransformation.) Werefertothism ultipletastheconforma lgravitino mult iplet. Themultipletofcomponent“eldsof Wisirreducibleandgaugeinvariantand shouldappearinthegaugeinvariantaction.However,intheabsenceofdimensional constantswecannotwriteafreeactionofthecorrectdimensionintermsof W .Wecan writeanonlocalgaugeinvariantactionintermsof,e.g., S =1 2 d4xd4 €i €1,1+ h c .,(4.5 .31) (1,1isthecorrectprojectorontothephysicalgaugeinvariantrepresentation)butit leadstoanonlocalcomponentaction. To “ndalocalaction,weaddmorerepresen tations.Clearlyremovingfromthe actionrestoreslocalitybutintroduces all therepresentationsinanddestroysthe gaugeinvariance.Wecan,however,restrictth erepresentationswh ichappear.Webegin withthegeneralexpression S =1 2 d4xd4 €i €(i cii)+ h c .,(4.5 .32) withthesumrunningoverallprojectors(3.11.38,39),andchoose citoobtainalocal

PAGE 220

2084.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSaction.Werequirethatthesuperprojector1,1bepres ent,and“ndtwosolutionscontainingtheleastnumberofadditionalsuperprojectors.Oneusesthesuperprojectors Š 1,1Š 20,1 2 ŠŠ 2,1 2 ,+theotheruses Š 1,1Š 20,1 2 Š+1,0.(Theoverall signischosentogivethephysicalvectorthecorrect kineticterm.)Theresultingactionsare (1) S(1)= d4xd4 [ Š ( D€)( D €) Š1 2 ( D2+ h c .)+1 4 ( D+ D€ €)2] (2) S(2)= d4xd4 [ Š ( D€)( D €) Š1 2 ( D2+ h c .)].(4 .5.33) However,thegaugegroupisnolongerdescribedby(4.5.27);theinvariancegroupsassociatedwiththeactionsabovearesmallerthantheinvariancegroupof W;theyare (1) = i D2DK1+ DK2, Ki= Ki, (2) =1 + D( D2 + D€ 2€), D€i =0,(4. 5.34) forthetwoactions. Ascomparedto(4.5.27),i nthe “rstcasetheinvariancegrouphasbeenreduced b ecauseisrestrictedto thesp ecialform i D2DK1, K1= K1,andisres trictedto bereal.Inthe secondcaseremainsunrestrictedbutisrestrictedtotheform D2 + D€ 2€.Inbothc asesthe“nalgaugegrouphasfewerparametersthanthe originalone.However,formanypurposes(e.g.,quantization),weneedtousetheoriginalgaugegroup;todothis,weintroducec ompensatingmultiplets(seesec.3.10). e.2.Compensators Forthegravit inomulti plet,twoinequivalentsetsofcompensatorscanbeintroduced.Wedothisby nonlocal “eldrede“nitionsofthebasicgaugesuper“eld.Thus,for thetwoloca lactionswemaketh ered e“nitions (1) + Š 1(1 2 D2W+ D2DG ), W= i D2DV V = V G =1 2 ( D+ D€ €), D€=0,

PAGE 221

4.5.Othergaugemultiplets209(2) + Š 1(1 2 D2W+ D D2 ), D€=0.(4 .5.35) Theyinducethefollowingchangesintheactions (1) S(1) S(1)Š d6zW2+ d8z [ Š ( W+ W€ €)+ G2Š G ( D+ D€ €)], (2) S(2) S(2)Š d6zW2+ d8z [ Š ( W+ W€ €) Š 2 Š ( D+ D€ €)].(4.5.36) Althoughtherede“nitionsarenonlocal,theactionsremainlocal.(Actually,incase(1) wecanalsou sesimply + i1 2 DV +.) Intheabove“eldrede“nitionsweintroducedavectormultiplet V andeithera tensormultipletorachiralscalarmultiplet.Thesechoicesareare”ectionofthe representationsthatwereintroducedbytheadditionalprojections:Avectormultiplet 0,1 2 Š,atensormu ltiplet2,1 2 +,andac hiralscalarmultiplet1,0.Inthe presenceofthecompensatingmultipletsthegaugevariationofisgivenby(4.5.27).The compensatingmu ltipletstransformasfollows: (1) V = i ( Š ), = Š + i D2DK3, (2) V = i ( Š ), = Š D2 ,(4.5 .37) Sincetheyarecompensators,theycanbeal gebraicallygaugedtozero.Intheresulting gauge,thetransformations(4.5.27)ofarerestrictedbackto(4.5.34). Thetwoinequivalentformulationsofthegravitinomultiplet,oneusingatensor mult ipletcompensatorandtheotherusingachiralscalarcompensator,leadtodierent auxiliary“eldstructuresatthecomponentlevel.IntheWess-Zuminogaugeforcase(1) thecomponentso fthegravit inomulti pletare

PAGE 222

2104.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS= D€D €+ D2+1 2 D( D+ D€ €)+ WŠ DG P = Š i ( D D2Š D€D2 €+ DW), V a= i ( D2 D€+ D2D €) Š a[ G Š1 2 ( D+ D€ €)], A a=( D2 D€Š D2D €), t = D( [ )+ )], t=2 W+ ( €B )€, = D2D[ G Š1 2 ( D+ D€ €)], B a=1 2 [ D, D€] V + i ( D €+ D€), a = i1 2 [ D, D€]+ i [ D D€+ D2 €+ W€].(4.5 .38) Forcase( 2)wehaveinstead = D€D €+ D2+ WŠ D, P = Š i ( D D2Š D€D2 €+ DW), J = D2( D+2 ), A a= Š i D2D €+ a, t=2 W+ ( €B )€, = D2 D2+ i €D D€Š i € D€ ,

PAGE 223

4.5.Othergaugemultiplets211B a=1 2 [ D, D€] V + i ( D €+ D€), a = i1 2 [ D, D€]+ i [ D D€+ D2 €+ W€].(4.5 .39) Incase(1),thegauge“eld t ofthetensormultipletreplacesthecomplexscalarcomponent “eld J ,w hi ch co rrespondstotheauxiliary“eldofthechiralscalarmultipletof case(2 ).Inthecomponentaction t onlyappearsas t €A €Š t€€€A €.This te rmisinvariantunderseparategaugetransformationsof t and A €.Alsoits hould benote dthatt he“eld A aisreal.Incase(2) A€ =( A€)hasno gaugetransformationsbecausethephysicalscalarsofthechir alscalarmultiplethavebecomethelongitudinalpartsof A€andcancelthetransformationin(4.5.30).Inbothcases,thephysical v ectorofthecompensatingvectormultiplethasbecomethephysicalvectorofthegravitinomultiplet,whilethephysicalspinorofthevectormultipletbecomesthespin1 2 part ofthegravitinoandcancelsthespinortranslationin(4.5.30). e.3.Duality Sincethetwoformulationsabovedierinthat(1)hasatensorcompensatorwhere (2)hasachiralcompensator,usingtheapproachofsec.4.4.c.2,wecanwrite“rstorder actionsthatdemonstratethedualitybetweenthetwoformulations.Forexample,we canstart with(1)andwrite S (1)= S(1)[, V ]+ d8z [ X2Š X ( D+ D€ €) Š 2 X (+ )].(4.5.40a) Varyingthe chiral “eldleadsto X = G andformulation(1),whereaseliminating X resultsinformulation(2).Similarly,wecanstartwith(2)andwrite S (2)= S(2)[, V ]+ d8z [ Š X2Š X ( D+ D€ €)+2 XG ].(4.5 .40b) Varyingt helinearsuper“eld G we “nd X =+ andform ulation( 2),whereaseliminating X leadsdirectlyto(1). Thereareotherinequivalentformulationswherewereplace V bythevariantvectormultipleta nd/orreplacebyeithertherealorcomplexthree-formmultiplets.This

PAGE 224

2124.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSsi mplyreplacessomeofthescalarauxiliary“eldswithgaugethree-form“eldstrengths. e.4.Ge ometricformulations Finally,wegivegeometricalformulationsofthetheories.Todescribeamultiplet thatgaugesasymmetrywithaspinorialgaugeparameter,weintroduceasuper1-form A withanadditionalspinorgroupindex.Theanalysisissimpli“ediftheirreducible mult ipletisconsidered“rst.Theirreducibletheorywasdescribedby Win(4.5.26). Todescribethismu ltipletgeometrically,weintroducemoregauge“elds(inparticulara complexsuper1-formA( = A))andenlargethegaugegroup.Whenwegettosupergravitywewill“ndthatthisprocesscanalsobecarriedout.TheretheirreduciblemultipletistheWeylmultiplet andtheenlargedgroupisthe conformalgroup.The“nal formofthe(3 2 ,1)multi pletwithmoreirreducib lemultipletsandcompensatorsisanalogoustoPoincar esupergravity. Wew illuset hewordsPoincar eandWe ylforthe (3 2 ,1)multi plettoemphasize thisanalogy. Thecompletesetofgauge“eldsandgaugetransformationsthatdescribetheWeyl (3 2 ,1)multi pletis: A = DAKŠ A L A€= DA K€Š A€ L A= DAL A= DA L .(4. 5.41) The K -termsaretheusualgaugetransformationsassociatedwithasuperformandthe L -termsaretheconformaltransformation.Recallthatwefoundthatthegaugetransformationsoftheirreducib lemultipletcontainan S -supersymmetryterm. L isthe super“eldparameterthatcontainsthesecomponentparameters.Thevectorcomponent ofthecomplexsuper1-formAisthecomponentgauge“eldwhose“eldstrength appearsin(4.5.28).The“eldstrengthsforAarethoseforanordinary(complex)vectormultiplet, butthos eforA anditsconjugate A€mustbe L -covaria nti zed: FAB = D[ AB ) Š TAB DD +[ AB ) FAB€= D[ A B )€Š TAB D D€+ [ AB )€.(4. 5.42) Wecannowimposeth econstraints:

PAGE 225

4.5.Othergaugemultiplets213F = F ,€ = F€ ,€= F a =0, F a€ + F a ,€=0, F = F ,€=0,( F€ ,€ =0).(4 .5.43) Evenwiththe L invariancethegeometricaldescript ionheredoesnotqu itere ducetothe irreduciblemultiplet W.Howev er,theseconstraintsreducethesuper1-formstothe irreduciblemultipletplusthecompensatingvectormultiplet,whicharethetwoirreducib lemultipletscommontobothformsofthePoincar e(3 2 ,1)multi plet,andthusare sucientfortheirgeneralanalysis. Theexplicitsolutionoftheseconstraintsisintermsofprepotentials ,, (complex),and V (real): = D Š ,€ = D€, a = Š i [ D€D + D D€+ (€Š D€)]; = D,€= Š D D€( Š ) Š D2( €Š €)+ D€ Š W€, a= Š iD2 D€( Š )+ a;(4.5 .44) where W= i D2DV .T heprepotentialstransformunder Kand L ,a sw ellasunder newpar ameters(complex)and(c om plex)underwhichthesareinvariant(thisis analogoustothe-groupparametersinsuper-Yang-Mills): = K+ D, = KŠ D€=0; = L Š D2, V = i ( Š ).(4. 5.45) Aswiththevectormultiplet,wecang otoachiralrepres enta tionwhere andonly appearasthecombination= Š ,with = ( Š )=+ D.(4.5 .46)

PAGE 226

2144.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSAtthispo int,bycomparisonwith(4.5.27),wecanidentifyand V withthecorrespondingquantitiesthere.TorecoverthefullPoincar etheory,wemu stbreakthe L invarian ce.Tobreakthe L invariance,weintroducea tensorcompensator Gor ,to obtainthetensor-multipletorscalar-multiplet,respectively.Thesetensor(scalars)are not prepotentials,andtransformcovariantlyunder all ofthegaugetransformations de“nedthusfar.Bycovariant,wemeanthatthesetransform without deriva tives DA. G = Š ( L + L ),(4.5 .47) = Š L .(4. 5.48) Wenowimp osethe L -covariantizedformoftheusualconstraints( D2G =0and D€=0)whichd escribetensorandchir alscalarmultiplets;1 2 D€[ D€ G +(€+ €)]+ h c .=0,(4 .5.49) D€ + €=0.(4. 5.50) Theinvarianceoftheseconstraintsfollowsdirectly(4.5.27,37,47,48).(Thehermitian conjugatetermaboveisnecessarytoavoidconstrainingitself.)Theseconstraints canbesolvedintermsofprepotentials: G = G Š (+ ) Š1 2 ( D+ D€ €),(4.5 .51) = Š ;(4.5 .52) where G andaregivenin(4.5.35)andtransformasin(4.5.37).Toobtaincase(1)as describedabove,weintroducethetensorcompensator G ,choos ethe L -gauge G =0,and solvefor+ inte rmsof G .Thequantity Š i s undetermined,butcanbegauged awaybyusingthere maininginvarianceparametrized L Š L .(R ecallgauging G to zero onlyusesthefreedomin L + L .)Toobtaincase(2),weintroducethechiralscalarcompens ator and gaugeittozerowhichgives= .Thus,gaugingeithertensorcompens ator or G to zeroforcesthestocontainthecorrectandcompletePoincar emultiplets.Alternatively,wecouldgaugeto zero,sothatthetensor(scalar)submultiplet iscontaine donlyin G ( ).Weshouldalsomentionthatotherchoicescouldbemadefor tensorcompensators.Anyofthevariantscal arornonminimalscalarmultipletscanbe

PAGE 227

4.5.Othergaugemultiplets215usedbygeneralizingthediscussionabove.Thesewillleadtoanumberofinequivalent o-shellformulationsofthePoincar e(3 2 ,1)theory. The“eldequations,obtainedfromtheaction(4.5.36),takethecovariantform D€X +(€+ €)=0, X = Gor + .(4.5 .53) (Incase(2),using(4.5.50),thesesimplifyto D€ +€=0.)

PAGE 228

2164.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS4.6.N-extende dmulti plets Sofarinthischapterwehavedescribedthemultipletsof N =1gl obalsupersymmetry.Forinteractingtheoriestherearetwosuchmultiplets,withspins(1 2 ,0),and (1,1 2 ),althoughtheirsuper“elddesc riptionmaytakemanyforms.For N -extended supersymmetry,global mult ipletsexistfor N 4.Theyarenaturallydescribedinterms ofexte ndedsuper“elds.Itispossible,however,todiscussthesemultiplets,andtheir interactions,intermsof N =1super“el dsdescribingtheir N =1s ubmultiplets.In manycasesofinterestthisisthemostcompletedescriptionthatwehaveatthepresent time. a.N=2multiplets Asdisc ussedinsec.3.3,thereexisttwoglobal N =2mult iplets:avectormultipletwithspins(1,1 2 ,1 2 ,0,0),andasc alarmultipletwithspins(1 2 ,1 2 ,0,0,0,0).There existson lyoneglobal N =4mult iplet:the N =4v ectormu ltiplet,with SU (4) representation spins(1 1,4 1 2 ,6 0).(Theonly N =3mult ipletisthesameas thatof N =4.)Webeginbyd iscu ssingthe N =2situation. a.1.Vect ormultiplet The N =2v ectormult ipletconsistsofan N =1 Ya ng -M illsmultipletcoupledtoa scalarmultip letinthesame (adjoint)represe ntationofthe internalsymmetrygroup. Theactionis S = 1 g2 tr ( d4xd4 + d4xd2 W2)(4. 6.1) inthevectorrepresentation.Inadditiontotheusualgaugeinvariance,itisinvariant underthefollowingglobaltransformationswithparameters : = Š W Š i [ 2( ) +( ) iW] = Š W Š [(1 2 [ €, ] ) €+( i 2 ) ], eŠ e= Š i + W€ € .(4. 6.2)

PAGE 229

4.6.N-extendedmultiplets217(Forthe transformationweuse(4.2.52),andalsoagaugetransformationwith K = Š i ( €D2 D€ Š D2D )+€1 2 [ D€, D] .)Duetotheidentity =[ ,( eŠ e)](4.2.77),thesecondtransformationcanbewrittenas = Š( i Š W€ € )= Š i + W€1 2 [ €, ] .(4. 6.3) Bothparametersare x -independentsuper“eldsandcommutewiththegroupgenerators (e.g., = D ).Theparameter ischiralandmixesthetwo N =1mult iplets, whereas istherealpar ameterofthe N =1supersy mmetrytransformations(3.6.13). Since hasthe( x -independent)gaugeinvariance = i ( Š ),theglobals uperparametersthemselves forman abelian N =2v ectormult iplet.Referringtothecomponentsof thisparametermultiplet( )bythe namesofthecorrespondingcomponentsinthe “eldmu ltiplet(, V ),we“ndthefollowi ng:Thephysicalbosonic“eldsgivetranslations(fromthevector a1 2 [ D€, D] | )and centralcharges(fromthescalars z | ); thephysicalfermionic“eldsg ivesupersymmetrytransformations( 1 i D2D | 2 D | );andtheauxiliary“eldsgiveinternalsymmetry U (2)/ SO (2)transformations( r 1 2 D D2D | q D2 ).(Thefull U (2)symmetryhas,inadditionto( r q q ) transformations,phaserotations = iu V =0). Thealgebraofthe N =2gl obaltransformationsclosesoshell;e.g.,thecommutatoroftwo transformationsgivesa transformation: [ 1, 2]= 12, 12= i [12]= i ( 12Š 21).(4.6 .4) Thetransformationstakeasomewhatdie rentforminthechiralrepresentation: = Š W Š i 2( ) eŠ V eV= i ( Š )+( W+ W€ €) ,(4. 6.5a) andhence = [ i ( Š )+( W+ W€ €) ].(4.6 .5b) Nowthe i ( Š )partofthe transformationdoescontri bute,butonlyasa“eld-dependent gaugetransformation= W .

PAGE 230

2184.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSWecanaddan N =2Fayet-I liopoulosterm(param etrizedbyconstants = 1+ i 2, 3= 3) d4xd43V +( Š i d4xd2 + h c .)(4.6 .6) to th ea bo ve ac ti on in th ea be lia n( free)case.Thisisinvariantunder(4.6.5)ifwe restricttheglobalparametersby Š D2 = D2 3 D2 = i (2 D2D2 + u ),(4.6 .7) where u istherealconstantparameterofthephase SO (2)par tof U (2).Theconstraint ontheparametersimpliesthatthe U (2)isbroke ndownto SO (2) U (1). Thismodelhassomeinterestingquantumproperties.Ithasgaugeinvariantdivergencesatone-loop,butexplicitcalculationsshowtheirabsenceatthetwo-andthreelooplevel.Insec.7.7wepresentanargume nttoes tablishtheirabsenceatallhigher loops. a.2.Hypermultiplet a.2.i.Freetheory The N =2scal armultipletcanbedescribedbyach iralscalarisospinorsuper“eld a(theahypermulti plet)withthefreeaction S = d4xd4 aa+1 2 ( d4xd2 amabb+ h c .),(4. 6.8) wherethesymmetricmatrix m satis“esth eco ndition macCcb= Cac mcb.(4. 6.9) (Theexplicitformis mab= iMCabb c, M = M ,with a a=0and b a= a b.Wit hout lossofgenerality, mabcanbechosenproportionalto ab.)Thefreeactionisinvariant undertheglobalsymmetries a= Š ( D2 Cab bŠ Z a) Š i D2[( D ) Da+( D2 )a],(4.6 .10) where Z isacentralcharge: Z a= Cabmbcc, Z a= Cab mbc c.(4. 6.11)

PAGE 231

4.6.N-extendedmultiplets219Onshell,wealsohave Z a= Š Cab D2 b.(4. 6.12) Wecanuse eitheroftheforms(4.6.11,12)inthetransformation(4.6.10),becauseofthe localinvariance a= CabSb, Sb S b ,(4. 6.13) forarbitrary x -dependentchiral .(This isaninvariancebecausethevariationofthe actionisproportionalto S SaCabSb=0.)Ifweus etheform(4 .6.12),thevariations donotdependontheparameters mab.Aninter estingfeatureofthealgebra(4.6.10)is thatitdoesnotcloseo-shellifweuserealization(4.6.11)for Z .Onthe otherh a nd,if weuserea lization(4.6.12)instead,thesymmetries(4.6.10)containpartofthe“eld equations,andhencebecomenonlinearand coupling-dependentwheninteractionsare introdu ced.Theseeectsareasignalthatinthedecompositionofthe N =2super“eld thatdescribesthetheoryinto N =1 su pe r“elds,someauxiliary N =1super“ eldshave b eendiscarded.Wediscussfurtheraspectsofthisproblembelow. Withoutthemassterm,theinternalsymmetr iesofthefreescalarmultipletarethe exp licit SU (2)thatactso ntheisos pinorindexofaandthe U (2)madeupofthe r and q transformationsin and ,andofthe uniformphaserotations a= iu a.Themass termbreakstheexplicit SU (2)tothe U (1)s ubgroupthatcommuteswith mac. a.2.ii.Interactions The N =2scal armultipletcaninteractwithan N =2v ectormult iplet,anditcan haveself-inter actionsdescribinganonlinear model.Aclassofsupersymmetric -modelscanbefoundbyco up linganabelian N =2v ectormult iplet(withnokineticterm butwithaFayet-I liopoulosterm)to nN =2scal armultipletsdescribedbythe n -vector.Thesupersymmetrytransformationsoft hevectormult ipletarethesameasthose givenabovein(4.6.2)or(4.6.5)fortheabeliancase.(Theyareindependentofthe“elds inthescalarmultiplets.)However,thetransformationsofthescalarmultiplets(eachof whichisdescribedbyapairofchiralsuper“eldsa)are gaugecovariantized: a= Š D2[ c(e V)c bCab]Š i D2[( D ) a+( D2 )a] Š i1 2 u a.

PAGE 232

2204.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDS(4.6.14) Thematrixisan SU (2)generatorthatbreakstheexplicit SU (2)ofthescalarmultipletdownto U (1).Because SU (2)preservesthealternatingtensor Cab,(e V)a b(e V)c dCac= Cbd.Theacti onthatisleftinvariantbythesetransformationsis: S = d4xd4[ a(e V)a bb+ 3V]+ d4xd2 i [1 2 aCabb ccŠ ]+ h c .(4. 6.15) provided(4.6.7)aresatis“ed.Thetheoryisalsoinvariantunderlocalabeliangauge transformations: a= i a b b, V = i ( Š ), =0;(4 .6.16) aswellasglobal SU ( n )rot ationsofa.Forexp licitcomputation,itisusefultochoose asp eci“c:Wechoose=3.Wewritea (+,Š) (+ i,Š i)where i =1 ... n isthe SU ( n )i ndex,+, Š arethe SU (2)iso spinindices,and+tr an sformsunderthe SU ( n )representatio nconju gatetoŠ.Thetransf ormations(4.6.14)andtheaction (4.6.15)become(using(4.6.7)) = D2( Š+eŠ+ V) Š1 2 3 ( D2 )Š i D2( D ) (4.6.17a) S = d4xd4[ + ieV+ i+Š ieŠ V Š i+ 3V], + d4xd2 i [Š i+ iŠ ]+ h c .(4. 6.17b) We nowproceedaswedidinthecaseofthe CP ( n )models(see( 4.3.9)):Weeliminate thevectormultipletbyits(algebraic)equationsofmotion.Inthiscase,actsasa Lagrangemultipliertoimposetheconstraint: Š i+ i= .(4. 6.18) Choosingagauge(e.g.,+ 1=Š 1),wecaneasilysolvethisconstraint;forexample,we canparametrizethesolutionas:

PAGE 233

4.6.N-extendedmultiplets221+ i=(1+ u+ uŠ)Š1 2 1 2 (1,u+), Š i=(1+ u+ uŠ)Š1 2 1 2 (1,uŠ).(4. 6.19) The V equationofmotiongives: + ieV+ iŠ Š ieŠ V Š i+ 3=0 ,( 4.6.20a) or Me V=1 2 [( 3 2+4 M+MŠ)1 2 Š+ 3];(4.6 .20b) where M= = | |2= | || 1+ u+ uŠ|Š 1(1+ | u|2).(4.6 .20c) Substituting,we“ndtheaction S = d4xd4{( 3 2+4 M+MŠ)1 2 + | 3| ln[( 3 2+4 M+MŠ)1 2 Š| 3|] }.(4. 6.21) Intermsoftheunconstrainedchiralsuper“elds u,t he transformations(4.6.17a)become u= D2[ eŠ+ V( )1 2 (1+ u+ uŠ)1 2 (1+ u+ uŠ)Š1 2 ( uŠ+Š u)]Š i D2[( D ) Du],( 4.6.22a) wheretheauxiliarygauge“eld V isexpr essedintermsof uby(4.6 .20).Thesupersymmetrytransformations(4.6.22a)includeacompensatinggaugetransformationwith parameter i = Š D2[ ( coshV )( )1 2 (1+ u+ uŠ)1 2 (1+ u+ uŠ)Š1 2 ](4.6.22b) thatmustbeaddedto(4.6.17a)tomaintainthegaugechoicewemadein(4.6.19). Asfort hefree N =2scal armultiplet,wecanaddaninvariantmassterm(which introdu cesanonvanishingcentralcharge).Themasstermnecessarilybreaks SU ( n )and hastheform Im= i1 2 d4xd2 aCabb cM c+ h c .,(4.6 .23) where M isanytra celess n n matrix( M sdie ringby SU ( n )trans formationsare

PAGE 234

2224.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSequivalent).Thesupersymmetrytransformati onsthatleavethisterminvariantarethe sameasbefore,includingthe Z termof(4.6.10).Therealizationof Z givenin(4.6.11) ispreferable,sinceitislinear,whereastherealization(4.6.12)mustbegaugecovariantized. Thesenonlinear -models liveonK¨ ahlermanifoldswith threei ndependentcomplexcoordinatesystemsrelatedby nonholomorphic coordinatetransf ormations(they havethreeindependentcomplexstructures(seetheendofsec.4.1);theconstants 3, parametrizethelinearcombinationofcomplexstructureschosenbytheparticularcoordinatesystem).ThusthesemanifoldsarehyperK¨ ahler.Justaswefoundthat foreveryK¨ ahlermanifoldthereisan N =1no n linear -model(andconversely),onecan showthatforeveryhyperK¨ ahlermanifoldthereisan N =2no n linear -model,andconversely, N =2no n linear -modelsarede“nedonlyonhyperK¨ ahlermanifolds.An immediateconsequenceofthisrelationisastr ongrestrictiononpossibleo-shellformulationsofthe N =2scal armultiplet: Noformulationcanexistthatconta insasphysicals ubmultipletstwo N =1sc alar mult iplets(e.g.,suchaswehaveconsid ered),thatcanbeusedtodescribe N =2 nonlinear -models,andthathassupers ymmetrytransformations independent of theformoftheaction. Ifsuchaformulationexisted,thenthesumoftwo N =2 in va riantactionswouldnecessar ilybeinvariant;however,thesumoftheK¨ ahlerpotentialsoftwo hyper K¨ ahlermanifoldsis not ingenera ltheK¨ ahlerpotentialofahyperK¨ ahlermanifold.Wewillsee belowthatwe can giveano-shellformulationofthe N =2scal armultipletthatavoids thisproblem. Wecang eneralizetheaction(4.6.15)inthesamewaythatwegeneralizedthe CP ( n )models(se e(4. 3.11)): S = d4xd4[ a(e V)a bb+ 3trV]+ d4xd2 i[1 2 a Cabb ccŠ tr ]+ h c ., (4.6.24a) where V = VATA,=ATA,and TAarethegeneratorsofsomegroup.The N =2 transformationsthatleave(4.6.24a)invariantaretheobviousnonabeliangeneralizations

PAGE 235

4.6.N-extendedmultiplets223of(4.6.14).Theequationsthatresultfromvarying(4.6.23a)withrespecttoA, VAare, choosing=3asabove, ŠTA+Š trTA=0, +eVTA+Š ŠTAeŠ VŠ+ 3trTA=0.(4. 6.24b) Asinthe N =1case,the sedonot,ingeneral,have anexplicitsolution. a.3.Tens ormultiplet Justasthe N =1scal armultipletcanbedescribedbydierentsuper“elds,wecan describethe N =2scal armultipletbysuper“eldsoth erthanthechiralisodoubleta. Wenowdiscussthe N =2tensorform ulationofthesc alarmultiplet.Thisisdualtothe previousdescriptioninthesamewaythatthe N =1tensor andscalarmul tipletsare dual(seesec.4.4.c).Wewritethetensorfor mofthescal armultipletintermsofone chiralsc alar“eld andachiralspinorgauge“eld withlinear“eldstrength G =1 2 ( D+ D€€), D2G = D2G =0.The N =2supers ymmetrytransformationsof thistheoryare = Š 2 D Š i D2[( D ) D+( D2 ) ], = Š D2( G ) Š i D2[( D ) D +2( D2 ) ].(4.6 .25) Incontrasttotheahypermulti pletrea lizationofthe N =2scal armultiplet,these transformationsclose o-shell; theyhavethesamealgebraasthetransformationsofthe N =2v ectormultiplet(4.6.4)(uptoagaugetransformationof ).However,although thesuper“eldsdescribeascalarmultiplet,thecentralchargetransformations z = | leavethe “elds in ert;thisgivesoneguidetounderstandingthedualitytothehypermult iplet. Thesimplestactioninvariantunderthetransformations(4.6.25)isthesumofthe usualfreechiralandtensoractions((4.1.1)and(4.4.34)): Skin= d4xd4 [ Š1 2 G2+ ].(4.6 .26) To “ndotheractions,weconsiderageneralansatz,andrequireinvarianceunderthe

PAGE 236

2244.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDStransformations(4.6.25).Actually,wecanconsideraslightlymoregeneralcasethat stillhasfullo-shell N =2invariance byrest rictingthechiralparameter by D2 =0, andtherealparameter by D2D2 =0.Thismea nsthatwedonotimpose SU (2) invariance.Anaction S = d4xd4 f ( G , )(4. 6.27) is in va riantunder(4.6.25)(with D2 =0)if f satis“es 2f G2 + 2f fGG+ f =0.(4. 6.28) ( f containsnoderivativesof G , .)Itdescribe sageneral N =2tensormu ltiplet interactingmodel.Wealsocanco nsidermorethanonemultiplet Gi, i, i,eachtransformingas(4.6.25);thenthemostgenera linvaria ntacti onis(4.6.27)wherethe Lagrangian f satis“es fGiGj+ fi j=0.(4. 6.29) (Actually,wecangeneralize(4.6.27)slightlybyaddingaterm d4xd2 hii+ h c wherethe hisarearbitraryconstants.) a.4.Duality To gaininsightintothephysicsofthesemodelswe“ndthedualtheoriesdescribed bytheahypermulti plet.Weconsiderthefollowing“rstorderaction(cf.(4.4.38)): S= d4xd4 [ f ( Vi, i, i) Š Vi(i+ i)].(4.6.30) Eliminating, gives(4 .6.27),whileeliminating V resultsinthedualtheory.We“nd the N =2transf ormationsoftheresultinga ihypermulti pletsfromthetransformations thatleavethe“rstorderaction(4.6.43)invariant.Since d4xd4 f ( V , )isinvarian t under(4.6.25)with G Vexcept forterms D2V or D2V ( V diersfrom G onlybecauseitdoesnotsatisfytheBianchiidentities D2G = D2G =0),weca ncan cel thesetermsbychoosingthevariationofappropriately.The“rstorderaction(4.6.43) is in va riantunder

PAGE 237

4.6.N-extendedmultiplets225 Vi= DiD + D€ i D€ + Vi,( 4.6.31a) i= Š D2( Vi)+ i,(4. 6.31b) i= Š D2[ ( fi+ Vj( f jViŠ f iVj))]+ i;(4. 6.31c) where istheusual N =1supersy mmetry(3.6.13)(with wV=0, w= Š 2, w=0). (T op rovetheinvarianceof(4.6.30)under(4.6.31),weneed(4.6.29)anditsconsequen ces,inparticular, fViV[ jk ]=0and f i[ jVk ]=0b ecauseoftheantisymmetrization, andtherefore,using thechain rulewe“nd D€f [ jVi ]=0.)Perfo rmingthedualitytransformations,wecanrewritethetransformati ons(4.6.31)andthecondition(4.6.29)in termsofthedua lvariables, andtheLegendretransformedLagrangian IK (+ , ).We“nd(dro ppingtheuninteresting terms) i= D2( IKi), (4.6.32a) i= Š D2[ ( IKi+ IKj(( IKk i)Š 1IKjkŠ ( IKk j)Š 1IKi k))],(4.6.32b) forthetran sformations,and IKi j=( IKi j)Š 1+ IKi m( IKm n)Š 1IKn j(4.6.33) fortheconditionthat theLagrangianmustsatisfytoguaranteeinvariance.Notethatin contrast withthe o-shell transformations(4.6.25),the on-shell transformations (4.6.31,32)dependexplicitlyontheformoftheaction.Furthermore,thecondition (4.6.29)need edforinvarianceoftheo-shellversionofthemodelis linear, andhence thesumoftwoinvariantactionsisautomaticallyinvariant,whereasthecondition (4.6.33)is nonlinear. TheLegendretransformationallowsthistooccur,andallowsusto complywith therestrictionono-shellformulationsthatwediscussedabove. Althoughitisalwayspossibletogofromtheo-shellformulation(intermsofthe tensormultiplet)totheon-shellformulation(intermsofthehypermultiplet),thereverse transformationisgenerallynotsostraightforward.Theimprovedform(see(4.4.45-5)) ofthefreemultipletcanbefoundbyexploitingananalogywiththenonlinear -models discussedabove(Actually,thetensormult ipletformoftheinteractingmodelscanbe foundinthisway).Alternatively,somesimplemodelscanbefoundbyusingthecentralchargeinvarianceofthetensormultiplet(seebelow).

PAGE 238

2264.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSByanalogywith(4.6.17),wecanwritedownthefollowing“rstorder N =2invariantactionbyintroducinganauxiliary N =2v ectormult iplet V ,: S= d4xd4[ +eV++ŠeŠ V ŠŠ GV]+ d4xd2 i [Š+Š ]+ h c .(4. 6.34) Thisactionisinvariantunderthetransformations(4.6.5,14,25).Varying G and ,we “ndthefreehypermu ltiplet(seediscussioninsec.4.6.a.2);varying V and,w e “nd thatdrop outoftheactionentirely,andtheimproved( N =2)tensorm ultiplet results: Simp= d4xd4[( G2+4 )1 2 Š Gln ( G +( G2+4 )1 2 )].(4. 6.35) Thiscomplicatednonlinearactioncorrespondstoafreehypermultiplet!Itis,however, an o -shellformulation,invariantunderthetransformations(4.6.25).Itgeneralizes directlytogiveano-shellformulationofthenonlinear -modelswediscussedabove. Analternativederivationoftheimprovedtensormultipletdoesnotrequirean N =2v ectormultiplet,butusesthecentralchargeinvarianceofthetensormultiplet. Webeginwit hthefreehyperm ultipletaction (4.6.8)(withoutlossofgenerality,wetake mab= imCab( 3)b c).WewishtoLegendr etransfo rmoneofthechiral“elds a=(+,Š),andkeeptheother“eldasthechiral“eld ofthetensormultiplet. However,though is in ertundercentralchargetransformations,arenot;wethereforede“netheinvar iantcombination i +Š,andinte rmsofitwritethe“rstorder action S= d4xd4 [ eŠ V+ eVŠ GV ]+1 2 m [ d4xd2 + h c .].(4. 6.36) Varying G ,wer ecoverthehypermultipletaction(4.6.8)with V = ln ( ++);varying V wer ecovertheimprovedtensormult ipletaction(4.6.35)witha linear termthatacts asamassterm.Thealgebraoftransformationsthatactonthemassivescalarmultiplet has a centralcharge;however,thedescri ptionofthe mult ipletgivenbythe N =2tensor mult ipletonlyinvolves“eldsthatare inert underthecentralcharge.

PAGE 239

4.6.N-extendedmultiplets227Finally,wenotethatthegaugeinteractionsofthe N =2tensormu ltipletareanalogoustothe N =1case(s eesec.4.4.c). a.5.N=2super“eldLa grangemultiplier Anotherformulationofthe N =2scal armultipletwitho-shell N =2supersymmetryisthe N =2L agrangemultipliermultiplet.Itisthe N =2genera lizationofthe mult ipletdiscussedinsec.4.5.d,andcontainsthat N =1mult ipletasasubm ultiplet. Unliketheo-shell N =2supersy mmetricscalarmultipletdiscussedabove(the N =2 tensormultipletofsecs.4.6.a.3,4),thismultipletcanbecoupledtothe( N =2)nonabelianvectormultiplet,thoughonlyinrealrepresentations.Byusingtheadjointrepresentation,thisallowsconstructionof N =4 Ya ng -M illswitho-shell N =2supersymmetry,asdiscussedbelowinsec.4.6.b.2. The N =2L agrangemultipliermultipletisdescribedbythefollowing N =1 super“elds:(1)1 and Y ,des crib ingan N =1L agrangemultipliermultipletasin (4.5.18),withthegaugeinvarianceof(4.5.19),and“eldstrength1= D€ 1€(forwhich F and G of(4.5.18)aretherealandimaginaryparts);(2)asecondspinor2 ,withthe sa medimensionandgaugeinvariance,butwhichisauxiliary;(3)acomplexLagrange mult iplier,whichco nstrainsallof2tovanish(insteadofjusttheimaginarypart,as does Y for1),andhasa“eldstrength D€with gaugeinvariance =(forchiral);(4)aminimalscalarmultiplet,describedbyacomplexgauge“eld1withchiral “eldstrength1(seesec.4.5.a);and(5)twomo reminimalscalarmultiplets2and3, but auxiliary.Wethushavean N =1L agrangemultipliermultiplet,aminimalscalar mu ltiplet,andassortedauxiliarysuper“elds. Theactionis S = Š d4xd4 [1 8 (1+ 1)2+i 2 Y (1Š 1)] + d4xd4 [ 11+(2+ 2)+(23+ 2 3)].(4. 6.37) Themostinterestingpropertiesofthistheoryappearwhenitiscoupledto N =2superYa ng -M ills.Wedothisby N =2 gaugecovariantizingthe N =2L agrangemultiplier mult iplet“eldstrengths.(Inthe absence ofYang-Millscoupling,thescanbe

PAGE 240

2284.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSconsideredasordinaryscalarmultiplets,r athert han“el dstrengths.)Th iscouplingis discussedins ec.4.6.b.2. b.N=4Yang-Mills Inmanyrespects,the N =2no n linear -models,whenstudiedintwodimensions,areanalogsof N =4 Ya ng -M illstheoryinfourdimensions.Despitepower-countingarguments,theyarecompletely“niteonshell,andtheyarethemaximallysupersy mmetricmodelscontainingonlyscalarmultiplets(thevectorisauxiliaryandcanbe e liminated).The N =4 Ya ng -M illstheoryisthe“rstandbest-studied4-dimensional theorythatisultraviolet“nitetoallordersofperturbationtheory,andthusscaleinvariantatthequantumaswellastheclassicallevel.(Its -functionhasbeencalculatedto vanishthro ughthreeloops;ar gumentsfortotal“nitenessaregiveninsec.7.7.Indepe ndentargumentsusinglight-conesuper“eld shaveb eengivenelsewhe re.)Itisselfconjugateandisthemaxima llyextendedgloballysupersymmetrictheory.Twosuper“eldformulationsofthetheoryhavebeengiven:Oneusesan N =2v ectormult iplet coupledtoaahypermul tipletandhasonly N =1supersy mmetryoshell,andthe otherusesan N =2v ectormult ipletcoupledtoan N =2L agrangemultipliermultiplet andhas N =2 su pe rsymmetryoshell(however,ithasalargenumberofauxiliary super“elds). b.1.Minimalf ormulation Atthecomponentleve lthetheory containsagaugevectorparticle,fourspin1 2 Weylspinor s,andsixspin0particles,allintheadjointrepresentationoftheinternal symmetrygroup.Itcanbedescribedbyonerealscalargaugesuper“eld V andthree chiralsc alarsuper“eldsi,andisthesameasan N =2v ectormult ipletcoupledtoan N =2scal armultiplet.Ifweuseamatrixrepresentationforthei,the(c hiralrepresentation)conjugatecanbewrittenasi= eŠ V ieV.The N =1supers ymmetricaction (inthechiralrepresentation)isgivenby S = 1 g2 tr ( d4xd4 eŠ V ieVi+ d4xd2 W2+1 3! d4xd2 iCijki[j,k]+1 3! d4xd2 iCijk i[ j, k]).(4.6.38)

PAGE 241

4.6.N-extendedmultiplets229InadditiontothemanifestSU(3)symmetryonthe i j k i ndicesofand ,ithasthe fo llowingglobalsymmetries: i= Š ( Wi+ Cijk 2 jk) Š i 2[( ) i+2 3 ( 2 )i], = [ i ( iiŠ ii)+( W+ W€ €) ];(4.6 .39) inthechiralrepresentation,an dinthev ectorrepresentation i= Š ( Wi+ Cijk 2 j kŠ i [ jj,i]) Š i [ 2( ) i+( ) iWi+2 3 2( 2 )i]; = Š( i i i+ W€ € ).(4.6 .40) The iarethegeneralizationofthosegivenforthe N =2mult ipletsabov e,butnow theyforman SU (3)iso spinor,asdoesi.Theidenti “cationofthe componentsof and isthesame:Thephysicalbosonic“eldsarethetranslationsandthecentralcharge parameters(3complex=6real,asfollowsfromdimensionalreductionfromD=10:see s ec. 10.6),thespinorsarethesupersymmetryparameters,andtheauxiliary“eldsare internalsymmetryparametersof SU (4)/ SU (3).Thealgebradoesnotcloseo-shell. Uponreductiontoits N =2s ubmultiplets,(4.6.39)(or(4. 6.40))reducesto(4.6.5)(or (4.6.2))and(4.6.14)(butwithdierentR-weights). Thecorrespondingcomponentactionhasaconventionalappearance,withgauge, Yukawa,andquarticscalarc ouplingsallgovernedbythesamecouplingconstant.In sec.6.4wediscusssomeofthequantumpropertiesofthistheory. b.2.La grangemultiplierformulation Wenowbrie” ydes cribeanother N =1super“eldf ormulationof N =4superYa ng -M ills;itemploysthe(unimproved)typeof N =1scal armultipletofsec.4.5.d. Althoughevenlessofthe SU (4)symmetryismanifest,thisformulationiso-shell N =2supersy mmetric:Itfollowsfromthe N =2super“eldformula tionofthetheory, asdescribedbyt hecouplingof N =2 su pe rY ang-Millstoan N =2(L agrangemultip lier)scalarmultiplet.Thisformulationhasanumberofothernovelfeatures:(1)

PAGE 242

2304.CLASSICAL,GLOBAL,SIMPLE(N=1)SUPERFIELDSrenormalizablecouplingsbetweennonminimalandotherscalarmultiplets,(2)thenecessaryappearance(ininteractionterms)oftheminimalscalarmultipletintheformofa gaugemultiplet(sec.4.5.a),and(3)lossof(super)conformalinvarianceoshell(this o ccursbecausethemodelincludesan unimpr oved Lagrangemultipliermultiplet). Theactioncanbewrittenas(inthesuper-Yang-Millsvectorrepresentation)the sumof(4.6.1)and(4.6.37).However,thede“nitionsofthe“eldstrengthsiandiare nowmodi“ed: i= € i€Š i [0,i] i= 2i( i =0,1,2),3= 23Š i [0,];(4 .6.41) where0(withprepotential0)isthechiralsu per“eldofthe N =2 Ya ng -M illsmultiplet.The Ainthesede“nitionsistheYang-Millscovariantderivative.Inadditionto theusual(adjoint,vectorrepresentation)Yang-Millsgaugetransformations,wehave manynewlocalsymmetriesoftheaction: i= € Ki€( i =0,1,2), 3= € K3€+ i [0, ]; (4.6.42a) i = Ki ( )+ i [ 0, Ki ]( i =1,2); =; Y = =0;(4.6 .42b) whereiscovariantlychiral( €= 0),andistheYang-Millsvector-representation prepotential.Underthesetransformationsthe“eldstrengthsi( i =0,. ..,3),i( i =1,2), €, Y ,and Wareinvariant.Intheabelian(orlinearized)case,thesumof (4.6.1)and(4.6.37)asmodi“edby(4.6.42)describesan N =2v ectormult iplet( Wand 0)plusan N =2scal armultipletconsistingofthe N =1L agrangemultipliermultiplet of(4.5.18)(1 and Y ),aminimal N =1scal armultiplet(1),andsomeauxiliary super“elds(2 ,,2,and3).However,intheinteractingcasetheformulationis somewhatunusualinthat3isnotj ust N =1covarian tlychiral( €3 =0)nor areiN =1covarian tlylinear( 2i =0), buttheysatisfythe N =2covariantB ianc hiidentities €3= Š i [0, €], 2i= Š i [0,i].(4.6 .43) Theinteractiontermsoftheauxiliarysuper“elds(introducedthroughthenonlinearities

PAGE 243

4.6.N-extendedmultiplets231ofthe“eldstrengths2and3)can celamongthemselves:Theirtermsintheactioncan berewri ttenas,inthe chiral representation, d4xd4 ( D€ 2€+ h c .)+[ d4xd2 2( D23)+ h c .].(4. 6.44) BycombiningtheBianchiidentities(4.6.43),theusualconstraint €i=0(for i =0,1,2),and €W=0, W+ €W€=0with the“eldequationswhichfollow fromtheaction,weobtaintheon-shellequationsforallofthesuper“elds D€=1Š 1=2=2=3=0, i W= Š i € W€=[0, 0]+[1, 1]+1 4 [(1+ iY ), (1+ iY )], €0= 2 0+ i [1,1 2 (1+ iY )]=0, €1= 2 1+ i [1 2 (1+ iY ),0]=0, €(1+ iY )= 2 (1+ iY )+2 i [0,1]=0.(4 .6.45) Wecanthuside ntifythisfor mulation onshellwiththatgivenaboveinsubsec.4.6.b.1. bytheco rrespondences W W,(0,1,1 2 (1+ iY )) i.(4. 6.46)

PAGE 244

Contentsof5.CLASSICALN=1SUPERGRAVITY 5.1.Reviewofgravity232 a.Potentials232 b.Covariantd erivatives235 c.Actions238 d.Conformalc ompensator240 5.2.Prepotentials244 a.Conformal244 a.1.Linearizedtheory244 a.2.Nonlineartheory247 a.3.Covariantderivatives249 a.4.Covariantactions254 b.Poincar e 255 c.Densitycompensators259 d.Gaugechoices261 e.Summary263 f.Torsionsandc urvatu res264 5.3.Covariantapproachtosupergravity267 a.Choiceofconstraints267 a.1.Compensators267 a.2.Conformalsupergravityconstraints270 a.3.Contortion273 a.4.Poincar esupergravityc onstraints274 b.Solutiontoco nstrai nts 276 b.1.Conven tionalconstraints276 b.2.Representationpreservingconstraints278 b.3.The gaugegroup279 b.4.Ev aluationofand R 281 b.5.Chir alrepresentation284 b.6.Densitycom pensat ors286 b.6.i.Minimal( n = Š1 3 )supergr avit y 287 b. 6.ii.Nonminimal( n = Š1 3 )supergr avit y 287 b. 6.iii.Axial( n =0)supergr avit y 288 b.7.De gauging289

PAGE 245

5.4.SolutiontoBianchiidentities292 5.5.Actions 299 a.Reviewofvectorandchiralrepresentations299 b.Thegeneral measure300 c.Tensorcompensators300 d.Thechiralmeasure301 e.Representationindependentformofthechiralmeasure301 f.Scal armultiplet302 f.1.Superconforma linterac tions303 f.2.Conformallynoni nvariantac tions304 f.3.Chir alself-interactions305 g.Vectormultiplet306 h.Generalmattermodels307 i.Supergravityactions309 i.1.Poincar e 309 i.2.Cosmologicalterm312 i.3.Conformalsupergravity312 j.Fieldequ ations313 5.6.Fromsuperspacetocomponents315 a.Generalconsiderations315 b.Wess-Zumino gaugeforsupergravity317 c.Commutatoralgebra320 d.Localsupersymmetryand componentgauge“elds321 e.Superspace“eldstrengths323 f.Supercovariantsupergravity“eldstrengths325 g.Tensorcalculus326 h.Componen tactio ns 331 5.7.DeSittersupersymmetry335

PAGE 246

5.CLA SSICALN=1SUPERGRAVITY 5.1.Reviewofgravity a.Potentials Ourreviewisintendedtodescribetheapproachtogravitythatismostusefulin understandingsupergravity.Wetreatgravityasthetheoryofamasslessspin-2particle describedbya gauge“eldwithanadditional vector indexasagroupi ndex(sothatit containsspin2).Byanalogywiththetheor yofama sslessspin1partic leitslinearized transformationlawis h a m= a m.(5. 1.1) SincetheonlyglobalsymmetryoftheS-matrixwithavectorgeneratoristranslations, wechoosepartials pacetimederivatives(momentum)asthegeneratorsappearingcontractedwiththegauge“eldsgroupindexinthecovariantderivative e a aŠ ih a m( i m) =( a m+ h a m) m e a m m.(5. 1.2a) Thus,incontrastwithYang-Millstheory,weareabletocombinethederivativeand gr ouptermsintoasingleterm.Thegauge“eld e a misthe vierbein, whichreducestoa Kroneckerdeltain”atspace.It isinvertible:Itsinverse e m aisde “nedby e m ae a n= m n, e a me m b= a b.(5. 1.2b) Finitegaugetransformationsarealsode“nedbyanalogywithYang-Millstheory: e a= ei e aeŠ i mi m.(5. 1.3) Thelinearizedtransformationtakestheformof(5.1.1),whereasthefullin“nitesimal formtakestheformofa Liederiv ative: ( e a m) m= i [ e a]= Š [ n n, e a m m],(5.1.4a) or,inmoreconventionalnotation, e a m= e a n n mŠ n ne a m.(5. 1.4b)

PAGE 247

5.1.Reviewofgravity233Thegaugetransformationofascalarmatter“eldis,againbyanalogywithYang-Mills theory, = ei ei eŠ i ,(5. 1.5) an di ni n“nitesimalform = i [ ]= Š m m .(5. 1.6) Equation(5.1.5)canalsobewrittenasthemorecommongeneralcoordinatetransformation ( x) ( x ), x= eŠ i xei .(5. 1.7) (Thiscanbeveri“edbyaTaylorexpansion.)Forthecaseofconstant ittakesthe fa miliarformofglobaltranslations.Orbital(global)Lorentztransformationsare obtainedbychoosing m= x€+€€x€(whichju stequals x min th ei n“nitesimal case);istraceless.Scaletransfo rmationsareobtainedbychoosing m= x m. Wecouldatt hispointde“ne“eldstrengthsintermsofthecovariantderivatives (5.1.2),buttheinvarianceg roupwehavede“nedistoosmallfortworeasons:(1)The vierbeinisareduciblerepresentationofthe(global)Lorentzgroup,somoreofitshould be gaugedaway;and(2)therearedicultiesinrealizing(global)Lorentztransformationsongen eralrepresentations,aswenowdiscuss. SinceunderglobalLorentztransformations ( x )transfo rmsasascalar“eld,its gradient m willtransformasacovariantvector.Ingeneral,wede“neacovariantvectortobeanyobjectthattransformslike m .Wecande “neacontr avariant v ectorto belongto theadjointrepresentationofourgaugegroup.Indeed,ifwede“ne V V mi mandrequirethat[ V ]= V mi m transformasascalar,i.e., V V mi m= ei VeŠ i ,(5. 1.8) then V mtr an sformscontravariantlyunderglobalLorentztransformations.However, th is proceduredoesnotallowustode“neobjectswhichtransformasspinorsunder globalLorentztransformations,andinfacti tisimpo ssibletode“nea“eld,transforming linearly underthe group,whichals otransfo rmsasaspinorwhenthe sarerestricted torepresentglobalLorentztransformation s.Itispossibletogetaroundthisdiculty byrea lizingthe transformationsnonlinearly,butthisisnotaconvenientsolution.

PAGE 248

2345.CLASSICALN=1SUPERGRAVITYComparing(5.1.3)to(5.1.8),wesee e a mtransformsasfourinde pe ndentco ntrava riant v ectorsundertheglobalLorentzgroup:The transformationsdonotactonthe a i ndices. Tosolvethe seproblemsweenlargethegaugegroupbyadjoiningtothe transformationsagroupof local Lorentztransformations,andde “nespinorsw ithresp ectto this group.Thisisaprocedurefamiliarintreatmentsofnonlinear models.Nonlinearrealizationsofagrouparereplacedbylinearrepresentationsofanenlarged(gauge)group. Thenonlinearitiesreappearonlywhenade“nitegaugechoiceismade.Similarlyhere, byenlarg ingthegaugegroup,weobtainlinearspi norrepresentations.Thenonlinear spinorrepresentationsofthegeneralcoordinategroupreappearonlyifwe“xagaugefor thelocalLorentztransformations.Itwillthusturnoutthatour“nalgaugegroupfor gravitycanbeinterpretedphysicallyasthedirectproductofthetranslation(general coordinate)groupwiththespin(internal)angularmomentumgroup. Wede “netheactionofthelocalLorentzgrouponthevierbeintobe e a m= Š e€ m+ h c ., =0.(5. 1.9) Thesetransformationsactonlyon thefreeindicesintheoperator e a( butnotthehidden i ndicescontractedwith ,sincewe wanttheope ratortotransform covariantly). From nowonwewillindicateindicesonwhicht helocalLorentztr ansformationacts (”at or tangentspace i ndices)byusinglettersfromthebeginningoftheGreekandRoman al phabets( ,... a b ,...),andi ndicesonwhichl ocaltr anslations(generalcoordinate transformations)act (curved or world i ndices)bylettersfromthemiddle ( ,... m n ,...).Transfo rmationsrepresentedbyamatri xmulti plyingthefreeindex of e aarecalledtangentspacetransformations. ThelinearizedformofthelocalLorentztransformationsis h a m= Š €€ + h c ..(5.1 .10) Itisthuspossibletogaugeawaytheantisymmetric-tensorpartofthevierbein(although notthescalarpart)witha nonderivative transformation.Tostayinthisgaugealocal coordinatetransformationmustbeaccompan iedbyarelatedlocalLorentztransformation;theLorentzparameterisdeterminedint ermsofthetranslationparameter.Atthe lin earizedlevelwe“nd,usingthecombinedtransformations(5.1.1)and(5.1.10),

PAGE 249

5.1.Reviewofgravity235€€h( € )€=0 = Š1 2 ( € )€.(5. 1.11) Inthisgaugethe(orbitalLorentzcoordinate)transformationde“nedaboveinduces thesameglobalLorentztransformationactingonthe”atindices.Wecanthusde“nea Lorentz spinor bychoosingitss pinorindextobea”atindex;”at,ortangentspace, i ndicestransformunderlocalLorentztransformationsbutnotlocaltranslations except whenagaugeischosen,e.g.,asin(5.1.11).Furthermore,wecande“ne all covariant objects exceptthevierbein tohaveonly ”atindices.Thecurved-indexvectorsde“ned abovecanberelatedto”at-i ndexonesbymultiplyingwiththevierbeinoritsinverse. b.Covariantderivatives Wenowde “neournewlocalgroupofPoincar etransfo rmations,derivatives covariantunderit,anditsrepresentationonall“elds.Theparameterofourenlarged localgroup isde “nedby = mi m+( iM + €€i M€€).(5.1 .12) Thegenerator M (andtheparameter )istra celessandactsonlyonfree”at indices.Itsactiononsu chi ndicesisde“nedby [ M ]= ,[ M €]=0, [ €€ M€€, ]=0,[ €€ M€€, €]= €€€.(5. 1.13) An yc ov ar ia nt “eldwithonly”atindicestransformsunderthisgaugegroupas: ...= ei ...eŠ i .(5. 1.14) Thecovariantderivativeisde“nedbyintroducingagauge“eldforeachgroupgenerator, sowemustnowaddto e aof(5.1.2)anewgauge“eldfortheLorentzgenerators: D D a= e a+( a M + a ,€€ M€€).(5.1 .15) Itstransformationlawtakesthecovariantform D D a= ei D D aeŠ i .(5. 1.16)

PAGE 250

2365.CLASSICALN=1SUPERGRAVITYInde“ningthiscovariantderivativewehaveintroducedagauge“eld awhich transformswithaderivativeoftheLorentzgaugeparameter( a = a + ... ). However,duetothevierbeinsLorentztransformationlaw(5.1.10),wecande“nethis Lorentzgauge“eldtobeaderivativeofourfundamental“eld(thevierbein),justasfor thenonlinear -model(seesec.3.10),ratherthanhavingitasanindependent“eld. Therearetwowaysto“ndthisexpressionfor a:(1)Comp arethefulltransformation lawsofthevierbeinandtheLorentzgauge“eld,andconstructdirectlyfromthevierbein aLorentz connectionthathasthecorrecttransformationproperties;or(2)constrain someofthe“eldstrengthsinsuchawaythattheLorentzgauge“eldisdeterminedin termsofthevierbein.Becaus ethe “eldstrengthsarecovariantthiswillautomatically leadtocorrectlytransforminggauge“elds a. The“eldstrengths t a b cand r a b( M )arede “nedby: [ D D a, D D b]= t a b cD D c+( r a b M + r a b ,€€ M€€).(5.1 .17) Wehaveex pandedtheright-handsideover D D and M insteadof and M b ecausethen thetorsion t andcurvature r arecovariant.Byexaminingth eresulta ntexpr essionsfor the“eldstrengthsintermsofthegauge“elds,we“nd t a b c= c a b c+[( a €€+ a ,€€ ) Š a b ], r a b =( e a b Š a b ) Š c a b e e + a ,( | b | ),( 5.1.18a) wherethe anholonomycoecientc isde “nedby [ e a, e b]= c a b ce c.(5. 1.18b) Weseethatcon strain ingthetorsiontovanishgivesasuitableLorentzgauge“eld: a = Š1 4 ( c a ,( € )€+ c( € )€ €).(5.1 .19) (Ifinsteadofthetorsionweconstrainedthecurvaturetovanish,theconnection awouldbe pureLorentzgauge,andunrelatedtothevierbein.However,theantisymmetricpartofthevierbeinwouldremainasacompensatorforasecond hidden localL orentz gr ou po ft hetheory,underwhich awouldtra nsformhomogeneouslyandnotasaconnection.Hence D D de“nedby(5.1. 15)wouldbe noncovariant underthenewtransformations,andinstead

PAGE 251

5.1.Reviewofgravity237 D D a= e aŠ [1 2 ( c a € €+ c€ € €) M + h c .] = e aŠ [1 2 ( t a € €+ t€ € €) M + h c .](5.1 .20) wouldbecovariant.S inceEinsteintheory isnowdes cribedintermsofacurvatureconstructedoutof D D ,theor iginal aanditsassociatedLorentzinvariancewouldbeirrelevantto thetheory.Co nstrainingthecurvaturetovanishisgaugeequivalenttonot introducinganyconnectionatall.Suchaformulationofgravityisoftenreferredtoasa telepara llelismtheory.Ofcourse,ifweweretoconstrainboth t and r tovanish, D D wouldbe gaugeequivalentto ,andwe wouldhavenogravity.) Intheabsenceofanyconstraint,wecouldalwaysexpressthecovariantderivative astheconstrainedcovariantderivative( t =0) plus Lorentzcovar ianttermsthatcontain onlythetorsion.Thetorsioncouldthusbe consideredasanindependenttensorwithno relationtogravity.Ourtorsionconstraintisthusaconventionalconstraint,justlike theconventionalconstraint(4.2.60)ofsuper-Yang-Millstheories. Allremaini ngtensors(i.e.,covariantobjectsthatarenotoperators)canbe expressedintermsofthecurvatureanditscovariantderivatives.Thecurvatureitselfis al ge braicallyreducible(undertheLorentzgroup)intothreetensors: r a b = C€€( w Š1 2 ( ) r )+ Cr€€,(5. 1.21) wherethetensorsaretotallysymmetricinundottedindicesandindottedindices(which isequivalenttobeingalgebraicallyLorentz-irreducible).Thetensors r and r€€arethe traceandtracelesspartsofthe Riccitensor, and wisthe Weyltensor. (Notethat ournormalizationoftheRicciscalardiersfromthestandard:Weusethemoreconvenientno rmalization,ingeneralspacetimedimensionD, r a b c d= Š [ a c b ] dr + ... ,rather than r a b a b= r .Thesignisch osensothat r isno nnegativeonshellinunbrokensupersymmetrictheories.)Thesetensorsare,ofcourse,relateddierentiallythroughthe Bianchiidentities(theJacobiidentitiesofthecovariantderivatives).Explicitlyfrom [[ D D a, D D b], D D c]+[[ D D b, D D c], D D a]+[[ D D c, D D a], D D b]=0,(5 .1.22) we “nd D D[ at b c ] dŠ t[ a b | et e | c ] dŠ r[ a b c ] d=0 ,( 5.1.23a)

PAGE 252

2385.CLASSICALN=1SUPERGRAVITYD D[ ar b c ] Š t[ a b | er e | c ] =0.(5. 1.23b) Theselasttwoequations(whichfollowfromthelinearindependenceof D D aand M )are the“rstands econdBianchiidentities,respectively. c.Actions IncontrasttoYang-Millstheory,ingravityonecannottraceoverthegroupwithoutintegratingoverthespacetimecoordinates,sincethetranslationgroupactsonthe coordinatesthemselves.Thus,onlyintegratedquantitiescanforminvariants.Furthermore,gravitydiersevenfromthegroupmanifoldapproachtoYang-Mills,wherethe groupgeneratorsaretreatedastranslationsinthegroupspace,inthatthelocaltranslationgroupisnotunitary:Although mis hermitian,anin“nitesimaltranslationisnot: ( mi m)= i m m= mi m+( i m m).(5.1 .24) Fromthereord eringofthetwofactors,wegetanadditionaltermproportionaltothe divergenceof .Thistermarise sb ecausesomecoordinatetransformationsarenot volume-pr eserving: e.g.,thetransformationgivenby m x misascaletransformation. Consequentlythevolumeelement d4x dx+€+/ / \ \dx+€Š/ / \ \dxŠ€+/ / \ \dxŠ€Šisnotcovariant.To covariantize,wesimplyreplace dx mwithanobjectthatisascalarundercoordinate transformations(aworldscalar): a= dx me m a.Theresult ingvolumeelementis 4= d4x eŠ 1,whereei sthe determinantof e a m. Theinvarianceofascalarintegratedwiththecovariantvolumeelementcanalso bes eenfromthetransformationlawofe,whichwewriteinthecompactandconvenient form e Š 1=eŠ 1ei .(5. 1.25) Here = mi mmeansthatthederivativeactsonallobjectstoitsleft.(ForthepresentdiscussionwemayignoreLorentztransformations.)Beforederivingthistransformationlaw,weshowhowitallowseŠ 1toforminvariantintegrals:Foranyscalar L d4x (eŠ 1L )= d4x (eŠ 1ei )( ei LeŠ i ) = d4x eŠ 1ei ( eŠ i Lei )

PAGE 253

5.1.Reviewofgravity239= d4x (eŠ 1L ) ei ,( 5.1.26a) wherew ehaveusedtheid entity (forany X ) [ X ]=[ X ] ei XeŠ i = eŠ i Xei .(5. 1.26b) Finally,using Xei = X +tot alderivati ve,w e “nd d4x (eŠ 1L )= d4x eŠ 1L .(5. 1.27) Toderivethet ransformationlaw(5.1.25),weneedtheidentity detX = detXtr ( XŠ 1 X ),(5.1 .28) whichfollowsfrom detX = etrlnX.T hu sw e “nd,from(5.1.4b), eŠ 1= Š eŠ 1( e m a( e a n n mŠ n ne a m)) = Š eŠ 1( m mŠ e m a n ne a m) = Š eŠ 1 m mŠ m meŠ 1= Š m( meŠ 1)=eŠ 1i .(5. 1.29) To “ndthe“nitetransformation,weiteratethein“nitesimaltransformation(5.1.29)and use ex=n lim(1+x n )n;wethusarriveatthedes iredresult(5.1.25).Anequivalentstatementofourresultisthat1 ei istheJacobiandeterminant ofthecoordinatetransformation ei .(1 ei meansthatderivativesacttotheleftuntilannihilatingthe1.) Wecannow constructinvariantactionsforgravityanditscouplingstomatter. Theonlypossibleactionthatgives h a bas econd-orderkineticoperatoris S = Š3 2 d4x eŠ 1r ,(5. 1.30) where r isthecurvaturescalarde“nedby(5.1.18)and(5.1.21).Theresultant“eldequationsare r = r€€=0.Coup lingtomatterisachievedbycovariantizationofthe de rivatives,asinYang-Millstheory,butnowthevolumeelementisalsocovariantized (witheŠ 1).AsinYang-Mills,wearealsofreetoaddnonminimalcouplingsdepending

PAGE 254

2405.CLASSICALN=1SUPERGRAVITYonthecurvature.(Fortheteleparallelismtheorywecanstillusethisactionif r is de“nedbythecommutatorof D D a.Thisactionl eads,byuseofthe“rstBianchiidentity, toanexpressionthat is purelyquadraticin t a b c.) d.Conformalcompensator In”atspace(i.e.,withoutgravity)certaintheories(e.g.,massless 4,orma ssless QCD)areinvariantunder(global)conformaltransformationsattheclassicallevel.On theother hand,whengravityispresent all theoriesareconformally invariantsinceconformaltransformationsareaspecialcaseof generalcoordinatetransformations.However,thistypeofconformalinvariancehasnophysicalsigni“cance,andispresentsimply b ecausethevierbeinautomaticallycompensa testheconformaltransformationsofother “e lds.ThisisanalogoustoglobalorbitalLorentztransformations: Any nonLorentz covariant”at-spacetheorycanbemadecovariantundertheseorbitaltransformationsin curvedspace,becausetheantisymmetricpartofthevierbeinactsasacompensator (e.g.,d4x ( 0 )2 d4x eŠ 1( e 0 m m )2).Aswesawabove,itisnecessarytointroduceadditional,local,tangent-spaceLorentztransformationstogiveameaningfulde“niti on of Lorentzinvarianceincurvedspace.Theoriesthatareinvariantunderthesetange nt spaceLorentztransformationswillautomaticallybeinvariantundertheusual Lorentztransformationsin”atspace,orwhenagaugeforlocalLorentzandgeneral coordinatetransformationsischosen. Sim ilarly,inthepresenceofgravityitispossibletogiveameaningtoglobalconformalinvariancebyobservingthatincurv edspaceitcorrespondstoanadditional in va rianceunder local scaletransformations e a= ee a, ...= ed ....(5. 1.31) Here ( x )isalocalpar ameterand d isthecanonicaldi mensionofthe“eld ...(usually 1forbosons,3 2 forfermions)whenwrittenwith ”at tangent-spaceindices.(Notethat e a,which hasnofreecurvedindicesanddescribesaboson,hascanonicaldimension1 sinceitcontainsaderivative.)Anytheoryincurvedspacethathaslocalscaleinvariancegivesa”atspacetheorywhichiscon formallyinvariant.Thetransformationin (5.1.31)isanotherexampleofatangentspacetransformation.

PAGE 255

5.1.Reviewofgravity241Thus,bothlocalLorentzandlocalscaleinvariancere”ect”atspaceinvariance propertiesofmattersystems.However,t hereisan importantdistinctionbetween Lorentzandconformaltransfo rmations:Conformalinvarianceisnotageneralproperty ofphysicalsystems,andconsequentlywedonotintroducelocalscalegeneratorsandcorrespondinggauge“eldsintoourcovariantderivatives. Asdescribedabove,theantisymmetricpartofthevierbeincanbegaugedawayby localLorentztransformations.Intheresultinggauge,generalcoordinatetransformationsmustbeaccompaniedbyrelatedlocalLorentztransformationsthatrestorethe gauge.ThelocalLorentzparameterbecomesanonlinearfunctionofthegeneralcoordinateparameter,makingconstructionofLore ntzcovariant actionsmoredicult.Similarly,localscaletransformationscanbeusedtogaugeawaythetraceofthevierbein.In fact, inlocallyscaleinvarianttheories, thedeterminantofthevierbeincanbegaugedto 1bylocalsc aletransformations.Intheresultinggauge,generalcoordinatetransformationsmustbeaccompaniedbylocalscaletransformationswithparameter determined by 1=(eŠ 1)=eŠ 1ei eŠ 4 =(1 ei ) eŠ 4 .(5. 1.32) Thelocalscaleparameterbecomesanonlinearfunctionofthegeneralcoordinateparameter.Inparticular,dimension d “elds ...nowtransformas densities undergeneralcoordinatetransfo rmations,i.e.,withanadditionalfactor(1 ei )d /4.Thefo rmalism b ecomesrathercumbersome.(Wenotethatevenintheoriesthatarenotinvariant underlocalscaletransformations,ecanstillbegaugedto1,atleastinsmallregionsof spacetime,bysomeofthegeneralcoordinatetransformations: e m m.Howev er,this resultsintheconstraint m m=0onfurtherc oordinatet ransformations,anddierenti a lly constrainedgaugeparametersareundesirablewhenatheoryisquantized(seesec. 7.3);suchgaugechoicesarepossibleuponquantization,buteshouldnotbesetto1 beforequan tization.) Wehavealre adyindicatedthateactsasacompensatorforthelocalscaletransformationsof“elds ...asgivenin(5.1.31).Infact,bymakingthe“eldrede“nition ... eŠ d /4...we canmakeall“eldsexcepteinertunderscaletransformations.In termsofthenew“eldslocalscaleinvarianc eofanactio nisequivalentt oi ndependence ofe.However,tomaintainmanifestcoordinateinvariance,itispreferabletokeep

PAGE 256

2425.CLASSICALN=1SUPERGRAVITYexplicitthedependenceone.Ontheotherhand,itisfrequentlyusefultodescribeto whatextentatheorybreakslocalscaleinvariance,bothbecauselocallyscaleinvariant theori esareinterestingintheirownright,andbecauseadecompositionintolocallyscale invariantplusscale-breakingpartscanbeh elpful.Thiscanbedonebyintroducingan additional compensating“eldintothetheory,butonewhichunlikeeisa scalar under generalcoordinateandlocalLorentztransformations.Todistinguishthistypeofcompensat orfromtheetype,wewillhenceforthrefertothemas tensorcompensators and densitycompensators, respectively.Densitycompensators(e.g., e[ a m ]forlocalLorentz, or ef or lo calscale)generallyoccuraspartsofphysical“eldsandarenottensorsunder thelocalsymmetrygroup(e.g.,generalcoord inatetransformations)oftheactionwithoutcompensators.Tensorcompensatorsarecovariant,andtheirpresenceallowsthe introductionofalocalsymmetryevenintheabsenceofacorrespondingglobal,”at spacesymmetry. Forlocalsc aletransformationsweintroduceasc alarcompensatortransformingas = e .(5. 1.33) Startingwith“eldsinvariantunder transformations,wenowmakethereplacements e a Š 1e a, ... Š d....(5. 1.34) Thenew“eldsstilltransformaccordingto(5.1.31).Thereplacement(5.1.34)isjusta -dependentscaletransformation.Hencelocalscaleinvarianceofagivenquantityis equivalent toi ndependencefrom .Forex ample,aftert herede“nition(5.1.34),the usualgravityaction (5.1.30)becomes S = Š3 2 d4x eŠ 1 ( + r ) .(5. 1.35) Thisactionisscaleinvariantbecause compensates thetransformationofeand + r Sinceitisnot -independenttheoriginalEinsteinactionwasnotscaleinvariant.Alternatively,(5.1.35)canbeinterpretedasascaleinvariantactionforthe“eld .Thescale invarianceallows tobegaugedtoone.InthatgaugeonerecoverstheusualEinstein action. Incontrast,theWeyltensor,whichistheonlypartofthecurvaturewhichis homogene ousin after(5.1.34),canforma i ndependent,locallyscaleinvariant(but

PAGE 257

5.1.Reviewofgravity243higher-derivat ive)action: SWeyl d4x eŠ 1( w)2.(5. 1.36) *** Weintro ducedthelinearizedvierbein h a masagauge“eldfortranslations;alternatively,wecanusetheanalysisofchapter3to “ndh a m.Lineari zedgravityisthetheory ofamasslessspin2“eld.Asdiscussedinsec.3.12,itisdescribedbyanirreducible onshell “eldstrength satisfying(s ee(3.12.1)) €=0.(5. 1.37) Usingtheresultsofsec.3.13,th eco rrespondingirreducible o-shell “eldstrengthisthe lineari zedWeyltensor wsatisfyingthebis ectioncondition( s +N 2 =2isan integer) w= K K w=€€€€ w€€€€(5.1.38) whichisequivalentto €€w= €€ w€€€€.(5. 1.39) By(3.13.2)appliedto N =0,thesolu tiontothisequationis w= ( €€V )€€,(5. 1.40) where h a b h€ € V( )(€€ )isatracelesssymmetrictensor.Themaximalgauge invarianceof(5.1.40)is: h a b= ( a b )Š1 2 a b c c.(5. 1.41) Thesearelinearizedcoordinatetransformationsidenticalto(5.1.1), except thatscale transformationsandLorentztransformati onsarenoti ncluded.Theycanbeaddedby introducingcompensators:thetraceandantisymmetricpartsof h a b.

PAGE 258

2445.CLASSICALN=1SUPERGRAVITY5.2.Prepotentials a.Conformal Asforsuper“eldYang-Millsandforgravity,onecandevelopaformulationfor super“eldsupergravityineitheroftwoways:(1)Studyo-shellrepresentationsto determinethelinearizedfo rmulationintermsofunconstrainedsuper“elds (prepotentials), andthenconstructcovariantderivatives,whichprovidethegeneralizationtothe nonlinearcase;or(2)startbypostulatingcovariantderivatives,determinewhatconstraintstheymustsatisfy,andsolvethemin termsofprepotential s.Inthiss ectionwe willdescribetheformerapproach,andinthefollowingsectionthelatter. a.1.Linearizedtheory Fromtheanalysi sins ec.3.3.a.1,weknowthatthe N =1superg ravitymultiplet consistsofmasslessspin2andspin3 2 physicalstates.Thecorrespondingon-shellcomponent “eldstrengthsare and (on-shellW eyltensorandRarita-Schwinger “eldstrength), totallysymmetricintheirindices,a sdiscu ssedinsec.3.12.a.Theselie inanirreducibleon-shellmultipletdescribedbyachiralsuper“eld(0) thatsatis“es theconstraint D(0) =(1) ,(5. 2.1) where(1)istotallysymmetricandisthesuper“eldcontainingtheon-shellWeyltensor (= w)atthe =0leve l.Theconstraintimplies D(0) =0.(5. 2.2) Bytheanalysisofsec.3.13,thecorresponding irreducibleo-shell super“eld strengthisachiralsuper“eld Wsatisfyingthebis ectioncondition( s +N 2 =3 2 +1 2 is aninteger) W= Š K K W= Š Š 1 2 D2€€€ W€€€,(5. 2.3) whichcanberewrittenas €DW= Š € D€ W€€€.(5. 2.4)

PAGE 259

5.2.Prepotentials245By(3.13.2)thesolutiontothisequationis W= Š i1 3! D2D( €H )€,(5. 2.5) where H€isreal.( H mightbeexpressedasaderivativeofamorefundamental“eld; thispossibilityiseliminatedwhenweexaminethe N =1theoryat thenonlinearlevel.) Were markinpassingthat Sconf= d4xd2 W2= d4x [ WD2W +( DW )2] | (5.2.6) containstheactionforlinearizedconformalgravityd4x ( w)2.Thus( 5.2.6)isthe extensionofconformalgravitytoconfo rmalsupergravityatthelinearizedlevel. Acareful examinationof(5.2.4)revealsthatthelargestgaugeinvarianceof W, writteninaformcontainingthefewestderivatives(andthusthecomponenttransformationscontainthefewestpossi blespacetimeder ivatives),is H a= D L€Š D€L.(5. 2.7) Togetinsight intothephysicalcontentof H anditstransformation,weconsidertheir componentsusing D projection. Thecomponentsof H aare h a= H a| h€= DH€| h(2) a= D2H a| h a b= Š1 2 [ D, D€] H€| a ,€= Š iD2 D€H€| A a= Š2 3 D D2DH a|Š1 6 a b c d b[ D, D€] H d| .(5. 2.8) where a b c d= i ( CCC€€C€€Š CCC€€C€€)(3. 1.22).Althoughitisconvenientto de“nethecomponent“elds h a b, a ,€, A aasabove,theseareonlythe linearized ,conformal de“nitionsofthesecomponent“elds.Inthe“nalPoincar etheorya dditi onal H aandcompensatorsuper“elddependentterms,aswellasnonlinearities,arepresent. Thecomponentsof D€L(therestof L neverenters)are: a= D€L| L1 €= D D€L| = D2L| ,

PAGE 260

2465.CLASSICALN=1SUPERGRAVITYL2 a= D2 D€L| = D D2L| =1 2 D( D2L )| = D2 D2L| ,(5. 2.9) andsimilarlyforthecom plexconjugate(butnote a= Š D L€).Thetransformations ofthei ndependentcomponentsof H are: h a=2 Re €, h€=1 2 C €Š L1 €, h(2) a= Š L2 a, h a b= Š ( C€€+ C€€)+ CC€€Re Š aIm b, a ,€= a €Š iC€€, A a=2 3 aIm .(5. 2.10) Wecantherefo regotoaWess-Zuminogaugebyusing Re L1, L2toalgebraicallygauge awayallof H aexcept h a b, a A a.Thesecanbei denti“edasthelinearizedvierbein, thespin3 2 Ra rita-Schwinger“eld,andanaxialvectorauxiliary“eld,respectively. West udytheremainingtransf ormations:Examining h a bwenote that canbe usedtoelim inatetheantisymmetricpartofthevierbein,whichidenti“esitasan i n“nitesimallocalLorentztransformation. Re removesthetrace,andisthereforea localscaletransformation.Finally, Im generatesacoordinatetransformation.Examining a ,€weidenti fythe €termasaRarita-Schwingergaugetransformation(alinearizedlocalsupersymmetr ytransfo rmation).The termisalocal S -supersymmetry transformation:itgaugesawaythetrace € ,€.From A aweidentify Im asanaxial gaugetransformation.(Notethatthelocal S -transformationofthespin3 2 “eldcontains nospacetimederivatives.Avo idingderivativesisimportantforquantization(seesec. 7.3)).

PAGE 261

5.2.Prepotentials247Thus,intheWess-Zuminogauge,the L -gaugegroupthatremainsconsistsoflocal superconformaltransformati ons:thesuper-Poincar es ubgroup(coordinate,Lorentz,and localsupersymmetry),andaxial, S -supersymmetry,andscaletransformations(seesec. 3.2.e).Localconformalinvarianceplaysamoreimportantroleinsupergravitythanin gravity:Whereastheno nconformalpartofthevierbein(itstrace)canbeprojectedout algebraically,theanalogousstatementdoesnotholdfor H (avectorisnotalgebraically reducibleinaLorentzcovariantway).Thesamedistinctionbetweenthereducibilityof thevierbeinand H applieswithregardtolocalLorentzinvariance(whichmustbemaintainedinsupergravitysimplybecausesup ersymmetrictheoriescontainspinors). a.2.Nonlin eartheory To generalizetothenonlinearcaseweexamine(asinsuper-Yang-Mills)theappropriate transformationsofthesimplestmultiplet,thechiralscalarsuper“eld.Sincegravitygaugestranslations,supergravitywillgaugesupertranslations.Wethereforelookfor themostgeneraltransformationoftheform = ei eŠ i ,=MiDM;(5. 2.11) (Wechoosetoparametrizewith DMratherthan Minordertokeepmanifestglobal supersymmetry.Thissimplyamountstoarede“nitionoftheparameters.)Wemaintain thechiralityof ( D€ =0),byrequ iringtosatisfy [ D€,] =0,(5. 2.12) whichimplies D€=0, D€ n= i €€,(5. 2.13) andhasthesolution m= Š i D€L,= D2L,€arbitrary ;( 5.2.14a) i.e., m m+D=1 2 { D€,[ D€, LD] } .(5. 2.14b) Notethattheparametersuper“eldMmustbecomplex. Inparticularthismeansthat m =( m), =(€),and€ =().Caremustb etakentodi stingu ishand

PAGE 262

2485.CLASSICALN=1SUPERGRAVITY€fromthehermi tianconjugatedquantities (=(€))and €(=()).We de“ne thetransformationofanantichiralscalarinasimilarfashion: = ei eŠ i = MiDM, m= Š iD L€, €= D2 L€, arbitrary .(5. 2.15) Thequantity Misthecomplex conjugateofM. At thispointitisclear,byanalogywithsuper-Yang-Mills,that H misthec orrect “eldtocovari antizethe mpartofthetransformationofthescalarmultipletkinetic term,sinceitslinearizedtransformationis(from(5.2.7)) H m= i mŠ i m.Wethereforecomplete H mtoasuper v ector HM=( H, H€, H m)andintro duceanexp onential eH, H = HMiDM.A sf orYang-Mills,thenonlineartransformationlawis eH= ei eHeŠ i .(5. 2.16) Wenote that H€canbetriviallygaugedawaybecausetheparameter€isarbitrary;cons equently Hisalsogaugedawayby .Toprese rvethisgaugechoice,the L gaugetransformations(5.2.14,15)mustbeaccompaniednowbycompensating€and tr an sformations.Forin“nitesimalwehave (eH mi m)= Š ( m m+ € D€+ D)(eH mi m) +(eH mi m)( m m+D+€ D€).(5.2 .17) (Thisequationistobeinterpretedasanoperatorequationactingonanarbitrarysuperfunction totheright.)Thisimpliesthatwemustcancel Dand D€termsontherighthandsideandhence = eHeŠ H= eH D2LeŠ H,€= eŠ H €eH= eŠ HD2 L€eH.(5. 2.18) However,wewillnotrestrictourselvesto thisgaugeinthesubseq uentdiscussion.

PAGE 263

5.2.Prepotentials249a.3.Covariantderivatives Ournexttaskistoconstructcovariantderivatives A=( €, a).Byanalogy withYang-Millstheorywewouldrequire( A)= ei ( A),i.e., A = ei AeŠ i or A= i [, A].However,sincew eexp ectlocalLorentztransformationstobepresent, wecang eneralizeto ( A)= LA Bei B, LA B=( L L€€, L L€€),(5.2 .19) Wede “ne,byanalogywithEinsteinstheory(5.1.15),covariantderivativesthattakethe form: A= EA+A M +A€€ M€€,(5. 2.20) wherethe M sareLorentzrotationoperators.Theiractionisde“nedin(5.1.13).Again inanalogywithordinarygravity,weadheretoalate-earlyindexconventiontodistingu is hb etweenquantitieswithcurvedindices(thattransformonlyunderthe-gauge gr o up)andquantitieswith”atindices(thattransformonlyundertheactionofthe Lorentzgenerators M and M€€).Theform(5.2.19)assumesthattheLorentztransformations LA Bactinthe usualmanner:Spinorsandvectorsdonotmix,andboth rotatewiththesameparameter.Foranin“nitesimalLorentztransformation, LA B= A B+ A B,thecovarian tderivat ivestransformas A=[ i A]+ A BB,(5. 2.21) where A B=( €€, a b)and(from( 5.2.19)) a b= €€+ €€( isthechiralrepresentationconjugateof €€:seebelow).Thisi mpliesthefollowingtransformationlawsforthe connections: A€€=[ i ,A€€] Š EA€€+ A DD€€+ €€A€€Š A€€€€.(5. 2.22) Thereisacertainamountofarbitrarinessinde“ningconnectionsthattransform properly:OnecanalwaysaddtoA anytensor KA thattransformscovariantly.As willbediscussedinthenextsection,thisarbitrarinessisphysicallyirrelevant.Astandardwayto“ndconnectionsistocomputethe anholonomycoecientsCAB Cde“nedby

PAGE 264

2505.CLASSICALN=1SUPERGRAVITY[ EA, EB} = CAB CEC. Suitableconnectionscanbede“nedaslinearcombinationsofthe C s. PushingtheanalogywithYang-Millsfurther,wecantrytoconstructthefollowing chiral-representation covariantd erivat ives: E€ D€, E eŠ HDeH, E mŠ i { E, E€} .(5. 2.23) However,atthispointtheanalogywithYang-Millstheorybreaksdown.Thesederivativesarenotcovariantfortworeasons:(1)Actingonnontrivialrepresentationsofthe Lorentzgroup,theyarenoncovariantbecausetheyhavenoconnections(thisiseasily cured);and(2)moreseriously,evenactingonscalarstheyarenoncovariantbecause€isnotc hiral.Thus, E€=[ i E€] Š ( E€€) E€=[ i E€]+ €€ E€+ E€,(5. 2.24) where €€= Š1 2 E(€€ )= Š1 2 D(€€ ),= Š1 2 E€€= Š1 2 D€€.(5. 2.25) Theterminvolving isharmless:itisjustaLorentzrotation,andwillbeperfectly covariantafterweintroduceLorentzconnections.(Thereisaslightproblem,however. Theindicesin(5.2.19,20)are”atspinorindiceswhereasthosein(5.2.24,25)arecurved indicesinanalogywithourdiscussionofordinarygravity.Thereforeitisnotquitecorrecttoidentifythe in(5.2.25)withtheonein(5.2.20).Wewill“ndasolutionforthis shortly.) Bycontrast,thetermproportionaltoisa supers pace scaletransformation whichisnotpartofouroriginalgaugegroupasde“nedby(5.2.19)and(5.2.20).For thetimebeingwein tro duceintothetheorya(density) compensator thattra nsforms as: =[ i ,] Š .(5. 2.26) Lateron,willbedeterminedintermsof H .Withthiso bj ect,wecanconstructa covariantspinorderivative:

PAGE 265

5.2.Prepotentials251 E€ E€= D€, E€=[ i E€]+ €€ E€.(5. 2.27) Thecomplexconjugateof E€iscovariantnotwi threspecttobutratherwithrespect to ; ho we ve r,justasintheYang-Millscase,wecanuse eHtoconv ertanyobject covariantwithrespectto intoanobj ectcovariantwithrespectto.Weobtain E eŠ H DeH=( eŠ H eH) E E,(5. 2.28) whereisthechiral-representationHermitianconjugateof(asinsuper-Yang-Mills: see(4.2.37)and(4.2.78)).Thecovarianttransformationof Eis E=[ i E]+ E,(5. 2.29) where = eŠ H eH. Thespinorvielbeinsthatwehaveconstructedtransformasin(5.2.21)butwith theimportantrestrictionthatthe parameterofLor entzro tations, A B,mustbed etermined(bythede“nitionin(5.2.25)andthosefollowing(5.2.29)and(5.2.21))intermsof theparameterofsupercoordinatetransformationsM.Inthe discussionofordinary gravity(see(5.1.10,11)),wesawthatananalogoussituationoccurredonlyiftheantisymmetricpartofthevierbeinwasgaugedaway.Wealsohavetherelatedproblemthat thefreei ndexonthevielbeiniscurvedwhereastheindexin(5.2.19)is”at(andconsequentlythe problemofidentifying €€with €€).Thissituationari sesbecausethevielbein E€€asde“nedin(5.2.27)isgivenby €€andthushasnosymmetricpart(i.e., E(€€ )=0).Thesolutionist orestore themissingpartbyin tro ducinganewsuper“eld N€€.InageneralL orentzframethespinorvielbein(5.2.27)ismodi“edto E€= N€€ D€,(5. 2.30) where N€€isanarbitrary SL (2 C )matrixsup er“eld( detN =1).Itac tsasacompensating“eldfortangentspaceLorentztransformations.(Thisisanalogoustogeneralizing fromaframewheretheusualvierbeinissymmetric.)The N -dependenceoftheother equationscaneasilybefoundbysimplyperf ormingthegeneralLorentztransformation whichtakes E€€from €€to N€€.Thequantity N mapsbetweencurvedand”at

PAGE 266

2525.CLASSICALN=1SUPERGRAVITYspinorindices.Thispermitsustosolvetheproblemofidentifying €€with €€.Since webe ganinthegauge N€€= €€,thetwoq uantitiesareequal.Furthermore,aslongas were maininthisgauge,weneednotbecarefultodistinguishcurvedand”atspinor i ndices. Adistinc tionmuststillbemadebetween”atandcurvedvectorindices. Wenowa ttempttoconstructthevectorcovariantderivativebyanalogywith Ya ng -M illstheory: E m= Š i { E, E€} .(5. 2.31) Thetransformationlawfollowsfrom(5.2.27,29): E m=[ i E m]+ E€+ €€ E€Š i ( E€€) E€Š i ( E€ ) E.(5. 2.32) De“ning EM ( E, E€, E m)wecanwrite( 5.2.27,29,32)as EM=[ i EM]+ M N EN.(5. 2.33) However,becauseoftermslike m€= Š i E€€,whicharen otpresentin(5.2.19), EMis notquitecovariant. Thetermswewanttoeliminateare(spinor) derivativesoftheLorentztransformationparameter €€;therefo re,theremedyistointroduce(spinor)Lorentzconnections into(5.2.31).Theseconnectiontermswillrede“ne E msothatittransformscovariantly. To “ndtheconnections,wede“nea(noncovariant)setofanholonomycoecients CMN Pby [ EM, EN} = CMN P EP.(5. 2.34) Fromthetransfo rmationsin(5.2.27,29,32)weobtain CMN P=[ i CMN P]+ E[ MN ) P+ [ M | R CR | N ) PŠ CMN RR P.(5. 2.35) Inparticularwe“nd C€€€=[ i C€€€]+ E(€€ )€+ (€ |€ C€ |€ )€Š C€ ,€€€€, C m ,€ r=[ i C m ,€ r] Š E€ m r+ m n C n ,€ rŠ C m ,€ n n r+ €€ C€ m r+ i m €€,(5. 2.36)

PAGE 267

5.2.Prepotentials253andcorrespondingequationsfortheconjugates C and C m r.From( 5.2.36)wesee that Š1 2 C( € ,€ )€transformsastheneed edspinorconnection: [ Š i1 2 C( € ,€ )€ E]= i ( E€ ) EŠ i1 2 C( € ,€ )€ EŠ i1 2 €€ C( € ,€ )€ E.(5. 2.37) Thereforewede“ne E a a m[ E mŠ i1 2 C( € ,€ )€ EŠ i1 2 C (€ € ) E€],(5.2 .38) where a m N N€€inthegauge N€€= €€.Thev ectorvielbein E atransformscovariantly. Wehavealre adyconstructedoneoftheLorentzconnectionsuper“eldsA (as notedabove ,inthe gauge N = wen eednotdistinguishcurvedand”atspinor i ndices).Wecanconstructtheremainingconnectionsinthestandardway(seesubsec. 5.3.b.1)fromtheanholonomycoecients CAB Cde“nedby EA ( E€, E, E a),where E€= €€ E€, E= EinourparticularLorentzgauge,and E aisgivenin(5.2.38). Alternativel y,wecanuse C directly.Aswesawabove, anappropriatelytransformingspinconnection€ isgivenby1 2 C( € ,€ )€.For€€€wehave achoi ce:Both Š1 4 C€ (€ € )andŠ 1 2 [ C€ ,€€+ C€ ,€ ,€Š C€ ,€ ,€]transfo rmappropriately.Ingeneral,any linearcombinationofthesecanbeus edasaspinconnection.Furthermore € (€ C€ ), d dis Lorentzcovariant,andcanbeaddedto€€€;sees ec.5.3.a.3.Wechoose €€€=1 4 [ Š C€ (€ € )+ € (€ C€ ), d d]. (5.2.39a) Wealre adyhad €€= Š1 2 C (€ € ).(5. 2.39b) Wealsohaveco rrespondingexpressionsforthecomplexconjugates ,€ Thevectorconnectionisde“nedby: a = Š i [ E€ + ,€€€ + E€ +€ + ( | € | )].(5.2 .40)

PAGE 268

2545.CLASSICALN=1SUPERGRAVITYasfollo wsfrom €= Š i {, €} a.4.Covariantactions InsupergravityLagrangianscannotbeinvariantbecauseallourquantities,includin gs calars,transformundertransformations.AtbestaLagrangiancantransformas atot alderiva tive: IL = Š ( Š )MDM(MIL ).(5.2 .41) Thiscanberewrittenas IL = iIL = i [, IL ]+ i (1 ) IL ,(5. 2.42) where = i MD M= i ( Š )M[ D MM+( DMM)].(5.2.43) Equation(5.2.42)isthetransformationlawfora density. Itiseasytocheckthata scalartimesadensityisalsoadensity. Byanalogywithgravity,wetakethevielbeinsuperdeterminant E = sdetEA Masa candidateforadensity.Indeed,from(3.7.17) EŠ 1= Š ( Š )MEŠ 1[( EŠ 1)M A EA M],(5.2 .44) where,from(5.2.21) EA M= i ( EA M)+( EAM)+ A BEB M.(5. 2.45) (HowevertheLorentzrotationtermstriviallydropoutof(5.2.44).)Consequently, EŠ 1= Š i ( Š )MEŠ 1[( EŠ 1)M A EA M] Š ( Š )MEŠ 1[ DMM] = i [, EŠ 1] Š ( Š )M( DMM) EŠ 1= iEŠ 1.(5. 2.46) Therefore,invariantactionscanbeconstructedasintegralsofproductsof EŠ 1and scalarquantitiesoftheappropriatedimension( E itselfisdimensionless): ( EŠ 1IL )= i ( EŠ 1IL ).

PAGE 269

5.2.Prepotentials255Forsupergravitywew anttohavetheusualEinsteinterminthecomponentaction; hencewerequire SSG=1 2 d4xd4 EŠ 1ILSG=1 2 d4x eŠ 1ILcomponent.(5. 2.47) Thisimpliesthat ILSGisadimensionlessscalar.Ingeneral,theonlydimensionless scalarsinthetheoryareconstants,sowemusttake(however,seebelow) SSG1 2 d4xd4 EŠ 1(5.2.48) asthelocallysupersymmetricinvariantactionforPoincar esupergravity. Thevielbeinsuperdeterminantcanbeworkedoutinastraightforwardmanner using( 3.7.15).We“nd E = sdet [ EA M]= sdet [ EA M]=22sdet [ EA M( H )].(5.2.49) However,variationoftheactionwiththe“eldconsideredasanindependentvariable leadstoasingular“eldequation:( E )Š 1=0.Thisisnot surprising:wasintroduced asadevicetosimplifytheconstructionofth ecovariantderivati ves,andits houldbe relatedtothefundamentalprepotential HM. b.Poincar e Wenowconsi derspeci“cformsforthecompensatorintermsof H .Aswediscussedearlier,the-gaugegroupincludessuperconformaltransformations:Thus,the covariantderivativeswehaveconstructedareappropriatefordescribingasuperconformallyinvarianttheory.Thesuperconformalactionisthenonlinearversionof(5.2.6).It isafunctionalof HMonly;dropsoutcompletely.TodescribePoincar esupergravity, wew illhavetobreaktheextrainvariance,i.e.,thecomponentsuperscaleinvariance. To “ndanappropri ateexpressionforwerecallthatittransformsasa(noncovariant) density( 5.2.25,26) =[ i ,] Š .(5. 2.50) Theonlyotherdimensionlessobjectthattransformsasadensitywithrespecttoand not is E (, H )(see(5. 2.46)).Wethereforeexpressintermsof HM(implicitly)by writing

PAGE 270

2565.CLASSICALN=1SUPERGRAVITY4 n= En +1.(5. 2.51) (Thisparticularparametrizationwillproveconvenientwhenwritingtheexplicitaction (5.2.65).)However,thisrelationisnotpreservedbythefull-group:Ifwetransform bothsidesof(5. 2.51),using(5.2.46,50)we“ndtherestriction Š 4 n = Š ( n +1 )(1 i ),(5.2 .52) or,moreexplicitly, (3 n +1) D€€=( n +1)( m mŠ D).(5.2 .53) Thisisanacceptablerestrictiononthegaugegroup:Wecanshowthatitcorrespondstoreducingthe component localsup erconformalgrouptothesuper-Poincar e group.We note thatwhen n = Š1 3 (5.2.53)setsthe chiral quantity m mŠ Dto zero:i.e.,using(5.2.14)therestrictioncanbewrittenas n = Š1 3 : DŠ m m= D2DL=0.(5. 2.54) Ontheotherhand,for n = Š1 3 theconditionrestricts€:(5. 2.53,54)imply n = Š1 3 : D2€=0.(5. 2.55) Wean alyzethecase n = Š1 3 “rst.Since€isunrestricted,wecanstilluseitto gaugeaway H€;wethenn eedonlyreconsiderourdiscussionoftheWess-Zuminogauge for H msubjecttotheres triction(5.2.54).Thisrestrictionimpliesthefollowingrelations amongthecomponentsof Lin(5.2.9): = i a a, = Š i €L1 €, aL2 a=0.(5. 2.56) Thusthelocalsuperconformaltransfo rmationsarereducedtothoseoflocal super-Poincar e: and havebeenremovedasindependentparameters.Further,a differential constrainthasbeenimposedononeoftheparameters,the L2thatweusedto gaugeawayextracomponentsof H m: B = mh(2) mcannolongerbeeliminated.Consequently( cf.thediscussionfollowing(5.2.10)),we“ndthattheminimalsetofcomponent

PAGE 271

5.2.Prepotentials257“eldsdes crib ing N =1Poincar esupergravityare thevierbein(then onlinearextension of h a b),thegravitino a ,anaxialv ector“eld A a,andina ddition thecomplexscalar “eld B (whichcouldbegaugedawayinthesuperconformalcase). For n = Š1 3 ,weca nnotuse€togaugeawayallof H€.Wecan gaugeawayparts ofitbyusingupallthecomponentsof D€€andhence,becauseoftheconstraint (5.2.52),thoseof D2DL.A gainthelocalsuperconfo rmalgrouphasbeenreducedto thesuper-P oincar egro up: and nolongerenterasindependentparameters.Wewill discussthecomponentcontentof n = Š1 3 supergravitylater. For n =0,(5. 2.51)impliesthat E =1andhen cethattheaction(5.2.48)vanishes. Itisclearfrom(5.2.53)thatfor n =0thepar ameterMsatis“es( Š )MDMM=0. Thisispreciselytheconditionthatthesupercoordinatetransformationparametrizedby aresupe rvolumepreserving.However,the n =0caseis uniquebecauseitcontains a(constrai ned)dimensionlessscalar V (anabeliangaugeprepotential)whichcanbe usedtoconstructanaction.Furthermore,theconstraint(5.2.51)isinvariantunderan arbitrarylocalphaserotationof:= eiK5, K5= K5= eŠ H K5eH.Thisinv ariance canbeusedtochooseagaugewhere=.Another consequenceof(5.2.51), E = E =1,isthatweh aveimposedap artialgaugeconditionon H :The hermiticity conditi onthat EŠ 1satis“es(s ee(5.2 .60)below)impliesthat(1 eŠ H)=1, i.e., 1 H=0.Thisisachievedbyc hoosingthe gaugewhere H= Š i D€H€insteadof zero,sothat H = DH€ D€+ D€H€Dwherethe D and D preceding H€actonall objectstotheright.Thecaseof n =0w illbediscu ssedinmoredetailinthefollowing sections. To “ndtheexplic itexpressionforintermsof H ,weuse(5 .2.49)and(5.2.51) andwrite 4 n= eŠ H 4 neH= eŠ H En +1eH.(5. 2.57) (Forsimplicitywehaveassumedthat n isreal;thegeneralizationtocomplex n is straightforwardbutnotinteresti ng).Therefore,wemustcompute E .Altho ughthis couldbedonebybruteforce,amoreelegantprocedureispossible: In(5.2.28)wede“ned

PAGE 272

2585.CLASSICALN=1SUPERGRAVITYE= eŠ H EeH,(5. 2.58) where Eisthehermitianconjugateof E€.Therighthandsidecanbeinterpretedasa coordinatetransformationwithimaginaryparameter iHMinsteadof i M.Byhe rmitian conjugation E€isthecoordinatetransformof E€(thehermitianconjugateof E). Therefore,anycovariantconstructedfrom Eand E€canbeobtainedbyacomplex coordinatetransformationfromthecorrespondingobjectconstructedoutof E€and E. Thisisthecasefor E a,andalsofo rthevielbe insuperdeterminant. Thefullnonlineartransformationofthesuperdeterminantfollowsfrom(5.2.46): ( EŠ 1)= EŠ 1ei =( ei EŠ 1eŠ i )(1 ei ).(5.2 .59) Bythesam emethodusedt oderivethisr esultfrom E A= ei EAeŠ i ,from EA= eHEAeŠ H(5.2.60a) wehave EŠ 1= EŠ 1eH(5.2.60b) andhence EŠ 1= eŠ H EŠ 1eH(1 eŠ H).(5.2 .60c) Substituting(5.2.60)into(5.2.57)we“nd 4 n=[ E (1 eŠ H)]n +1,(5. 2.61) or,usingtheoriginalconstraint(5.2.51), 4 n=4 n(1 eŠ H)n +1.(5. 2.62) Finally,substituting(5.2.62)and(5.2.51)into(5.2.49)we“nd 4 n=4( n +1)(1 eŠ H)( n +1)22 n En +1,(5. 2.63) or =[(1 eŠ H)( n +1)28 n En +1 4 ]Š 1.(5. 2.64)

PAGE 273

5.2.Prepotentials259Thuswehavesolvedforintermsof H .Usingtheser esults,wecanrewritethe n =0 supergravityaction(5.2.48)intermsoftheunconstrainedsuper“eld HM: SSG= 1 n 2 d4xd4 [ En(1 eŠ H)n +1 2 ].(5.2 .65) (Thefactor1 n givestheappropriatenormalizationforthephysicalcomponentactions andforthesupersymmetric-gaugepropagators: En=(1+)n 1+ n ;see,e.g., (7 .2 26) .) Thisactionisinvariantunderthegroupoftransformationsrestrictedby (5.2.53). c.Densitycompensators Formany purposes,e.g.,quantization,itisawkwardtoworkwiththeconstrainedgaugegroup.Asdescribedinsec3.10wecanenlargetheinvariancegroupofa theorybyintroducingcompensating“elds.Inthiscase,theconstraint(5.2.53)was in tr o ducedbytherelation(5.2.51),whichisnotcovariantunderthefullgaugegroup. We ch oo seourcompensatorstorestorethecovarianceof(5.2.51)underthefullgroup. Forthe n = Š1 3 caseasuitablecompensating“eldisa chiraldensity thattransformsas =[ i ]+1 3 ( DŠ m m) D€ =0.(5. 2.66) (Thefactor1 3 gives thesameweightasadensitymatte rmulti plet:seebelow.)From (5.2.46,50,66),itfollowsthatthecovariantversionof(5.2.51)is: Š4 3 = 2E2 3 .(5. 2.67) Eq.(5.2.66)canberewrittenas ( 3)=( 3) ei ch,ch= mi m+iD.(5. 2.68) Thechiralparameter(1 i ch)= Š m m+ Dcanbeusedtoscale arbitrarily:In particular,ifwe choosethegauge =1,from(5. 2.67,68)werecovertheconstraints (5.2.51,54),r espectively. Inthe n = Š1 3 case,theconstrainedobjectisthelinearsuper“eldexpression(cf.

PAGE 274

2605.CLASSICALN=1SUPERGRAVITY(5.2.53,54)) D€€+(n +1 3 n +1 ) D2DL.Hen ce,asuitablecompensatorisacomplexlinear super“eld, D2=0.Wedete rmineitstransformationpropertiesbyrequiringthatits variationis linearandthatcanbescaledto1.(I tsho uldberecalledfromthediscussioninsec.4thatacomplexlinearsuper“eldcanalwaysbeexpressedintermsofan unconstrainedspinorsuper“eld:= D€ €(4.5.4).)Sincetheproductofachiralanda linearsuper“eldislinear,wecanalwayshaveaterm( DŠ m m)in .Toscale to1,wen eedaterm( D€€), butsincetheproductoftwolinearsuper“eldsis not linear,suchatermmustcomefromthecombination( D€€) Š € D€= D€(€). Thisleadsuniquelyto =[ i ,]+( D€€)+(n +1 3 n +1 )( DŠ m m)(5.2 .69) and 4 n=3 n +1En +1.(5. 2.70) Asintheminimal n = Š1 3 case,choosingthegauge=1leadsbacktotheconstraints (5.2.53,51).Weobservethatfor n =0,(5. 2.70)implies=(1 eŠ H).(Wec ande “ne a linearcompensator Š3 n +3 3 n +1 intermsofbothand ;thise nlar gesthegauge groupbyachiralscaletransformation,andresultsin n -independenttransformationlaws forboth and). Wecanr epeatthecomputationsofeqs.(5.2.57-64)includingthecompensators;we “nd n = Š1 3 ,=[n Š 1n +1]Š (3 n +1 8 n )[(1 eŠ H)n +1 E2 n]Š (n +1 8 n ), n = Š1 3 ,= Š 11 2 (1 eŠ H)1 6 EŠ1 6 .(5. 2.71) The n =0superg ravityaction(5.2.48,65)takestheform n = Š1 3 SSG( H ,)=1 n 2 d4xd4 En(1 eŠ H)n +1 2 []3 n +1 2 n = Š1 3 SSG( H )= Š3 2 d4xd4 EŠ1 3 (1 eŠ H)1 3 .(5. 2.72)

PAGE 275

5.2.Prepotentials261Theseareinvariantunderthefullgaugegroup;after -integration,theybecomethe usualPoincar es up er gravitycomponentactionforthegraviton,gravitino,andauxiliary “elds.The n =0casew illbediscu ssedlater. d.Gaugechoices Thecomponent“eldcontentoftheactions(5.2.72)ismanifestinaWess-Zumino gaugewherethetransformationshavebeenusedtoremovealgebraicallythegauge componentsofthesuper“elds.Sincethegroupisnowunconstrained,wecanchoose thegauge H=0and H mwithonly h , A asdescribedfollowing(5.2.10).Therewe foundthattheremaini ng gaugefreedomisparametrizedbythe Im , Re + iIm and componentsof L;theseco rrespondtoLorentz,coordinate,localsupersymmetry, scale+ i chiral,and S -supersymmetrytransformationsrespectively. For n = Š1 3 ,wede“nethe(lineari zed)componentsof by u = | u= D | S =S+ i P= D2 | .(5. 2.73) UnderthegaugetransformationsthatremainintheWess-Zuminogauge(weneedthe compensatingtransformations(5.2.18)andinaddition L1 €=1 2 C €in(5.2.10)), thesecomponent strans formas u = Š1 3 ( + i a a), u=1 3 ( + i €L1 €), S = i1 3 aL2 a.(5. 2.74) Inwritingthesetransformationlawswehavelinearizedthefullin“nitesimaltransformationof(5.2.66)andkeptonlythosetermsindependentof H aand .( Thisisanalogous to th ea pproximationgivenin(5.2.7)ascomparedtothefullin“nitesimaltransformationgivenin(5.2.17).)Thescaleandaxialtransformationsparametrizedby canbe usedtoscale u to1,andthe S -supersymmetrytransformationcanbeusedtogauge uaway.Thusweseethat actsasacompensatorfortheconstrainedpartofthe group,i.e., u and uarecompensatorsforthecomponentsuperscaletransformations (scale,chiral U (1),and S -supersymmetry).Setting u Š 1= u=0restoresthe

PAGE 276

2625.CLASSICALN=1SUPERGRAVITYconstraintandleavesonlythecomponentsuper-Poincar etransfo rmations.Theremaining“elds S S alongwith h ,and A from H marethecomponent“eldsofminimal supergravity. Withintheframeworkofthe n = Š1 3 theory,thech oiceofcompensatorisnot unique.Anysuper“eldthatcontainsonlysuperspin0(components u and u)canbe usedasacompensator.Inglobalsupersymmetrywefoundanumberofsuchmultiplets: thevariantrepresentationso fs ec.4.5.d.Thuswecanreplace 3Š 1in(5. 2.66-72)by thechiral“eldstrengths= D2V of(4.5.56)or= D2Dof(4.5.66).Thenew components u uofthenewmultipletsarecompletelyequivalenttothecorresponding componentsof .Onthe otherhandthe S componentischanged.Wethushavetwo casesina dditi onto S = D2 | =S+ i P (2) S = D2 | = D2 D2V | =1 2 { D2, D2} V | +1 2 [ D2, D2] V | =1 2 { D2, D2} V | + i €[ D, D€] V | =S+ i aP a,( 5.2.75a) or (3) S = D2 D2D| = Š i €D2 D€| = aS a+ i aP a.(5. 2.75b) Incase(2)theauxiliary“eldsaretherealscalarSandthedivergenceoftheaxialvector P a(insteadofthepseudoscalarP).Alsoi ncase( 2)wecanmakethereplacement V = iV.Thee ectofthisatthecomponentlevelisthat S takestheform S = aS a+ i P. Theauxiliary“eldsarethedivergenceofavector S aandthe pseudo sc alarP.Incase(3),bothauxiliary“eldsaredivergences. For n = Š1 3 ,beforeweintro ducedcompensators,therestrictedgaugegroup (5.2.54)couldbeusedtoeliminateallbutthe h , A and S componentsof H m,where S mD2H m| (cf.thediscussionafter(5.2.56)).Inthepresenceofthecompensators, thefullgaugegroupwasusedtogaugeaway S from H m; S appearedinthecompensatorinstead.However,onlyincase(3)aboveis S (inthecomp ensator)adivergence; ther eforethisi stheonlyca seinwh ichthetheory with thecompensatoriscompletely equivalentto thetheory without thecompensator. For n = Š1 3 ,0,wecan alsogototheWess-Zuminogaugeand“ndthecomponents

PAGE 277

5.2.Prepotentials263ofthenonminimaltheor y.Wede “nethe(linearized)componentsof: u = | u= D | S = D2 | = D |Š (4 n +1 3 n ) D | V a= D D€ | €= D2 D€ | .(5. 2.76) Thetransformationsanalogousto(5.2.74)are u = Š Š (n +1 3 n +1 )( Š i m m), u= S = i (n +1 3 n +1 ) mL2 m, = Š i € €+ i (n +1 3 n +1 ) €L1 €, V a= Š i1 2 a Š i € €€, €= €.(5. 2.77) Asbeforewe canscale u to1andgauge uto zero.Theremainingcomponents V a, aretheadditionalauxiliary“eldsofnonminimalsupergravity. On ceagainwenotethatthecase n =0isdiere nt;thetran sformation u isindepe ndentof Im (theaxialrotations)sothatthisgaugeinvariance survives intheW essZuminogaugefor.Since=(1 eŠ H),wehave S = €= €=0and V a= bT[ a b ], where T[ a b ]isarealantisymmetrictensorgauge“eld. e.Su mmary Wesu mmarizeheresomeofthequantitie sthatwehavecon structedsofar: E€= D€, E= eŠ HDeH, H = HMiDM, E€= E€, E= E,= eŠ H eH, E€= Š i { E, E€} ,[ EM, EN} = CMN R ER.( 5.2.78a)

PAGE 278

2645.CLASSICALN=1SUPERGRAVITYE€= N€€ E€, E= N E, E€= N€ €( E€+ i1 2 C€ ,( € )€ E+ i1 2 C (€ € ) E€), N€ €= N N€€, det ( N€€)=1.(5.2 .78b) Thefactor N isequalto inasuitableLorentzframe.Thesuperscalecompensator takesth eform n = Š1 3 := Š 11 2 (1 eŠ H)1 6 EŠ1 6 n = Š1 3 ,0:=[n Š 1n +1]Š (3 n +1 8 n )[(1 eŠ H)n +1 E2 n]Š (n +1 8 n ).(5. 2.78c) Thetildequantitiesarechiralrepresentationhermitianconjugatesde“nedbyanalogy withthewayisde “nedfrom.TheLorentzconnectionsuper“eldsA andA€€takesimpleformswhenexpressedasfunction softhea nholonomycoecientsde“nedby [ EA, EB} = CAB CEC.Theyaregivenby = Š1 4 [ C ,( € )€Š ( C ), d d], €€= Š1 4 C (€ € ), a = Š i [ E€ + E€ Š C ,€ Š C ,€€€ + € +€ ],(5.2 .78d) and€€€,€ a€€areobtainedbychiral-representationcomplexconjugation(i.e., thetildeoperation). f.Torsionsandcurvatures Fromthecov ariantderi vatives A= EA+A( M )wede “netorsionsandcurvatures [ A, B} = TAB CC+ RAB( M ).(5.2 .79)

PAGE 279

5.2.Prepotentials265Usingtheexplicitform,we“ndthatthetorsionsandcurvaturesidenticallysatisfythe followingconstraints: T €= T ,€ cŠ i €€= T c=0, R ,€= R €€= T ,( b c )Š1 2 b cT d d=0 .( 5.2.80a) Wealso “ndfurtherconstraintswhos eformdep endsonthevalueof n .Theseare n = Š1 3 : T b c=0, n = Š1 3 ,0: T b c=1 2 €€ T d d, R = Šn 3 n +1 ( €+n Š 1 2(3 n +1) T€) T€,(5. 2.80b) where R = i1 4 T€ ,€ and T= T b b.Weref erto theseas conformalb reaking constraints.Ina treatmentthatdoesnotusecompensators,theseconstraintshavetobe imposeddirectly,tobreakconformalsupergravitydowntoPoincar esupergravity.Inthe compensatorapproachtheyarisenaturallywhenagaugechoiceismadetobreakthe conformalinvariance;seesec.5.3.b.6,7. Anotherwaytoex presstheconstraintsonthetorsionsandcurvaturesistowrite thegradedcommutatorofthecovariantderivatives.Fortheminimaltheory( n = Š1 3 ) we “nd {, } = Š 2 RM, {, €} = i €, [ b]= Š iC[ R €Š G€] + iC[ W€€€ M€€Š ( G€) M ] Š i ( € R ) M, [ a, b]={[ C€€W + C( €G€) Š C€€( R ) ] + iCG€€

PAGE 280

2665.CLASSICALN=1SUPERGRAVITY+[ C€€( W +( 2 R +2 R R ) C ) Š C( €G€)] M }+ h c .,(5.2 .81) where W, G a,and R arei ndependentLorentzirreducible“eldstrengths de“ned by theseequations. Fornonmi nimaltheories( n = Š1 3 ,0)thegr adedcommutatorstaketheforms {, } =1 2 T( )Š 2 RM, {, €} = i €, [ b]=1 2 T€Š iC[ R +1 4 T] €+ i [ CG€Š1 2 C(( +1 2 T) T€)+1 2 ( €T) ] Š i [ C( G€) M +(( €Š T€) R ) M] + iC[ W€€€ M€€+ i1 3 W€ M€€],(5.2 .82) where W, G a,and Tarethei ndependenttensors.Intheseequations, R and Ware de“nedintermsof T.Thequantity R wasde “nedin(5. 2.80b)and Wisgivenby W= i [1 2 €( €+1 2 T€)+ R ] T(5.2.83) TheexpressionforthecommutatoroftwovectorialcovariantderivativesinthenonminimaltheorycanbecalculatedfromtheBianchiidentity; [ a, {, €} ]= { €,[ a, ] } + {,[ a, €] } .(5. 2.84)

PAGE 281

5.3.Covariantapproachtosupergravity2675.3.Covariantapproachtosupergravity a.Choice ofconstraints Intheprevioussectionweshowedhowtostartwithunconstrainedprepotentials andconstructoutofthemacovariantsuper“eldformulationofsupergravity.Herewe dothereverse:Startingwithamanifestlycovariantbuthighlyreduciblerepresentation ofsupersymmetry,weimposeconstraintsonthegeometryandsolvethemintermsof theprepotentials.Prepotentialsareessentialforsuper“eldquantization,whereasmanifestlycovariantf ormulationsmakecouplingtomatterstraightforwardandallowusto developanecientandpowerfulbackground“eldmethodforthequantumtheory. a.1.Compensators Asdiscussedinsec.3.10,itisoftenusefultointroduce(additional)localsymmetriesrealizedthroughcompensators.Asdiscussedinsec.5.1,ingravitationaltheories therearetwotyp esofcompensators:(1)densitycompensators,whichtransformnoncova riantlyunderthefulllocalsymmetrygroup,andthusappearincovariantquantities onlyincombinationwithother“elds;(2)tensorcompensators,whichtransformcovariantly,andthusallowtherealizationofsymmetriesthatmaynotbeinvariancesofthe entiretheory.Densitycompensatorsallo wthe linearrealizationofsymmetriesthat wouldotherwisebereali zednonlinearly(as,forexample,innonlinear models,or Lorentzi nvarianceingravitywith spinors).Whensuchsymmetriesareglobalsymmetriesofsometheory(atleaston-shell),the (super)s pacetimederivativeofthedensity compensatorcanappearasagaugeconnection.(Ifsome“eldtransformsas = then itiseasytoconstr uctaconnectionas since = .) Ontheotherhand, ifthesymmetriesonewantstorealizelinearlyandlocallyare not evenglobalsymmetriesoftheentiretheory,itisnecessarytointroducetensorcompensat orstocancelarbitrarygaugetransformationsofthistypeintermsthatarenot invariant.Thisphenomenonwasillustratedinsec.3.10.binthediscussionofthe CP (1) model,andin(5.1.35),wherethecompensator pe rmitsageneralizationoftheEinstein-Hilbertactiontoanact ionwithanadditionallocalscaleinvariance.(Thisdoes not implyanynewphysics.Thegaugeinvariancewithrespecttoscaletransformations mustbe“xedju stasanygaugeinvarianceandthemostconvenientchoiceis =1.)

PAGE 282

2685.CLASSICALN=1SUPERGRAVITYIngravitationaltheories,somesymmetriesneed both typesofc ompensators.This isbecausethesymmetriesarere alizedtwice.Forexample,inordinarygravity,thedensitycompensatoreŠ 1transformsas eŠ 1= Š m mŠ 4 ,where m mrepresentsthelocal scaletransformationpartofthegeneralcoordinatetransformationparametrizedby m, and isanindependentlocaltangentspaceparameter.ThequantityeŠ 1isthusa gauge “eldforsca letransformationsin m, butalsoa compensator forthetransf ormations parametrizedby .Ita llowsthefull minvariancetoberealizedlinearly;italsoallows localscaletransformation stoberea lizedlinearlyvia .Howev er,mostgravitational theori esarenotlocally(tangentspace)scaleinvariant:Thus,ifwestillwanttorepresent localscaletransformations,wemustintroduceatensorcompensator = ,tocancelthe transformationofnoninvarianttermsinanaction.Tosummarize:(1)eŠ 1isa density compensatorforlocalscaletransfo rmations,allowingthemtoberealized linearly inlocallyscale-invarianttheories;while(2) isa tensor compensatorforlocalscale transformations,allowingthem toberealizedlin early(wheneŠ 1isalsopr esent)in noninvariant theori es.NotethateŠ 1tr an sformsunderboth m mand ,whereas transformsonlyunder (inthelinearizedtransformation ).Thereisalsothecombination eŠ 14,t ransformingonlyunder m mas eŠ 14= Š m m,whichis usefulinconstructinginvariantactions.ItshouldbenotedthateŠ 14is,inasense,a“ eldstrengthfor the gaugetransformations.Anothersuch“eldstrengthistheintegrandofthe expressionin (5.1.35). We “ndasim ilarsituationinsupergravity.Thereweintroducenotonlylocal (real)scaleinvariance,butalsolocal(chiral) U (1)invariance(asageneralizationofthe globalR-invarianceofpuresupergravity)tosimplifytheanalysisofconstraintsand Bianchiidentitiesasmucha spo ssible,andweincludeitsgeneratorinthecovariant derivatives.(Inextendedsupergravity,thecorrespondingextrainvarianceis U ( N ):i.e., thelargestinternalsymmetryoftheon-shell theory.S eesecs.3.2and3.12.)Inthepresentapplication,theresultofsuchanapproachisthatthetorsionsandcurvaturescontainfewertensorsthantheywouldwithout theenlargedtangentspace.(Themissing tensorsrea ppearas“eldstrengthsofa tensor compensator.)Theselattertensorsare generallythetensorsoflowestdimension,sotheireliminationfromthetorsionsandcurvaturesa llowsgreatsimpli“cationintheanalysisoftheBianchiidentities,asdiscussed below(s ec.5.4),andsimpli“esouranalysisofconstraints.

PAGE 283

5.3.Covariantapproachtosupergravity269Sinceneither U (1)(R-invariance)norscaleinva rianceisasymmetryofgeneral theoriesofsupergravity+matter,weintroduceacomplexscalar tensor compensatorto compensateforboth(exceptfor n =0,whereR-inva rianceismaintained,sothecompensat orisreal).Inanalogytogravity,thereisalsoacomplexscalar density compensatorforthesetransformations:Itisthedensityoftheprevioussection,now unconstrained ,whichw illappearwhensolvingtheconstraints.Thequantityisthedirect analogofeŠ 1ofgravity,andthetensorcompensatoristheanalogofgravityscomponent “eldcompensator .Wew illfurthermore“ndaspecialsi gni“cancefortheanalogof gravityscombinationeŠ 14:Itisthesupe rspacedensitycompensator or,whichsatis“esasimple(noncovariant)constraint.T hetensorcompensatorsatis“esthedirect covariantizationofthisconstraint.Theanalogofgravitys misM,andth atof is thescaleparameter L and U (1)parameter K5.Afeatureofs upergravityn otappearing ingravityisthatofaglobalsymmetry,namely U (1),withbothdensityandtensorcompensat ors,whosedensitycompensator()isusedtoconstructa U (1)-gaugeconnection thattriviallygaugesthesymmetryinsuperspace(asinnonlinear models).(However, asingravity,itisnotusefultointroducegaugeconnectionsforscaletransformations, sinceglobalscaletransformations,unlikeR-symmetry,arenotanunbrokeninvarianceof theclassicaltheory(evenwithoutmatter)). Aftercompletingouranalysisofthe U (1)-covariantderivat ivesandt ensorcompensators,w ew illobtainthe( n =0) U (1)-noncovariant derivativesoftheprevioussection, whicharemoreconvenientforsomeapplications.Thisisachievedby“rstgaugingthe tensorcomp ensatorto1,whichexpressesintermsof H and or,andthenby droppingthe U (1)connection,shoulditnotdisappearautomatically. Webeginwit hthecovariantd erivat ives(cf.(5.2.20)) A= EA+A( M ) Š i AY [ A, B} = TAB CC+ RAB( M ) Š iFABY ,(5. 3.1) where Y = Yisthe U (1)gen erator,whosetangentspaceactioncanbesummarizedby [ Y A]=1 2 w ( A ) A,orexp licitly: [ Y ]= Š1 2 ,[ Y €]=1 2 €,[ Y a]=0.(5 .3.2)

PAGE 284

2705.CLASSICALN=1SUPERGRAVITYThecovariantderivativestransformas A= eiKAeŠ iK, K = KMiDM+( K iM + K€€i M€€)+ K5Y .(5. 3.3) a.2.Conf ormalsupergravityconstraints Thecovariantderivativesde“nearealizat ionoflocalsupersymmetry.However,it ishighlyreducibleandcontainsmuchmorethanthesupergravitymultiplet.Therefore, by analogywithYang-Mills,weimposecovariantconstraintsonthesederivativesto e liminateunwantedrepresentations.Thesupergravityconstraintscanbeexpressedin thesimpleform €= Š i {, €} ,(5. 3.4a) T = T b b= T (€ € )=0,(5. 3.4b) {, } =0 when =0;(5. 3.4c) (andtheirhermitianconjugates)or,intermsofthe“eldstrengths, T€ c= i €€, T€ = R€ = F€=0,(5. 3.5a) T = T b b= T (€ € )=0,(5. 3.5b) T c= T€=0.(5. 3.5c) Wehave dividedtheconstraintsintothreecategories:(a)conventionalconstraints thatdeterminethevectorLorentzcomponentofthecovariantderivative, a,inte rmsof thespi norcomponents €;(b)conv entionalconstraintsthatdeterminethespinor connectionsand(andtheirhermitianco njugates)intermsofthespinorvielbein E;and(c)representatio n-preservingconstraintsthatareneededforconsistencywith thede“nitionofchiralsuper“eldsincurvedsuperspace.Asforsuper-Yang-Millsand ordinarygravity,conventionalconstraintscanbeinterpretedaseithersettingcertain “eldstrengthstozero,oraseliminatingthemfromthetheoryby“eldrede“nitions. The“rstsetofconventionalconstraintsisofthesameformasforsuper-Yang-Millstheory,whilethesecondisanalogoustotheconstraintsofordinarygravity.The

PAGE 285

5.3.Covariantapproachtosupergravity271representation-preservingconstraintsarealsoofthesameformasforsuper-Yang-Mills theory.Althoughwehaveonlyrequiredtheexistenceofchiralscalars(i.e.,scalarmultiplets),thissetofconstraintsisalsosucienttoallowtheexistenceofchiralundotted sp inors(forexample,the“eldstrengthsofsuper-Yang-Mills),witharbitrary U (1) char ge.(ThesecondtypeofconstraintalreadydeterminesthespinorialLorentzand U (1)conn ectio ns.) Theconstraintsactuallyhavealargerinvariancegroupthanthatimpliedby (5 .3 .3 ): in a dditiontobeinginvariantunderthetransformationsgeneratedby(5.3.3), th eyareinvariantunderlocal superscale transformations.Inthecompensatorapproach, weuseconstra intsthatdetermineonlythe conformal partoftheP oincar esupergravity mult iplet.Therestofthemultiplet(thesuperscalepart)iscontainedinthecompensatori tself,andthereforetheparticularformofthePoincar esupergravit ymulti plet dependsonthechoiceofcompensatormultiplet. Todiscoverthee xplicitformoftheadditionalinvariance,we“rstnotethatthe i n“nitesimalvariationofthespinorialvielbeinunderscaletransformationsmustbeof theform LE=1 2 LE(5.3.6) where L isarealunconstrainedsupe r“eldwhichparametrizes thescaletransformation (see(5.3.4c)).Next,to“ndthesuperscalevariationof A,weuse(5.3 .6),vary( M ), ,and E aarbitrarily,and demandthat(5.3.5)issatis“ed.Thisdeterminestheremainingvariations.Theres ultscanbesummarizedas L=1 2 L +2( L ) M +3( L ) Y ,(5. 3.7a) L€= L €Š 2 i ( €L ) Š 2 i ( L ) €Š 2 i ( €L ) M Š i 2( €L ) M€€+3 i ([ €] L ) Y ,(5. 3.7b) andconsequently LEŠ 1= Š 2 LEŠ 1.(5. 3.8)

PAGE 286

2725.CLASSICALN=1SUPERGRAVITYThegaugesymmetriesofthetheory,(5.3.3,7),andtheconstraints(5.3.4)aresucienttoreducethecovariantderivativessothattheydescribean irreducible mult iplet: conformalsupergravity.Toseethis,westudythescalingpropertiesoftheremaining “eldstrengths.Thesearenotallindependent;UsingtheBianchiidentities,asweshow insec.5.4,allnontrivial“eldstrengthscanbeexpressedintermsofthreetensors R G a, and W.Forconv enience,wealsointroducethe(dependent) U (1)“eldstrength W. Theseobjectscanbede“nedby R =1 6 R€€€€=1 4 iT€€ , G a= iT a W=1 12 iR€ ( € )= Š1 12 T( € € ), W=1 2 iF€ €.(5. 3.9) From(5.3 .2,7)wehavethe U (1)andsuperscaletransformationsof (and hence €by hermitianconjugation),and a.Wecanthende terminethetransformationsofthese “eldstrengthsbyeva luatingcommutators.Theresultis: [ Y R ]= R LR = LR Š 2 2L ; [ Y G a]=0, LG a= LG aŠ 2[ €, ] L ; [ Y W]=1 2 W, LW=3 2 LW; [ Y W]=1 2 W, LW=3 2 LW+6 i ( 2+ R ) L .(5. 3.10) Thusthesuperscaleand U (1)transformationscanbeusedtogaugeawaypartsofthese tensors,leavin gonlyt he“eld Wofconformalsupergravity.Atthelinearizedlevel, thiscontainsth e puresuperspin3 2 projectionof H mdiscussedins ec.5.2.a.1.

PAGE 287

5.3.Covariantapproachtosupergravity273a.3.Cont ortion Thereisnothinguniqueaboutthesetofconventionalconstraintsweuse.Anyset thatallowsustoexpr essthevectorderivativeandspinorconnectionsintermsofthe spinorvi elbeins E, E€isequallysuitable.Forexample,wecoulduse T a b c=0instead of R€ c d=0tode termine a b c.Thiswou ldgivea a b cwhosecorresponding aisan equallygoodcovariantderivative.Thedierence Š isatensor(the contor tion tensor).Addingcontortionstoconnectionsdoesnotchangethephysicsandsimply amountstoarede“nitionofminimalcoupling.Indeed,formostfamiliarmodelstheconnectionsdonotenteratall:Forthe scalarmultiplettheLagrangian andthechirality constraint € =0arei ndependentoftheconnection.The“eldstrengthsofsuperYa ng -M illstheoryare FAB= [ AB )+[ AB )Š TAB CC= E[ AB )+[ AB )Š CAB CC(5.3.11) andarealsoindepe ndentofthesupergravityconnections.Finally,thesupergravity Lagrangian(for n =0)is EŠ 1,alsoi ndependentoftheconnections. Furthe rmore,anyothersetofcons traintsthatd etermines E aiscorr ect:Wecan always rede“ne E a Mbywrit ing E a M= E a M+ g a E M+ g a€E€ M,(5. 3.12) where g a isacovariantobjectconstructedoutofthe“eldstrengthsof A.There fore, insuperspace,inadditiontothecontortiontensorfortheLorentzconnection,wehavea contortionthatchanges E a M.H ow ever,thisdoesnotaectthephysicsas,onceagain,it amountssimplytoa rede“nitionofminimalcoupling. Thereisanotherambiguityinthechoiceofconstraints,which,however,leadsto nomodi“cationofthetheoryatall:SincetheBianchiidentitiesrelatevarious“eld strengths,therearemanywaystoexpressanyparticularconstraint.Forexample,since allcurva turesand U (1)“eldstrengt hscanbeexpressedintermsoftorsions(seesec. 5.4),anyconstraintonacurvatureor U (1)“eldstrengthcanbe expressedintermsof torsions.W hichformischosenispurelyamatterofconvenience.

PAGE 288

2745.CLASSICALN=1SUPERGRAVITYa.4.Poincar esupergravit yconst raints Thesuperspin3 2 superconformalmultipletwith“eldstrength Wisnots ucient todescribeo-s hellPo incar esupergravity.Wemu stincludelowersuperspinsuper“elds toobtainaconsistentactio n.Wecandothisintwoways:Byintroducingextraconformalrepresentationsascompensators,orbydirectlyrestrictingthegaugegroupsothat somelowersuperspinconformalrepresentationscontainedin HMcannotbegauged away.Suchrestriction sonthe gaugegroupareintroducedbyimposingconstraintsthat ar en ot in va riantunderthefullgroup.Theseconstraintsappearnaturallywhenweuse thefullsuperconformaltransformationstogaugethecompensatorsawayandrequire thattheremainingtransformationspreservetheresultingsuperconformalgauge. Therearethreetypesoftensorcompensa torsthatcanbecoupledtoconformal supergravityandcanbeusedtoreduceittoPoincar esupergravity.Th epossiblecompensat orsarerestrictedbytherequirementth attheymustha vedime nsionlessscalar “eldstrengthstocompensatefor L of(5.3.6,7,8,10).(Thus,thecompensator“eld strength X hastheusuallinearizedcompensatortransformation X = L X | isthena scalarwithaction(5.1.35).Theremainingtypeofconformalmattermultiplet,thevectormultiplet,c annotbeusedasacompensatorbeca useitsonly scalar“eldstrength Whasthewrongdimensionanditsprepotentialisinertundersuperscaletransformations.)Theyareparametr i zedbythecomplexnumber n :(1)thescala rmulti plet ( n = Š1 3 ),(2)thenonminimalscalarmultiplet(any n except0or Š1 3 ),and(3)the tensormultiplet G ( n =0).Thesem ultipletscanbede“nedbyconstraintsandcanbe expressedexplic itlyintermsofunconstrainedsuper“elds(prepotentials): €=0,=( 2+ R );(5.3.13a) ( 2+ R )=0,= € €;(5. 3.13b) ( 2+ R ) G =0, G = G =1 2 ( 2+ R )+ h c .;(5.3 .13c) where R isa“eldstrength(see(5.3.9)andsec.5.4)and 2+ R givesachiralsuper“eld whenactingonasuper“eldwithoutdottedspinorindices(seebelow).The U (1)and superscaletransformationsforwhichtheycompensateare

PAGE 289

5.3.Covariantapproachtosupergravity275[ Y ,]=1 3 L= L ; (5.3.14a) [ Y ,]= Š2 n 3 n +1 L=2 3 n +1 L ;(5.3 .14b) [ Y G ]=0, LG =2 LG .(5. 3.14c) Insec.5.3.b.7wewillbreakthesuperconformalsymmetryby“xingthecompensators. Intheresultingsuper-Poincar etheory( 5.3.13)becomeadditional,conformalbreaking constraintsonthecovariantderivatives(i.e.,onthetorsionsandcurvatures). Thescaleweightofisarbitrary(sincewecouldreplacebymandstillsatisfy (5.3.13a)).However,theratioofthe U (1)chargetot hescalewei ghtforachiralsuper“eldis“xed.Thiscanbeseenbyasimpleargument.Consideranarbitrarychiral super“eld € =0.Wewriteitss caletransformationin termsofthedilatational generator d d d d (see(3.3.34)): L = L [ d d d d ].Ifweperformascalevariationofthede“ning conditionforachiral“eld,anduse(5.3.7a),we“nd: 0=( L €) + €( L ) = Š 3( €L )[ Y ]+ €( L [ d d d d ]) =( €L )[ Š 3 Y + d d d d ], (5.3.15a) andhence 0=[ Š 3 Y + d d d d ].(5.3 .15b) Thusthe U (1)chargeandthedilatationalcharge always satisfytherule d d d d Š 3 Y =0for chiralsuper“elds .Thisiss eenfor W, W,and R in(5.3.10)andforin(5.3.14a). (Actuallyfor R thisisonlyclearifthetransformationlawiswrittenintheform LR =3 LR Š 2( 2+ R ) L .)Therelationofthechiralchargetothedilatationchargefor chiralsupe r“eldsin N =1supersy mmetryisaspecialcaseofthegeneralrelationnoted insec.3.5. Inpreciselythesamemanner,startingfromthede“ningcondition(5.3.13b,c)fora linearsuper“eld,wecan showthat thecondition d d d d Š 3 Y =2mustbe satis“edfor all linearsuper“elds.Thisisseenforand G in(5.3.14b,c).Inthecaseofin(5.3.14b)we havechosenac onvenientparametrizationforitsscaleweight.Thetensormultipletis

PAGE 290

2765.CLASSICALN=1SUPERGRAVITYneutralbecauseits“eldstrengthisreal( Y =0),andthusthe n =0theory (withthe G compensator)retainsitslocal U (1)invarianceaftersuperscaleinvarianceisbroken. Therelationofthedilatationalchargeandthechiral U (1)chargeforchiraland linearsuper“eldsimp liesthatc ombi ned L and K5transformationsonarbitrarychiral andlinearsuper“elds, andrespectively,taketheforms = L [ d d d d ]+ iK5[ Y ], = d ( L + i1 3 K5) ,( 5.3.16a) = L [ d d d d ,]+ iK5[ Y ,], =[ dL + i1 3 ( dŠ 2) K5].(5.3.16b) Thequantities d and darethescaleweightsofthesuper“elds. b.Solutiontoconstraints b.1.Conven tionalconstraints The“rstconstraintintheform(5.3.4a)isalreadyexplicitlysolved(aswasthe ca se fo rs up er -Y ang-Mills). Webeginour analysisofthesecondconstraintbyextractingfrom(5.3.1)the explicitformofthetorsion.Insections5.1,2wede“nedthecoecientsofanholonomy CAB Cby [ EA, EB} = CAB CEC(5.3.17) Theycanbeexpressedexplicitlyintermsofthe EA Mandtheir deriva tives.Wethen have TAB C= CAB C+[ AB ) CŠ i1 2 w ( C )[ AB ) C;(5. 3.18) i.e., T€= C€, T c= C c,

PAGE 291

5.3.Covariantapproachtosupergravity277T€ c= C€ c, T b€= C b€, T a b = C a b€, T = C +( ) +1 2 i ( ), T€€= C€€+€€Š1 2 i €€, T a = C a + a +1 2 i a, T b c= C b c+ €€+€€ T a b c= C a b c+( a €€+ h c Š b a );(5.3 .19) aswellasthecomplexconjugates. Byusingt heseequationsthe“rstconstraintof(5.3.4b)canbesolveddirectly(notingthatA istra celessinitslasttwoindices): T =0 =1 2 ( CŠ C ( )) Š1 2 iC ( ).(5. 3.20) However,solvingthelasttwoequationsof(5.3.4b)forand€€,resp ectively, isless straightforward,since C b citselfdepe ndsonthemthrough a.(Ontheo ther hand, (5.3.4a)introdu cesnodependenceof aon .)Tosolvetheseconstraintsweintroduce,asins ec.5.2.a.3, EA=( E, E€, E a) ( E, E€, Š i { E, E€} ).(5.3 .21) Wede “ne CAB Cby[ EA, EB} = CAB C EC.Weem phasizethat,since E aisstilldependenton€€and, CAB Cisalso.Incontrast, CAB Ciscompletelydeterminedinterms of Eand E€.Webeginby expressing E aintermsof E aan d (th ea sy etundetermined) C€ M: E a= Š i [ { E, E€} +(€€Š1 2 i €€) E€+(€ +1 2 i € ) E]

PAGE 292

2785.CLASSICALN=1SUPERGRAVITY= E€Š i (€€Š1 2 i €€) E€Š i (€ +1 2 i € ) E,(5. 3.22) whereweuse(5.3.4a).Thencomputingthecommutator[ E, E b]= C b DEDwe “nd C b c= C b cŠ (€€Š1 2 i €€)+ i (€ +1 2 i €) C c.(5. 3.23) Aswewillseeshortly,thenextconstraintweimpose(eq.(5.3.4c))willset C c=0,and ther eforethenonlineartermdropsout.Consequently,thelasttwoconstraintsof (5.3.4b)(inc ombi nedform) 0= T € €= C € €+2€€=( C € €Š €€+1 2 i €€)+2€€(5.3.24) give €€= Š1 2 C (€ € ),= i C b b.(5. 3.25) Thiscompletesthesolutionoftheconventionalconstraints(5.3.4a,b).Wehavenow determined E a,A,andAintermsof Eand E€.Furthe rmore,fromtheformof (5.3.22)weimmediatelyobtain E = sdetEA M= sdet EA M.(5. 3.26) (Thelasttermsin(5.3.22)givenocont ributiontothesuperdeterminant.) b.2.Representationpreservingconstraints Havingdeterminedallquantitiesintermsof E,wehaveare alizationoflocal su pe rsymmetrywith512ordinarycomponent“elds.IntheYang-Millscase,further reductionwasachievedbyimposing repres entation-preserving constraints:Toensurethe existenceof(anti)chiralscalarsuper“elds(de“nedby =0),werequired {, } =0.(5. 3.27) Insupergravity(assuming[ Y ]=0forsimp licity),we“nd(5.3.27)implies T DD = T + T€€ + T c c =0.(5. 3.28) Therefore,toallowtheexistenceofchirals calarsinsupergravitywemustenforcethe

PAGE 293

5.3.Covariantapproachtosupergravity279constraints(5.3.5c): T€= T c=0.(5. 3.29) From(5.3 .19),thisimplies: { E, E} = C E.(5. 3.30) (Equivalentlysince = E =0itfo llowsthat { E, E} =0whic hi mmediately leadsto(5.3.30).)Thus E= E MDMisabasisfortangentvectorsthatlieinacomplextwo-dimensionalsubspaceofthefullsuperspace:Alloperators Egenerate complex translationswithanalgebrathatcloses.We canalsoparametrizethesetranslations byabasis ofderivativeswithrespecttocoordinates( 1, 2): E= A ,where A is anarbitrarymatrixor zweibein. Wecanalwaysexpre ssthecoordinates as complex supercoordinatetransformsoftheusual -coordinates: = eŠ De,where =MiDM = isarbit rary.Ourfullsolutionoftheconstraints(5.3.4c)isthus E= A eŠ De eŠ A De; =MiDM, A = N ;(5. 3.31) wherew ehavesplit A intoacomplexscalefactor andaLoren tzrotation N ( detN =1 ).ThissolutioniscloselyanalogoustotheYang-Millssolution = eŠ Deto {, } =0,ex ceptfortheintroductionof A .I nf act,theofYang-Millscanbe interpretedasacomplextra nslationinthegroupmanifold.Wenowhaveadescription ofsupergravityintermsof,,and N .Howev er, N canbegaugedawaybya Lorentztransformation(withparameter K ). b.3.The gaugeg roup Atthispo int,wecanmakecontactwiththeprevioussection:Thesolutionofthe constraintsimposedsofarhasintroducedanewgaugegroupasaninvarianceof E= eŠ A De.Thevielbein Eremainsunchangedunderthetransformations ( e)= ei e,

PAGE 294

2805.CLASSICALN=1SUPERGRAVITY( A D)= ei ( A D) eŠ i .(5. 3.32) with = MiDM,[ D]= Š i ( D ) D;(5. 3.33) providedthat D €= D €Š i €=0, arbitrary ;(5. 3.34) or €= D2 L€, €= Š iD L€.(5. 3.35) Thefactthat iscompletelyarbitraryimpliesthatthepartofthe -gaugegroup parametrizedby canalwaysbecompensatedawaybyarede“nitionof N .Thus asintheYang-Millstheorysolvingaconstraint( F=0)givesrise toanewgauge group.Thetransformationon A canberewrittenas = ei (1 eŠ i )Š1 2 eŠ i ( N D)= ei (1 eŠ i )1 2 N DeŠ i ;(5. 3.36) wherethefactor(1 eŠ i ),= iD ,isthesuper-Jaco bian ofthetransformation(c.f. (5.2.59)). West illhavethe realK = KMiDMcoordinatetransforma tionsofthetheory(see (5.3.3)),aswellasthetangentspaceLorentzand U (1)rotations: E = eiKEeŠ iKis realizedby ( e)= eeŠ iK,( A )= A ;(5. 3.37) while E = eŠ1 2 iK5K Eisrealizedby ( e)= e,( A )=( eeŠ1 2 iK5K eŠ ) A .(5. 3.38) The K transformationscanberewrittenas = Š iK + O (, K ).(5.3 .39)

PAGE 295

5.3.Covariantapproachtosupergravity281Since K = K ,thisimp liesthat Š canbe gaugedaway.Intheresultinggauge, =1 2 H H = H = HMiDM.(5. 3.40) Sim ilarly,theLorentztransformationcanbeusedtogauge N to .Inthere sult ing gaugewehave E= eŠ1 2 H De1 2 H, E€= e1 2 H D€eŠ1 2 H.(5. 3.41) However,itismoreconvenienttoeliminatet herealpartofbygoingtoachiralreprese nt ation(asforsuper-Yang-Mills),asdiscusedinsec.5.3.b.5below. b.4.Eval uati onof and R Wecannow “ndsimpleformsfor(5.3.25)and R (5.3.9).Theresultsarecontainedin(5.3.52,53,56).Thedetailsofthederivationarenotessentialforfurtherreading, butpresentsomeusefulgeneraltechniques.Tosolvefor,weusetheidentity EŠ 1 A= Š EŠ 1( Š )BTAB B,(5. 3.42) whichholdsindependentlyofanyconstraints,foranysuperspace,foranytangentspace. Incaseswhere( Š )BTAB Bvanishes(ash ere),itallowscovariantintegrationbyparts, since dzEŠ 1AX = Š dzEŠ 1 AX =0.(5. 3.43) Toderivethis identitywewillsaveourselvesalotoftroublebynotingthatattheendof acalculati onthesignsresultingfromgradedstatisticscaneasilybedeterminedifthe i ndicesofeachcontractedpairareadjacent,withthecontravariantindex“rst.Thenet signchangeisthenjustthatresultingfromthegradedreorderingoftheindicesofthe initialexpression.Usingthisfacttoignorethesignsfromgradingatintermediatesteps ofthecalculation,wehave(inthebasis EA= EA MM) ( Š )BTAB B= EM B[ A, B} zM= EM B[ AEB ) M

PAGE 296

2825.CLASSICALN=1SUPERGRAVITY= EM BAEB MŠ EM BBEA M.(5. 3.44) We evaluatethesecondtermbyuseoftheidentity(againignoringgradingsigns)for arbitrarysup erfunctions X and Y XY A= X [ Y A]+ X AY = X AY + X AY ,(5. 3.45) wherewehaveused(5.1.26b)toevaluatethecommutator.(5.3.44)nowbecomes EM BAEB MŠ EM BEA M B+ EM B BEA M= AlnE Š 1 A+0,(5. 3.46) whichleadsto(5.3.42).Theevaluationofthe“rsttermusedtheusualexpressionfor thederivativeofthelogarithmofadeterminant(see(5.1.28);fortangentspacegroups whichincludescaletransformations,( Š )BAB B =0,sothatt hescalegener atoractsnontriviallyon E ).Thelasttermv anishesb ecause EM B B= EM BEB N( N+N( M) Š i NY).Ther efore M N( N+N( M) Š i NY) = M N N=0. Actually,itissimplerforourpurposestousetheformof(5.3.42)intermsof EAinsteadof A.Using E = E ,wehave: EŠ 1 E A= Š EŠ 1( Š )B CAB B.(5. 3.47) Fromtheexp ression(5.3.25)for, E= E,and C€ = C€€=0weobtain Š i =( Š )B C B B+ C = Š EŠ 1E E + C .(5. 3.48) Usingtheexpression(5.3.31)for EintheLorentzgauge N = (thegen eralLorentz gaugewillbeeasilyrestoredattheend),we“nd C = ( E )ln ,sothisexpression b ecomes Š i = Š EŠ 1E E +3 Eln = Š 1 eD eŠ + ElnE 2,(5. 3.49) wherew ehaveused E= eŠ De(5.3.50a) whichimplies

PAGE 297

5.3.Covariantapproachtosupergravity283 E = eD eŠ .(5. 3.50b) Weuset heidentity ,va lidforanyfunction f andlinearoperator X feX=(1 eXeŠ X) feX=(1 eX)( eŠ XfeX) =(1 eX)( eXfeŠ X) =(1 eX)( eXf ), (5.3.51a) toderivetherelation 1=(1 eŠ X) eX=(1 eX)[ eX(1 eŠ X)].(5.3.51b) Thesetworesultsmakeitpossibletorewrite(5.3.49)as Š i = Š (1 eŠ ) eŠ D(1 e)+ ElnE 2= Š (1 eŠ ) eŠ De(1 eŠ )Š 1+ ElnE 2= T T,(5. 3.52) wherewehaveintroduceda(noncovariant)scalar densityT : T ln [ E 2(1 eŠ )].(5.3.53) Animmediateconsequenceof(5.3.52)is F=0(see (5.3.1)). Wenowsolvefor R ,where {, } = Š 2 RM,(5. 3.54) asfollowsfromtheBianchiidentities(sec.5.4).Usingthesameformfor Easinthe previouscalculation,andusingtheresultfor,we “ndfrom(5 .3.20) = Š1 2 ( ( E )ln 2+ i ))= Š1 2 ( E )ln ( eŠ T 2) = Š1 2 ( E )ln [(1 eŠ )Š 1EŠ 1].(5.3 .55)

PAGE 298

2845.CLASSICALN=1SUPERGRAVITYWethen “nd(€€doesnotcontribute) R = e T( E€)2( eŠ T2)= e T( E€)2(1 e )Š 1EŠ 1= EŠ 1 ( E€)2e T(1 e )Š 1.(5. 3.56) where( E€)2=1 2 ( E€)( E€). *** Itisusefultoderivetheexplicitformoftheoperatorthatgivesachiralscalar fromagener alscalar f .Forth ecase[ Y f ]=0,asimpl ecalculati onusing(5.3.55)for theconnectiongiv estheresultthat( 2+ R ) f = fEŠ 1( E€)2e T(1 e )Š 1iscovariantly chiral.Thisres ultcanthenmosteasilybeextendedtoarbitrary U (1)charge [ Y f ]=1 2 wf byusingt heexpression (5.3.53)fortowrite ( 2+ R ) f =[ e1 2 w T( 2+ R ) eŠ1 2 w Tf ](1 e )Š 1,(5. 3.57) where 2istheformof 2onaneutralscalar(asimpliedby(5.3.57)for1 2 w =0).We thusobtain ( 2+ R ) f = feŠ1 2 w TEŠ 1( E€)2e(1+1 2 w ) T(1 e )Š 1.(5. 3.58) Thisquantityiscovariantlychiralwith U (1)charge1+1 2 w b.5.Chiralrepresentation Duetotheformof Ein(5.3.31),itispossibletode“nelocalrepresentationsthat arechiralwithrespecttothesupergravity“elds.(Theseareanalogoustochiralreprese nt ationsinsuperYang-Mills(4.2.78)aswellasinglobalsupersymmetry(3.4.8).)On allquantities F weperforma(n onunitary)similaritytransformation F(+)= eŠ Fe .(5. 3.59) (Antichiralrepresentatio nscanalsobede“ned,with Š .)Inthisrepresentation, asforsuper-Yang-Mills,allquantitiesareinvariantunder KMtransformations,andthe covariantderivativestransformexplicitlyundertransformations.Furthermore,we

PAGE 299

5.3.Covariantapproachtosupergravity285choosetheL orentzgauge N = ,which forces K€€toequal €€of(5.2.27)tomaintainthegauge.Inthechiralrepresentation,thevielbeinbecomes: E(+)€= D€, E(+) = eŠ H DeH=eŠ HDeH; eH= ee .(5. 3.60) Thisispreciselywhatwehadconstructedintheprevioussection((5.2.27)and(5.2.28)). Thetransformationof H canbeobtainedfromthatof: ( eH)= ei eHeŠ i .(5. 3.61) (Notethat,asinsuper-Yang-Mills,(5.3.60)canbeusedtode“ne H inany K -gauge;it is K invariant.Alternatelythe and€transformationscanbeusedtogaugeaway Hand H€.) Itispossiblet ogotoarepresentationthatisalsochiralwithrespectto U (1). From(5 .3.53)wehave = iET ,€= Š i E€T;( 5.3.62a) where T= eŠ H TeHisthechiral-representati onhermitianconjugateof T (cf.(5.2.28)). Using(5.3.2),wecanwrite EŠ i Y = eŠ TYeŠ1 2 TEeTY, E€Š i €Y = eTYeŠ1 2 T E€eŠ TY;(5. 3.62b) Inadditiontothetransformation T = Š iK5,t heseexpressionsareinvariantunder T = i 5,where5ischiral.Wecanusethisgaugefreedomtoreplace T by 5= T +3 ln ,(5. 3.62c) thusintro ducingforsubsequentusethechiraldensity D€ =0.Wenowgo t oachiralrepresentationnotonlywithrespecttoMDMand N , butalsowithrespectto5, bymakingth ea ppropri atenonunitary U (1)transformation,andobtain: E€Š i €Y = EŠ1 2 D€,

PAGE 300

2865.CLASSICALN=1SUPERGRAVITYEŠ i Y = eŠ V5YEŠ1 2 (1 eŠ H)Š1 2 EeV5Y= E EŠ1 2 E+( E EŠ1 2 EV5) Y eV5 e5e5= E3 EŠ 1(1 eŠ H) 33=(€€EŠ 1E )3,€€EŠ 1 EŠ1 3 (1 eŠ H)1 3 ;(5. 3.63) whereweh aveused E =2 2 E (5.2.49)toreplacewith E .Thise ectivelyreplaces with E assuperscaledensitycompensator:Inthe U (1)-chiralrepresentation,the U (1)densi tycomp ensator5nolongerappears.Forthisreasonthischiralrepresentationisus efulfor n =0superg ravity,whereatruelocal U (1)invariancer emains,butnot veryus efulforother n .Howev er,itdoesbearacloserelationshiptothe n = Š1 3 results oftheprevioussection: n = Š1 3 canbeobtainedbyconstrainingAtovanishidentically.Inthisrepresentation,theresultissimplythat V5vanishes,a ndhence EŠ 1=€€EŠ 1, inagreementwith(5.2.72).Insection5.5,thi sresultw illbeusedtowritea“rst-order formalismfor n =0combin edwith n = Š1 3 b.6.Densitycompensators Aftergaugingaway Hand H€,wehaven owdete rminedallthegeometrical super“eldsin Aintermsof H mand ,whichconta in64and32component“elds, respectively.Theaxialvectorprepotential H mcontainsthecomponentgauge“elds. Thesuper“eld isthesupercon formal(densitytype)compensator:Byscaling arbitrarily (witho uttransforming H m),wegeneratecomplexscaletransformations(realscale U (1))ofthevielbein.Thus,thecomplexscaletransformationpropertiesofanyquantityexpressesits dep endence.Thesetransformationsmustberestrictedandtherepresentationreducedfurther,sinceEinsteintheoryisincludedinPoincar esupergravity andis not scaleinvariant.Wenowconsiderthe(scalar)tensor-typecompensators ,, G (5 .3 13) ,w hichalsotransformunderthesecombinedtransformations.Fixing thegaugesofthesetransformationsby“xingthecompensatorisaconvenientwayof determining,sinceitseparatesthelowersuperspinmultipletsfromtheconformal supergravitymu ltipletinacovariantway. Itisconvenienttosolvetheconstraints(5.3.13)onthetensorcompensatorsin termsofcorresponding density compensatorsthatsatisfythecorresponding ”at-space

PAGE 301

5.3.Covariantapproachtosupergravity287constraints; thisallowsusto“ndanexplicitsolutionfor. b.6.i.Minimal( n = Š1 3 )supergravity We“rstco nsiderthe(covariantly)chiralscalarcompensator.Theconstraintis, us ing(5.3.62a), €=( E€Š1 3 i €)= e1 3 T E€eŠ1 3 T= 0( 5.3.64a) andissolvedby = e1 3 T =[4 2 E (1 e )]1 3 E€ =0.(5. 3.64b) Ittransformsunderscaletransformationsasin(5.3.14).Here isa”atspacechiral super“eld,inthechiralrepresentation,asfollowsfromthede“nitionof E.Ifwechoose thegauge=1,then T = Š 3 ln andweobtain = Š 1 1 2 EŠ1 6 (1 eŠ )1 6 (1 e )Š1 3 .(5. 3.65) (Wehaveagainused E =2 2 E .)Inthisgauge €=0imp lies €=0;bycomplex conjugationand(5.3.4a)A=0,andth usthe“elds trengthoftheaxial U (1)transformations(see(5.3.9))vanishes: W=( 2+ R )=0.Therela tion(5.3.58)becomes 3( 2+ R ) f = fEŠ 1( E€)2(1 e )Š 1.( 5.3.66a) Inthechiralrepresentation,thissimpli“esto 3( 2+ R ) f = D2( EŠ 1f ).(5.3 .66b) Thetheoryisnowdescribedby H mand andtheonlysuperspacegaugefreedom leftisthatofsuper-Poincar etransfo rmations.Asin(5.2.75),thedensitycompensator canbereplacedbyoneofitsvariants. b.6.ii.Nonminimal( n = Š1 3 )supergravity Fo rt he no nm inimalscalarmultiplet,weagain“ndasolutionintermsofadensity compensator.Theconstraintis,using(5.3.58), ( 2+ R )= EŠ 1e2 n 3 n +1 T( E€)2eŠ2 n 3 n +1 T(1 e )Š 1=0 (5.3.67a)

PAGE 302

2885.CLASSICALN=1SUPERGRAVITYandissolvedby = eŠ2 n 3 n +1 TE (1 e )=[4 2 E (1 e )]n +1 3 n +1 Š 2,( E€)2=0.(5.3 .67b) Hereisa”at-space-linearsuper“eld(asopposedtothecovariantlylinearsuper“eld ):Inthechiralrepresentation D2=0.A gain,inthegauge=1,weobtainthe solutionfor: =[n Š 1 n +1]Š 3 n +1 8 n [ E2 n(1 eŠ )n +1(1 e )n Š 1]Š n +1 8 n .(5. 3.68) Inthisgaugewehave T i asanewtensor(appearinginarbitrarygaugesas ),intermsofwhich R and Waredetermined.Thesolution(5.3.68)doesnot a pplytothefollowingcases:(1) n = Š1 3 ,forwhichthemi nimalscalarmultipletisused instead;(2) n =0 ,f orwhichthesolutionof(5.3.67a)ismoresubtleandwillbediscussednext;and(3) n = ,which doesnotleadtoasensibletheory.Theparameter n canalsobegeneralizedtocomplexvalues,buttheconstraintsthenviolateparity(o shell),andwedonotdiscussthemhere. b.6.iii.Axial( n =0 )supergravity Theconstraint(5.3.13c)for n =0ismostea silysolvedbyexpressingthecompensator G intermsofacovariantlychiralspinor =( 2+ R ).Usingtherelation = E ( EŠ 1E )(asfo llowsfromintegrationbypartsond4xd4 EŠ 1f =d4xd4 EŠ 1Ef forany f ),wehaveinthechiralrepresentation(using(5.3.63)): G =1 2 ( + € €)=1 2 E ( EŠ1 2 E + h c .) E G ,(5. 3.69) sothat G isafunctionofonly H and .Inthechiral representation,butinthe Lorentzgauge€€=0, is”at-spac echiral: D€=0.(This gaugeexistsbecause R€€=0,asfo llowsfromtheBianchiidentities,seesec.5.4,whichimpliesthat€€anditsconjugatearepuregauge.)Insuchagauge N = X = , butdependsonly on H .Ontheot herhand,inth eLoren tzgauge N = ,where€€dependsonlyon H upt oafactoro f(see (5.3.25)), is( XŠ 1) timesa”at-spac echirals pinor.

PAGE 303

5.3.Covariantapproachtosupergravity289IntheWeylgauge G =1weobtain EŠ 1= G ( H )=1 2 ( EŠ1 2 E + h c .).(5. 3.70) Inthisgauge,(5.3.13c)impliestheconstraint R =0.Furth ermore,inthechiralrepresentation(5.3.70)maybecombinedwith(5.3.63)toreplacewith E asthecompensatorf orthe n =0covariantd erivat ives: = EŠ1 2 = G1 2 ,= E EŠ1 2 = GŠ 1 EŠ1 2 .(5. 3.71) b.7.De gauging Thetheoryandthecovariantderivati veswehav econstru ctedsofarcontain exp licit U (1)gen eratorsandconnections.For n =0,the U (1)symmetryisagenuine local symmetryofthetheory attheclassicallevel(thereareanomaliesatthequantum level,seesec.7.10)andtherefore n =0superg ravityonlycouplestoR-invariantmatter systems(3.6.14,4.1.15).For n =0,thesupersca lecomp ensatorcanbeusedto remove the U (1)chargeofanymultiplet:Bymultiplyingthesuper“eldbyanappropriatepower ofthecompensator(see(5.3.12a,b))wecanalwaysconstructa U (1)neutral object.If wedothis toallquantities(vielbein,connections,matter),the U (1)generatorsdonot actandcanbedro ppedfromthetheory.Theresultingformalismisapplicabletomattermultiplets without de“nite U (1)chargea ndhencetosystemswithoutglobalR-invariance(seesec.5.5).Theprocedurewearefo llowingissimilartowhatonedoesinordinaryspontaneouslybroken gaugetheories.Onegoestoa U-gauge eitherbyusingthe Goldstone“eldtode“negaugeinvariantquantitiesaswejustdid,orbygaugingthe Goldstone“eldaway,aswedonow:Insteadofrescaling“eldsbythecompensator,we cangaugeitaway,and“xthe U (1)(andsuperscale)gaugeasdiscussedinsec.b.6. above.For n = Š1 3 ,thissetsA=0andthe Y generatordropsfromthecovariant derivati ves.For n = Š1 3 ,0,the U (1)conn ectionbecomesacovarianttensorwith respecttotheremaining(super-Poincar e)group.Thereforewecaneliminate Y by a ddingacontortionterm AAŠ i AY andthus,by(5.3.18), TAB C TAB CŠ i1 2 w ( C )[ AB ) C( butwithnochangeinthecurvatures).Theonlymodi“edtorsi onsare(where iT,see(5.3 .52)):

PAGE 304

2905.CLASSICALN=1SUPERGRAVITYT T Š1 2 ( T ), T€€ T€€+1 2 €€T, T a T a Š1 2 i ( T€Š €T+ T T€).(5.3 .72) Furthe rmore, R and Whavethefollowing explicitexpressionsintermsof Tandthe Y i ndependentor degauged : R = Š Š 1 2 Šn 3 n +1 ( €+n Š 1 2(3 n +1) T€) T€,(5. 3.73) W=( 2+ R ) i [1 2 €( €+1 2 T€)+ R ] T.(5. 3.74) Weem phasizethatfor n = Š1 3 wecansimplydropa llreferenceto U (1)wit hout anyothermodi “cations. Althoughwehaveem phasizedthecompensatorapproachtothebreakingofthe superconformalinvariance,weshouldpointoutthat“xingtheconformalgaugebysettingthecompensatorto1iscompletelyequivalenttoimposingadditional,conformalbreakingconstraintsonthecovariantderivatives.After U (1)degauging,theconstraint equations(5.3.13)or(5.3.64a,67a)become,w he nt he compensatorsare“xed,conditions onthecovariantderivatives.Theseconditionsaretheconstraintsontorsionsandcurvaturesgivenin(5.2.80b). Atthispo intwehaveadescri ptionofPoincar esupergravit yinte rmsof H andone ofthedensitycompensators.Theycompensatefor component conformaltransformations.Forexample,the -independentof canbeidenti“edwiththecomponent of (5.1.33).Similarly,thelinear component,aspinor,compensatesfor S -supersymmetry. Afterdegauging,thesuperconformalinvarianceofthesupergravityconstraintsis destroyedforarbitrarysuper“elds L and K5.Nev erthelessaremnantofsuperconformal invarianceremains.Thisisbecausetheuncon strainedsuper“eldsthatdescribePoincar e supergavityare H mandsomescalarsuper“eldcompensator.Thussuper“eldsupergravity,unlikeordinarygravity, always containsacomponent compensatorof(5.1.33). (Thisist hereasonwhyatthecomponentlevelsu perconfo rmalsymmetryissouseful.)

PAGE 305

5.3.Covariantapproachtosupergravity291Obviouslyarede“nitionofthedensitycompensatingmultiplet(onceitstypehasbeen speci“ed)cannotaectthePoincar esupergravityc onstraints.Thisisrealizedbyan invariancegroupoftheconstraintsinadditiontothatparametrizedby KMand K Thetransformationsofthisinvariancegroupareexactlythesameasthoseoftheconformalgroup,butwiththeimpor tantrestrictionthat L and K5arenolongerarbitrary. Thesimplestwaytoobtaintheformofthesere strictedconformalt ransformationsisto usethetensortypecompensators,,and G Beforegaugingthecompensatorsto1wecansimplymakeanarbitraryrede“nition ofthetensortypecompensators.Wehave(i): = Š ,( ii) = Š ,a nd (iii) G = Š G ,whereisaco variantsuper “eld.Ineachcasetherede“nitionmustbe suchthattheproductofthecompensatortimessatis“esthesamedierentialequation astheoriginalcompensator(5.3.13)(i.e.,(i) €=0,( ii)( 2+ R )()=0,and(iii) ( 2+ R )( G )=0,= ).Thetransformationsaect only thecompensators,not thecovariantderivativesnormattersuper“elds,whereas L and K5transformations a ectall“elds.Thecombinedtransformationofthetensorcompensatorsunder L K5, andisthus n = Š1 3 : =( L + i1 3 K5) Š ,( 5.3.75a) n = Š1 3 ,0: =[(2 3 n +1 ) L Š i (2 n 3 n +1 ) K5] Š ,(5.3.75b) n =0: G =2 LG Š G .(5. 3.75c) Wenowde gaugebysettingthecompensatortoone.Inordertomaintainthis gaugeconditionwemustsetthetotalvariationofthecompensatortozeroandwe“nd that L and K5satisfytheconstraints L =1 2 (+ ), K5=3 2 i ( Š );(5.3.76a) L =3 n +1 4 (+ ), K5=3 n +1 4 n i ( Š );(5.3.76b) L =1 2 ,= ;(5.3 .76c) respectively.

PAGE 306

2925.CLASSICALN=1SUPERGRAVITY5.4.SolutiontoBianchiidentities Inanygaugetheorythe“eldstrengthssatisfyBianchiidentitiesthatareaconsequenceoftheJacobiidentitiesforthecovariantderivatives,ormoregenerally,forsuperforms,aconsequenceofthePoincar etheorem.Asexp lainedinsec.4.2,theBianchi identities containnousefulinformation unless someofthe“eldstrengthshavebeenconstrained.Inthatcasetheymakeitpossibletoexpressallthe“eldstrengthsintermsof anirreducibleset(thatmaystillsatisfy dierential constraintsthatarealsocalled Bianchiidentities).Insec.4.2wegaveadetailedexampleofthisprocedureforsuper Ya ng -M illstheories;hereweconsidersupergravity. Webeginina generalcontext,withcovariantderivativesforarbitrary N andarbitraryinternalsy mmetrygeneratorsi(cf.(5.2.20,5.3.1)): A= EA MDM+A M +A€€ M€€+A ii(5.4.1) where EA M,A,andAarethevielbein,Lorentzconnection,andgaugepotential, respectively.Wede“ne“eldstrengths:torsions TAB C,curvatures RAB and RAB€€,and gauge“eldstrengths FAB iintermsofthegradedcommutator [ A, B} = TAB CC+ RAB M + RAB€€ M€€+ FAB ii.(5. 4.2) Thegeometryofsuperspaceimplicitin(5.4.1)givesnontrivialrelationsamongthe “eldstrengths T R F .Wehavechosent heacti onoftheLorentzgrouptobereducible intangents pace:Itdoesnotmixthe( V, V€, V a)parts ofasupervector VA,andit rotatesthevectorandthespinorpartsbythesametransformation;i.e., VA A BVB,where A B=( €€, €€+ €€ )(cf.(5. 2.19,5.3.3)).Consequently,thereareno connectionssuchasA corA €,andA b c=A €€+ h c .,et c. Wecanviewthi srestric tionasaconstraint:Ithasthe consequencethattheBianchi identitiesnowgivealgebraicrelationsamo ngthe“eldstrengths.BecausewehavechosentheactionoftheLorentzgrouptobere ducib le,wehaveimposed theconstraints RAB c = RAB = RAB€ = RAB d=0, RAB = RAB c d, RAB c d= RAB €€+ RAB€€ ,(5. 4.3) andtheir hermitianconjugates.Theseconstraintsaresucienttoexpressallofthe

PAGE 307

5.4.SolutiontoBianchiidentities293curvatures R andgauge“eldstrengths F intermsofthetorsions T ifwea ssumethat EA Mtr an sformsundertheactionofi;otherwise Firemainsasanindependentobject. Inparticular,inthepresenceofcentralchargesthecorresponding“eldstrengthscan remainasindependentquantities.Wewri tethetransformationintermsofmatrices (i)A B: [i, A]=(i)A BB(5.4.4) wheretheonlynonvanishing(i)A Bare (i)a b ,(i)a b€€,(i) a b,(i)a b=((i)a b)(5.4.5) TheBianchiidentitiesfollowfromtheJacobiidentities0= [[ [ A, B} C )} = BABC EE+ BABCandare: BABC EŠ ABC E+ R[ AB C ) E+ F[ AB C ) E=0,(5. 4.6a) BABC( M ) Š[ ARBC )( M )+ T[ AB | DRD | C )( M )=0,(5 .4.6b) BABC iŠ[ AFBC ) i+ T[ AB | DFD | C ) i=0(5.4 .6c) where ABC E[ ATBC ) EŠ T[ AB | DTD | C ) E, FABC E FAB i(i)C E.(5. 4.7) TheBianchiidentitiesaresatis“edidenticallysimplybecausethe“eldstrengthsareconstructedoutofthepotentials EA M,A,andA.In(5. 4.6a)bydecomposing BABC Einto i rreduciblepiecesundertheLorentzandinternalsymmetrygroups,weexpress F and R intermsof T ;thissol utionautomatic allysatis“es BABC( M )= BABC i=0,sothat (5.4.6b-c)containnousefulinformation. Toor ganizetheanalysisoftheBianchiidentities,weclassifytheidentitiesby (ma ss)dimension.Thelowestdimensionidentitieshavedimension1 2 : B d= B € d=0andhe rmitianconjugates.Thehighestdimensionidentitieshave dimension3: B a b c i= B a b c( M )=0.Ordi nar ily,westartwiththelowestdimensionidentitiesandworko urwayup;however,thedimension1 2 B sarerelationsamongthetorsionsonly(theyareindependentofthecurvaturesand“eldstrengths).Therefore,to determine R and F westartwitht hedimension1 B s.Forexample, B € €=0imp lies

PAGE 308

2945.CLASSICALN=1SUPERGRAVITYR ,€€c e+ F c e€€= € €(5.4.8) (Thegradedsymmetrizationdropsoutbecauseoftheconstraints(5.4.3)and(5.4.5)). Toextract F c eand R ,€€from(5.4 .8),wedecomposeitintoLorentzirreducible pieces.Thisgives: F c e=1 2 ,€ c€ e(5.4.9a) R ,€€=1 2 N ,(€ c€ ) c(5.4.9b) Proceedinginasimilarmanner,wedetermine all theremainingcurvaturesandgauge “eldstrengthsinte rmsofthetorsions: R =1 4 ,( € )€,(5. 4.9c) F c d=1 2 , d€ c€.(5. 4.9d) Furthe rmorewe“nd RAB =1 2 N ABd ( d ),(5. 4.9e) FABc d=1 2 ABc d ,(5. 4.9f) for( A B )=(€ ,€ ),(€ b ),and( a b ),and “nally R B =1 4 [(1 N +1 )(B ,( a | | c )( c )+1 3 + B a ( C ) ) +(1 N Š 1 )(B ,[ a | | c ]( c )+Š B a ( C ))],(5. 4.9g) F Bc d=1 2 [1 3 B ,( a | | c ) d +B ,[ a | | c ] d ] Š1 2 [1 3 R( a | B | c ) dŠ R[ a | B | c ] d],(5.4 .9h) + B =B ,( a | | c ) c ,Š B =B ,[ a | | c ] c for( B )=(€ )and( b ).Thissolutionautomaticallysatis“es(5.4.6b-c),ascanbeveri“ed

PAGE 309

5.4.SolutiontoBianchiidentities295bydir ectcomputation.FromnowonweneedonlyconsidertheBianchiidentitiesin (5.4.6a)thatareindependentof R and F Wenows p ecializeto N =1superg ravity:OurtangentspacetransformationscontainonlytheLorentzgroupand U (1);i.e.,i= Y .Weimpose: T = T€= T c= T€€= T€ cŠ i €€=0 ,( 5.4.10a) T ,( € )€= T (€ € )= T b b= F€=0.(5. 4.10b) We proceedasabove,startingwiththelowestdimensionidentities.Forexample, B ,€ d= T ,€ d+ T€ d+ €T d+ T ETE ,€ d+ T ,€ ETE d+ T€ ETE d(5.4.11) but T E=0and T ,€ E= i €€whichimplies 0= B ,€ d= Š i €€T € €Š i €€T € €.(5. 4.12) Nextwed ecompose T b cintoirreduciblerepresenta tionsoftheLo rentzgroup, T b c= C€€[ f1 + C ( f2 )+ Cf3 ] +[ f4 ( )(€€ )+ C ( f5 )(€€ )+ Cf6 (€€ )](5. 4.13) andnote T b b=0imp lies f3=0, T ,( € )€=0imp lies f2= f1=0,and “nally T (€ € )=0imp lies f6=0.Nows ubsti tuting(5.4.13)into(5.4.12)leadsto 0= Š i 2 f4 ( )(€€ )Š i 2 C ( f5 )(€€ ),(5. 4.14) whichyields f4 ( )(€€ )= f5 (€€ )=0.Inotherwords T b cvanishesid entically. ThisexampleshowshowwedecomposethetorsionsintoirreduciblerepresentationsoftheLorentzg roup,andthensolvetheconstraintsbylookingatwhattheyimply aboutthevariousirreduc ibleparts.Fortwocomponentspinorsthisdecompositionsimplyconsistsofsymmetrizingandantisymmetr izinginallpossibleways.Otherexamples are

PAGE 310

2965.CLASSICALN=1SUPERGRAVITYT€ € = C€€CX1+ CX2 (€€ )+ C€€X3 ( )+ X4 (€€ )( ), T€ € ,€= X1 (€€€ )+ X2 (€C€ )€+ X3 €C€€.(5. 4.15) TheseexpressionsarenowsubstitutedintotheBianchiidentitieswhichseparateinto severalequations.Thesearethensolved,withtheresultthatsomeoftheirreducible partsarezero,whileothersareexp ressibleintermsofaminimalset. Thecompleteanalysisisstraightforwardbuttedious.We“ndthatalltheBianchi identitiesandconstraintsaresatis“edby {, €} = i €, { €, €} = Š 2 R M€€, [ €, i €]= C€€[ Š R Š G€ €+( €G€) M€€Š iWY Š1 3 iWM + W M ]+( R ) M€€, [ i €, i €]= C€€fŠ h c .(5. 4.16) wheretheoperator fisde “nedby f= Š i1 2 G( € )€Š1 2 ( ( R Š i1 3 W( ) )+ W Š1 2 ( ( G )€) €Š i1 2 ( ( W )) Y Š WMŠ i1 8 [( ( €G )€) M + ] +( 2 R +2 R R + i1 6 W) M+1 2 ( ( €G )€) M€€,(5. 4.17) and W1 4! ( W ). Theindependenttensors R G a,and W,andthedep endentone Wsatisfytherelations G a= G a, €R = €W= €W=0,

PAGE 311

5.4.SolutiontoBianchiidentities297 €G€= R + iW, W+1 3 i ( W )=1 2 i ( €G )€, W+ € W€=0.(5. 4.18) Therefore,allthetorsions,curvatures,and“eldstrengthsof(5.4.2)areexpressiblein termsofthethreecova riantsuper“elds W, G€,and R ,andthei rderivat ives. Byconsid eringthecoecientof Monbothsideso ftheeq uationforthecommutatoroftwovectoriald erivatives,weconcludethatthesuperspaceanalogofthedecompositionin (5.1.21)takestheform R a b = C€€[ W Š1 2 ( ) ( 2 R +2 R R + i1 6 W)+1 2 ( ( X ) )] + C1 4 ( (€( G )€ )),(5.4 .19) where X isde “nedby X= Š i1 8 ( €G )€.(5. 4.20) Bycomparingthisto(5.1.21),weseethereisarepresentation Xpresentinthesupercovariantc urvature R a b thatwasabsen tinthecompon entcurvature r a b .This o ccursbecausetheconstraintsthatwehavechosenimplythereisnontrivial x -spacetorsion T a b c a b c dG dpresent. Wenowc hoosethescale+U (1)gaugewherethecompensatorequals1.For n =0 theonlyresultingmodi“cationof(5.4.16)isthatweset R =0 (thus,for n =0,the spinor derivatives(butnotthevectorderivatives!)obeytheglobalsupersymmetryalgebra).F orother n itisnecessarytodropthe Y partofthecov ariantderiva tives.However,for n = Š1 3 ,Avanishesid entically inthisgauge,sotheonlymodi“cationof (5.4.16)istoset W=0(seedisc ussionbefore(5.3 .72)).Inthiscase,(5.4.16)reducesto (5.2.81).For n =0, Š1 3 themodi“cationsareslightlymorecomplicated:(1)Thespinor U (1)conn ectionisnowcovariant;toavo idconf usion,wede“ne TŠ i ( degauged ).(2)Notensorsar esetto zero,butnow R isdetermi nedbythe compensatorconstraint(see(5.3.73)),and Wbyitsexp licitform(see(5.3.74)).(3)

PAGE 312

2985.CLASSICALN=1SUPERGRAVITYSeparatingoutthe Y partscausesshiftsinafewofthetorsions(see(5.3.72))bythe covariantquantity T.Theresult ingformof(5.4.16)is {, €} = i €Š1 2 ( T €+ T€) { €, €} =1 2 T(€ € )Š 2 R M€€[ €, i €]= C€€[ R + G€ €] Š1 2 ( T€Š €T+ T T€) €+ C€€[ Š ( €G€) M€€Š W M + i1 3 WM ]+(( + T) R ) M€€,(5. 4.21) with(5.4.18)modi“edby + TY and € €Š T€Y .Thisisj ustthe U (1) degaugingdescribe dins ec.5.3.b.7. Thisresultcorrespondstooneofthemanyformsofnonminimal n = Š1 3 N =1 supergravity.Asexplainedinsec.5.3,covariantderivativescanberede“nedbyshifting withcontortions.Asanexample,wenotethatthetwoanticommutatorsin(5.4.21)can besimp li“edbytheshift €€Š i1 2 ( T €+ T€),(5.4 .22) whichgives(5.2.82).Fortheremainderofthebook,unlessotherwisestated,ourcovariantderivativesfor N =1superg ravitywillbeinthegaugewiththetensorcompensator settoone,andfor n =0,withthe Y partsdropped.

PAGE 313

5.5.Actions2995.5.Actions Inthissectionweconstructanddiscuss superspaceaction sforma ttersystems coupledtosupergravity,andforsupergravityitself. a.Reviewofvectorandchiralrepresentations Inthevectorrepresentation(where,e.g., €=( )),acovariantsuper“eld X€ ...tr an sformsunderthegaugetransformationsoflocalsupersymmetry(hermitian supercoordinateandtangentspacetransformations)as X= eiKXeŠ iK(5.5.1a) with K = K = KMiDM+ K iM + K€€i M€€(cf.(5.3.3)).Inchiral( € D€)or antichiral( D)representatio ns,thetransfo rmationlawsare X(+) = ei X(+)eŠ i ,(5. 5.1b) X( Š ) = ei X( Š )eŠ i ,(5. 5.1c) respectively,with, givenby,e .g.,(5.3.33-35),and X(+)= eŠ HX( Š )eH= eŠ Xe .(5. 5.2) (Recallthatthehermitianconjugateofano bj ectinthechiralrepresentationisinthe antichiralrepresentationandtransformswith (5. 5.1c).Therefore,justasin(5.2.28) andinYang-Millstheory(4.2.21),wemustconverttheconjugatetoanobjecttransformingwith( 5.5.1b),andde“nethechiralrepresentationconjugate X(+)= eŠ H( X(+))eH.)Thetransformationpropertiesof EŠ 1inthevector,chiral,and antichiralrepresentationsare EŠ 1 = EŠ 1eiK(5.5.3a) E(+) Š 1 = E(+) Š 1ei (5.5.3b) E( Š ) Š 1 = E( Š ) Š 1ei (5.5.3c) respectively.

PAGE 314

3005.CLASSICALN=1SUPERGRAVITYb.Thegeneralmeasure Usingtheresultsoft heprevioussections,itisstraightforwardtoconstruct locallysupersymmetricactions:Wecovariantizeallderivatives(withpossiblysome ambiguityinwhetherweuseminimalcouplin goraddcontor tionterms),includingthe derivativesusedtode“neconstrainedmatter“elds(suchaschiral“elds),andwecovariantizethemeasure.Forintegralsoverallsuperspace,byanalogywithordinaryspace (see5.1.23-5),weuse EŠ 1asadensitytode“neacovariantmeasure,andwriteactions oftheform: S = d4xd4 EŠ 1ILgen,(5. 5.4) where ILgenisageneralrealscalarsuper“eldconstructedoutofcovariantmatter“elds, derivatives,etc.Sincebyconstruction ILgenmusttran sformasin(5.5.1),andsince EŠ 1transformsasin(5.5.3),theexpressionin(5.5.4)isinvariant.(Recallthatcoordinate invarianceisde“nedonlyuptosurfaceterms(5.1.27).)Thistypeofexpressionisthe integratedversionofwhatisreferredtoasaD-typedensityformula. c.Tensorcompensators Justasingravity(sec.5.1.d),wecangeneralizethecoordinateinvariantmeasure (5.5.4)toascale(and U (1))invariantmeasurebyintroducingtensorcompensators.For an ILgenthathasscaleweight d (therealityoftheactionimpliesthatitmustbe U (1) invariant),wehave(see(5.3.8,14)) n = Š1 3 : S = d4xd4 EŠ 1( )1 Šd 2 ILgen,(5. 5.5a) n = Š1 3 ,0: S = d4xd4 EŠ 1( )3 n +1 2 (1 Šd 2 )ILgen,(5. 5.5b) n =0: S = d4xd4 EŠ 1G1 Šd 2 ILgen.(5. 5.5c) Theseactionsreduceto(5.5.4)inthegaugewherethecompensatoris1.(Theanalogousexpressioninordinarygravityisd4xd4 eŠ 14 Š dL where isthetensor-type componentscalecompensatorintroducedin(5.1.33)and L isascaleweight d Lagrangian.)

PAGE 315

5.5.Actions301d.Thechira lmea sure Inthechiralrepresentation,acovariantlychiralscalarsuper“eld (+), €(+)=0isch iralinthe”atsuperspacesense D€(+)=0.There fore,its transformation (5.5.2b)canbewrittenas (+) = ei (+)eŠ i = ei ch(+)eŠ i ch(5.5.6a) where ch=(for€=0)= mi m+iD.(5. 5.6b) Sincefor n = Š1 3 thetransformationof 3is(5.2.68) 3 = 3ei ch, ch=(for€=0)=( mi m+iD ),(5.5 .7) thequantity 3isasuitablechiraldensitytocovariantizethe”atspacechiralmeasure (sees ec.5.2.c). Sn = Š1 3 = d4xd23ILchiral.(5. 5.8) ThisistheintegratedversionofanF-typedensitymultiplet.Forothervaluesof n nodimensionlesschiraldensityexists.Wedescribehowthissituationishandledbelow. e.Representationindependentformofthechiralmeasure For n = Š1 3 ,wecanwritethe chiralme asureintermsoftherealmeasure.From (5.3.66b)wehave,in chiralrepresentation, D2EŠ 1IL = 3( 2+ R ) IL .(5. 5.9) Thus d4xd4 EŠ 1ILgen= d4xd23( 2+ R ) ILgen.(5. 5.10) Fr om(5.3.66a)wecan“ndthevectorrepresentationof(5.5.10): d4xd4 EŠ 1ILgen= d4xd2 eŠ 3( 2+ R ) ILgen.(5. 5.11) Ifwechoose ILgen= RŠ 1ILchiral,since ILchiral= R =0,

PAGE 316

3025.CLASSICALN=1SUPERGRAVITY d4xd23ILchiral= d4xd4 EŠ 1RŠ 1ILchiral.(5. 5.12) Theformofthechiralmeasure d4xd4 EŠ 1RŠ 1isvalidin all representationssinceitdoes notdependontheexistenceofachiraldensity.Itismanifestlycovariantand,inprinciple,couldbeusedforall n =0( R =0for n =0).ThusFtypede nsitymultiplets alsoexistfor nonminimal supergravity.However,unless n = Š1 3 ,(5. 5.12)leadstocompo ne nt actionscontaininginversepowersoftheauxiliary“elds(withtheexceptionofRinvariantsystems:seebelow). The U (1)-covariantformofthechiralmeasure( n = Š1 3 )issomewhatm ores ubtle: U (1)invariancealonegivestheanalogof(5.5.12)as S = d4xd4 EŠ 1RŠ 13(1 Š1 2 w )ILchiral(5.5.13) when[ Y ILchiral]=1 2 wILchiral.Inparticular,su perconfo rmalactionsalwayshave w =2. However,scaleinvarianceisnotsostraightforward:Using(5.3.10),weseethat R hasan inhomogeneousterminitstransformationlawproportionalto 2L butbecauseofthe chira lityof R ,,and ILchiralthistermvanishesuponintegrationbyparts.Thusthechiralityofthecompensatorisessentialforconstructinggeneralchiralactions.Sinceinthe ( U (1)-)chiralrepresentation= ,u sing(5.3.64a),wereobtain(5.5.8).Theexpression (5.5.13)canbeusedfor n = Š1 3 ,0if w =2,sinceth entheactionisindependentand consequentlysuperconformal. f.Scalarmultiplet Todisc ussspeci“ccouplingstomatter,we“rstconsiderthe U (1)-covariantform. Accordingtoourgeneralprescription,thedir ectcovariantizationoftheaction(4.1.1)for thefre escalarmul tipletis S = d4xd4 EŠ 1 ,( 5.5.14a) withcovariantlychiral € =0,(5. 5.14b) Thisformoftheactionisvalidforany n .Ifwea ssignscaleweight d =1to (see

PAGE 317

5.5.Actions303(5.5.5)),theactionissuperconformalsinceitisindependentofthetensortypecompensators.Inparticular,itisinvariantundertherestrictedsuperconformaltransformationsthats urvivein Poincar esupergravity( 5.3.76).Atthecomponentlevelthisaction leadstoconformallycoupledscalar“eldswithactionsasin(5.1.35)butwiththeoppositesign.(Compensatorsgenerallyhaveacti onswithanoverallminussignrelativeto physicalsystems).Sincetheactionissuperconformalevenwithoutthecompensators,it isclearthatthescalarsofthemultipletareconformallycoupledtogravitywithoutthe needforacomponentcalculation. Afterdegauging,theaction(5.5.14a)andde“ningconditionremainunchangedfor n = Š1 3 .Forthen onminimaltheories, weuset hesameactionbutif wewantthecomponentscal arstobeconformallycoupledtogravit ywemustc hangethede“ningcondition((5.5.18)with w =2 3 ;seebelow).A lternativelyifthede“ni ngconditionisnotmodi“edandthe operatorin(5.5.14b)isforadegaugednonminimaltheory,thenthe actionof(5.5.14a)doesnothaveconformallycoupledscalars. f.1.Superconformalinteractions Forsuper conformallyinvariantactions,thedensitycompensator canbegauged away(i.e .,removedbya“eldre de“nitionwhichisasupersca letransformation).Inthis case,theaction(5 .5.8)writtenfor n = Š1 3 makessenseforany n .Forex ample, since EŠ 1= EŠ1 3 (1 eŠ H)1 3 (5.2.72),theconformallyinvariantactionforacovariantlychiral scalarsuper“eldis S = d8zEŠ 1 +( 1 3! d6z 33+ h c .).(5. 5.15) Atthecomponent level,thetermsproportionalto describequartics elf-interactions andYukawacouplingsforthecomponent“el dsofthematterchiralmultipletjustasin theglobalcase. Therescaling removes fromtheactio nentirely,an dtheresult isvalidforany n ( isachiraldensityofweight w =2 3 ).Thiscanbeg eneralized s lightly:Toremove fromthechiralintegrands,fullsuperconformalinvarianceisnot required;R-invariance(3.6.14)issucient.Then appearsonlyinthe fullsuperspace integrand,andonlyi nthecomb ination ;inthatcase,the n = Š1 3 compensator can

PAGE 318

3045.CLASSICALN=1SUPERGRAVITYberepl acedby n =0orno nminimalcompensators.Forexample,inthecaseofasingle chiralmu ltiplet,wecangeneralizetoarescaledchiraldensity witharbitraryweight w ; see(5.5.20)andtheparagraphafter(5.5.32)below. f.2.Conformallynoninvariantactions Nonconformalcouplingsofthescalarmultipletarealsopossible.Wecanalways addthesupersymmetricterm Snonconf= d4xd4 EŠ 1( 2+ 2)(5. 5.16) Thisactuallyvanishesfor n =0.(Itisalso po ssibletowriteaCPnon-conservingterm bytaking i timethedierenceinsteadofthesumin(5.5.16).)Atthecomponentlevel, (5.5.16)generatesd is-improvementterms r (A2Š B2)+ ... withoppositecontributions forthescal arandpseudoscalar“elds.Therefore,(5.5.16)cannotbeusedtoeliminate theimprovementtermsofboth“elds.For n = Š1 3 ,wecanrewrit e(5. 5.16)asthechiral integral Snonconf= d4xd23R 2+ h c .(5. 5.17) Fornonmi nimal( n = Š1 3 ,0)supergravit y(5. 5.16)alsointroducesdisimprovem enttermsbutthereexistsanotherwayofintroducingsuchnonconformal termsforthescalarmultiplet.Beforedegauging,ifthescaleweightof isnot1 ,then theonlywaytowriteasuperconformalkineticactionforthechiralmultipletistointroduceoneofthedensitycompensatorsof(5.5.5).Thustheactionwithoutthecompensatorsisnotsuperconformal.Afterdegaugingthe U (1)invariance(seesec.5.3.b.8),we canusethenewtensor Ttode “neamodi“edchiralcondition.Wecanreplace (5.5.14b)by ( €+1 2 w T€) =0.(5. 5.18) Thekineticactionisstillgivenby(5.5.14a),andif w =2 3 ,itisnotsuperc onformalwithoutoneofthecompensatorsof(5.5.5).Eventhoughthe U (1)groupis nolonger gauged,itstillexistsasaglobalR-invarianceoftheaction(5.5.14),andtheconstraint (5 .5 18) is covariantevenunderlocaltransformations

PAGE 319

5.5.Actions305[ Y ]=1 2 w .(5. 5.19) Themodi“edchiralityconditionof(5.5.18)intermsofunconstrainedsuper“elds leadstoamorecomplicatedactionfort hescal armultiplet.Wecanexpress interms ofa”at chiraldensity eŠ1 2 w T E€ =0(inthechiral representation D€ =0).In termsofunconstrainedsuper“elds,theaction(5.5.14a)becomes(inthegaugewithcompensat orssetequaltoone) S = d4xd4 En [(1 eŠ )(1 e )]n+1 2 ,(5. 5.20) where nisde “nedintermsof w by w =2n Š n3 n +1 .Only n= Š1 3 issuperconformaland hastheconventionalconformalimprovementterms;then w =2 3 .Thusinthe nonminimaltheories,conformalcouplingforthescalarsisachievedbyreplacing(5.5.14b)by (5.5.18)with w =2 3 f.3.Chiralself-interactions Thecovariantizationofanyglobalchir alpolynomialself-interactiontermsP( )is straightforward.Fromourgeneralprescription(5.5.12)wehave(for n =0) Sint= d4xd4 EŠ 1RŠ 1P( )+ h c ..(5.5 .21) Theexpression(5.5.21)remainslocallysupersymmetricfor“eldssatisfyingthemodi“ed chira litycondition(5.5.18)orinthe U (1)-covariantforma lismwitht heusual € =0 forarbitrarychiralweight.For n = Š1 3 Sintispolynomialinthecomponent“eldsafter theeliminationofthesupergravityauxiliary“eldswheneverP( )ispol ynomial.For n = Š1 3 ,itisingen eralnonpolynomial,exceptforP( )= 2 w (forasinglechiralmultiplet;formoremultiplets,theconditionisgivenbelow).Thus,asmentionedearlier, althoughF-typedensitiesexistfornonminimaltheories,ingeneralthesewillleadto no np olynomialityaftertheeliminationofauxiliary“elds. Inthechiralrepresentation,for n = Š1 3 ,(5. 5.21)canberewrittenintheform (5.5.8),andfor n = Š1 3 ,intheform

PAGE 320

3065.CLASSICALN=1SUPERGRAVITYS = d4xd2 eŠ TP( )+ h c .(5. 5.22) whenthechiralchargeofP( )is1 2 w =1.Thisfo llowsfrom(5.5.13),sincefor n = Š1 3 S mustbei ndependent.(Forthespecialinteractiongivenabove,thiscanbewritten asd4xd2 2 w ,withnodep endenceonthesupergravity“elds.) g.Vectormultiplet Thevectormultipletcanbecoupledtosupergravitybysimplyde“ningderivativesthatareco variantwithresp ecttothelocalinvariancesofbothsupergravityand su pe r-Yang-Mills: A= EA+(A M +A€€ M€€) Š i AATA; A = eiKAeŠ iK, K = KMiDM+( K iM + h c .)+ KATA;(5. 5.23) whereAAistheYang-Millspotentialand KAitsgaugeparameter.Fieldstrengthsfor bo thsupergravityandYang-Millsarede“nedbythegradedcommutatorsofthecovariantderivativesasusual,andthesamesupergravityandYang-Millsconstraintsare imposed(see(4.2.66)and(5.3.4,5,13)).Thesolutiontotheconstraintscanbegivenby expressing Aintermsofthepures upergravityc ovariantderivatives Aandthe usual Ya ng -M illssuperpotential=ATA: = eŠ e, €= Š i {, €} .(5. 5.24) Alternatively,thesolutioncanbewrittenwith ofthesameformas butnowwith =MiDM+(inanalogyto U U + V intheglobalcase).TheYang-Mills“eld strengthis W= i [ €, { €, } ](5. 5.25) andtheactionis S = gŠ 2tr d4xd4 EŠ 1RŠ 1W2.(5. 5.26)

PAGE 321

5.5.Actions307For n = Š1 3 intheYang-Millschiralrepresentation,usingtheBianchiidentities (5.4.16,18)wecanrewrite(5.5.25)as W= i ( 2+ R ) eŠ VeV,(5. 5.27) andtheactionis S = gŠ 2tr d4xd23W2.(5. 5.28) Asfortheconformalcouplingofthescalarmultipletin(5.5.8),theactionsin(5.5.26,28) areinvariantwithrespecttotheconformaltransformationsparametrizedbyarbitrary L and K5super“elds.Thisensuresthatthe dependenceof W issuchthatit cancelsin theactionof(5.5.28).Moregenerally,alsoasaconsequenceofconformalinvariance, (5.5.26)isindependentof or T For n =0thefor m(5. 5.26)cannotbeusedsince R =0and( 5.5.28)cannotbe usedsince onlyoccursinthe n =1 3 theory.T hecorrectactionis S = gŠ 2tr d4xd4 EŠ 1( WŠ1 6 [ €,€])+ h c .(5. 5.29) where Wisgivenby(5.5.25)and,€areobtainedfrom(5.5.24)using Š i A= AŠA.The gaugeinvarianceoftheaction(5.5.29)followsfromtheBianchi identity W+ € W€=0.Wenotetha tthisform( va lidforall n andinallrepresentations)issimilartothethree-dimensionalgaugeinvariantmassterm(2.4.38). Alternatively,itispossibletouse(5.5.28)forall n ,ifweareon lyinterestedinthe explicitdependenceonthesupergravityprepotential H m.The H mdependenceanddensitytypecompensatorindependenceofany truly conformalactionisindependentof n h.Generalmattermodels Wenowconsi derageneralclassofmattermulti plets(chir alandgauge)coupled to n = Š1 3 supergravity.A globally supersymmetricgaugeinvariantaction,restricted onlybytherequirementth atnobosonictermswithmorethantwoderivativesor fermionictermswithmorethanonederivat iveappearinthecomponentLagrangianis

PAGE 322

3085.CLASSICALN=1SUPERGRAVITYS = d4xd4 [ IK (i,j)+ trV ] + d4xd2 [P(i)+1 4 QAB(i) WAWB]+ h c .(5. 5.30) where j= k( eV)j k, WA= i D2( eŠ VDeV)A(5.5.31) andP(i)and QAB(i)= AB+ O ()arechir al.Theterm trV isthe(global)FayetIliopoulosterm(4.3.3).Asexplainedinsec.(4.1.b), IK canbeinterpretedastheK¨ ahler potentialofa ninterna lspacema nifold. Thecorresponding locally supersymmetricaction, including ( n = Š1 3 )supergravity, is S = Š3 2 d8zEŠ 1eŠ 23 [ IK (i,j)+ trV ]+ d6z 3[P(i)+1 4 QAB(i) WAWB]+ h c .(5. 5.32) Inthelimit 0, E and 1,thisreducestotheglobalaction(5.5.30).ThecovariantFayet-Iliopoulostermis(Yang-Mills)gaugeinvariantonlyifthechiralactionisgloballyR-invariant.Underagaugetransformation ( trV )= itr ( Š ), EŠ 1exp ( Š1 3 2 trV )isinvaria ntifwesimu ltaneouslyperformthe(restricted)complex superscaletransformationdiscusseda ttheendofs ec.5.3withchiralparameter L + i1 3 K5= Š i26 tr .Theinvarianceofthe chiral integralin(5.5.32)followsfromRinvarianceof(thechiralpieceof)theglobalaction. Ingeneralthecouplingsofthescalarmultipletinsuperspaceinvolveconformal couplingofthespinzerocomponent“eldstogravity.Thereisonespecialchoiceofthe K¨ ahlerpotential,however,whereallsuchconformalcouplingcanbeeliminatedforthe componentscalar“elds.Thisspecialchoiceisgivenby IK (i,j)=ii. ForR-invaria nttheori essuperscaletransformationscanbeusedtorescalethematter“eldsan dremove fromthe chiral integral;asmentionedabove,theresultingaction dependsonlyonthecombination ,andweca nrewriteitforany n ,e.g .,usingduality

PAGE 323

5.5.Actions309transformationsofthecompensatoraswillbe describedinsec.5.5.ibelow.Inparticular,ifweperformadualitytransformationtothe n =0theory,theac tion(5.5.32) b ecomes S = d8zEŠ 1[ Š1 2 V5+ IK (i,j)+ trV ] + d4xd2 [P(i)+1 4 QAB(i) WAWB]+ h c .(5. 5.33) whereiand Waresuita blyde“neddensities(the -independentquantitieswede“ned tomakethedualitytransformationpossible). Althoughwehaveconcentratedhereonthe n = Š1 3 theory coupledtovectorand chiralsc alarmultiplets,moregeneralsystemsalsocanbeconsidered.Aswestated above,couplingofotherversionsofsuperg ravitycanbeobtainedbyperformingduality transformations.Asdescribedinchapter4,therearealargenumberofscalarmultipletsandmanyothermattermultiplets.Thesemaybecoupledtosupergravitybyuse theprescriptionof(5.5.4,12). i.Supergravityactions i.1.Poincar e For n =0,thePoincar esupergravitya ctionisobtainedfrom(5.5.4)(or(5.5.5), for d =0)bychoosing Lgen=( n 2)Š 1.For n = Š1 3 ,thiscanbere writtenas S = Š 3 Š 2 d4xd23R .(5. 5.34) For n =0,theobvi ouschoice S = d8zEŠ 1,(oritss cale invariantformwiththe tensorcompensator(see(5.5.5c))and Lgen= Š 2)vanishes:With thecompensator G =1,thechiral curvature R =0 (seesec.5.3.b.7.iii),and(e.g.,inthechiralrepresentation)(5.3.56)implies D2EŠ 1=0.Iftheacti onvanishesinonegauge,itmustdosoin allgauges,includingoneswherethecompensatorhasnotbeengaugedaway.However, the n =0theoryha sadime nsionless U (1)prepotential V5thatallowsustowritean action:Since D2EŠ 1=0,int he gauge G =1 th ef o llo wingactionisinvariantunder

PAGE 324

3105.CLASSICALN=1SUPERGRAVITYU (1)gaugetransformations E =0, V5= i ( 5Š 5): Sn =0= Š1 2 d4xd4 EŠ 1V5.(5. 5.35) Inthechiralrepresentation,thiscanberewritten,using(5.3.63),as Sn =0=3 2 d4xd4 EŠ 1ln [ EŠ 1 E1 3 (1 eŠ H)Š1 3 ].(5.5 .36) Sinceinthegauge G =1wehave EŠ 1= G (5.3.70),thisisthecovariantizationofthe ”atspaceaction(4.4.46)fortheimprovedte nsormult iplet.Wesawthat(4.4.46)could bewri ttenina“rst-orderformthatmademani festthedualitybetweenthescalarand tensormultiplet.Thisconstructioncarriesovertothelocalcase,andwe“ndthat n =0 supergravity(withatensorcompensator)isdualto n = Š1 3 supergravity(withachiral scalarcompensator). Wewritea“rst-o rderactionas S = Š3 2 d4xd4€€EŠ 1( e€€XŠ€€G€€X ),€€G =1 2 (€€+ €€ € €), €€ €=0;(5. 5.37) where€€X isanindepe ndent,unconstrained,realsuper“eld,and all objects(€€EŠ 1,€€)are thoseof n = Š1 3 .Thisi sjust n = Š1 3 supergravitycoupledtothe“rst-orderformof theimprovedtensormultiplet(4 .4.45).Ifwevarywithrespectto€€X ,ands ubsti tutethe resultbackinto(5.5.37),we“ndthe n =0acti on;ontheotherha nd,ifwevarywith respectto ,we “ndthe n = Š1 3 action.Indetail,wehave,fromthevariationwith respectto€€X€€X = ln€€G (5.5.38) andhence(5.5.37)becomes Sn =0=3 2 d4xd4€€EŠ 1(€€Gln€€G Š€€G )(5. 5.39) B ecause€€G islinear,thesecondterm canbedropped.Since€€EŠ 1€€G = G = EŠ 1G ,(see

PAGE 325

5.5.Actions311(5.3.63)),using(5.3.63)weobtaintheaction(5.5.35)withthecompensator G inageneralgauge: Sn =0=3 2 d4xd4 EŠ 1G ( lnG Š1 3 V5).(5.5 .40) Thisactionisscaleand U (1)invar iant. Alternatively,vari ationwithrespectto gives( €€ 2+€€R )€€€€X =0andh ence€€X = ln + ln €€ €= 0,sothatagainusingthelinearityof€€G toeliminatetheterms€€Gln + h c .,weobtain S = Š3 2 d4xd4€€EŠ 1 ,(5.5 .41) i.e.,the n = Š1 3 action(5.5.5a). Thedualitytransformationfromthe n = Š1 3 supergravitytheorytothe n =0theory,asdescribedabove,canbereversedthro ughastraightforwardc ovaria ntizat ionof thereversedualtransform(4.4.38)(compareto(4.4.42)).Bothformsoftheduality transformcanbeperformedeveninsystemswh erethesupergravity mult ipletiscoupled tomattermultiplets(justasins ec.4.4.c.2);however,thoughany n =0sy stemcanbe conv ertedtoan n = Š1 3 system,thereversetransformationispossibleonlyifthe n = Š1 3 systemisR-invariant,andhencetheactioncanbewrittensothatitdepends onthe n = Š1 3 compensator(tensorordensitytype)inthecombination or Analogousdualitytransformationsthatarethecovariantizationofthosedescribed attheendofsec.4.5.b.canbeusedtorelate n = Š1 3 andnonminimalsupergravitysystems. Theformofthesuperspaceactionfor n =0revealsach aracteristiccommonto mostextendedsupersymmetrictheories.Naively,wemightexpectactionstotakeageometricalformdzEŠ 1IL ( fieldstrengths ).However, wecaneas ilyseethatfor N 3, evenif dz isachiralmeasure,therearenoquantitiesofproperdimensionstoformsuch anactionforglobalorlocalsupersymmetry.Ourexperiencewiththe n =0theory showsthatitispossible,aftersolvingconstraints,to“ndquantitieslike V5thatwemay

PAGE 326

3125.CLASSICALN=1SUPERGRAVITYcallsemiprepotentialsorprecurvatures,withoutwhichtheactioncannotbewritten. Thus,theunconstrainedsuper“eldapproachbecomesincreasinglyimportant,sincesuch precurvaturesareactuallyfoundasintermediatestepsinsolvingconstraints. i.2.Cosmologicalterm TothePoi ncar esupergravitya ctionwecanaddasupersymmetriccosmological term(foradiscussionofglobaldeSittersupersymmetry,seesec.5.7).For n = Š1 3 ,we have Scosmo= Š 2 d4xd23+ h c .(5. 5.42) For n = Š1 3 ,0wecouldwriteatypeo fcosmol ogicaltermusingtheform(5.5.12), but thattermcontainsinversepowersofthescalarauxiliary“eld;for n =0, R =0and henceitisimpossibletowriteacosmologicalterm(thesedicultiesarisebecausethe cosmologicaltermis not R-invarian t).Oneotherinterestingfeatureofthecosmological termfornonminimalsupergravityisthatthesumof(5.5.12)(with ILchiral=1)and (5.5.5b)(with ILgen=1)leadst oaspontaneous breakingofsupersymmetry.Theresulting“eldequationsaresuchthatitisnotpossibletoconstructananti-deSitterbackgroundwhichissupersymmetric(seesec.5.7). i.3.Conformalsupergravity Nextweconsidertheactionforconformalsupergravity.Itisjustthecovariantizationofthelin earizedexpression(5.2.6): Sconf= d4xd23( W)2.(5. 5.43) Theconformal“eldstrength Wdependson onlythroughaproportionalityfactor Š3 2 ,soall dependencecancels.Thef ormof(5.5.43)isvalidonlyfortheminimaltheory.Itcanbeextendedtothenonminimaltheorybytheuseof(5.5.12).For n =0ev en thisinsucient,againbecause R =0.Thisdoe snotimplyt hatconformalsupergravity doesnotexist;itis n -independent.Insteadtheactionforconformalsupergravitytakes aformsim ilartothethree-dimensionalsupergravitytopologicalmasstermwith Wtakingtheplaceofth ethr ee-dimensional G,and G€and R theplaceof thet hree-

PAGE 327

5.5.Actions313dimensional R (see(2.6.47)). j.Fieldequations Toobtaincovari ant“eldequationsfromtheactionbyfunctionaldierentiation withrespecttothesupergravitysuper“elds,whicharenotcovariantthemselves,we de “ne am od i“edfunctionalvariation,aswedidforsuper-Yang-Mills(see(4.2.48)): H eŠ H eHor eŠ e, ( e ) eŠ ;( 5.5.44a) ( 3).(5.5 .44b) Theequationsofmotionforsupergravitywithactiongivenby(5.5.4)andthecosmologicalterm(5.5.42)canthenbeshowntobe S H a = Š Š 2G a=0, S = Š Š 2( R Š )=0.(5 .5.45) Thecovariantized“eldequationfor H€isthesam easthato btainedbythebackground “eldmethod(thevariationisthesameasthebackground-quantumsplittinglinearizedinthequantum“eld). Thederivationofthis“eldequationwillbedescribedin moredetailwhenwedescribethissplittinginsec.7.2.The equationisea s ilyobtained using(5. 2.71,5.3.56,5.5.34,42). To obtaincovariant“eldequationsforacovariantlychiralsuper“eld,itisnecessarytode“neasuitablefunctionalderivative.Thiscanbedoneinanyofthreeways: (1)by“rstusingthe” at-space de“nitionfordierentiationby ,and usingtherelation (5.5.2);(2)bycovariantizingthe”at-spaceforminawaythatsatis“esthecorrect covariantchiralitycondition;or(3)byexpressingthechiralsuper“eldasthe“eld strengthofageneralsuper“eld.Theresultis: ( z ) ( z) =( 2+ R ) 8( z Š z),(5.5 .46) where,for n = Š1 3 is U (1)covariant.Theresulting“eldequationsforascalarmultipletarethusthesameasintheglobalcaseexceptthat D2isrepla cedwith 2+ R The“eldequationsforsupergravitycoupledtoascalarmultipletare(for n = Š1 3 ,and usingtheacti on(5.5.14))

PAGE 328

3145.CLASSICALN=1SUPERGRAVITYŠ 2G€= 1 3 [ i€ Š ( € )( )+ G€], Š 2R =1 3 ( 2+ R ) =0.(5. 5.47) Whenaself-interactiontermisadded,the G equationisunchanged,but R b ecomes nonzero(exceptforthesuperconformalcoupling 3).Inthelastequat ionwehav eused theequationofmotionofthescalarmultiplet.Alternatively,termsin“eldequations prop ortionaltoother“eldequationscanberemovedingeneralevenoshellby“eld rede“nitionsintheaction.(Toremovetermsproportionaltothe“eldequationsof 2fromthe“elde quationsof 1,a“eldre de“nitionof theform 1= 1 2= 2 + f modi“esthe“eldequationsto S 1 = S 1 + S 2 f 2 .)Inthiscase,the appropriate“eld rede“nitionis = Š 1= Š + ... ,whichre movesall dependencefromthe scalar-multipletaction(seesec.7.10.c). Fo rt he co upl ed su pe rgravity-Yang-Millssystem(sec.5.5.h),the“eldequationsfor Ya ng -M illsarestill {, W} =0,wh ilethesupergravityequationsare Š 2G€= gŠ 2tr W€W, Š 2R =0.(5. 5.48) Wehave droppedtermsinthe G eq ua tionproportionaltotheYang-Mills“eldequation. Theseterms,whichinthiscasearenotYang-Millsgaugecovariant,canagainbeelimina tedbya“eldrede“nition(againseesec.7.10.c). Althoughwehaveonlyconsidered n = Š1 3 forsimp licity,covariantvariationwith respecttothecompensatorsfortheotherversionsofsupergravitycanalsobede“ned analogously.Inboth n =0andn onminimaltheoriestheimportantpointtonotein de“ningthecovariantvariationsisthattheunconstrainedcompensatorsforboththeoriesarespinors(= D€€forthenonminimaltheoryand for n =0).Thus functionaldierentiationinthesecasesle adtospinorialequationsofmotion.

PAGE 329

5.6.Fromsuperspacetocomponents3155.6.Fromsuperspacetocomponents a.Generalconsiderations Sofarinourdiscussionofsupergravitywehaveconcentratedexclusivelyon superspaceandsuper“elds.Ontheotherhand,somewhereinthisformalismasupergravitytheoryinordinaryspacetimeisbei ngdescribed.Thequestionariseshowto extractfromasuperspaceformulationinformationaboutcomponent“elds.Weknow howtodothisintheglobalsupersymmetryc ase,andherewewilldescribethecorrespondingprocedureinlocalsupersymmetry,andderivethe tensorcalculus ofcomponent supergravity.Wecannotuse D and D tode “nethecomponentsofsuper“eldsbyprojectionasinglobalsuperspace,sincethiswouldnotbecovariantwithrespecttolocal supersymmetry. Todisc usscomponentsupergravity,wemust“rstchooseaWess-Zuminogaugein whichthe K -transformationshavebeenusedtosettozeroallsupergravitycomponents thatcanbegaugedawayalgebraically.AWess-Zuminogaugeisnecessarysothat resultsfornoncovariantquantities(i.e.gauge“elds)canbederivedalongwiththosefor covariantquantities.Wecanthenderivetra nsformationlawsfortheremainingsupergravitycomponentsaswellascomponentsofothersuper“eldsandexhibitsupercovariantizationandthecommutatoralgebraofloc alsupersymmetryatthecomponentlevel. Wederivemult iplicationrulesforlocal(covariantlychiral)scalarmultiplets,andwrite thecomponentformoftheintegrationmeasures(densityformulae),fromwhichcomponentactionscanbeobtained.Alltheresultsre”ecttheunderlyingsuperspacegeometry andcanbeobtainedforany N ,imposingasf ewconstraintsaspossible(preferably none).Thisimpliesthatsuperspacegeometryismoregeneralthanacomponenttensor calculuswhichfo llowsfromachoiceofconstraintsonsuperspacetorsionsand/orcurvatures.The “nalformofthetensorcalculusisdeterminedbywhichsolutionofthe Bianchiidentitiesisutilized. Webeginwitha generalsuperspacefor N -extendedsupergravity.(Wewillspecializeto N =1when everneeded.)Insuchasuperspacewehaveavielbein EA Mwhich describessupergravity.Wealsointroduceanumberofconnectionsuper“eldsA for tangentspacesymmetriessuchasLorentzrotations,scaletransformations, SU ( N )-rotations,centralcharges,et c.Thesesuper“eldsarecombinedwithoperators DMand M

PAGE 330

3165.CLASSICALN=1SUPERGRAVITYtoformasuper covariantderivative A= EA+A M EA= EA MDM,(5. 6.1) where M arethegeneratorsofthetangentspacesymmetries.The -subscripti salabel thatrunsoverallthegeneratorsofthetangentspacesymmetries.Forinstance,in N =1, n =0superg ravity M =( M, M€€).Therealizationofthesegeneratorsis speci“edbygivingtheiractiononanarbitrarytangentvector XA.Thus,fors omesetof matrices( M )A Bwehave [ M XA]=( M )A BXB.(5. 6.2) Wewrite DM= DM N zN +M M for “xed matrices DM NandM where DM NŠ M NandM vanishat =0(sees ec.3.4.c).Weassumethevielbeinisinvertible;speci“cally,weassumethatwecanalways“ndacoordinatesystem(orgauge)inwhichwecan write A= A+A.(5. 6.3) Thegaugetransformationsof Aaregivenasusualby A= eiKAeŠ iK.(5. 6.4) Theparameter K isasuper“eldwhichisalsoexpandedover iDMand iM K = KMiDM+ K iM ,(5. 6.5a) andissubjecttoarealitycondition K = ( K ).Wecanequallywell expandtheparameter K overthecovariantderivatives Aand M : K = KAi A+( K Š KAA ) iM = KAi A+ K iM .(5. 6.5b) A gaugetransformationofan arbitrary covariantsuper“eldquantityisalwaysgenerated byacti ngwith iK as in (5 .6 .4 ). Fo ri n“nitesimaltransformationsofsupercovariant quantitiesthisimpliesthatwesimplyactonthequantitywiththeoperator iK .

PAGE 331

5.6.Fromsuperspacetocomponents317b. We ss-Zuminogaugeforsupergravity Wede “necomponentsof covariantq uantit ies (matter“elds,torsionsandcurvatures)usingthelocalgeneralizationofthe covariantprojectionmethodintroducedina globalcontext:thesecomponentsarethe i ndependentprojectionsofthesuper“elds andtheircovariantderivatives.Wede“necomponentsof gauge“eldsEA M,A by choosingasp ecialgaugeandthenprojectingasoncovariantquantities.ThisWessZuminogaugechoicereducesthesuperspacegaugetransformationstocomponentgauge transformations:Itusesallbutthe i ndependentpartof K toalgebraicallygauge aw ay th en oncovariantpiecesofthehighercomponentsofthegauge“elds(thelowest componentsremainasthespacetimecomponentgauge“elds).Weusethenotation X | tomeanthe i ndependentpartofanysuper“eldquantity X ;if X isanoperator XMiDM+ X iM ,then X | istheoperator XM| i M+ X | iM ;weuse DM| = Manddo not set , € to zero.Inparticular,wede“nethecomponentsofthecovariantderivativesby C| C| C| €C| ,etc. Wede “netheusualco mponentgauge“eldsby a| e a m m+ a + a€ € + a M D D a+ a + a€ € ,(5. 6.6) where e a misthecomponentinversevierbein, a , a€ arethecomponentgravitino “elds,and a arethecomponentgauge“eldsofthecomponenttangentspacesymmetries.(For M =( M M€€)these gauge“eldsaretheLorentzspinconnections a and a€€.) Fr om th ei n“nitesimaltransformationlaw a=[ iK a]= Š i aK + ... weseethatt hesecomponentstransformasspacetim egradientsof thegaugeparameters, whichjusti“esthede“nition.Wehavealsointroducedtheordinaryspacetimecovariant deriva tive D D a= e a+ a M (cf.5.1.15).Covariantlytransformingcomponentsofthe su pe rgravitymultiplet(e.g.auxiliary“elds)appearascomponentsofthetorsionsand curvatures. Wenowderivet he“rstfewcomponentsofthecovariantderivatives.Webeginby exploitingtheexistenceofaWess-Zuminogauge.Fromthein“nitesimaltransformation law =[ iK ],using(5.6.3)we“nd

PAGE 332

3185.CLASSICALN=1SUPERGRAVITY | = Š i K + ... =[ iK ] | K(1) (5.6.7) andhence,byusingthe componentof K i.e., K(1) ,wecanchoo seagauge | =0 or | = .(5. 6.8) Wehaveth usdetermined A| .W ec anproceedto“ndthehigher-ordertermsina straightforwardmanner.Thusto“nd | westartwith ( )=[ iK ].(5.6 .9) Then ( ) | = Š i K + ... .(5. 6.10) Since { } =0weca n gaugeaway[ ] butnot { } .Howev er,the latter iscovariant:Itcanbeexpressedintermsof torsionsandc urvatures.Henceinthis gaugewe“ndthe componentof : | =1 2 { }| =1 2 T CC| +1 2 R | M .(5. 6.11) Inthesameway,we“ndthe € componentof : € | =1 2 { € }| =1 2 T€ CC| +1 2 R€ | M .(5. 6.12) Similarly,we“ndthenextcomponentof b;we“rstobserve thatb ecause | = ,wehave b| = D D b+ b | + b€ € | .(5. 6.13) Thenwecompute b| =[ b] | + b | =[ b] | + D D b | + b | + b€ € | .(5. 6.14) Using(5.6.8,11,12)weobtain

PAGE 333

5.6.Fromsuperspacetocomponents319 b| = T b CC| + R b | M + b [ M | ] +1 2 b [T CC| + R | M ]+1 2 b€ [T€ CC| + R€ | M ](5.6.15) Thuswehavefound B| .Wecan “ndhighercomponents,butwhatwehaveissucientfortheapplicat ionswegivebelow.Wehaveobtainedtheseformulaewithout imposing any constraints. Theprocedurewehavedescribedusesallthehighercomponents(projectionswith more s)of K to eliminatethenoncovariantpiecesof and € andde “nestheWessZuminogauge.Theremaininggaugetransformations,determinedbythe i ndependent term K | ,arejusttheu sualcomponenttransformations.Coordinatetransformationsare determinedby iKGC| = Š m( x ) m(5.6.16) (orequivalentlycovarianttranslations iKCT| = Š a( x ) D D a= Š m mŠ a( a M )). Tangents pace gaugetransformationsaredeterminedby iKTS| = Š ( x ) M ,(5. 6.17) andsupersymmetrytransfo rmationsaredeterminedby iKQ| = Š ( x ) Š € ( x ) € = Š ( x ) |Š € ( x ) € | .(5. 6.18) However,tostayintheW ess-Zuminogauge,the K transformationsmustbe restricted:thehighercomponentsareexpressedintermsof K | .Forex ample, | = implies: =0=[ iK ] | ,( 5.6.19a) sothat 0= Š [ KBB, ] |Š [ K M ] | = Š KB[ B, }| +[ KB}B|Š K [ M ] | +[ K ] | M ,(5. 6.19b) andhence

PAGE 334

3205.CLASSICALN=1SUPERGRAVITY KB| = KCTC B| + K | ( M ) B K | = KCRC | .(5. 6.20) Similarly,wecan“ndthehighercomponentsof K fromthehigherc omponentsof andtherequirementt hattheWess-Zuminogaugeismaintained.Itturnsoutthatinthe We ss-Zuminogauge(5.6.16)holdstoallordersin i.e., KGChasnohighercomponents, whereasboth KTSand KQhavehighercomponentsdependingonthecomponent“elds, thegaugeparameters ,andin general,the gradients oftheparameters.Thus,inthe localcas e,wecannotwrite Š iKQ= Q + € Q€ forsomeoperator Q .The higher ordertermsin iKTSarealwaysproportionaltothematrices( M ) ,( M )€ € ;hen cefor internalsym metries ascomparedtotangentspacesymmetries, iKTShasnohighercomponentsa nd(5.6 .17)isexact. c.Commutatoralgebra Asanotherapplicationoft heuseoftheWess-Zuminogaugesupersymmetrygenerator,wederivethecommutatoralgebraoflocalcomponentsupersymmetry.Inthis gaugeweusethe dierentialoperator iKQasthelocalsupersymmetrygeneratorforthe componentformulationofsupergravity.Sincethesupersymmetrygeneratoris“eld dependent,wecanindicatethisbywriting iKQ( ; )where denotesallofthe x -space “eldscon tainedin iKQ.Thismea nsthatcaremustbe takeninde“ningthecommutator oftwosuchtransformations.Letusi magineperformingsequentiallyon twosupersymmetrytransformationswithparameters 2and 1.The “rsttransforma tionisobtained from iKQ( 2; ) ,wherewe havedroppedthecommutatornotation,keepinginmind that iKQisanoperator.Thesecondtransformationisimplementedby iKQ( 1; + 2 ) iKQ( 2; ) .Therefo re,thecorrectwaytocomputethecommutator algebraisfro mthe de“nition [ iKQ1, iKQ2] iKQ( 1; + 2 ) iKQ( 2; ) Š (1 2) iK12.(5. 6.21) However,bylookingattheformofthesupersymmetrygeneratorin(5.6.18)wenote that iKQ| hasno“elddependentterms.Thisimplies[ iKQ1, iKQ2] | correspo ndsto the usualcommutator[ iKQ( 1, ), iKQ( 2, )],andthisisallweneedto“ndthecomponent commutatoralgebra.Takingtheexpressionfor iKQfrom(5.6 .18)andusingtheWess-

PAGE 335

5.6.Fromsuperspacetocomponents321Zuminogauge-preservingcondition[ iK ] | =0(see (5.6.19)),weobtain [ iKQ1, iKQ2]= 1 2 { } + 1€ 2€ { € € } +( 1 2€ + 1€ 2 ) { € }| (5.6.22) Comparingtherighthandsideoftheaboveequationto iKGC| iKTS| and iKQ| we “nd iK12 iKGC( m)+ iKTS( )+ iKQ( ), m= Š [( 1 2€ + 1€ 2 ) T € c+ 1 2 T c+ 1€ 2€ T€ € c] e c m, = Š [( 1 2€ + 1€ 2 )( R € + T € c c ) + 1 2 ( R + T c c )+ 1€ 2€ ( R€ € + T€ € c c )], = Š [( 1 2€ + 1€ 2 )( T € + T € c c ) + 1 2 ( T + T c c )+ 1€ 2€ ( T€ € + T€ € c c )].(5. 6.23) Theseresultsshowhowthecommutatoralgebraoflocalsupersymmetryiscompletelydeterminedbysuperspacegeometry.Inparticularthe“elddependenceofthe localalgebraisaconsequenceofonlyconsideringcomponent“eldswhicharepresentin theWZgauge.Thefullresultfor(5.6.21),toallordersin isgivenby [ iKQ1, iKQ2]= iKGC( m)+ iKTS( ; )+ iKQ( ; ), + 2 Š 1 .(5. 6.24) d.Localsupersymmetryandcomponentgauge“elds Wenowderivet hesupersymmetryvar iationofthecomponentgauge“elds.We obtainthesebyevaluatingasuper“eldequationat =0 a| =[ iKQ, a] | .(5. 6.25) From(5 .6.13)wehave

PAGE 336

3225.CLASSICALN=1SUPERGRAVITY[ iK a] | = iK a|Š aiK | = Š ( + € € ) a|Š D D aiK |Š a iK |Š a€ € iK | .(5. 6.26) UsingtheWess-Zuminogaugecondition(5.6.19a,17),werewritethisas [ iK a] | = Š ( + € € ) a|Š D D aiK |Š a iK |Š a€ iK € | = Š ( a| + € € a| )+ D D a( + € € ) | + a ( + € € ) | + a€ ( + € € ) € | .(5. 6.27) Expa ndingover m, | and M ,a ndusing(5.6.10,11,15)we“nd: Qe a m= Š [ T a d+ € T€ a d+( € a + a€ ) T€ d+ a T d+ € a€ T€ € d] e d m, Q a = D D a Š ( T a + T a e e ) Š € ( T€ a + T€ a e e ) Š ( € a + a€ )( T € + T € e e ) Š a ( T + T e e ) Š € a€ ( T€ € + T€ € e e ), Q a = Š ( R a + T a e e ) Š € ( R€ a + T€ a e e ) Š ( € a + a€ )( R € + T € e e ) Š a ( R + T e e ) Š € a€ ( R€ € + T€ € e e ).(5.6 .28) Theseresultscanbespecializedto N =1, n = Š1 3 superspace,andusingthesolutionto constraintsandBianchiidentitieswecandeducethetransformationlawfor e a mand a :

PAGE 337

5.6.Fromsuperspacetocomponents323Qe a m= Š i ( a€+ € a ) e€ m,( 5.6.29a) Q a = D D a+ i €S Š i A€Š i ( € a + a€) € .(5. 6.29b) Thetransformationlawofthegravitinocanbesimpli“edsomewhatbyconsideringthe supersymmetryvariationof m .(Thelasttermin( 5.6.29b)isabsenta sacons equence of(5.6.29a).)Theauxiliary“elds S and A aarede “nedas R | and G a| respectively;consequently,theirtransformationscanbefounddirectlybecause R and G aarecovariant (seebelow).Thesecovariantde“nitionsoftheminimalauxiliary“eldsarethegeneralizationsofthelinearizedexpressionsof(5.2.8)and(5.2.73). Weshouldpo intoutthattheresultsfor Q a arevalidforgaugedinternalsymmetries(suchas U (1), SU (2),etc.)also.Inthiscase( M ) =0,( M )€ € =0andthe quantities RAB arethe “eldstrengthsfortheinternalsymmetrygaugesuper“eld. Thereforetheformulaein(5.6.28)containspartofthetensorcalculusforamattervectormultiplet.T hecovariantcomponentsofsuchamultipletaretreatedjustlikethose ofanycovariantmultiplet,e.g.,achiralscalarmultiplet. e.Superspace“eldstrengths Tosimp lifycalculationswithcomponentgauge“eldsitisconvenienttode“ne supercovariant“e ldstrengths(quantitieswhichtransformwithoutderivativesofthe localsup ersymmetryparameter).Webeginbycomputing [ a| b| ]=[ D D a, D D b]+( D D[ a b ] ) +( D D[ a b ]€ ) € +( [ a | | b ] ) +( [ a | €€ | b ]€ ) € .(5. 6.30) Theordinaryspacetimetorsionsandcurvaturesarede“nedby(see(5.1.17)) [ D D a, D D b]= t a b cD D c+ r a b M ,( 5.6.31a) where t a b c= c a b c+ [ a ( M ) b ] c,(5. 6.31b)

PAGE 338

3245.CLASSICALN=1SUPERGRAVITY[ e a, e b]= c a b ce c,(5. 6.31c) r a b = e[ a b ] Š c a b c c + a 1 b 2f 1 2 ,(5. 6.31d) and[ M 1, M 2]= f 1 2 3M 3.Wede “neacurvaturefor a by t a b D D[ a b ] Š t a b d d = e[ a b ] Š c a b d d Š [ b a ] .(5. 6.32) Wenowhaveallthe x -space“eldstrengths.Fromthefactthat e a m, a ,and a are gauge“elds, t a b c, t a b ,and r a b aretheappropriate“eldstr engths.The“eldstrength associatedwith M(and M€€), r a b = Š r a b ,€€istheRiemanncurvaturetensor.With thesede“nitionswecan expressthesuperspacetorsionandcurvatures at =0as T a b C= t a b C+ [ a T b ] C+ [ a€ T€ b ] C+ [ a b ] €T € C+ a b T C+ a€ b €T€ € C,(5. 6.33) R a b = r a b + [ a R b ] + [ a€ R€ b ] + [ a b ] €R € + a b R + a€ b €R€ € .(5. 6.34) Wehave used [ a, b] | =[ a| b| ]+ [ a b ]| + [ a€ € b ]| (5.6.35) and(5.6.15).Weseethesetensorsdierfromtheir x -spaceanalogs(5.1.17,18)byadditionalgravitinoterms.Thesuperspace“e ldstrengthsarecovariant:Thereforethe =0proj ectionsof(5.6.33,34)arethesupercovariant x -space“eldstrengths. Fo rt hegauge“eldsofinternalsymmetries,covariant“eldstrengthsarealsonecessary.These“eldstrengthsarede“nedbyexactlythesameformulaeasthecurvatures above.

PAGE 339

5.6.Fromsuperspacetocomponents325f.Supercovariantsupergravity“eldstrengths Wenowusethee xplicitsolutionofthe n = Š1 3 Bianchiidentitiestoobtainfrom (5.6.33,34)thecomponent“eldstrengths.ThesolutionoftheBianchiidentitiescontains allofthenecessaryinformationaboutthetors ionsandc urvatures.Forthetorsionsand curvatureswithatleastonelowerspinorialindex,wesubstitutefrom(5.2.81)intothe leftha ndsideof(5 .6.33). Considering“rst T a b we “nd T a b = t a b + i ( a G€Š b G€) Š i ( a€ Š b€ ) R .(5. 6.36) Thisequationiscorrectto -independentorderandthusasupercovariantgravitino“eld strength, f a b ,isde “nedby f a b = T a b | .(5. 6.37) For T a b cweuse(5. 6.33)inaslightlydierentway.Alongwiththetorsionswith atleastonelowerspi norialindex,wealsosubstitutefor T a b contheright side.This yields t a b c+ i [ a b ]€= i ( C€€G€Š C€€ G€).(5.6 .38) Nowwecantakethisresult,useittosolveforthecomponentspin-connection,andthus obtainasecondorderformalism.Beforedo ingthisiti sconvenientt oobservethat i ( CC€€G€Š C€€CG€)= Š a b c dG d,(5. 6.39) sothat a b ccanbeexpressedas a b c= (e) a b c+ i1 2 ( [ b c ]€+ [ a c ]€Š [ a b ]€) Š1 2 a b c dA d.(5. 6.40) where (e) a b cisde“nedin(5.1.19). FinallythesupercovariantizedRiemanncurvaturetensoristreatedanalogouslyto T a b .Webeginbys ubsti tutingfrom(5.2.81)forthecurvatureswithatleastonespinorialindex. R a b = r a b + R a ( b )+ i { [ a€W

PAGE 340

3265.CLASSICALN=1SUPERGRAVITYŠ1 2 a ( C( G )€+( € R ) C ( C ) )] Š ( a b ) } .(5. 6.41) However,wemustcarryoutonefurtherste pbeforewehaveanexpressionwhichcanbe evaluatedintermsofcomponent“elds.Wemusteliminate G€, € R ,and Wfrom thisexpr ession.Thiscanbedonebyconsideringthecoecientsof onbothsidesof (5.2.81).Onthele fthandsidewe“nd f a b ,wh ileontherig htha ndside W, €G€, and R appear.Wecanthereforesolveforthesequantitiesintermsof f a b whichis expressedintermsofcomponent“eldsin(5.6.36). W=1 12 f( € € ),( 5.6.42a) G b= Š1 2 [ f€ € ,€Š1 3 C f€ € ,€],(5.6 .42b) R = Š1 3 f€ € .(5. 6.42c) Theseexpressionscannowbesubstitutedinto(5.6.41)whichresultsinawellde“ned(at thecomponentlevel)supercovariantizedRiemanncurvaturetensor. Asaby-pro ductofthisprocesswehavealsoderivedthecomponentsupersymmetrytransformationlawoftheauxiliary“elds A aand S where S R | and A a G a| Thesearesupercovariantsandhencetheir supersymmetryvariationsaregivenby, QA a= iKQG a| = Š1 2 [ f€ € ,€+1 3 f€ €€]+ h c ., (5.6.43a) QS = iKQR | = Š1 3 f€ € .(5. 6.43b) g.Tensor calculus Thecomponentrulesforthemanipulationoflocallysupersymmetricquantities arecalledthetensorca lcul usforsupergravitytheories.Theserulesgiveacomponentby componentdescriptionofsupersymmetrictheories.Super“eldsontheotherhandprovideaconcise descriptionofthesetheoriesinmu chthesam ewaythat v ectorno tation pr ov idesamoreconcisedescriptionofMaxwellsequations.Super“eldscanalwaysbe reducedtotheircomponent“eldcontentinthecaseofglobalsupersymmetryandinthis sectionwediscusstheanalogousprocedureinthelocallysupersymmetriccase.

PAGE 341

5.6.Fromsuperspacetocomponents327Asanexample,letusconsiderfor N =1superg ravityalocalscalarmultiplet describedbyacovariantlychiralsuper“eld € =0.Thecomp onent“eldsofthis mult ipletarede“nedbyprojection A | | F 2 | .(5. 6.44) Thein“nitesimalsupersymmetrytransformationsof all quantitiesareobtainedbycommuta tionwith iKQ( ).Thus,usi ng(5.6 .18) QA = iKQ( ) | = Š | = Š Q= Š ( + € €) | = Š[(1 2 {, } + C2)+ €{, €}] | .(5. 6.45) At thispoint,nospeci“cchoiceofauxiliary“eldsforsupergravityhasbeenmade.The onlyconstraintsonthesupers pacetorsionsnecessaryarethosewhichfollowasconsistencyreq uirementsfortheexistenceofchiralsuper“elds,i.e.,therepresentation-preservingconstraints.UsingthesolutiontotheBianchiidentitiesforthecaseof N =1, n = Š1 3 supergravityweobtain Q= F Š i €( € ) | .(5. 6.46) Using(5.6.6),wehave Q= F Š i €( D D€A + € ).(5.6 .47) Thelastexpressionillustratestheconceptofa supercovariantderivative atthecomponentlevel.Thecombination[ D D aA + a ],whichgeneralizestheordinarycovariant deriva tive D D aA ,transfo rmswithoutatermproportionalto D D a.Thus,this combinationof“eldsiscovariantwithrespecttoal ocalcomp onentsupersymmetrytransformation.Finally,forthetransformationlawfortheauxiliary“eldwe“nd

PAGE 342

3285.CLASSICALN=1SUPERGRAVITYQF = Š ( + € €) 2 | = Š €[ i D D€Š A€Š i €F Š €€( D D€A + € )]+ S = Š €[ i D D€Š A€] + S .(5. 6.48) (Analogoustransformationsforachiralscalarmultipletcanbefoundfor n = Š1 3 by usingtheappropriatesolutiontotheBianchiidentities.)Onthesecondlineabovewe haveintroducedthenotation D D aforthesupercovariantderivativeofthespinormatter“eld. Wecanalso “ndthecomponentsoftheproductoftwodierentmultipletsin termsofthecomponentsoftheoriginalmultiplets.Thus,forexample,aproductoftwo chiralsc alarmultipletsdescribe dbychirals uper“elds 1, 2isthescalarmultiplet describedbythechiralsuper“eld 3= 12: A3= 12| = A1A2, ( 3)= ( 12) | =([ 1] 2+ 2[ 2]) | =( 1)A2+ A1( 2), F3= 2( 12) | =([ 21] 2+[ 1][ 2]+ 1[ 22]) | = F1A2+( 1)( 2)+ A1F2.(5. 6.49) Thecomponentsof3transformaccordingto(5.6.45,47,48).Thismultiplicationlawis just likeintheglobalcase(3.6.11). Anotherpossi bleproductoftwoscalarmultipletsisfoundbytakingtheproductof achirals uper“eld 1andanantichi ralsuper“eld 2;thisgivesthec omplexgeneral scalarsuper“eld= 1 2: | = A1 A2, | = 1 A2, 2 | = F1 A2,

PAGE 343

5.6.Fromsuperspacetocomponents329[ €] | =2 1 2€, ( 2+ R ) | = F21 + i 2€( D D€A1), ( 2+ R ) | = Š 1 ( i D D€+ A€) 2€+ 2€( Š i D D€+2 A€) 1 +2 F2F1Š ( D D€ A2)( D D€A1).(5.6 .50) whereweh aveused( 2+ R ) =0whichcan beobtained from(5.4 .16).Notethe appearanceofthesupercovariantderivative D D a. Wecanalsog ivethecompon entsofachiralsuper“eldmadeoutofanantichiral one 1=( 2+ R ) .Thisi ssometim escalledthekineticmultiplet.Itscomponents are: A1=( 2+ R ) | = F + S A ( 1)= Š ( i D D€+ A€) €Š1 3 f€ € A F1=(( D D a+ i 3 A a) D D aŠ1 3 R€ €€€Š 4 S S ) A Š 4 SF Š1 3 f€ € ,€ €.(5. 6.51) wherew ehavemadeuseoftheresult 2R +2 R R = Š1 6 R€ €€€.(5. 6.52) The x -spacesupercovariantcurvature R€ €€€in(5.6.51)isgivenby(5.6.41).The computationoftheseresultsisstraightforwardbuttedious.Alloftheaboveresultshave madeextensiveuseofthecommutatoralgebrain(5.2.82). Asafurther exampleofcomponenttensorcalculus,weconsiderthevectormultiplet.Thelocalcomponentsarede“nedinthesamewayasintheglobalcase(4.3.5), butwith Areplacedby A,t he su pe rgravityandYang-Millscovariantderivative.The “eldstrengthsandBianchiidentitie sforthev ectormult iplettaketheforms FYM = FYM €=0,

PAGE 344

3305.CLASSICALN=1SUPERGRAVITYFYM b= Š iC W€ YM, FYM a b= C€€f+ C f€€, €W YM= W YM+ € W€ YM=0.(5. 6.53) Thequantity fisasuper covariant“eldstrength(seebelow).Thelocalcomponentsof themultipletar ethusde “nedby v a= a| = W YM| ,D= Š i1 2 W YM| .(5. 6.54) Thesupersymmetryvariationsofthecovariantcomponents, andD,areobta inedas withthecomponentsofthechiralmultiplet(see(5.6.46)). Q = Š f+ i D, QD=1 2 ( D D€ €Š €D D€),(5.6 .55) where D D€ €isthesup ercovariantderivativeof D D€ €= D D€ €Š €€( f€€Š iC€€D).(5.6 .56) Thisfollowsfrom(5.6.13)andtheBianchiidentitiesofthevectormultiplet(4.2.90) whicharevalidinacurvedsuperspace.Thequantity f(anditsconjugate f€€)canbe calculatedinthesamewayas(5.6.41)from(5.6.34,53) f=1 2 [ fYM € €+ i ( ( € ) €)+ i ( ( € ,€ ))],(5. 6.57) where fYM a bistheordinary x -s pa ce Ya ng -M ills“eldstrength.Forthetransformation lawof v a,weuse(5.6 .28).(Eventhoughthederivationofthatresultwasforthegauge “eldsfortangentspacesymmetries,italso appliestothegauge“eldsforinternalsymmetries.) Qv a= i ( €+ €) Š i ( a€+ € a ) v€.(5. 6.58) Justasforthegravitinotansformationlawin(5.6.29b),thelasttwotermsaboveare

PAGE 345

5.6.Fromsuperspacetocomponents331absentifweconsiderthetransformationlawof v m. Wehave presentedtheaboveresultsfor N =1, n = Š1 3 supergravity;theycanbe generalizedtoall n byusingt heappropriateset ofBianchiidentities(5.4.16-17). Inourdiscussionofglobalsupermultipletswefoundalargenumberofgaugemultipletsw herethecomponent gauge“eldwasnotaspin-one“eld(forexamplethetensor mult iplet).Sincewegaveacompletelygeometri caltreatmentofthesemultipletsusing p -formswithinglobalsupersy mmetry,theirextensiontothelocallysupersymmetriccase (i.e.,transformationlaws,supercovariant“eldstrengths,etc.)isobtainedbythe straightforwardgeneralizationofthemethodswhichweusedtotreatthespin-onecase. Theonlycomplicationthatcanoccuristhattheexistenceoftheunconstrainedprepotentialmustbecon sistentwiththesetofconstraintsthatdescribethesupergravity background.Anexampleofan N =1mult ipletforwhichthesuper“eldextensionto localsupersymmetryisnotknownisthemattergravitinomultiplet.Thisisnotsurprisingsinceasecondsupersymmetry(i.e., N =2supersy mmetry)isrequiredfortheconsistencyoftheequationsofmotionforthemattergravitino. h.Componentactions Finallywegiveformulaetoobtaincomponentactionsfromthe N =1 n = Š1 3 superspaceactions S1= d4xd23ILchiral,( 5.6.59a) S2= d4xd4 EŠ 1ILgeneral.(5. 6.59b) We“rst have S1= d4x eŠ 1[ 2+ i €€+3 S +1 2 (€ |€ |€ )€] ILchiral| (5.6.60) Toderive(5.6 .60)thereareseveralsteps.Fir st,using(5.5.9)andchoosing IL = RŠ 1,we seethat 3= D2EŠ 1RŠ 1.T hisisadensityunder x -spacecoordinatetransformations. Butin x -space,adensityiseŠ 1mult ipliedpossiblybya dimensionless x -spacescalar.

PAGE 346

3325.CLASSICALN=1SUPERGRAVITYNosuchdime nsionlessscalarscanbeconstructedintheminimaltheory.Therefore 3at lowestorderin mustbepr oportionaltoeŠ 1.(Itshouldbenote dthatt hereareno exp licitfactorsof anywhereexcepttha tmulti plyingthesupergravityaction.)Thissituationisnottrueforthenonminimaltheor ies,inwhichitispo ssibletoconstructa di mensionlessscalarfromsomeoftheadditionalauxiliary“elds.Thisispreciselywhat happensfortheF-typedensityforthenonminimaltheory,andisresponsibleforthe nonpolynom ialitydiscussedinsubsec.5.5.f.3. Onceweknowth atthelowestcomponentof 3iseŠ 1,wederive(5. 6.60)bymultiplyingeŠ 1bythehighest component F ofachirals uper“eldandperformingasupersymmetrytransformation.Thisgen eratesatermproportionalto F timesthegravitino, whichwecancancelbyaddingtoeŠ 1F atermpropo rtionalto €€.Thisnewterm generatessupersymmetryvariationsproportionalto D DA timesthegravitino.Thesecan becan celedbyaddingatermproportionalto 2A tothestartingpoint.Finally,we determinethecontributionofthe SA termbycancelingvariationsproportionalto S Bydime nsionalanalysis,therecanbenoothercontributions,andwehaveobtainedthe densityformula of(5.6.60). To “ndthecorrespondingexpressionfor S2,weuse S2= d4xd23( 2+ R ) ILgeneral(5.6.61) andtheformula(5.6.60)forthechiralcase.Thecovariantderivativesactonthesuper“eldsintheL agrangianandprojectoutthecomponents. Asasimpleexample,wecomputethemasstermforachiralsuper“eld : S =1 2 m d4xd232=1 2 m d4x eŠ 1[2 FA + + i 2 €€A +(3 S +1 2 (€ |€ |€ )€) A2].(5.6 .62) Asas econdexamplewecomputethe N =1, n = Š1 3 componentsupergravityaction andcosmologicalterm.

PAGE 347

5.6.Fromsuperspacetocomponents333From(5.5 .4,34)wehave SSG=1 2 d4x eŠ 1[1 2 r€ € ,€€Š a b c d a ,€D D c d Š 3 | S |2].(5.6 .63) Thisisthecomponentformofthesupergravityactionwiththeimprovedspinconnection.Theaxialvectorauxiliary“eldispresentimplicitlyinthe“rsttermsincethespin connection,asde“nedin(5.6.40),dependson A a.Ifwesepar ateoutfrom ( e ) a b cthe contributionof A aitappears only quadraticallyintheaction.Inparticular,thereisa cancellationamongtermsoftheform A a b d whichcomefromthe“rsttwotermsin theaction. Forthecosmol ogicaltermfrom(5.5.42)andusing(5.6.60),wehave Scosmo= Š 2 d4xd23+ h c = Š 2 d4x eŠ 1[3 S +1 2 (€ |€ |€ )€+ h c .](5.6 .64) Thecosmologicaltermcontainsatthecomp onentlevelanapparentmasstermforthe gravitino.However,inthedeSitterbackgroundgeometrythegravitinoisactuallymassless,sinceitisstillagauge“eld. Inclosingwemaketwoobservations:Although(5.6.61)wascomputedafterthe constraintswereimposedonthecovariantderivatives,inprincipleonecancomputesuch anactionformulaw itho utimposing any constraintsa ta ll.Thisfollowsbecausethe transformationlawsforthecomponentsofthetotallyunconstrainedsuperspaceare di rectlyobtainablefrom(5.6.28)and,formattermultiplets,fromequationsanalogousto (5.6.45-48).Alargenumberofauxiliary“eldsde“nedasthe =0valueoft hevarious su pe rspacetorsionswillentersuchaconstruction.Amongtheseoccursanauxiliary“eld whichisaLagrangemultiplierthatmultiplieseŠ 1.(Thevariati onofthisLagrangemultiplierwillconstrainthegeometryof x -space.)Clearlythisisunacceptable,andwehave seenhow,for N =1 supergravity,thiscanbeavoided.However,understandingtherole of such“eldsmaybenecessarytounderstand N > 4o-shellt heories. Thesecondpointisthatwelackatpresentadirectmethodforcomputingdensity fo rmulaeanalogousto(5.6.60).Wecanalwayscomputesuchaformulabyhand:We startwitheŠ 12 N(whereisan arbitrarysuper “eld)andperformsupersymmetry

PAGE 348

3345.CLASSICALN=1SUPERGRAVITYvariationsto obtainanentiredensitym ultiplet.Whatislackingisawaytoobtainthis resultwithoutlaboriouscalculation.

PAGE 349

5.7.DeSittersupersymmetry3355.7.DeSitters upe rsymmetry Insec.3.2.f,wediscussedthesuper-deSitteralgebra(3.2.14).Herewedescribe howsupersymmetricdeSittercovariantderivativescanbeobtainedfromsupergravity covariantderivatives.We“rstdiscussthenonsupersymmetricanalog.NonsupersymmetricdeSittercovariantderivativescanbeobtainedfromgravitationalcovariantderivativesbyeliminatingall“eldcomponentsexceptthe(density)compensating“eld(i.e., thedeterminantofthemetricorvierbein).T hisfo llowsfromthefactthatindeSitter spacetheWeyltensorvanishes,whichsaysthatthereisnoconformal(spin2)partto themetric:Itisconformally”at.Ontheotherhand,thescalarcurvaturetensoris anon zeroconstant r =2 2(thisisthegravity“eldequation). Wecanwrite e a m= Š 1 a mwhere isthecompensatorof(5.1.33).Aftersetting theothercomponentstozero,theactionfordeSittergravity(Poincar epluscosmo logical term)isjusttheactionforama sslessscalar“eldwithaquarticself-interactionterm. (Therestofgravity,theconformalpart,issimplythelocallyconformalcouplingofgravitytothisscalar.)T heequationofmotioncorrespondingtothecovariantequation r =2 2, =2 23,(5. 7.1) hasthesolution,withappropriateboundaryconditions, Š 1=1 Š 2x2.(5. 7.2) ThedeSittercovariantderivativesarenowobtainedfromthegravitycovariantderivativesofsec.5.1bysubstituting e a m= Š 1 a m,with Š 1givenby(5.7.2). Inthesupersymmetriccase,westartwiththesupergravityactionandacosmologicalterm(5.5.16).Weset H tozero,andsolveforthechiraldensitycompensator :In super-deSitterspace Wvanishes(asdoes G a),while R = TheactionforthecompensatoristhemasslessWess-Zuminoaction(againaconformalaction,whosesuperconformalcouplingto H givesthe deSittersupergravity action).The“eldequationsinthechiralrepresentation D2 = 2(5.7.3) havethesolution

PAGE 350

3365.CLASSICALN=1SUPERGRAVITYŠ 1=1 Š x2+ 2.(5. 7.4) The(realpartofthe) =0compon entof isthusthegravitycompensatorof(5.7.1,2). Thesuper-deSittercovariantderivativesareobtainedbysubstitutingthissolutionfor (with H =0)intotheex pressionsf orthesupergravitycovariantderivativesgiveninsec. 5.2. Theprecedingdiscussioninvolvedthe n = Š1 3 compensator .For other n ,we “ndstrangepathologies:deSitte rspace cannotbedescribedfor n = Š1 3 inaglobally (deSitter)supersymmetricway.For n = Š1 3 ,emptyde Sitterspaceisdescribedby R = G a= W=0, butfornonminimal n wewouldrequire G a= W=0with T .Thisfo llowsfromthefactthatthecommutatorsofcovariantderivativesmust takethefollowingformtode scribedeSittersuperspace {, } = Š 2 M, {, €} = i €,[ b]= Š i C €, [ a, b]=2 ( C M€€+ C€€M).(5.7 .5) Thisrequires spontaneousbreakdown of N =1supers ymmetry,since Tisatensor: T| =0wouldimply T=0ifgl obal(deSitterorother)supersymmetryweremaintained.( Tmustbeno nzerofor R tobeno nzerointhenonminimaltheory.See (5.2.80b)).For n =0, G a= W=0alre adyimpliesMinkowskispace.

PAGE 351

Contentsof 6.QUANTUMGLOBALSUPERFIELDS 6.1.Introductiontosupergraphs337 6.2.Gauge“xingandghosts340 a.OrdinaryYang-Millstheory340 b. Sup er sy mmetricYang-Millstheory343 c.Othergaugemultiplets346 6.3.Supergraphrules348 a.DerivationofFeynmanrules348 b.Asamplecal culation353 c.Theeectiveaction357 d.Divergences358 e.D-algebra360 6.4.Examples364 6.5.Thebackground“eldmethod373 a.OrdinaryYang-Mills373 b. Sup er sy mmetricYang-Mills377 c.CovariantFeynmanrules382 d.Exam ples 389 6.6.Regularization393 a.General393 b.Dimensionalreduction394 c.Othermethods398 6.7.AnomaliesinYang-Millscurrents401

PAGE 353

6.QUANTUMGLOBALSUPERFIELDS 6.1.Introductiontosupergraphs Aswehaves eeninpreviouschapters,atthecomponentlevelsupersymmetric modelsaredescribedbyordinary“eldtheoryLagrangians,andtheirquantizationand renormalizationusesconventionalmethods.Evidentlythequantumtheoryshouldbe renormalizedinamannerthatpreservessupe rsymmetry.Unlessamanifestlysupersymmetricregularizationmethodisused,thisrequiresapplyingtheWard-Takahashiidentitiesofsupersymmetryateachorderofperturbationtheory. Supersymmetricmodelsareingenerallessdivergentthannaivecomponentpower countingindicates,andthiscanbetracedtotheequalityofnumbersofbosonicand fermionicde greesoffreedom,togetherwithrelationsbetweencouplingconstantsthat areimposedbysupersymmetry.We“ndthatthevacuumenergy(or,when(super)gravit yi sp resent,thecosmologicalterm)receivesnoradiativecorrections,andthat,in renormalizablemodels,acommo nwave-f unctionrenormalizationc onstantissucientto renormalizetermsinvolvingonlyscalarmultiplet“elds(the norenormalization theorem).Arelatedresultisatheoremthatifthe classicalpotentialhasasupersymmetric minimum(nospontaneoussupersymmetrybreaking),sodoestheeectivepotentialto allordersofperturbationtheory(noCole man-Weinbergmechanism:seesec.8.3.b). Improvedconvergenceduetosupersymmetryisalsoevidentinsupergravity.For all N ,the S-matrixof(extended)supergravityis“niteatthe“rsttwoloops;weargue insec.7.7thatitisalso“niteatlessthan N Š 1loops.Ins uitablesupersymmetric gaugesthis“nite nessalsoholdsfortheo-shellGreenfunctions. Insupersymmetrictheoriestheone-loo psuper conformalanomalies(traceofthe energy-momentumtensor, -traceofthecomponentsupersymmetrycurrent,andthe divergenceoftheaxialcurrent)formasupers ymmetricmultiplet,thesupertrace,so thattheircoecientsareequal.Thereexistotheranomaliesaswell.Weshowinsec. 7.10thatinnonminimal N =1superg ravity( n = Š1 3 ),anomaliesmaybepresentinthe Wardidenti tiesoflo calsupersymmetry.Thus,ingeneral,onlyminimal N =1supergravityisconsistentatthequantumlevel( butextendedtheoriesthathavenonminimal N =1superg ravityasasubmultipletareconsistentbecauseofanomalycancellation

PAGE 354

3386.QUANTUMGLOBALSUPERFIELDSmechanisms). Super“eldsgreatlysi mplifyclassicalcalculations:Supersymmetricactionscanbe easilyconstructed,andthetensorcalculusofsupersymmetrybecomestrivial.However, thegreatestadvantagesofsup er“eldsappearatthequantumlevel.Therearealgebraic simpli“cationsinsupersymmetricFeynmang raph(supergraph)calculationsfora numberofre asons:(1)compactnessofnotation,(2)decreaseinthenumberofindices (e.g.,thevector“eld A aishi ddeninsidethescalarsuper“eld V ),and(3)automaticcancellationofcomponentgraphsrelatedbysup ersymmetry(whichwou ldrequireseparate calculationincomponentformulations).Furthermore,theuseofsuper“eldsleadsto power-counti ngruleswhichexplainmanycomponentresultsandcanbeusedtoderive additional“nitenesspredictions,especiallywhencombinedwithsupersymmetricbackgr o und-“eldmethods. Renormalizat ionismuchsimplerinthesuper“eldformalism.Supersymmetryis manifestand,aswediscusslater, any regularizationmethodthatpreservestranslational invarianceinsuperspacewillmaintainit.Forgaugetheorieswecanusesupersymmetric gauge-“xingterms.BycontrastcomponentWe ss-Zuminogaugecalcul ationsexplicitly breaksupersymmetryandhavethedisadvantagethattheWard-Takahashiidentitiesfor globalsupersymmetrycannotbedirectlyappliedduetotheirnonlinearity. Inthischapterandthenextonewediscussthequantizationof N =1super“eld theori es.Weconsiderclassicalsuper“eldactions S ()and usefunctionalmethodsto constructthegeneratingfunctional Z ( J )andthee ectiveaction().Ifisagauge “eldwequantizecovariantl y,intro ducinggau ge-“xingterms,gaugeaveraging,and super“eldFaddeev-Popovghosts.WethenderiveFeynmanrulesforsupergraphsusing superspacepropagators( x x, ).Themethodsarecompletelyanalogoustothose forcomponent“elds,butsomenewfeaturesarepresent:Wemustdealwithconstrained (chiral)super“elds,a ndweencounternotonly i a= p aoperators,butalsospinor deriva tives Dactingontheargumentsofpropagatorsorexternallines.Weshowhow theseoperatorsaremanipulatedandhow,foranygraph,the -integralsateachvertex canbedone,leavingu swithoneoverall -integralforthewholegraph(theeective actionislocalin ),andordinaryloop-momentumintegrals.Atallstepsofthecalculationsmanifestsupersymmetryismaintained.

PAGE 355

6.1.Introductiontosupergraphs339We di sc us sn ex tt he ba ck ground“eldmethodforsupersymmetricYang-Millstheories.Thisissimilartothatforcomponenttheories,withonesigni“cantdierence:The quantum-backgroundsp littingisnonlinear,re”ectingthenonlinearitiesofthegauge transformationsofthesuper“eld V .Themethod simpli“esma nycalculat ionsandcan beusedtost udyhigher-loop“nitenessquestions. Forsupergr a phsthesimplestregularizationprocedureistousedimensionalregularizationofmomentumintegrals after thecontributionfromagraphhasbeenreduced toasingle integral.Theresultingeectiveaction,whichisa(localin )f unction alof theexternals uper“elds,ismanifestlysupersymme tric.How ever,thisregularization methodcorrespondsto(component)regular izationbydimensionalreduction,whichis knowntobeinconsi stent.Thesuper“eldresults,alth oughsupersymmetric,mayre”ect thisinconsistencybyexhibitingambiguitiesassociatedwiththeorderinwhichsomeof the -integrationshavebeencarriedout.Wealsodiscussalternativeregularizationprocedures.Besidesgivingpowercountingruleswedonotdiscussthedetailsoftherenormalizationofsuper“eldtheories.WeworkwithWick-rotatedtimecoordinates: d4x id4x ,so eŠ iS eS.(Themetric a bhassignature( Š + ++) (++ ++),so + ,etc.Notethatino urconventions, i isoppositeinsignfromusualconventions:Thuspositive-energystatesaredescribedby ei tandpropagatorsare ( p2+ m2+ i )Š 1.Wefurtherwarn thereaderthatthegaugecouplingconstant g is 2 timesthe usual g (seepage55).)

PAGE 356

3406.QUANTUMGLOBALSUPERFIELDS6.2.Gauge“xi ngandghosts Thequantizationofsupersymmetricgaugetheoriesissimilartothatofordinary gaugetheories.Therearetworelatedaspectsofthesituation:(a)Theactionisinvarian t undergaugetransformationsandthereforethefunctionalintegrationshouldbe restrictedtothesubsetofgaugeinequivalent“elds.(b)Thekineticoperatorisnot invertibleoverthespaceofall“eldcon“gurationssothatthepropagator,neededfor doingperturbationtheory,cannotbede“nedunlessthesetof“eldsisrestricted.In componentgaugetheories,imposinganalgeb raicrestrictionexplicitlyinthefunctional integralleadstoanaxialgaugewhichbreaksm anifestLorentzinvariance.Alternatively, wecanquantizecovar iantlyusingtheFaddeev-Popov procedure:Weintroducegauge “xingfunction(s),weightedgaugesandFadd eev-Popovghosts.Insupersymmetricgauge theoriestheanalogoftheaxialgaugeistheWess-Zuminogauge.Inthisgauge,quantizationbreaksmanifestsupersymmetry.Inco ntrast,covariantsuper“eldquantization maintainsmanifestsupersymmetry. a.OrdinaryYang-Millstheory Fo ro rientationwebrie”yrecallthequantizationmethodforordinaryYang-Mills theory.TheYang-Millsgaugeactionis SYM=1 g2 tr d4x [Š 1 8 f a bf a b], f a b= [ aA b ]Š i [ A a, A b],(6.2 .1) withgaugeinvarianceunderthetransformation A a A a = ei [ A a+ i a] eŠ i ,(6. 2.2a) or ,i n“nitesimally, A a= a = a + i [ A a].(6.2 .2b) Here isanelementofthegaugealgebra.Both A and arematricesintheadjoint representation.Weobservethatthekinetic(quadratic)partoftheLagrangiancanbe written(afterrescaling A gA )intheform1 2 A TA where(T) a b= a bŠ1 2 a b Š 1isatransverseprojectionoperator(see(3.11.2)). Westartwiththe normalizedfuncti onalintegral

PAGE 357

6.2.Gauge“xingandghosts341Z = N IDA aeSinv,(6. 2.3) wherewehaveincludedin Sinvpo ssibletermswithsourcescoupledtogaugeinvariant operators.Wede“nethegaugeinvariantintegraloverthegroupmanifold F( A a)= ID [ F ( A a ) Š f ( x )],(6.2.4) where f ( x )isanarbit rary“eld-independentfunction,and F isagauge-variantfunction suchthat F = f forsomev alueof .Itisimportant toverifythatthisisthecase.We introduceafactorof1inthefunctionalintegral,intheformF Š 1F: Z = N IDA aF Š 1( A a) ID [ F ( A a ) Š f ]eSinv= N IDA aF Š 1( A a) ID [ F ( A a) Š f ]eSinv,(6. 2.5) wherethelastformfollowsfromachangeofvariablesthatisagaugetransformation, andthegaugeinvarianceofFand S .The in te gr al nowgivesaconstantin“nitefactorthatwea bsorbintothenormalization N,leadi ngto theform Z = N IDA aF Š 1( A a) [ F ( A a) Š f ]eSinv.(6. 2.6) Byconstr uction Z isindepe ndentof F and f ,and hencewecan averageover f withan arbitrary(normalized)weightingfactor.Inparticular,ifweintroduceafactor 1= NIDfexp(Š1 g2 trd4xf2),the ( F ( A ) Š f )factorc anbeusedtocarryout theintegrationandleadstotheform Z = N IDA aF Š 1eSinv+ SGF, SGF= Š1 g2 tr d4x [ F ( A a)]2.(6. 2.7) wherew ehaveabsorbed Ninto N. Wecanpar ametrizethegaugegroupbyagaugeparameter ( x )sucht hat F ( Aa )= f ( x )for =0.Then

PAGE 358

3426.QUANTUMGLOBALSUPERFIELDSF( A a)= ID [ F ( A a ) Š f ]= ID [ F ]Š 1 ( ) = ID [ F ]= ID ID e F dx,(6. 2.8) wherew ehavewrittenaninteg ralrepresentationforthefunctional -function.Inthe secondlineof(6.2.8),and intheequationsbelow, F isevaluatedat =0.Toobtain F Š 1wereplace and byreal anticommuting(Faddeev-Popovghost)“elds c ( x )and c( x )(sees ec.3.7).Finally,wecanchooseforthegauge“xingfunctiontheform F ( Aa)=1 2 aA a.Then F =1 2 a a ,andwe have Z = N IDA aIDcIDceSeff, Seff=1 g2 tr d4x [ Linv( A a) Š1 F ( A )2+ ic F c ] =1 g2 tr d4x [ Linv( A a) Š1 4 ( aA a)2+ ic1 2 a ac ].(6.2 .9) (The i isforhermiticity.)Thegauge-“ xingtermcanbew ritteni ntheform1 2 A LA whereL=1 Š Tisthelongitudi nalprojectionoperator(L) a b=1 2 a b Š 1.The totalkineticop eratorbecomes (1+(1 Š 1)L),whichisinvertible:Minusitsinverse (thepropagator)is Š Š 1(1+( Š 1)L).IntheFermi-Feynmangauge, =1,the pr opagatoris Š Š 1. Thegauge-“xedLagrangian,includingghosts,isinvariantunderthe global BRST transformations: A a= i ac c=2 F =1 aA a, c = Š c2,(6. 2.10) withconstantGrassmannparameter .Thesetran sformationsarenilpotent: 2onany

PAGE 359

6.2.Gauge“xingandghosts343“eldvanishes(whentheantighostequatio nsofmoti onareimposed).TheWardidentitiesforthisglobalinvariancearetheSlavnov-Tayloridentitiesofthegaugetheory. b.SupersymmetricYang-Millstheory Fo rs up er sy mmetricYang-Millstheorywequantizefollowingthesameprocedure. Westartwiththef unction alintegralforagaugerealscalarsuper“eld V = VATA,where TAarethegeneratorsofthegaugegroup: Z = IDVeSinv( V ).(6. 2.11) Notethatinsupersymmetrictheoriesthenormalizationfactorof(6.2.3) N=1(see sec. 3.8.b).Theactionis Sinv=1 g2 tr d4xd2 W2= Š1 2 g2 tr d4xd4 ( eŠ VDeV) D2( eŠ VDeV) =1 2 g2 tr d4xd4 [ VD D2DV + higher Š orderterms ].(6.2 .12) Itisinvariantunderthegaugetransformations eV= ei eVeŠ i ,( 6.2.13a) or ,f or i n“nitesimal(see(4.2.28)), V = L1 2 V[ Š i ( +)+ cothL1 2 Vi ( Š )], LXY =[ X Y ].(6.2 .13b) Intheabeliancase,thisis V = i ( Š ).Thekineticoperatoris 1 2 withthesuperspin1 2 projectionoperator1 2 = Š Š 1D D2D,andis notinvertiblebecauseitannihilatesthechiralandantichiralsuperspinzeropartsof V : V0=0V = Š 1( D2 D2+ D2D2) V Wemustnowc hoose gauge-“xingfunctions.Correspondingtothechiralgauge parameterweneeda gauge-variantfunctionthatcanbemadetovanishbyasuitable gaugetransformation.Thereforeitmusthavethesamespinandsuperspinasthegauge

PAGE 360

3446.QUANTUMGLOBALSUPERFIELDSparameterandhenceshouldbechosenachiralscalar.Thegauge-variantquantity F = D2V isasuitablegauge-“ xingf unction.Foranychiralfunction f ( x ),weverify thatgaugetransformationscanbefoundtomake F = f .Forex ample,intheabelian case,underagaugetransformation F ( V ) F ( V)= D2V + i D2 ;ifwechoose i = Š 1D2( f Š D2V ),we “nd F= f Wede “nethefunctionaldeterminant ( V )= ID ID [ F ( V ,, ) Š f ] [ F ( V ,, ) Š f ].(6.2 .14) We“rstwrite( cf.(6.2.5)) Z = IDV Š 1( V ) [ D2V Š f ] [ D2V Š f ]eSinv.(6. 2.15) Asin(6.2.7),weaverageover f and f withaweightingfactorIDfID fexp ( Š1 g2 tr d4xd4 ff ),andobtaintheform Z = IDV Š 1( V )eSinv+ SGF,(6. 2.16) where SGF= Š1 g2 tr d4xd4 ( D2V )( D2V ).(6.2 .17) Wewrite ( V )= ID ID ID ID ed4xd2 F + F +d4xd2 F + F (6.2.18) wherewehavereplacedthe -functionsinvolving(anti)chiralquantitiesbyintegralrepresentationsinvolving(anti)chiralparamet ersandintegrationmeasures.Thevariational derivati vesof F F ,areevaluatedat= =0.Inth ef unction alintegral,where Š 1( V )appears,wer eplacetheparameters,byanticommutingc hiralghost“elds c c.Fina lly,we“nd Z = IDVIDcIDcID cID ceSinv+ SGF+ SFP,(6. 2.19) where

PAGE 361

6.2.Gauge“xingandghosts345SFP= itr d4xd2 c D2( V )+ itr d4xd2 cD2( V ) = tr d4xd4 ( c+ c) L1 2 V[( c + c )+ cothL1 2 V( c Š c )].(6.2.20) Integratingbyparts,wecanwrite( D2V )( D2V )=1 2 V ( D2 D2+ D2D2) V =1 2 V 0V Thequadraticpartofthegauge“eldactionhasnowtheform Š1 2 V (1 2 + Š 10) V = Š1 2 V [1+( Š 1Š 1)0] V ,(6. 2.21) andtheoperatorisinvertible.Toavoid Š 2termsinthepropagatorandthusbad infraredbehavior,wechoosethesupersymmetricFermi-Feynmangauge =1,which leadstoasimple Š 1pr opagator.(The Š signin(6 .2.21)leadstotheusualkinetic termforthecomponentgauge“eld: Š d4 V V A a A a.) Thequadraticpartoftheghostactionhastheform S(2) FP= tr d4xd4 ( c+ c)( c Š c )= tr d4xd4 ( cc Š c c ).(6.2 .22) Thechiralandantichiral cc and c c termsvanishwhenintegratedwith d4 andhave b eendropped.(Suchtermscannotbedroppedinthepresenceofsupergravity“elds: see,forexample,(5.5.16)). Thetotalactionisinvariantundersuper“eldBRSTtransformations.Thesetake theform V = V |= i c= LV[( c + c )+ cothL1 2 V( c Š c )], c=1 D2 F =1 D2D2V c=1 D2F =1 D2 D2V c = Š c2, c = Š c2,(6. 2.23) andtheinvariancecanbeusedtoderivetheSlavnov-Tayloridentitiesofthetheory. Beforeperformingperturba tionexpa nsions,werescale V gV .Thenall quadratictermsare O ( g0),c ubictermsare O ( g ),et c.Werescaleback gV V inthe eectiveaction.Alternatively,wesimplyprovideeachgraphwithafactor( g2)L Š 1,

PAGE 362

3466.QUANTUMGLOBALSUPERFIELDSwhere L isthenumbe rofloops. c.Othergaugemultiplets Wegivetwoother examplesofthegauge-“xingprocedure:Forachiralsuper“eld ,thesolutionofthechiralityconstraint,= D2,givesthekineticaction Sinv= d4xd4 D2 D2,(6.2 .24) andintroducesthegaugeinvariance = D€ €, = D,(6. 2.25) foranarbitraryspinorparameter (see(4.5.1-4)).Suitablegauge“xingfunctionsare thelinears pinorsuper“elds F= D, F€= D€ .(6.2 .26) Toobtainac onvenientgauge“xingtermweaveragewith f€M€f,where M€= D D€+3 4 D€D.(6. 2.27) Thisleadsto Sinv+ SGF= d4xd4 ,(6.2 .28) andastandard pŠ 2pr opagator. As econdexampleisfortheactionofthechiralspinorsuper“eldthatdescribes thetensormultiplet(4.4.46): Sinv= Š1 2 d4xd4 G2= Š1 8 d4xd4 ( D+ D€ €)2,(6. 2.29) withgaugeinvarianceunder = i D2DK K = K .(6. 2.30) Asuita blegauge“xingfunctionis F = Š i1 2 ( DŠ D€ €),(6.2 .31)

PAGE 363

6.2.Gauge“xingandghosts347where F islinear.Thegauge-“xedaction Sinv+ SGF=1 2 d4xd4 [ Š G2+1 F2] = Š1 4 d4xd2 [1 2 (1+ K K )+1 1 2 (1 Š K K )]+ h c =1 2 d4xd4 [ Š1 2 (1+ )(1 2 D2+ h c .)+1 2 (1 Š ) €i €] (6.2.32) (with K K asinsec3.11)takestwoconvenientforms: For =1, Sinv+ SGF= Š1 4 d4xd4 (D2+ € D2 €);(6.2 .33) for = Š 1, Sinv+ SGF=1 2 d4xd4 €i €(6.2.34) (cf.(3.8.36)).

PAGE 364

3486.QUANTUMGLOBALSUPERFIELDS6.3.Supergraphrules Givenanaction S (),wede “nethegeneratingfunctionalforGreenfunctions Z ( J )= ID eS ()+ J ,(6. 3.1) where J isasourceofthesametypeasthe“eld (gen eralifisgeneral,chiralifis chiral,etc .).Thege neratingfunctionalof connected Greenfunctionsis W ( J )= lnZ ( J ).(6.3 .2) Theexpectationvalueofthe“eldortheclassical“eld inthe presenceofthe sourceis ( J )= W J .(6. 3.3) Thisrelationcanbeinvertedtogive J ( ).Theeectiveaction( ),thegenerating functionalofoneparticleirreduciblegraphs,isde“nedbyafunctionalLegendretransform ( )= W [ J ( )] Š J ( ) .(6.3 .4) InthissectionwederivetheFeynmanrulesforthepertubativeexpansionofthe eectiveaction.ThederivationoftheFeynmanrulesforunconstrainedsuper“eldspresentsfewsurprises.Insteadofhaving d4x integralswehave d4xd4 integrals.Propagatorsareobtainedfro mtheinversesofthek ineticoperators,an dverti cescanberead directlyfromtheinteractionterms.However,forchiralsuper“eldstheFeynmanrules re”ectthechiralityconstraints. a.Derivati onofFeynmanrules Webeginbyderiv ingtherulesfortherealscalargaugesuper“eld.Thegauge “xedacti on(intheFermi-Feynmangauge, =1)reads SV= tr d4xd4 [ Š1 2 V V +1 2 [ V ,( DV )]( D2DV )+ ].(6.3 .5) TheFeynmanrulescanbereaddirectlyfromthisexpression:Thepropagatorisminus

PAGE 365

6.3.Supergraphrules349theinverseofthekineticoperator, Š 14( x Š x) 4( Š )or,in momentumspace, Š pŠ 24( Š ).Si ncethespinorderivatives D containe xplicit s,wedo not Fourier transformwithrespecttothe variables.(Ifone doesFouriertransformwithrespectto ,thereis littlecha ngeintheFeynmanrules.)Vertices canbereadfromtheinteraction terms.Thus,thecubicterm1 2 tr [ V ,( DV )]( D2DV )leadstoat hree-pointvertexwith factorsof Dand D2Dactingontwoofthelin es,andagrouptheoryfactor.Inadditionweintegrateover x sand sateachvertexor,equival ently,overloopmomentaand over sateach vertex. Theserulescanalsobeobtainedbystartingwiththefunctionalintegral: Z ( J )= IDVe[ Š1 2 V V + ILint( V )+ JV ]=eILint J IDVe[ Š1 2 V V + JV ]=eILint J e1 2 J Š 1J,(6. 3.6) whereinthelaststepwehaveperformedtheGaussianintegralover V .TheFeynman rulescanbeobtainedusing J ( x ) J ( x, ) = 4( x Š x) 4( Š )andex pandingtheexponentialsinpowerse ries.Thus,weobtainfactorsof ILint( J )co rrespondingtovertices,and the J operators,whenactingonthefactorsof J Š 1J removethe J sandp roduce pr opagators Š 1connectingthevertices.Theresultisexactlyasforordinary“eldtheory,withtheadditionalfeatureof d4 integralsateachvertex,andadditional 4( Š ) fa ctorsineachpropagator. Chiralscalarsuper“eldsusuallyhavea kinetic action(withchiralsources j j )of theform S(2)= d4xd4 Š1 2 d4xd2 m 2Š1 2 d4xd2 m 2+ d4xd2 j + d4xd2 j .(6.3 .7) ToperformtheGa ussianinte gration,werewritechiralintegralsasintegralsoverfull

PAGE 366

3506.QUANTUMGLOBALSUPERFIELDSsuperspace.Achiralintegral Ic= d4xd2 FG ,(6. 3.8) where F and G are arbitrary chiralexpr essions,canberewrittenas Ic= d4xd4 F Š 1D2G ,(6. 3.9) using Š 1 D2D2G = G (3.4.10)and d4xd4 = d4xd2 D2.TheGa ussianin tegral canberewrittenaseW0( j ) ID ID exp d4xd4 [1 2 ( )O O +( ) Š 1D2j Š 1 D2 j ](6. 3.10) where O O = Š mD2 1 1 Š m D2 .(6. 3.11) Theinverseof O O is O OŠ 1= m D2 Š m2 1+ m2D2 D2 ( Š m2) 1+ m2 D2D2 ( Š m2) mD2 Š m2 .(6. 3.12) Perfo rmingtheintegralweobtain W0( j )= d4xd4 [ Š j 1 Š m2 j Š1 2 ( j mD2 ( Š m2) j + h c .)].(6 .3.13) Forageneralinter actionLagrangian ILint(, )wecanwrite Z ( j )=ed4xd4 ILint j j eW0( j ),(6. 3.14) andtheFeynma nrules canbeobtainedfromthisexpression.Since j ( x ) j ( x, ) = D24( Š ) 4( x Š x)(3. 8.10),thereisanoperator D2actingoneachchiral

PAGE 367

6.3.Supergraphrules351“eldlineleavingavertex.Sim ilarly,thereisanoperator D2actingoneachantichiral lineleavingav ertex.However,atapurelychiralvertex,e.g.d2 n,weuseoneof thesefactorstoconvertthe d2 integraltoa d4 integral.Thereforeatsuchverticeswe omitonefactorof D2. Wenows ummarizetheFeynmanrulesforintera ctinggaugeandchiralsuper“elds. (a)Propagators: VV : Š 1 p2 4( Š ), (6.3.15a) : 1 p2+ m2 4( Š ),(6.3 .15b) : Š mD2p2( p2+ m2) 4( Š ),(6.3 .15c) : Š m D2p2( p2+ m2) 4( Š ).(6.3 .15d) Inthemassivecase,the pŠ 2factorsint heand p ropagatorsarealwayscanceled bynumeratorfacto rs(e.g.,fortheverticesgive D2factorsand,aswediscusslater, weobtain D2D2 D2= Š p2 D2).Inthemasslesscasethesepropagatorsareabsent. (b)Vertices:ThesearereaddirectlyfromtheinteractionLagrangian,withthe additionalfeaturethatforeachchiralorantichirallineleavingavertexthereisafactor D2or D2actingonthecorrespondingpropagator,andtherulethatatpurelychiralor antichiralverticesweomitone D2or D2factorfrom amongtheonesactingonthepropagators. (c)Weintegrateover d4 ateachvertex,andinmomentumspacewehaveloopmomentumintegralsd4p (2 )Š 4foreachloop, andanoverallfactor(2 )4 ( kext). (d)Toobtaint heeectiveaction,wecomputeone-particle-irreduciblegraphs. Foreachexterna l linewithoutgoi ngmomentum ki,wemulti plyb yafactord4ki(2 )Š 4( ki)wheresta ndsforanyofthe“eldsintheeectiveaction.Foreach

PAGE 368

3526.QUANTUMGLOBALSUPERFIELDSexternalchiralorantichiralline,wehaveaor facto r,butno D2or D2factors. (e)Finally,theremaybesymmetryfactorsassociatedwithcertaingraphs. AnalternativederivationoftheFeyn manrulesforchiralsuper“eldscanbe obtainedbysolvingthechiralityconstraintsintermsofanunconstrained“eld(seesec. 4.5a): = D2, = D2 ,(6.3 .16) whereisageneral,complexscalarsuper“eld.Theaction,includingsourceterms, b ecomes S = d4xd4 [( D2 )( D2)+ ILint( D2, D2 )] + d4xd2 ( D2)( Š1 2 m D2+ j )+ h c ..(6.3 .17) Chiralintegralscanberewrittenasfullintegralsbyusingupa D2factor.Werecall thatintermsofwehaveanabeliangaugeinvariance, + D€ €(4.5.4).Consequently,thekineticoperatorappearingintheaction, D2 D2,isnotinvertible.Asdiscussedinsec.6.2,wecan“xthegaugeandarriveataninvertiblequadraticaction(the ghostsdecouple) S(2)= d4xd41 2 Š mD2 Š m D2 .(6. 3.18) TheFeynmanrulesarenowthenaiveonesandareidenticaltotheoneswehave obtainedbefore(afterusingthe D2, D2fa ctorsattheverticestosimplifythepropagators,obtainedfrom(6.3.12)).Inparticular,from ILint( D2, D2 ) ,w e again“ndfactors of D2, D2actingonthepropagators,excepttha to ne suchfactorismissingatpurely (anti)chiralvertices,since weconv erteverywheretofull d4 integrals. Itissimpletoobtainthesupergraphrulesforthetensormultiplet,withgaugeinvariantaction S = d4xd4 f ( G ), f ( G )= Š1 2 G2+ ... ,(6. 3.19) withpropagator Š 2 pŠ 4 D24( Š )(andthe hermitianco njugat efor )from

PAGE 369

6.3.Supergraphrules353(6.2.33).However,thereisamuchsimpler formoftheruleswhichresemblestherules forthescal armultiplet(towhichthetensormultipletison-shellequivalentbyaduality transformation:seesec.4.4.c.2).We“rs tnotethatthevertex ateitherendofa pr opagatorhasa D atthevertex(from f ( G )with G = D + D ),andnexttoita D2(asoccursattheendofanychiralpropagator,rule(b)above;thiskillsthe D partof thevertex).Alsonotethatt hespinorindexatthevertexcontractsdirectlywiththe correspondingspinorindexofthepropagator(becausethevertexisafunctionofonly G =1 2 D+ h c .).Contractingtheses pinorindices,andintegratingbypartsall D s fr omtheverticesontothepropagators,weobtainthesameexpressionfortheand p ropagators(withthesamevertices),whichcannowbeaddedtogether(i.e.,the totalcontributionfromgraphswithbothtypesofpropagatorsisthesameasthatfrom onlyonetype,butwithanoverallfactorof2foreachpropagator).Therulesarethus castintothefollowingform:Allverticesarenowsimply consta nts, readfromtheexpansionof f ( G )in G .T he reisonlyonetypeofpropagator,withnospinorindices,whichis Š 1 p2 D D2D4( Š )= Š 1 2 4( Š ).(6.3 .20) (Thealgebrafromthevariouscontributingfactorsis D D2D2 D2D = D D2D .)Each externallinegetsafactorof G .Ifweweretop erformthesamerearrangementofvertex factorsforthesupergraphsofthedualscalar-multiplettheory,wewouldobtain0insteadof1 2 ,theexternal linefactorswouldbe+ ,andtheconstantsatthevertices wouldbeo btainedfrom f(+ )intermsofthefunction fdualto f (seeagainsec. 4.4.c.2).Theon-shellequivalencethenfollowsfromthefactthatthecombinatorics resultingfromusing1 2 =1 Š 0,w ithapropagator1collapsingtoapoint(in andx ), perfo rmstheduality,wherefortheexternallines G =+ ons he ll. b.Asamplecalculation Wenowgiveanexamp leinatheoryofamasslesschi ralsuper“eldinteracting withagaugesuper“eld V .Wecom putetheone-loopcontributionfromthechiralsuper“eldto the V two-pointf unction.Therelevantinteractionisobtainedfrom eV= + V + ... .We “ndacontributiontotheeectiveaction,accordingto ourrulesandFig.6.3.1,

PAGE 370

3546.QUANTUMGLOBALSUPERFIELDS p k + p V ( Š k 2) V ( k 1) Fig.6.3.11 2 d4k (2 )4 d41d42V ( Š k 2) V (+ k 1) d4p (2 )4 D1 24( 1Š 2) D 2 2p2 D2 24( 2Š 1) D 1 2( p + k )2 .(6. 3.21) Notethatintheaboveexpression D1 = 1 +1 2 1€p€, D1€= 1€ Š1 2 1 ( k + p )€, D2 = 2 +1 2 2€( k + p )€, D2€= 2€ Š1 2 2 p€.(6. 3.22) Althoughwedonotindicatethemomentumdep endenceexplic itly,itisimplicitthatthe momentumisthat leaving thevertexthroughthepropagatoronwhichtheoperatorsact. (Fromthe V2interacti ontermwealsoobtainatadpole-typediagram;itscontributioncancelsasimilarcontributionfromthediagramweareconsidering,orvanishesifwe usedimensi onalregularization). The D scanbemanipulatedlikeordinaryderivatives.TheyobeyaLeibnitzrule, andatransferrule 4( 1Š 2) D€ 2( p )= Š D€ 1( Š p ) 4( 1Š 2),(6.3 .23) whichcanbecheckedbyexaminingtheexplic itformoftheoperators.Anotherexample ofthetrans ferruleis

PAGE 371

6.3.Supergraphrules355D1 24( 1Š 2) D 2 2= D1 2 D1 24( 1Š 2) = D1 2D1 24( 1Š 2).(6.3 .24) Insideintegralsthe D scanbeint egratedbyparts(seesec.3.7).Thus d4 [ D( p ) f ( p )] g ( Š p )= Š d4 f ( p ) D( Š p ) g ( Š p ).(6.3 .25) Thiscanbemosteasilyunderstoodin x -space.Since D= +1 2 i €€weared oing integrationbypartsin andin €. Armedwiththesefactswereturntotheevaluationoftheexpressionin(6.3.21). Wecon centrateonthe dependenceandwritetherelevantpartas d41d42V ( Š k 2)[ D1 2 D1 212][ D1 2D1 212] V ( k 1).(6.3 .26) Wehave a bbreviated 4( 1Š 2)= 12.Wenowinte gratebypartsand“nd“rstofall [ D2 D2 ][ D2D2 ] V = D2 D2[( D2D2 ) V ] = D2[( D2 D2D2 ) V +( D D2D2 ) DV +( D2D2 ) D2V ] = D2[ Š p2( D2 ) V + p€( D€D2 ) DV +( D2D2 ) D2V ],(6.3 .27) whereweh aveused( D )3=0andthe anticommutationrelations { D, D€} = p€when actingonthepropagatorwithmomentum p Be foreproceedingwemakethefollowingimportantobservation:Since 4( )= 2 2,multi plyingtwoidentical -functionstogether,ormultiplyingoneby giveszero.Wehavethereforethefollowingrelations: 2121= 2112=0, 21D21=0, 21D221=0, 21D D€21=0, 21D D221=0,

PAGE 372

3566.QUANTUMGLOBALSUPERFIELDS21D2 D221= 21 D2D221= 211 2 D D2D21= 21, 21D D2D21= C21.(6. 3.28) Intheserelations,weobtainanonzeroresultonlyifallthe sinthesecond -function areremovedbydierentiation.Hencetwo D sandtwo D saren eededandonlytheir momentumindependentpartscontribute.Expressionsofthiskind,butwithmore D s, canbereducedtooneoftheaboveformsbyusingtheanticommutationrelations.In theexpressionswithfour D stheorderisirrelevant(exceptforproducingsomeminus signs). Returningtoour calculation(6.3.27),andletting D2actonthefactorstoitsright, weseethato utoftheaprioripossiblesixterms,onlythreesurvive: [ Š p2( D2D2 ) V Š p€( D€ D€D2 ) D€DV +( D2D2 )( D2D2V )].(6.3.29) Finally,using D€ D€= €€ D2we “nd 4( 1Š 2)[ Š p2Š p€ D€D+ D2D2] V ( k 1).(6.3 .30) Insertingthisresultintotheoriginalintegral(6.3.21),weusetheremaining -functionto dothe 2integral,and“nallyobtain1 2 d4k (2 )4 d4 V ( Š k )[ d4p (2 )4 Š p2Š p€ D€D+ D2D2p2( k + p )2 ]V ( k ).(6.3.31) Theresultconsistsofanordinaryloopmomentumintegral,withusualpropagatorsand somemomentumfactorsinthenumerator,andoperators D D ,actingont heexternal super“el ds( i.e., D and D dependon k ,not p ).The p2termiscan celedbythetadpole di agrammentionedabove,(orgiveszeroindimensionalregularization)sothatthe“nal contributiontotheone-loopself-energyislogarithmicallydivergent.Thisisaconsequenceofgaugeinvariance.Theremainingtermsinthenumerator,inagauge-invariant regularization(suchasdimensional),combinetoform1 2 D D2D,giv ingaresultproportionalto W2.

PAGE 373

6.3.Supergraphrules357c.Theeectiveaction Intheexampleabove,the -functionhasreducedtheexpressiontoonewhich involvesasingle .Thisi sagenerala ndimportantresultofsuper“eldperturbationtheory.Theeectiveactionisasumoftermsinvolvingproductsof“eldsevaluatedatdifferentpoints,andaGreenfunctionwhichisanonlocalfunctionofitsarguments: =n d4x1... d4xnd41... d4nG ( x1... xn; 1... n)( x1, 1) ... DV ( xi, i) ... (6.3.32) Itturnsout,however,thatbymanipulationofthecontributionsfromanygraph,wecan reduceittoanexpressionthatislocalin i.e. =n d4x1... d4xnd4 G( x1... xn)( x1, ) ... DV ( xi, ) ... .(6. 3.33) Wedothisasfo llows:Consideranarbitrary L -loopcontributiontotheeective action.Itconsistsofpropagators,withfactors 4( iŠ i +1)and D operatorsactingon them,ext ernalsuper“eldfactors,and d4iintegrals.Wechooseanypropagatorfroma particularvertex v toanothervertex v,andintegr atebypartstoremoveallthe D s fromits -function.Theoriginalcontributionnowbecomesasumofterms.Ifthereare otherpropagators,eachofwhichconnects v and v,weusetherelatio ns(6.3.28):The termsvani shunless each oftheother -functionshasexactlytwo D sandtwo D sacting onit,inwhichcasetheycanbereplacedby1.Wenowusethefree -functiontodothe -integralat vandshrinkallthepropagatorsbetweenthetwoverticestoapointin -space.Werepeattheprocedure,choosingapropagatorleadingtoanewvertex v, untilweh averemovedall -functionsandperformedall -integralsexcepttheoriginal oneat v .Whe neverweh avemorethantwo D sandtwo D sonalineweusetheanticommutationrelationstoreplace D D pairsbymomenta.Wea releftwithasumof terms,allwithasingle integral,andvariousfactorsofloop-momentacomingfromthe anticommutatorsof D s,asw ellas D factorsactingontheexternalsuper“elds,coming fromtheintegra tionbyparts. InthecourseofevaluatingFeynmandiagrams,wemayencounterloop-momentum ultravioletdivergences,andasuitableregularizationprocedureisneededtohandle them.Wediscussregularizationissueslateron.Forthetimebeingweassumethat

PAGE 374

3586.QUANTUMGLOBALSUPERFIELDSthereexistsapro cedurethatallowsustocarryoutthemanipulationswehavedescribed aboveinsidemomentumintegrals. Theexpressionfortheeectiveactionin(6.3.33)revealsoneimportantfact:We haveendedupwitha d4 integral,eventhoughintheoriginalclassicalactionwemay havehad d2 integrals.ThisisaconsequenceofourFeynmanrules:Allourvertices carry d4 integrals,andnowhereinourmanipulationsdoesa d2 appear.Inparticular, iftheoriginalactio nhad purelychiral d2 massorcubicinteractionterms2or3, radiativecorrectionsdonotinduce“niteorin“nitemodi“cationsoftheseterms.Thisis the no-renormalizationtheorem forchiralsuper“elds.Massesandcouplingconstants are renormalized,butonlyasaconsequenceofwavefunctionrenormalization.(Any d4 integralcanbewrittenasa d2 integralanda D2operatoractingontheintegrand; however,thiswillnotproducetheabovete rms.)Thistheoremisvalidinperturbation th eory.Sofarnoonehassucceededingivingexamples,infourdimensions,whereit mightfailnonperturbatively,butaproofofitsgeneralvaliditydoesnotexist.Even withinperturbationtheorythereexiststhepossibilityofapathologicalinfrared-type behaviorwhichm ightinvalidateit.Forexample,ifinthecourseofevaluatingtheeectiveactionatermd4 2D2 werepro duced,the D2operatorwhichcomesfromconvertingthe integraltochiralform,whenactingonthechiral“eld,wouldgive D2D2 Š 1=andwe woulde ndupwithacontributiontothechiralcubicvertex. Whethersuchpathologicalbeh aviorcanbe obtainedinanycalculationwithasensible infraredregularizationisdoubtful. d.Divergences Wenowdiscussth edivergencestr uctureoftheeectiveaction.Therearetwo issuesinvolved:Wem ustdeterminewhichtermsintheeectiveactionaredivergent (powercounting),andwhichtermsinthecla ssicalactionleadonlytodivergencesthat canbeabsorbedinarenormalizationofthepa rameters(renormaliz ableinteractions). Werest rictourdiscussiontointeractinggaugeandchiralscalarsuper“elds(withnonegative-dimensioncouplingconstants). Thepossibledivergencesoftheeectiveactioncanbeunderstoodbystraight powercounting(sees ec.6.6)orsimplybyadimensionalargument:Thedivergentparts ofgraphsthatcontainnosubdivergencesgiverisetolocaltermsintheeectiveactionof

PAGE 375

6.3.Supergraphrules359theform = d4xd4 IP (, V D, ...),(6.3.34) where IP isapolynomialinthe“eldsandtheirderivatives.Sincetheeectiveaction mustbedi mensionless,and d4 hasdimen sion2, IP mustal sohavedimension2.has dimension1, Dhasdimension1 2 ,and V isdime nsionless.Therefore,graphswithmore thantwoexternalsareconvergent.Aor p ropagatorproducesanumerator fact orof m whichcontributestothedimensionof IP andthereforereducesthedegreeof divergence.If IP ismadeupofonlychiralsuper“eldsthe integrationwillgivezero unlesssome D s(atleasttwoofthem,to contractindices)arepresenttomaketheintegrandnonchiral,andagainthe D scontri butetothedimensionof IP reducingthe numberof“eld sthatcan appear.Finally, if gaugeinvariancerequires V toappear throughits“e ldstrength W= i D2DV (oritsnonabeliangeneralization),thislimits thepossibledivergencesinvolving V “elds.(Thisisanoversim p li“cation:Wemustuse thefullmachineryofSlavnov-Tayloridentities,atleastinthenonabeliancase,oruse th eb ac kg round-“eldmethod(seesec.6.5)toanalyzethedivergencesinvolvinggauge super“elds.) Thenetresultoftheanalysisistoestablishthattheonlylocaldivergentterms containatmostoneandone a nd,whiletheycontainanarbitrarynumberof V factors,theseenterinama nnerwhichiscontrolledbytheSlavnov-Tayloridentities.Fora renormalizabletheoryofchiralscalarmultipletsinteractingwithavectormultiplet(we omittheghostterms)therenormalizedclassicalactionhastheform d4xd4 [ RegRVRR+ tr RVRŠ tr R Š 1( D2VR)( D2VR)] +1 gR 2 d4xd2 trWR 2+[ d4xd2 IP (R)+ h c .],(6. 3.35) wherethesubscript R labelsrenormalizedquantities.(Intheexponentialwehavewrittenexplicitlythegaugecouplingconstant g thatwenormallyabsorbinto V .) Since V isdime nsionless, VRisingenera lano nlinearfunctionof V i.e.,thewavefunctionrenormalizationf actormaybeafunctionof V :Wecanhave functionalrenormalizations VR= f ( V ),whereeachcoecientintheTaylorexpansionof f isa

PAGE 376

3606.QUANTUMGLOBALSUPERFIELDSrenormalizationconstant.Sinceallsuchrenormalizationsareproportionaltothe“eld equations( S = S V V ),theyvanishonshell.(Suchnoncovariantrenormalizations areavoidedinthebackground“eldgaugesthatwediscussbelow.) Ghostsaredescribedbychiralsuper“elds whichfollowthesamerules.Thedivergencesofthetheoryarealllogarithmic,exceptthatoftheFayet-Iliopoulosterm,which isquadratic.(Howev er,asweshalldiscussinsec.6.5,thistermisnotproducedby radiativecorrections.) Ingeneral,renormalizableinteractionsareassociatedwithdimensionless(orpositivedime nsion)couplingconstants.ForFeynmangraphs,sinceateachvertexwehavea d4 integralwithdimension2anda d4x integralwithdimension Š 4,wemayallowupto theequivalentoffour D sateachvertex.Thisisindeedthecasewiththegauge“eld self-couplings,andalsotheusualverti cesinvo lvingchiralsuper “elds,wherethe D factorscomefromourFeynmanrules. Ontheotherhandatermsuchas 2,or4, wouldleadtoa nex cessof D sattheverti cesandanonrenormalizabletheory. e.D-algebra Inthenextsectionwegiveanumberofexamplesofevaluationofsupergraphs. Aspreparationwediscussseveralsimpli“cationsthatweuseinperformingthemanipulationofthe D sandthe integration.Thenumerousintegrationsbypartsthathaveto beperfo rmedcanleadtolongintermediateexpre ssions,andalotofeort(andpaper) canbesavedbydoingthemanipulationsdirectlyonthegraphs. Wedrawthes upergraphandindicateonit,adjacenttothevertices,the D factors actingonthepropagatorsintheorderinwhichtheyact.Weignoresignshavingtodo withtheorderingofthe D s:Thesewillbedete rminedlater.Thus,anexpressionsuch as D2D4( Š ) D D€D ,witht helastthree D sacting backwardsonthe argumentofthe -function(andthusintheorder D“rst,then D€next,etc.) ,wouldbe representedonthegraphasshownin“g.6.3.2:

PAGE 377

6.3.Supergraphrules361 D2DD D€DFig.6.3.2 Thetransferrulecanbeimplementedbyslidingthe D stotheleft,keepingtheorder. Thiscorrespondstowriting Š D2DD D€D4( Š )with Dacting “rstonthe argumentofthe -function.Wemustkeeptrackofthe Š signcomingfromtransferring anoddnumberof D s.(Notetheorderintheseexpressions,e.g., D1 12D 2 = D1 D2 12= Š D1 D1 12.) Weuset hecommutationrelationstoreplacetherightmost D D by p€.(Si nce p ishermitian p€= p€butwemainta inthedistinctiontokeeptrackoftheorderin whichthe D sappeared.)When ween counterexpressionssuchas D2 D2D2wereplace themwith Š p2D2. Theintegrationbypartscanalsobecarriedoutdirectlyonthegraphs.Forexample,weshowin“g.6.3.3theintegrationbypartsonavertexcomingfromthe V interaction: D2 D2 D€D2 D2D2 D2D2 D€Fig.6.3.3 Startingfromagiven graphingeneralweobtainseveral,becauseintegrationby partsgivesseveralcontributions.Weremovethe D sfromanygivenlineandusethe -functiontodooneofthe integrals,thuscontractingthelinetoapointin space. Wen eednotindicateexplicitlythiscontract ion:Alinewithoutanyoperatorsonitis understoodtobecontracted.Wheneverseverallinesconnectthesamepairofvertices, ifallthelines(otherthantheonewehaveclearedof D s)haveexactlytwo D sandtwo D seach,weuse(6.3.28)toreplacethemby1.Ifanylinehasfewer D sor D s,the contributionvanishe s.Ifanylinehasmore D sor D s,weusetheanticommutation

PAGE 378

3626.QUANTUMGLOBALSUPERFIELDSrelationstoreducetheirnumber.Intheendwehaveasumofgraphs,withmomentum factorsfromt heanticommut ators,and D sactingontheexternallinesonly. Tomakethispro cedureclear,weredotheexampleconsideredabove(“g.6.3.1, (6.3.21-31)),workingdirectlyonthegraph,asshownin“g.6.3.4: D2 D€DD2D D€D D2 D2 D2 D2 D2 D2 D2 D2 D2 D2 D2 D2D2D2D2D2D2D2D2D2D2D2D2DFig.6.3.4 Inthelaststepwehaveonlyindicatednonzerocontributions(theothersvanishtrivially b ecauseof(6.3.28).) Inthecourseofthemanipulationsonthegraphs,wemustkeeptrackof Š signs comingfromtransfersandfromintegrationbyparts.However,weneednotkeeptrack of Š signsthatcomefrompassinga D pastan other D ,norfr omsignsthatcomefrom raisingorloweringindices.(Onthegraphs,wedonotindicatetherelativeorderof D s ondierentlines,norwhichindicesareupo rdown).T hesesignscanbedeterminedat theendofthecomputationint hefollowingmanner:Ontheoriginalgraph,wehavefactorssuchas D2 D2=1 4 DD D€ D€,andalso,froma vertexsuchas V ( DV )( D2DV ), adjacent factors Dand1 2 D€ D€Dwherewedeterminetheinitialsignbyrequiringthat inanycontractedpair,the“rst D or D hastheupperindex.Thesevariousfactorsmay endupinadierentorderinthe“nalexpression,e.g., D... D€... D... D€,po ssibly actingondierentsuper“elds;however,westillwriteacontractedpairwiththe“rst i ndexraised.Todeterminethe“naloverallsign,wecountthenumberoftranspositions neededinthe“nalexpressiontobringthe D sbacktotheorigina lorder.Thi sistrue evenifsomeofthe D shaveb eenreplacedbymomenta:Anexpressionsuchas p€will

PAGE 379

6.3.Supergraphrules363correctlykeeptrackofthetranspositions.Wedo not counttranspositionsofcontracted pairs,sincetheconventionforraisingandloweringindicescancelssuchsigns: XY=+ YX.Aquickwaytoco untthetranspositionsis todrawlines connecting allcontractedpairswithlines,andcountthenumberofintersections:anoddnumber meansanoddnumberoftranspositions,andhencea Š sign,whereasanevennumber meansno Š sign. Therearemanyothertricksthatonecanusetosimplifythemanipulations.We givethefollowingtwinglingrulewhichisoftenuseful: D2DDDDDD Fig.6.3.5 Wearenowre adytoconsiderfurtherexamplesofgraphevaluation.

PAGE 380

3646.QUANTUMGLOBALSUPERFIELDS6.4.Examples Inthissectionwegiveanumberofexamplesofsupergraphcalculations.Mostof theexamplesw ereencounteredinvariouscalculationsthathavebeenperformed.For morecomplicatedoneswereferthereadertothecalculationsofthe3-loop -functionin N =4 Ya ng -M ills,andthe3-and4-loop -functionintheWess-Zuminomodel. Wedoourm anipulatio nsdirectlyonthegraphsuntilonlyanordinarymomentum integralremains.Fornotationalconveniencewesometimesindicateafactor p2mult iplyingapropagatorwiththesamemomentumbya drawnonthecorrespondingline.In thecaseofalinewithno D sactingonit,wesometimesleaveitinthegraph,whileat othertimes,w henwedraw -spacegraphs,wecontractitout.Toestablishtheprocedurewebeginwithsomesimpleexamples. Forthema ssiveWess-Zuminomodelweconsider“rstsomeselfenergygraphs: (1) D2 D2 Fig.6.4.1 d4 ( Š p )( p ) d4k (2 )4 1 ( k2+ m2)[( k + p )2+ m2] .(6. 4.1)

PAGE 381

6.4.Examples365(2) D2 D2 Š mD2( k + p )2 Š mD2( k + p )2 Š mD2p2 m D2Fig.6.4.2 Wehave used D2D2 D2= Š p2 D2.Atachiralvertexthe D2factorsc anbeputoneither linebyintegra tionbyparts. Theresultisd4xd4 =0b ecausetheintegrandischiral.Weconsidernextatrianglediagramwith3and 3verti ces: (3) m D2Š mD2p2 m D2D2 D2 D2 Fig.6.4.3 Thus,theone-loopcontributionstothisthree-pointfunctionarezerointhemassless caseandnonlocalinthemassivecase.

PAGE 382

3666.QUANTUMGLOBALSUPERFIELDS(4)For m =0,with3and 3verti ces, D2 D2 D2 D2 D2 D2 D2 D2D2D2D2D2D2D2D2D2 Fig.6.4.4 Inthesecondgraph,thesmallloopiscontractedtoapointin -space,afterwhichthe D2D2operatorscanbetransferredacrossitandallactonthesameline.The“nal graphshowstheactualmomentum-spacediagramonewouldhavetoevaluate.Apropagatorhasbeencanceledby (5)Againinthemasslesscase, A B C E F D2D2D2D2D2D2D2 D2 D2 D2 D2 D2 D2D2 Fig.6.4.5a Wehave placedthe D sand D soncertaintwoofthethree linesforconve nience,and havelabeledthevertices.Inthesecondgraphwehavetransferred D2, D2fromC,B,to E,F,respectively.Wenowintegratebypartsthe D2factor atE.Thiswillgenerate threeterms.

PAGE 383

6.4.Examples367 D2 D2 D2 D2 D2 D2 D€ D€D2D2D2D2D2D2D2D2D2D2 Fig.6.4.5b Byexaminingthe -spaceloopsAECBAandEFBCE,itisclearthatineachgraphthe D2sonABandBF,respectively,mustgiveasingletermwhenintegratedbypartsat theirr espectivevertices(AandF).Weobtain k q D D D2 D2 k€q€Fig.6.4.5c Wehave used D€D2 D2D = k€ D D2D = Š k a .(6. 4.2) Ournextexamplehas eVinterac tions:

PAGE 384

3686.QUANTUMGLOBALSUPERFIELDS(6) A B C E F D2D2D2D2 D2 D2 D2 D2Fig.6.4.6a IntheloopBCFwehavejusta D2 D2factor,sow econtr actittoapoint.Similarly,we contracttheAElinetoapoint.Forclaritywedrawa -spacegraphwhere q and h are themomentaoftheABandEFlines: h q A B C E F D2D2D2 D2 D2 D2Fig.6.4.6b Weintegr atethe D2factorothemiddleline.Itcannotgoon soitmustg ooneither thetoporbottomline,orsplit.Becauseof(6.3.28)the D2factormustfo llowit.This isthesam eascom puting D2D2[ ( q ) ( h )]wherethe sarechiral.Theresultissimply Š ( q + h )2 ( q ) ( h ).Ther efore,the D manipulationis“nishedandweobtain Š ( Š p )( p )( q + k )2mu ltiplyingastandardmomentumintegralwiththepropagatorsoftheoriginaldiagram. We givenowanexampleinnonabelianYang-Millstheory:

PAGE 385

6.4.Examples369(7) k V DD D2 D2D2 D2D2 D2D D€D2 D2DD D D€ ( p ) ( p) Fig.6.4.7 V ( Š p Š p) D( p ) D€ ( p) d4k (2 )4 k€k2( k Š p )2( k + p)2 .(6. 4.3) (8) N =4 Ya ng -M illstheory. Insec.4.6.bwehavegiventheclassicalactionofthistheoryintermsof N =1 super“elds.Herewediscusssomeofitsquantumproperties. Thetheoryisdescribedbysuper“elds V ,i( i =1,2,3),allinthe adjointrepresentationofanarbitrarygroup,andwithinteractionsgovernedbyacommoncoupling constant g .Itiscla ssicallyscaleinvariant,andbothcomponentandsuper“eldcalculationshaveestablishedthatits -functionvanishestothreeloops,sothat,perturbatively, thescaleinvariancesurvivesquantization.Proofsexistthatextendthisconclusiontoall ordersofpertur bationtheory.Herewediscusssome oftheexplicitsupergraphcalculationsfores tablis hing ( g )=0, andleavethegeneralargumentstosec.7.7. Weaddtothecl assicalaction(4.6.38)thegauge“xingandghostterms (6.2.17,20-22)withgaugeparameter =1+ O ( g2).To O ( g0)thisc hoiceg ivesthe Fe rmi-Feynmangaugeandapropagator Š 1thatavoidsseriousinfraredproblems. Ho we ve r,thetransversepartoftheself-energyreceives(local)radiativecorrections, whereasthelongitudinalpartdoesnot(asfollowsfromtheWardidentities).Tostayin theFermi-Feynmangauge,wemustmaintaint heequalityofthelongitudinalandtransverseparts,a ndwedothisbyadjustingthe O ( g2)par tsof ineachordero fperturbationtheory(actually,theradiativecorr ectionsvanishatoneloop,andonlyariseat O ( g4)).

PAGE 386

3706.QUANTUMGLOBALSUPERFIELDSAtthecla ssicallevelthe O (4)invarianceofthetheoryrequirestheequalityofthe gaugeand123couplingconstants.Althoughthegauge“xingprocedurebreaksthe O (4)symmetry(thiscouldbeavoidedifwehadan N =4super“eldfo rmalism),gauge in va rianceshouldinsurethatthecouplingconstantsreceiveacommonrenormalization. Therefore,thetheoryhasonlyone -function,whichwecancompute,forexample,by comparingtherenormalizationofthe Cijkijkvertexf unctionandthe iiwave functionrenormalization.However,thevertexbeingchiral,receivesnoradiativecorrections(seesec.6.3),sothattoestablish ( g )=0itissu cienttoshowthatthe iiself-energyis“nite.(Weobservethatif O (4)invarianceofthequantumeectiveaction werenotspo iledbythegauge“xingpro cedure,“nitenesstoallorderswouldfollow immediately:The “nitenessofthe Cijkijkvertexwouldi mplythe“nitenessofall otherlocaltermsintheeectiveaction.Inprinciple,thedesiredresultshouldstillfollowfromthe O (4)Wardidentities,butinpracticethenonlinearityofthetransformations(4.6.39,40)makesthemdiculttoapply.) Tolowordersin V (suci entforthethree-loopcalculation)theactionis S = tr d4xd4 { iiŠ1 2 V V + cc Š c c + g [ i, V ]i+1 2 gV { DV D2DV } +1 2 g ( c+ c)[ V c + c ] +1 2 g2[[ i, V ], V ]i+1 8 g2[ V DV ] D2[ V DV ] +1 6 g2( D2DV )[ V ,[ V DV ]]+1 12 g2( c+ c)[ V ,[ V c Š c ]]1 3! g3[[[ i, V ], V ], V ]iŠ1 2 (1 Š 1) V 0 V + ... } + tr { d4xd2 ig1 3! Cijki[j,k]+ h c } (6.4.4) with V = VATA,i=iATA, c = cATA, [ TA, TB]= ifAB CTC, fAB CfDC B= Š k AD.(6. 4.5)

PAGE 387

6.4.Examples371UsingtheFeynmanrulesitistrivialtoseethatattheone-looplevel,intheFermiFeynman gaugede“neda bove,the selfen ergyisidenticallyzer o.Thecontr i butions fromthetwo graphsbelowcancel: D2D2 D2 D2 Fig.6.4.8a Itiseasytoverifythattheone-loopcorrectionstotheghostandvectorself-energiesalso completelyvanish.Fortheformer,thisistrueinanytheory,butforthelatteritisdue tothemultiplicity(3)ofthechiralmultiplets,whichleadstocancellationsamongthe threegraphsbelow: Fig.6.4.8b Thisresultistriviallytrueinthebackground“eldmethod:(seesec.6.5).Then the“eld V doesnotcontributeandthethreechiral“eldsexactlycancelcontributions from three chiral ghosts.Thisalsooccursforthe V thr ee-pointfunction,andissucienttoestablishinanindependentwaythat (one-loo p)=0.Wereferthereaderto theliteratureforotherone-andhigher-loopcalculationsandsummarizethesupergraph results:(a)One-loopthree-pointfunctionsare“nite.The V -“eldfour-andhigher-point functionshaveadivergencethatcanberemovedbyanonlinear V -“eldreno rmalization. (The divergenceneverarisesinthebackground“eldmethod.)(b)Atthetwo-looplevel theghost self-energ iesarestillzero,wh ereasthosefor V andiareonly “nite.(Consequently,thehigher-ordercontributionstothegaugeparameterare O ( g4).)(c)Atthe three-loopleveltheiself-energyis“nite,thusensuringthevanishingofthe -function. Wepres entargumentsforprovingtheresultstoallordersinsec.7.7.

PAGE 388

3726.QUANTUMGLOBALSUPERFIELDSThevanishingofthe -functiontoallordersofperturbationtheoryleadstothe conclusionthat N =4 Ya ng -M illsisa“nitefour-dimensional“eldtheory(uptogauge artifacts;e.g.,exceptinsupersymmetricba ckgr oundorlight-conegauges,divergences arepresent,butonlyingauge-dependentquantities). Anothert heorywithinteresting“nitenesspropertiesis N =2 Ya ng -M illstheory. Ithasone-loopdivergences,butis“nitetoallhigherordersofperturbationtheory(as exp licitlyveri“edattwoandthreeloops),makingitsuperrenormalizable.Bycoupling anappropriatenumberof N =2hypermul tipletsto N =2 Ya ng -M illstheory,onecan arrangefortheone-loopdivergencestoca ncel,andthusconstructacompletely“nite theory(inperturbationtheory).(Inthespecialcaseofoneadjoint-representation hypermul tiplet,w eobtain N =4 Ya ng -M illstheory.)

PAGE 389

6.5.Thebackground“eldmethod3736.5.Thebackgro und“eldmethod a.OrdinaryYang-Mills Thebackground“eldmethodisextremelyusefulinsupersymmetricYang-Mills theory,and essentialinthequantumtheoryofsup ergravity.In thissectionwereview thebackground“eldmethodforordinaryYang-Mills,andthenextendittothesupersymmetriccase.Theextensioninvolvessomesubtleties,primarilybecauseofthenonlinearityofthegaugetransformations. Ingaugetheorieswestartwiththegauge-“xedfunctionalintegral(6.2.9)and intro ducesourcescoupledtothe“elds,de“ning Z ( J )= IDAIDcIDceSeff+ JA, JA d4xJ aA a.(6. 5.1) Weintr o duce W ( J )= lnZ ( J )and de“netheeectiveactionbyaLegendretransform ( A )= W ( J ) Š J A A a= W J a .(6. 5.2) Thisquantityisnotgaugeinvariantingeneral.Physicalquantitiescomputedfromit aregaugeinvariant,andtheGreenfunctionssatisfySlavnov-Tayloridentitiesthat expresstheunderlyinggaugeinvariance,butmanifestgaugeinvarianceislostbecauseof thegauge“xingprocedure. Ontheotherhand,theeectiveactioncomputedinthe background“eldmethodismanifestlygaugein variant.Itisequiva lenttotheusualone, butismoreconvenienttohandle. Inthebackground“eldquantizationofYang-Millstheorieswefollowaprocedure similartothatofsec.6.2a.Westartwiththegauge-invariantLagrangian ILinv( A a) (other“eldsmaybepres entbutwedonotindicatethemexplicitly)and split the“eld intoabackgroundan dquant umpart: ILinv( A A a+ A a). Theactionisinvariantundertwo kindsoftransformationsthatgivethesame ( A A a+ A a): Quantum: A A a=0, A a= a + i [ A a],(6.5 .3) a = a + i [ A A a],

PAGE 390

3746.QUANTUMGLOBALSUPERFIELDSBa ck gr o und: A A a= a A a= i [ A a].(6.5 .4) Weconsid ernowthefunctional Z Z ( A A )= IDA aeILinv( A a+ A A a),(6. 5.5) andquantizeasbeforeto“xthe quantum gaugeinvarianceexceptthat,tomaintain manifestinvariancewithrespecttothe ba ckground gaugetransformations,wechoosethe gauge-“xingfunctionsothatittransforms covariantly underthesetransformations. Thisrequiresinparticularthatwecovariantizethederivativesthatappeartherewith respecttothebackground“eld: aA a aA a= aA aŠ i [ A A a, A a].Therem ai nderof thequantizationpro cedureisthesame.WerequiretheFaddeev-Popovghoststotransformcovariantlyunderbackgroundgaugetransformations,andwechoosetheweighting function exp ( Š1 g2 tr f2)tobeinvar iant.Wet husobtai nthefo llowing expression: Z Z ( A A )= IDA aIDcIDceSinv( A + A A ) Š1 4 g2 tr ( A )2+ SFP.(6. 5.6) Z Z is manifestlyinvariantunderbackgroundgaugetransformationsbutitssigni“canceis notobvious.Toelucidateitsmeaningweconsideranobjectde“nedexactlylike Z Z exceptthatwealsocouplethequantum“eldtoasource: Z( J A A )= IDA aIDcIDceSeff( A a, A A a)+ JA.(6. 5.7) Wecannowpassfrom Z( J A A )to( A, A A )byaLege ndretransformationinthepresence ofthe“xed“eld A A .Ontheot herhand,returningto Zitself,wecanmakeachangeof variables A Š > A Š A A whichgives Z( J A A )= eŠ J A A IDA aIDcIDce[ ILinv( A a)+ ILGF+ ILFP+ JA ]= eŠ J A AZ ( J A A ).(6.5 .8) Itcontainstheusual ILinv( A a) butunusualgauge“xingandghosttermsthathaveadditionaldepe ndenceon A A a.There fore, Z ( J A A )isthe usualgeneratingfunctionalbut

PAGE 391

6.5.Thebackground“eldmethod375with A A -dependentgauge“xingandghostterms.Nowwehave W( J A A )= lnZ( J A A )= lnZ ( J A A ) Š J A A = W ( J A A ) Š J A A ,(6. 5.9) and ( A, A A )= W( J A A ) Š JA=[ W ( J A A ) Š J A A ] Š JA= W ( J A A ) Š J ( A+ A A ), A= W J = W J Š A A .(6. 5.10) Therefore ( A, A A )=( A+ A A A A ),(6.5 .11) istheusual eectiveaction( A ),evaluatedinanunusual A A -dependentgauge,andat A = A+ A A .Inparticula r,ifintheevaluationof( A, A A )werestricto urselvesto graphswith noexternal Alines(vacuumgraphs),i.e.,set A=0,wew illobtain( A A ), theusual eectiveaction.Butthese A-vacuumgraphsaresimplytheone-particleirreducible subsetofthegraphsobtainedfrom Z(0, A A )= Z Z ( A A ).(Actually,thisisan oversimp li“cation.Whatoneobtainsisnotex actlytheeectiveaction,because A A lines fromthe gauge-“xingtermgiveadditionalcontributions.However,becauseofgauge invariance,it canbeshownthatthesehavenoeectonS-matrixelementssothatthe identi“cation,thoughstrict lyspeaking notcorrect,canbeusedwhencomputingphysicalquant ities.) Ourconclusionisthattheeectiveactionisobtainedfrom Z Z ( A A )byevalu atingin pertur bationtheoryone-particle-irre duciblegraphswi thonlyinternal A alinesande xternal A A alines(asw ellasghost,andothernon-gauge“eldlines).Inparticular,ifwe expand ILinv( A A + A )= IL ( A A )+ IL( A A ) A + IL( A A ) A2+ ... ,the “rsttermdoesnotcontri butetoloopgraphs(itistheclassicalcontributionto),andthesecondcanbe droppedbecauseitdoesnotcontributetoone-particle-irreduciblegraphswithnoexternal A lines.The A2termgivesthecompletecontribution(fromthegauge“eld)tooneloopgraphs.Forhigher-loopgraphs,inter nalverticesarereadfromthehigherorder expansion,andallthetermscontributetoverticesthatinvolvetheexternal A A lines.

PAGE 392

3766.QUANTUMGLOBALSUPERFIELDSThereisanadditionalfeatureofthebackground-“eldquantizationthatisnotusuallyencounteredintheYang-Millscasebutthatisimportant.Thisistheappearanceof theNielsen-Kalloshghost.Inthegauge-ave ragingprocedureweusedthesimplestexponentialfactortoproducethegauge-“xingte rmintheeectiveLagrangian.However,a morecomplicatedaveragingfunctioncouldbeused,e.g., exp ( fMf )where M isany operator(matrix).Toproperlynormalizet heaveragingprocedure,wemustdivideby detM .If M is“eldindependent,thisisatrivialfactor.However,if M isafunctionof thebackground“eld,wenormalizethegauge aver agingbyintr o ducingintothefunctionalintegralafactor IDfIDbefMfebMb,(6. 5.12) where b isa ghost “eld,w ithoppositestatisticsto f .Whenweca rryoutthe f integrationus ingthe -functionofsec.6.2.a,weareleftwiththe b “eld.Thus,the“nalformis Z Z ( A A a)= IDA aIDcIDcIDbe[ ILinv( A + A A )+ ILGF( A A A )+ ILFP( c c, A A A )+ ILNK( b A A )](6.5.13) where ILNK= bMb .If M isindepe ndentofthebackground“eld,theadditionalghost givestrivialcontributionsandcanbedropped;otherwise,sincetheghost“eld b hasno interactionswithothe rquant um“eldsandsinceitentersquadratically,itonlycontributesattheo ne-looplevel. Tomotiva tetheprocedureweuseinthebackground“eldquantizationofsupersymmetricYang-Millstheory,wepointoutt wo aspectsofthebackground-quantum spli tting A a A A a+ A aofordinaryYang-Mills.Thissplittinghasthevirtuethatthe transformations ( A A a+ A a)=( a Š i [ A A a, ])+( Š i [ A a, ]),whichleavetheaction invariant,canbeinterpretedasordinarygaugetransformationsofthebackground“eld accompaniedbycovariantgaugerotations(linearandhomogeneous)ofthequantum “eld.Furthermore,inanexpansion IL ( A + A A )= [IL( n )( A A )]( A )n,(6. 5.14) eachterminthepowerseriesisseparatelyinvariantunderthesetransformations,since A atransformslinearlyandhomogeneously,asdothefunctionalderivativesof IL ( A A ).

PAGE 393

6.5.Thebackground“eldmethod377Thus,ifwetruncatetheseries,aswedoinaperturbativeloop-by-loopevaluationofthe eectiveaction,wema intainthebackgroundgaugeinvariance. Thiswouldnotbetrueif thetransformationofthequantum“eldwerenonlinear. b.SupersymmetricYang-Mills InsupersymmetricYang-Millstheoryth ec lassicalactionisinvariantunder nonlinear gaugetransformations eV eV= ei eVeŠ i ,andthes p litting V V + V V is unsuitable.Tomotivatethesubsequentprocedure,we“rstreexaminethebackgroundquantumsplittingofordinaryYang-Millstheoryfromadierentpointofview.Westart withtheoriginalgaugeandmatteraction,invariantunderthelocaltransformations Aa= a Š i [ A a, ]a nd,forsomematter“eld, = i [ ].U nder global transformationswithconstant west illhaveinvariance,withthegauge“eldsrotatinglikethe matter“elds.Forlocal ,wecanintr o duceanewinvariancebykeepingthecovariant transformations i [ A a]and i [ ]for all“elds,a ndintroducingaseparategauge“eld, thebackg round“eld A A atocovariantizetheder ivatives.Sinceintheoriginalactionall derivati vesenter edintheform a= aŠ i [ A a,],thiscovar iantizationamountstothe replacement a= aŠ i [ A a,] aŠ i [ A a,]= aŠ i [ A A a+ A a,],(6.5.15) whichisequivalenttotheordinaryquantum-backgroundsplitting.Wenowhavetwo in va riances:theoriginalonewherethebackground“eldisinert,andthenewone,under whichallthe“eldstransform. WeobtainalinearsplittingA A + A A becausethe gauge“eldenterslinearlyinthecovariantderivative. InsupersymmetricYang-Mills theorythisissointheabeliancase,butnoti nthe nonabeliancase.However,thephilosophyisthesame.Westartwiththelocallyinvariantgaugetheory,observethatitis invariantforglobaltransformations(with= = aconstant matrix,notasuper“eld),underwhichthegauge“eldstransformcovariantly(linearlyandhomogeneously, since eV ei eVeŠ i implies V ei VeŠ i )and now gaugethistransformationby covariantizingwith theaidofabackground “eld.Thisamountstothereplacement DA Awhere Aisabackgroundcovariantderivative.Wealsohavetotreatthe covariantlychiralsuper“eldsproperly. We recallthatsupersymmetricYang-Millstheorycanbeformulatedintermsof constrainedcovariantderivatives.Thereasonforsolvingtheconstraintsandintroducing

PAGE 394

3786.QUANTUMGLOBALSUPERFIELDSthegaugeprepotentialsisthatonlytheseunconstrainedobjectsaresuitableforquantization.Insolvingtheconstra intswehavethechoiceofwor kinginth ev ectorrepresentationorinthechiralrepresentation.Thelatterismoreconvenientforquantization, expressingthetheoryintermsoftherealsuper“eld V ,rathert hansuper“elds, with ar edundantgaugeinvariance.Wewanttomaintainthisadvantageinthebackground “eldmethod andworkwithaquantum V .Ontheot herhand,whenweintroducethe backgroundcovariantderivatives,itisusefultothinkoftheminthevectorrepresentation.Infactitispossibletoexpressallour resultsintermsofthe(constrained)backgroundcovariantderivativesthemselves,withouteverintroducingexplicitlythebackgroundgaugesuper“elds,i.e.w ithoutsolvingthe constraints,andinthatcasetheonly representationthatisavailableisthevectorrepresentation.Theadvantageofworking withthebackgroundderivativesdirectlyisthatbackgroundcovarianceismanifestand weobtainsi gni“cantsimpli“cationsandimprovementinthepowercountingrulesfor Feynmang raphs. Wealsoexpr esscovariant lychiralsuper“eldsin termsoftheq uantum“eld V and background-covariantlychiralsuper“elds.Thelatterthereforedependimplicitlyonthe background“eldsandwouldseemnottobesuitableforquantization.However,thisis notalwaysthecase:Atmorethanoneloop,a ndevenatonel oopforre alrepresentationsofthegaugegroup,weformulatecovariantFeynmanrulesdirectlyforcovariantly chiralsuper“eldsth atleadtoconsiderableimprov ementovertheordinaryones. Startingwith theordinarycovariantderivativesweperformthesplittingbywritingthem,inthe quantum-chiralbutbackground-vector representation,as = eŠ V eV, €= €, a= Š i {, €} ;(6. 5.16) where and €arebackgroundcovariantderivativessatisfyingtheusualconstraints. The s tr an sformcovariantlyundertwosetsoftransformations: (a)Qua ntum: eV ei eVeŠ i A A,(6. 5.17) withbackgroundcovariantlychiralparameters = €=0, i.e.,

PAGE 395

6.5.Thebackground“eldmethod379A ei AeŠ i .(6. 5.18) (b )B ac kg round: eV eiKeVeŠ iK, A eiK AeŠ iK,(6. 5.19) witharealparameter K = K i.e., A eiKAeŠ iK.(6. 5.20) Thebackground“eldtransformationsof V canberewrittenas V eiKVeŠ iK,(6. 5.21) i.e., V transformscovariantly. *** Whilethisprocedurehasgivenusacorrectquantum-backgroundsplitting,incontrasttothecomponentYang-Millscaseitresultsindierenttransformationsof Aunderquantumandbackgroundtransformations.However,thetransformationofthe unsplitgauge“eld is th es am e.Tounderstandthesplittingofthegauge“eldwesolve theconstraintsonthebackgroundcovariantderivatives: = eŠ De €= e D€eŠ .(6. 5.22) Hencet hesplittingof thefullderivativesis = eŠ VeŠ De eV, €= e D€eŠ .(6. 5.23) Wetransf ormtoabackgroundchiralrepresentationbypre-andpost-multiplyingall quantitiesby eŠ and e ,resp ectively.Then eŠ eŠ VeŠ De eVe € D€,(6. 5.24) andthespli ttingiseq uivalenttoreplacing eVby eV ( split )= e eVe .(6. 5.25) Inotherwords,wesplitthefull V intoaq uantum V andbackground and ina

PAGE 396

3806.QUANTUMGLOBALSUPERFIELDSparticular,non linearfashion.(Intheabeliancasethisreducesto V V + + which isjusttheordinarysplittingsinceby(4.2.72) + = V V .) Theusualchiralrepresentationtransformationsof(6.5.25) ( e eVe )= ei 0( e eVe ) eŠ i 0,(6. 5.26) (where0isordi narychiral, D€0=0),canbew ritteni ntwo ways: (a) ( e eVe )= e [( eŠ ei 0e ) eV( e eŠ i 0eŠ )] e = e ( ei eVeŠ i ) e ,( 6.5.27a) i.e.,thequantumtransformations(6.5.17), withbackgroundcovariantlychiral ,or (b) ( e eVe )=( ei 0e eŠ iK)( eiKeVeŠ iK)( eiKe eŠ i 0),(6.5 .27b) i.e.,thebackgroundtransformations(6.5.19)(cf.(4.2.70-71);recallthatthe0partof thetransformationof doesnot aectthetransformationofthebackgroundcovariant de rivatives).ThisisverysimilartothesituationincomponentYang-Mills. ThegaugeLagrangianhastheform trW2= Š tr (1 2 [ €, { €, } ])2.(6. 5.28) Whenwesubsti tute(6.5.16)into(6.5.28),weobtainasplittingoftheactioninto exp licitquantum V sandbackgroundcovariantderivatives.Sincethe stransform covariantly,theLagrangianwillbeinvariantunderbothbackgroundandquantumtransformations.Furthermore,since V transformshomogeneously,expandingtheLagrangian inpowersof V willmaintainthebackgroundinvari anceterm-by-term,whichisoneof therequiredpropertiesofagoodsplitting. Whencovariantlychiralsuper“elds, €=0arepr esent,we“rstexpressthem intermsofbackgroundcovarian tlychirals uper“eldsby=,= eV(inthe quantum chiralrepres enta tion) €= =0, andthen linearly sp litthemintoasumof backgroundandquantum“elds.Thequantum“eldstransformunder

PAGE 397

6.5.Thebackground“eldmethod381(a)Quantumtransformations: = ei = eŠ i .(6. 5.29) (b)Backgroundtransformations: = eiK, = eŠ iK.(6. 5.30) Thechiral“eldactionisinvariantunderbothquantumandbackgroundtransformations. We examinenowthebackground“eldquantization.Weproceedasintheconventionalapproac h,butcompute Z Z = IDVIDcIDcID cID c ( 2V Š f ) ( 2V Š f )eSinv+ SFP.(6. 5.31) We havechosenbackground-covariantlychiralgauge“xingfunctions,andthismeans thattheFaddeev-Popovghosts,introducedasinsec.6.2,arealsobackgroundcovariantlychiral.Finally,wegaugeaveragewith IDfID fIDbID beŠd4xd4 [ ff + bb ],(6. 5.32) wherethebackgroundcovariantlychiralNielsen-Kalloshghosts b b havebeenintroducedtonormalizeto1theaveragingover f f .Thislea dstothe“nalform Z Z = IDVIDcIDcID cID cIDbID beSeff, Seff= Sinv+ SGF+ SFP+ bb ;(6. 5.33) which,exceptfortheNielsen-Kalloshghosts,islike(6.2.19),butwithbackground covariantderivativesandcovariantlychiralsuper“elds.TheN.-K.ghostsinteractwith thebackground“eld,andonlygiveone-loopco ntri butions.Ifwecoupleexternalsources tothequantum “elds,e.g.,d4xd4 JV ,thege neratingfunctional Z Z ( J )willstillbe in va riantunderbackgroundtransformations,providedwerequirethesourcesto

PAGE 398

3826.QUANTUMGLOBALSUPERFIELDStransformcovariantly, J = i [ K J ]. Whatwemustdonowisarguethatthebackground“eldfunctional,obtainedby settingsourcestozeroandcomputingone-particle-irreduciblegraphswithonlyinternal quantumlinesandexternalbackgroundlines,i sequivalentt othe usualeectiveaction, exceptforbeingcomputedinadierentgauge.Thisislessdirectthanintheordinary casebecausethesplittingishighlynonlinear.Wepresentthefollowingargument: Thesplitting(6.5.25)is V V + + + nonlinearterms .(6. 5.34) Inagaugeforthebackground“eldswhere = =1 2 V V V V wewritethisas V f ( V V V ) where f (0, V V )= V V and f ( V ,0)= V .Ifnowintheori ginalfunctionalintegralweadda sourcetermd4xd4 J [ f ( V V V ) Š f (0, V V )]tode “nea Z( J V V )wew illhavea JV coup ling,andcouplingtohigherordertermsin V and V V (whichareirrelevantwhencomputingtheS-matrix),butnolinearcoupling J V V .Whenweset V V =0weobta inthe conventional Z ( J ).Asin(6.5.8)wemakeachangeofvariablesofintegrationwhich involvesthe inverse ofthefunction f .U nderthistransformati ontheinvariantgauge actiongoestoitsusualformintermsof V ,the gauge“xingandghosttermschangeina complicated,butphysicallyirrelevantmanner,andthecouplingtothesourcebecomes simply JV Š J V V .Furthe rmore,theJacobianofthetrans formationis1(se es ec.3.8.b). We no wh av et he sa meformasinordinaryYang-Millstheory,andweconcludethatthe background“eldfunctionalcomputedbysetting J =0, i.e.,evaluating graphswithonly internalquantumlines,doesgivetheusualeectiveactionasafunctionofthebackground“eld,albeitinanunconventionalgauge.Thereforeallphysicallyrelevantquantumcorr ectionscanbeobtainedfromthebackground“eldfunctional.Wenowdiscuss howtoevaluateitinp erturbationtheory. c.CovariantFeynmanrules Weconsid er“rstcontributionsfromonlythequantumgauge“eld V .The eectiveLagrangianis Š1 2 g2 tr [( eŠ V eV) 2( eŠ V eV)+ V ( 2 2+ 2 2) V ].(6.5 .35) Allthe dependenceonthebackground“eldsisthroughtheconnectioncoecientsand

PAGE 399

6.5.Thebackground“eldmethod383neverthrough thegauge“eldsthemselves. Thequadraticactionhastheform Š1 2 g2 trV [ Š 2 Š i W W +1 2 i ( W W)+ 2 2+ 2 2] V .(6. 5.36) Usingthecommutationrelations [ €, €]= C€€W W,(6. 5.37) thiscanberewrittenas Š1 2 g2 trV [ Š i W W Š i W W€ €] V ,(6. 5.38) where =1 2 a aisthebackgroundcovariantdAlembertianand Wisthebackground “eldstrength.Introducingconnectioncoecients(dependingonthebackground “elds)by A= DAŠ i A,wecanseparate outafreekineticterm,andinteractionswith th eb ac kg round: Š1 2 g2 trV [ 0Š i a aŠ1 2 i ( a a) Š1 2 a aŠ i W( DŠ i ) Š i W€( D€Š i €)] V .(6. 5.39) Thisexpressionissucientfordoingone-loopcalculationsusingconventionalpropagatorsforrea lscalarsup er“eldsandtheusual D -manipulations.Sincetheinteraction withthebackground“eldsisatmostlinearin D s,andatleastfour D saren eededina loop, the“rstnonvanishingone-loopcontribut ionfromVisinthefour-pointfunction. Self-interactionsforcomputinghigher-loopcontributionscanbeobtainedfromthe higher-orderin V termsintheLagrangian(6.5.35). Wenowt urntocontributionsfrom(fully)cov ariantlychiralphysicalsuper“elds andbackgroundcovariantlyc hiralghostsuper“elds.Inprinciplewehavetosolvethe chira lityconstraint €=0(bywr iting= e 0intermsofanordinarychiralsuper“eld),butthisintroducesexplicitdependenceonthebackgroundgaugeprepotentials whichwewishtoavoidifpossible.Instead,wereexaminethederivationoftheFeynman rulesforchiralsuper“elds.

PAGE 400

3846.QUANTUMGLOBALSUPERFIELDSWeconsid erthegeneratingfunctionaloftheform Z ( j j )= ID ID eS +(d4xd2 j + h c .),(6. 5.40) whereand j arecovariantlychiral €= €j =0.Forthet imebeingwe neednot specifywhetherthesearefullcovariantderivativesorjustbackgroundcovariantderivatives.Inprin ciplewede“neID astheintegr aloverthecorrespondingchiral-representation“eld0(antichiralfor integration ), butinpracticewesimplyde“neitby theGaussianintegral ID ed4xd21 2 2=1.(6. 5.41) Additional “eldsmaybepresentbutweneednotindicatethemexplicitly. We de“ne covariantfunctionaldierentiationby ( z ) ( z) = 28( z Š z).(6.5 .42) Thisformcanbederivedfrom(3.8.3),orbywriting= 2,intermsofageneral super“eld,andcovariantizing(3.8.13).Manifestlycovariantrulesforchiralsuper“elds cannowbefoundbyadirectcovariantizationoftheusualmethod.Thecovariantizationoftheidentity D2D2= 0(where 0denotesnowthefre edA lembertian) b ecomes 22= +, += Š iWŠ1 2 i ( W),(6.5 .43) withthecovariant .Weconsider“rstt hemasslesscase. Weca rryoutthefunctionalintegrationoverbyseparatingouttheinteraction termsanddoingtheGaussianintegral,andobtain Z = eSint( j j )eŠd4xd4 j + Š 1j,(6. 5.44) whereisthefunctionaldeterminant = ID ID eS0, S0= d4xd4 .(6. 5.45)

PAGE 401

6.5.Thebackground“eldmethod385Ingeneraltheaboveexpressionmuststillbeintegratedoverotherquantum“eldsthat maybepresent. Beforeweevaluate,whichwillgiveaseparate,one-loopcontributiontothe eectiveactionfromthe“eld,weexaminetherestofthecontributions.Theexpressionfor Z isidenticaltotheonein(6.3.14),exceptforthepresenceofcovariantderivativesandco variantlychi ralsources,andthefactor.Theperturbationexpansiontakes thesameform,exceptthatfromthefunctionaldierentiationwegetfactorsof 2or 2actingonchiralandantichirallines.Thepropagatorsaregivenby Š + Š 1, butinaperturbativecalculationweseparate +intoafreepart,whichleadsto pŠ 2pr opagators, andtheremainder,whichgivesadditionalinteractionvertices.However,atnostagedo ween counterexp licitgauge“elds,onlyconnectionsand“eldstrengths.(Theexplicit dependenceonthequantumgauge“eldswillbeneededonlywhenwefunctionallyintegrateoverthem.) Wenowevaluate.I tgivesthecom pleteone-loopcontributionofthechiral super“eldtographswithonlyexternal V linesandcouldbeevaluatedbyusingstandard Feynmanrules, butthiswewishtoavoid.Thisturnsouttobepossibleonlyfor real representationsoftheYang-Millsgroup.Ofcourse,realrepresentationsarefrequently theonesofinterest:e.g.,theYang-Millsghostsareintheadjointrepresentation,which isalwaysreal.Wethereforeconsider“rstth ecaseofrealreprese ntations,a ndreturn latertothecomplicationscausedbycomplexrepresentations.Wearestillconsidering thema sslesscase. Theaction S0leadstotheequationsofmotion(inthepresenceofsources) O + j j =0, O O 0 2 20 .(6. 5.46) Wede “nean actionwhoseequationsofmotionare O O2 Š j j =0, O O2= 220 0 2 2 ,(6. 5.47) intermsofthesquareof O O .Thisactio nisgivenby S 0+ S 0,where S 0= d4xd21 2 += d4xd41 2 2.(6.5 .48) Intermsofitwecanwritethefunctionalintegral

PAGE 402

3866.QUANTUMGLOBALSUPERFIELDS2= ID ID eS 0+ S 0= | ID eS o|2=( ID eS o)2.(6. 5.49) Wehave usedthefactthat S 0anditshermitianconjugatecontributeequallyto,as canbeseen,forexample,byexaminingtheresultingFeynmanrulesbelow.(Thisprocedur ei sa na logoustothedoublingtrickinQED,wheretheanalogueof O O is €and of +is C + f,with ftheel ectromagnetic“eldstrength.) Wenowintegrate S 0bysepara tingout D2D2from 22andtreating ( 22Š D2D2)asaninterac tionterm.Theresultis =ed4xd21 2 j [ 22Š D2D2] j eŠd4xd21 2 j oŠ 1j.(6. 5.50) (Writinginstead 22= D2eŠ VD2eVinthechiralrepresentationgivestherulesforthe one-loopexpressionintheusualnoncovariantformalism.)Therefore,acalculationofthe one-loopcontributionconsistsinevaluatinggraphswithpropagators pŠ 24( Š )and vert i ces 22Š D2D2givingrise t oastring ... [ 22Š D2D2]i4( iŠ i +1)[ 22Š D2D2]i +1... ,(6. 5.51) withd4iintegralsateachvertexandoneloop-momentumintegral.Wecarryoutthe evaluation inthechiralrep resent ation, sothat €= D€.Wecon centrateonthe i vertex,andfromthenextvertexwetemporarilytransferthe D2= 2factor acrossthe -function.Wenowusetheidentity( 22Š D2D2) D2=( +Š 0) D2(inthechiral representation).Havingperformedthismaneuverwereturnthe D2toitsoriginalplace, andproceedtomanipulatethenextvertexinthesameway.Thisprocedurecanbecarriedoutatallverticesbutone,whichretainsitsoriginalform.Theresultingrulesfor theevaluationofare,withtheusualpropagator, onevertex: D2( 2Š D2),(6.5 .52) othervertices: +Š 0,(6. 5.53) withthecovariantderivativesinchiralrepresentation.Nowonlyonevertexcontributes any D sandasaconsequencetheevaluationofone-loopgraphcontributionsfromchiral

PAGE 403

6.5.Thebackground“eldmethod387super“eldsisconsiderablysimpli“ed.Higherloopsaregivenbytherestoftheexpressionin(6 .5.44). Uptonowwehavenotsp eci“edwhetherthe sarefullorbackgroundcovariant derivatives.Ifbothquantumandbackgroundgauge“eldsarepresent,itismoreconvenienttoca rryouttheaboveprocedureatanearlystage,beforewewrite= eV, i.e.,workwithfullycovariantlychiralsuper“elds(butnotfortheghosts,whichareonly backgroundcovariantlychiral).Theresultofthecalculationisexpressibleintermsof thefullcovariant derivatives,andonlyatthatstage,havingintegratedoutthechiral super“elds,doweneedtomakethebackgro undquantumsplitting onthegauge“elds. Thedoublingtrickcannotbeappliedcovariantlywhenthescalarmultipletisin ac omplexrepresentationoftheYang-Millsgroup.Ifwewritethecovariantlychiralin termsofordinarychiral0( D€0=0)inthev ectorrepresentation = e 0,(6. 5.54) wehave = e 0,(6. 5.55) and 2 = eŠ D2e*e 0,(6. 5.56) isnotint hesamerepresentationas(doesnotsatisfythesamechiralitycondition) exceptwhentherepresentationisreal(inwhichcase*= Š ).Therefore,theoperator O O in(6.5.46)cannotbesquared,sinceitisnotrepresentation-preserving.Asa result,wemustuserules atoneloop whicharenotexpressedman ifestlyint ermsofconnectionsA, butinvo lveexplicitgauge“elds. In(6.5.45)weexpressintermsof0,andintro duceordinarychiralsources j0( D€j0=0).Wehaveinsteadof(6. 5.46)thefollowingequationsofmotioninthepresen ce ofexternalsuper-Yang-Mills: O O 0 0 + j0 j0 =0, O O = 0 D2eV D2eV *0 .(6. 5.57)

PAGE 404

3886.QUANTUMGLOBALSUPERFIELDSThenoncovariantobject O O canbesquared,sinceitpreserves( D€-)chir ality: O O2= D2eV *D2eV0 0 D2eV D2eV .(6. 5.58) Theaction S0 obtainedfrom O O2againgivesacontributionequaltothatofitshermitianconjugate.Asin(6.5.50)weseparatea D2from eV *D2eVandtreattherestasan interaction.Thepropagatorisasbefore,butthevertexisnow D2( eV *D2eVŠ D2).(6.5 .59) Notethat,for real representations, V *= Š V = Š V ,sothisver texisjust D2( 2Š D2), andtherulesof(6.5.52,53)canbeobtained.Ingeneral,foragroupcontainingfactors forwhichisinarealrepr esentation,wecanwrite V = V1+ V2,where V1*= Š V1, but V2* = Š V2([ V1, V2]=0),andwrit ethevertexas D2( eV2*1 2eV2Š D2), 1 = eŠ V1DeV1= D+1 .(6. 5.60) Then V1appearsintherulesonlyas1 Awhile V2appearsexplicitly. Thenetresultisthattheeectiveactionisexpressedmanifestlyintermsof Afo rY ang-Millsfactorsthatoccurcoupledonlytorealrepresentations,and always for hi gher-loopcontributions.However,atoneloop,andonlyforYang-Millsfactorscoupled tocomplexr epresentations,thecontributionmustbecalculatedinawaywherethe covarianceisnotmanifest. Ourmethodscanalsobeappliedtomassivechiralsuper“elds.Inthatcasethe term j + Š 1j of(6.5.44)isreplacedwith j 1 +Š m2 j +1 2 [ j m 2 +( +Š m2) j + h c .],(6. 5.61) adir ectcovariantizationoftheresult(6.3.13) .Inperfo rmingthedoublingtrick,weuse ( m )=( Š m ).Wereplac etheki neticope rator O O ( m )= m 2 2m (6.5.62) by

PAGE 405

6.5.Thebackground“eldmethod389O O ( Š m ) O O ( m )= 22Š m20 0 2 2Š m2 .(6. 5.63) Aftermakingthecorresp o ndingreplacementsin(6.5.50),weobtain Š ( 0Š m2)Š 1for thepropagator,whiletheverticesarethesameasbefore. Wesu mmarizetheprocedureforevaluatingtheeectiveactioninthebackground “eldforma lism:One-loopgraphswithonlyexternalgauge“eldlinesareobtainedfrom thequadraticLagrangianfor V in(6.5.39),andbyevaluatingforeachchiralsuper“eld.Higherloopsareobtainedwithverti cesinvo lvinginteractionsofthequantum “elds,eitherfromthehighe r-orderexpansionofthe V Lagrangian,orfromtheperturbativeevaluationof(6.5.44).Therulesforloopswith(some)externalchirallinesfollow fr om(6.5.44)andaretheusualonesbutwithcovariantpropagatorsandvertices. d.Examples Wenowpresent someresults.Webeginbyinvestigatingtheradiativegeneration ofaFayet-Iliopoulosterm(4.3.3)foranabeliangauge“eld,andconsider“rsttheonelooptadpolegraphwithachiral“eldinside(Fig.6.5.1). Fig.6.5.1 Ifthechiral“eldismassless,wecandropitwhenusingdimensionalregularization.In thema ssivecase,accordingtotheusualrules,itwouldseemtocontributebutgauge invariancerequiresthattherebetwochiral“eldsofoppositecharges,andtheircontributionscancel.Therefore,gauge-invariantPauli-Villarsregulatorscannotcontributeto thisgrapheither.Asaresult,thegraphmu stbede“nedtovanishinthemasslesscase inanygauge-invariantsupersymmetricregu larizationpro cedure,dimensionalorPauliVillars. However,inthecaseofrealr epresentations,withthecovariantrules,thereisno needforsuchanargument.Atthevertexwehaveacontribution(see(6.5.52);to

PAGE 406

3906.QUANTUMGLOBALSUPERFIELDSlineari zedorderweneednotdistinguishbetweenfullandbackgroundderivatives) D2( 2Š D2)= D2[ Š i DŠ i (1 2 D)](6.5.64) (tolineari zedorder),andwedonothavetwo D sintheloop.Thus,f orrealrepresentations,thegraph vanishes justby D algebra. Actually,eventhiscalculationisunnecessary,becausewecangiveasimpleproof thattheFayet-Iliopoulostermisnevergener atedinperturbationtheoryforrealrepresentations:Thistermcorrespondstoacontributiontotheeectiveactionoftheformd4xd4 V .Howev er,accordingtoourcovariantFeynmanrules,a V neverappearsata vertex,onlyco nnectionsand“eldstrengths,sothatnosuchtermcanbeproduced. Wenextcalc ulatetheone-loopcontributiontothe V self-energyfroma ma ssive chiralsuper“eldi nareal representation.Ifweusetheordinarynon-backgroundrules therearethreegraphstocompute(becausewehavemassivepropagators),andthey havetobecombinedtoexhibitthegaugeinvarianceofthe“nalresult.Also,thereare some D manipulationstobeperformed.Here,thereisessentiallynothingtodo.We consideragaintherelevantgraph,showninFig.6.5.2. Fig.6.5.2 On ev ertexisgivenby(6.5.64),whileattheothervertexwehave(againtolinearorder) +Š 0=[ Š i a aŠ i (1 2 a a)]+[ Š iWDŠ i (1 2 DW)].(6.5.65) Sincewerequiretwo D sandtwo D sintheloop,th ereisauniquetermwithcontributions Š i D2Dfrom onevertex,and Š iWDfromtheo ther.The answeris1 4 ktr d4 W( p )( Š p ) d4k (2 )4 1 ( k2+ m2)(( k + p )2+ m2) .(6. 5.66) (Thefactor k wasde “nedin(6. 4.5).)Weobservethatwiththecovariantrulesthereis noseagull-tadpolecontribution.Thiswouldhavetocomefromthenonlinearpartofthe

PAGE 407

6.5.Thebackground“eldmethod391one-vertexformula(6.5.64) butitdoesnothaveenough D stocont ribute. Wecanobt ainthe V se lfenergyinnonabelianYang-Millstheorywithoutanycalculation.Accordingtothedi scussionfollowing (6.5.39),ifwelookatgraphswithtwo externallines,therearenotenough D sintheloop.Theonlysourceof D sarethe W terms,andea chfact orof W bringswith itjustone D .Thusthewholec ontri butionto theselfenergycomesfromthethreechiralghosts,andthereforeweobtainananswer whichisjust Š 3timesthatfr omthechiral“eldw econsideredabove (withthe “elds nowbeingbackground).Thisisageneralfeature:Asalreadymentioned, V sstartcontributingatonelooponlybe gi nningwiththefour-pointfunction.Toseetheimplicationsofthisremark,wegivenowacomputationoftheone-loop,four-particleS-matrix in N =4 Ya ng -M illstheory. Theone-loopcontributionswithexternal V lines comefroma V loop,fromthe threechiral“elds,orfromthe threechiralghosts. B ecauseofthestatisticsoftheghosts thechiralcontributionscancelexactly.Thisistrueforagraphwithanarbitrarynumber ofexternalvectorlines.Inparticular,itimpliesthatthetwo-andthree-pointfunctions areidenticallyzeroattheone-looplevel.Therefore,weneedonlycomputethe V -loop contribution.Wehavejustaboxdiagram,withfactors Š i ( WD+ W€ D€)ateach vertex,andwemust k eeptermswi thtwo D sandtwo D s.The D -algeb raistrivial, andweobtainforthefourV amp litude =1 2 tr d4p1... d4p4(2 )16 d4 (2 )4 ( pi) Go( p1... p4) [ W( p1) W( p2) W€( p3) W€( p4) Š1 2 W( p1) W€( p2) W( p3) W€( p4)],(6.5.67) where G0isthecontributionfromthefour-pointscalarboxdiagram G0= d4k (2 )4 1 k2( k Š p1)2( k Š p1Š p2)2( k + p4)2 .(6. 5.68) Thetraceisoverinternalsymmetryindices,andallthesuper“eldshavethesame argument.

PAGE 408

3926.QUANTUMGLOBALSUPERFIELDSThisresultisvalido-shellandisultraviolet“nite.On-shellitgivestheone-loop S-matrix,butitisinfrareddivergent.(ToobtaintheS-matrixwedropthe piintegrals andsumover pipermutat ions .The W sgivekinematicalfactorsproportionalto momentaandpolarizations).Thesimplicityofthecalculationisdueinlargeparttothe absenceofc hiralsuper“eldcontributions.Intheparticulargaugeweareusingthereare noself-energyortrianglegraphstoconsider,andthewholeS-matrixisgivenbythebox graph.Wewillencounterasimilarsituationinsupergravity(seesec.7.8).

PAGE 409

6.6.Regularization3936.6.Regularization a.General Theperturbativerenormalizationofsuper“eldtheoriesisinprinciplenodierent fromthatofordinary“eldtheories.Weneedaprocedureforregularizingdivergentintegrals,andaprescriptionforsubtractingultravioletdivergences.Wemustdealwith renormalizablenon-polynomialLagrangians(e.g.,supersymmetricYang-Millsinasupersymmetricgauge),andusethesupersymmetryWardidentitiesinthecourseofrenormalizationor,alternatively,us earegula rizationschemethatmanifestlypreservessupersymmetry.Wedonothavemuchtosayaboutrenormalization.Forrenormalizable models,weintroducerenormalizationconstantsintheclassicalactionandusethemto cancel,orderbyorderinperturbationtheory,thedivergencesweencounter. Aswehavealreadymentioned,insupersy mmetrictheoriestherearefewerdivergencespresentthaninnonsupersymmetricones.Ingeneral,thedegreeofdivergenceof anysupergraphcanbedetermi nedbythedimensionalargumentofsec.6.3orbythefollowingpowercountingrules:Inrenormalizab letheori esallsupersymmetricverticeshave four D s(eitherfromthe D2and D2ofchiralsuper“elds,orthe D, D2Dofgauge super“elds).Innonrenormalizabletheoriesthereareadditionalfactors,butwe“rstconsidertherenormalizableca se.Allvert i ceshavea d4 factor. Wecons ideran L -loopgraphwith V verti ces, P pr opagatorsofwhich C areor m assivechiralpropagators,and E external linesofwhich Ecarechiralora ntichi ral. Fromthev erticesthereare V factorsof D2 D2 q2.T he pr opagatorsproduce qŠ 2factors, butor p ropagatorsgiveanadditional D2qŠ 2 qŠ 1factor.Eachloopproducesa d4q q4anduse supa D2 D2 q2factorfrom D2 D2 = .Eachexterna lchira l line accounts forone D2 q missingatthecorrespondingvertex.The super“cial degr eeof divergenceis(using L Š P + V =1) D=4 L Š 2 L Š 2 P +2 V Š C Š Ec=2 Š C Š EcTherefore,forgraphswithonlyexternal V sthesuper“cialdegreeofdivergenceistwo (butgaugeinvarianceimprovesthis),andzeroiftherearetwoexternalchirallines.Furthermore,iftheexternallinesareallchiralanadditional D2mustcomeouto ftheloop:d4 n=0soonemusth aveatleastd4 n Š 1D2foran onzeroresult.Thereforethe

PAGE 410

3946.QUANTUMGLOBALSUPERFIELDSconvergenceisimprovedandonlygraphswithoneandone linemaybedivergent. Forreno rmalizabletheoriesweobtaintheresultsofsec.6.3.Insupergravityonthe otherhand,whereateachvertexwehavetheequivalentofsixfactorsof D ,the degree ofdivergenceofagraphis2 Š C Š Ec+ V .Thisres ultcan alsobeobtainedbya dimensionalargument(seesec.7.7). Regularizati onisanimportantpartofanyrenormalizationscheme.Althoughin principleanyregularizationmaybeused,inpracticeitispreferabletouseaschemethat iscomputationallysimpleandmaintainsasmanypropertiesoftheclassicaltheoryas po ssible.Thissimpli“estheren ormalizationprocedure,whichmustnotonlymakethe quantumtheory “nitebysubtractionofdivergences,butalsomustmaintainunitarityby po ssiblesubtractionofadditional“nitequanti ties.Fortheorieswith(globalorlocal) symmetries,suchadditionalsubtractionsaredeterminedbytherequirementthatrenormalizedGreenfunctionssatisfyWard-Takahashiidentities,andinthecaseofnonabelian gaugetheories,Slavnov-Tayloridentities.However,itispreferabletoemployaregularizationschemethatmanifestlypreservesallsymmetries;thisallowsarenormalization schemethatrequiresthesubtractionofonlythedivergentparts,sotheapplicationof Ward-Tak ahashi-Slavnov-Tayloridentitiesisunnecessary. b.Dimensionalreduction Dimensionalregularizationhasproven tobethemostpracticalmethodofregularizationincomp onent“eldtheoriesbecauseithasthreeproperties:(1)Itmanifestly preserves(almost)allsymmetries,thusbypassingtheWard-TakahashiorSlavnov-Taylor identities;(2)theregularizedgraphsarenohardertocalculatethantheunregularized onesandrequireonlyoneregulator,thedimensionalityofspacetime;(3)renormalization isasimpleprocedure,requiringonlyminimal subtraction.Theprescriptionfordimensionalregularizationis:(1)Writetheactioninaformwhichisvalidforanydimension Dofspa cetime;(2)calculateFeynmangraphsfo rmallyinarbitraryspacetimedimensions,integratingoverDcomponentsofeachloopmomentum,giving“eldsofany LorentzrepresentationthenumberofcomponentsappropriatetothatvalueofD,and perfo rminganyalgebraicmanipulationsthatwouldbevalidfor“niteintegrals(i.e.,performingtheintegralindimensionsDforwhi chit is“niteandanalyticallycontinuingin D);(3)renormalizebysubtractingf romdivergentcontributions(asD 4)onlytheir

PAGE 411

6.6.Regularization395polepart s(prop ortionalto1 D Š 4 ),andnoadditional “niteparts,“rstinsubdivergences andthe nforthesuper “cialdivergence(inamputatedone-particle-irreduciblegraphs, i.e.,theeectiveaction). Thisprocedurehastwodrawbacks:(1)Itmustbesupplementedbyaprescription forhandlingsymmetrieswhichdonotc ommutewithparity,i.e.,involving 5or a b c d, andinparticularforcorrectlyobtainingchiralanomaliesforthosecaseswheretheyare present.(2)Itdoesnotmaintainsupersymme try:Theprescriptio nfor giving “eldsof anyLorentzrepresentationthenumberofc omponentsappropriat etoDdi mensionsdoes notkeepFermiandBosedegreesoffreedombalanced.Amodi“cationoftheprescription,whichwouldcontinueafour-dimensionaltheorytoatheorysupersymmetricinD dimensions,isnotpossibleeither.Forexa mple,ifDisincreasedpast10,aglobally supersymmetrictheorywouldhavetobecon tinuedtoalocallysupersymmetricone.We wouldhav espi ns 2b ecausethenumberofsupersymmetrygeneratorsincreaseswith incr easingD.Wenowdescribeamodi“cationofdimensionalregularizationintendedto preservesupersymmetry,andreturnlatertothe“rstdiculty. Sincethechangeinstructureofsupersymmetrictheoriesasthenumberofsupersymmetrygenerators(4 N )isincreasedis notuniform,weconsiderkeepingthisnumber “xed.Forregularizingultr avio letdivergencesitisonlynecessarytocontinueto lower dimensions,anditisthenpossibletokeepthenumberofsupersymmetrygenerators “xedattheirfour-dimensionalvalue.Ingeneral,ourprescriptionforcontinuingtolower dimensionsistocontinue only thedimensionalityofspacetime,butkeeptherangeofall Lorentzindicesthesame,asiftheywereinternalsymmetryindices.AswereduceD,an N -extendedsupersymmetryc anbereinterpretedasan N-extendedsupersymmetry, N> N .Forex ample, N =1inD=4d imensionscanberegardedas N =2inD=3 dimensions.Inthisway,thenumberofbosonicandfermionicvariablesstayequal.Such acontinu ationiscalleddimensionalreduction.Hereweconsidercontinuationonly fromD= 4toD < 4. OurrulesforapplyingdimensionalreductiontoregularizecomponentFeynman graphsare:(1)Allindiceson the“elds,a ndcorrespondingmatrices,comingfromthe actionaretreatedas4-dimensionalindices;(2) asinordina rydimensionalregularization, allmomentumintegralsareintegratedoverD-componentmomenta,andallresulting Kronecker sareD-dimensional;(3)sinceD < 4always,any4 -dimensionalKronecker

PAGE 412

3966.QUANTUMGLOBALSUPERFIELDScontractedwithamomentumequalsthatmomentum a bp b= p a,andany4-d imensional contractedwithaD-dimensionalonegivestheD-dimensionalone a b b c= a c,where thei ndicatesD-dimensionalquantities.The“rstruleisnecessarytopreservesupersymmetry,sinceitkeepsthenumberofcomponentsthesame;thesecondrulepreserves alltheusefulpropertiesofdimensionalregularization(e.g.,gaugeinvariance);thelast rulede“nestheregularizationasdimensionalreduction. Unlikeinordinarydimensionalregularization,both4-dimensionalandD-dimensionalquantitiesoccur.Therefore,whenappliedtocomponents,dimensionalreduction requireshandlingmoretypesof“elds:e.g.,a4-dimensionalvectorbecomesaD-dimensionalvectorand4-Dscalars.Thiscancause dicultiesinnonsupersymmetrictheories, sincealargervarietyofdivergencescanoccur,butinsupersymmetrictheoriessupersymmetryallowsonlydivergencescontainingthefullsetof4-dimensional“elds.Forexample,([ A a, A b])2 ZA 2([ A a, A b])2+ ZAZ([ A a, i])2+ Z 2([ i, j])2, butinsupersymmetrictheoriestheD-dimensional extended supersymmetrythatresultsfromreducing 4-dime nsionalsupersymmetryensuresthat ZA= Z.(Intheori eswithonlyscalarsand spinors,theonlydierencefromusualdimensi onalregularizationisinthenormalization ofthespinortrace,andhencetheseproblemsdonotarise.) Whenappliedtosuper“elds,dimensionalreductionistheuniqueformofdimensionalregularizationthatallowsthenaivealgebraicmanipulationofthe4-dimensional spinorderi vatives Dindivergentasw ellasconvergentsupergraphs.Thisrequirement leadstothefollowing de“nitionofregularizationbydimensionalreductiononsupergraphs:(1)PerformallalgebraasinD=4,obtainingaformwhereall -integrationhas b eenperformedi.e.,thegraphisexpressedasanintegraloverasingle d4 N ofpro ducts ofsuper“eldsofvariousmomentatimesanordinarymomentumintegral andistherefore man ifestlysupersymmetric; (2)performtheremainingmo mentumintegralinD-dimensions.Instep(1),weusethe4-dimensionalidentity p€p€= p2(r ecall p21 2 p ap a(3.1.16,18)).NoD-dime nsionalKroneckerdeltasariseatthisstage.Instep(2),symmetricintegrationsgenerateD-dimensionalKroneckerdeltas,e.g., p€p€2 D p2 € €,(6. 6.1) where € €= a bhastheproperties

PAGE 413

6.6.Regularization397 € € =D 2 €€, € € € €= € €,(6. 6.2) andspi norindicesarestillmanipulatedasin4dimensions. Ontheotherhand,adimensi onalregularizationscheme,whichlikeordinary dimensionalregularizationscontinuedspinorindices(includingtheoneon )toDdimensionaloneswith k =2D 2 Š 1components,wouldha veproble ms:e.g.,dk dk = Dk Dkwouldnolon gerhavewellde“nedstatistics,andwouldintroduce higherderivatives(for k > 2)intotheaction,requiringsomenonminimalsubtraction scheme(suchasanalyticregularization). Unfortunately,althoughitpreservessupers ymmetry,regularizationbydimensional reductionleadstoambiguities.Forexample,letusconsidertheexpression [ a fp bq cr ds e ]=0(6.6 .3) whichvanishesinD < 5b ecauseitistotallyantisymmetr icin5indiceswhichtakeless than5values.(Thisalsofollowsifwewritethevectorindicesintermsofspinorindices anduse4-dimensio nalspinormanipulations.)Ifwenowcontractwith f aweobtain (DŠ 4) p[ aq br cs d ]=0(6.6 .4) Since p[ aq br cs d ]doesnotvanishinD=4,andwemustrequireitnottovanishinD =4to avoidgen eratingarbitrarycoecientsforsuchtermsuponcontinuation,wehavean inconsistency.Thiscanalsobeviewedasanambiguity:Byevaluatingasupergraphin twodier entways,wemayobtainresultsthatdierbythelefthandsideof(6.6.4).If thesupergraphisconvergent,thisambiguitydisappearsinthelimitD 4.However,if itisdivergent, a “nitedierencebetweenthetwowaysofevaluatingthegraphmay result. Thesameambiguityispresentincomponenttheories(supersymmetricorotherwise)thathavechiralanomalies,wherethecorrespondingexpressionis 0= f a[tr ( 5 f ap / q / r / s /)+ tr ( 5 ap / q / r / s / f)]=(DŠ 4) tr ( 5p / q / r / s /)(6.6.5) Toderivethis result,wehaveusedtheidentities

PAGE 414

3986.QUANTUMGLOBALSUPERFIELDS{ a, b} =2 a b, { 5, a} =0, a a=D,(6. 6.6) whichareequivalenttotheprescription(6.6.2)combinedwiththe4-dimensionalspinor algebra. Thisproblemarisesbecausewehaverequiredthatourregularizationrespectslocal gaugeinvariance:Whenweconsidertheorieswithaxialcouplings,wemustuseaprescriptionsuchas(6.6.6)thatrespectschirali nvarian ce.Thismakesitimpossibletocalculate(unambiguously)anomaliesthatshouldbethere.Modi“cationsof(6.6.6)exist thatgivethecorrectanomalies,butunfortunately,thesealsogivespuriousanomalies thatmustbeeliminatedbyusingWard-Takahashi-Slavnov-Tayloridentities,whichis justwhatwe weretrying toavoid. c.Othermethods Theredoexistalternativeschemesforsupersymmetricregularization,atleastfor specialsystems.Fortheoriesthatallowtheintroductionofmassterms,wecanuse su pe rsymmetricPauli-Villarsregularization.Thisisthecase,forexample,intheWessZuminomodel(ormodelswithseveralchiralscalarsuper“elds)whereonecanworkwith theregularizedLagrangian ILR= d8z ( + ci ii) + d6z [1 2 ( m 2+ Mi2 i)+1 3! (+ i)3]+ h c .,(6.6 .7) orinmodelsofchiralmultipletscoupledtoaYang-Millsmultiplet, forregularizingchiral lo ops, ILR= d8z ( eV+ ci ieVi) +1 2 d6z ( m 2+ Mi2 i)+ h c ..(6.6 .8) Heretheiarechiralregulator“elds,andthelimit Mi istobetakenattheend ofthecalculations.

PAGE 415

6.6.Regularization399Fo rs up er sy mmetricYang-Millstheorieswecanusehigherderivativeregularization.Forexample,theusualcovariantactioncanbemodi“edtoread tr d4xd21 2 W(1+ Š 2 +) W,(6. 6.9) andsimilarmodi“cationscanbemadeinthegauge“xingandghosttermstoproduce pr opagatorswith kŠ 4behavior forl arge k .A si no rdinaryYang-Mills,allmultiloopdiagr am sa resuper“ciallyconvergent.Howeverthisproceduremustbesupplementedbya dierentoneloopregularization. St raightforwardPauli-VillarsregularizationcannotbeusedforYang-Millstheories b ecauseitdestroysgaugeinvariance.However,inthebackground“eldmethoditseems pe rfectlyacceptable.Inthismethodtheeectiveactionismanifestlycovariant,and sincethequantum“eldstransformcovariantly(rotateliketensors),onecanaddamass term,andthereforemassiveregulators,wit houtdestroyingthe gaugeinvariance.What isnotentirelyclearisthatthiscanbedoneingeneralinamanifestlyBRSinvariant way, i.e.,withoutdestroyingtheunitarityoftheS-matrix.Butthereseemtobeno problemsattheone-looplevel,sothatacombinationofhigher-derivativeandone-loop Pa uli-Villarsregularizationis ap erfectlyacceptableprocedur ef ormaintainingmanifest supersymmetryinthebackground“eldformalism.Arelatedprocedureforone-loop graphscanbeusede veninano n-backgroundformalism. Anotherregularizationp rocedure,whichhasbeenusedforone-loopgraphs,is pointsp litting.We“rstcon siderthenonsupersymmetriccase.Theregularizationis appliedtoone-loopgraphsbyexpressingthemastracesofpropagatorsinexternal“elds, andseparatingthecoincidentendpointsofthepropagator: d4xG ( x x ) d4xG ( x x + ),(6.6 .10) where is an i n“nitesimalregulator.WritingtheGreenfunction G asafunctionalaverageof“elds, G ( x y )= < ( x ) ( y ) > = ID eS ( ) ( x ) ( y ),(6.6 .11) wecanexpressthepoint spli ttinginthefollowingformintermsofanexplicitoperator: G ( x x + )= < ( x ) e ( x ) > .(6. 6.12))

PAGE 416

4006.QUANTUMGLOBALSUPERFIELDSThisproceduremustbemodi“edinordertopreservegaugeinvariances.However,for theform(6.6.12),gaugecovariantizationistrivial:Wereplacethepartialderivative withacovariantone = Š iA : < ( x ) e ( x ) > .(6. 6.13) Thisisequivalenttotheform: < ( x ) IP { exp [ Š ix + x dx A ( x)] } ( x + ) > ,(6. 6.14) wherethelineintegralisalongastraightlineand IP meanspathordering.Theequivalencecanbe proven,evenfor“nite ,bywriting theexponentialin(6.6.13)asaproduct of exponentialsofin“nitesimals,andthenreorderingallthe stotheright(which translatesthe A s).Incalculations,itismoreconve nienttohavethemanifestlycovariantform(6.6.13)intermsof s. Thesupersymmetricgeneralizationisstraightforward:For ausethesuperspace covariantderivative.(Inprincipleonecouldalsotranslatein with butthe integrationisalready“niteanddoesntneedregularization.)Theaboveequivalencetothe path-orderedexpressionalsoholdsinsuperspace. Whilesuchregularizationme thodsmaintainsupersymmetry,theyarecumbersome. Someformofdimensionalregularizationispreferable,forallthereasonswegaveearlier. Aswehavealreadydiscussed,atthesupergraphlevelthisamountstodoing“rstallthe D -algebrainfourdimensions,andthendimensionallycontinuingthemomentumintegrals.Theresultsaremanifestlysupersymmetric,butpresumablytheinconsistencieswe havediscussedearlierwillgiverisetosomeambiguitiesintheresults.

PAGE 417

6.7.AnomaliesinYang-Millscurrents4016.7.AnomaliesinYang-Millscurrents Asanexampleofourrulesforchiralsuper“eldsandregularizationmethods,we calculatethesupersymmetricversionofthe Adler-Bell-Jackiwanomaly.Weconsidera chiralmu ltipletcoupledtobothpolarandaxialvectorgaugemultiplets.Theonlyphysicalcomponentintheanomalymultipletis theanomalyinthechiralsymmetrycurrent correspondingtophasetransformationsofthechiralsuper“elds.Thissymmetrycommuteswithsupersy mmetrytransformationsandshould bedistingu ishedfromR-symmetry,whichdoesnotcommutewithsupersy mmetry.TheR-symmetrychiralanomaly appearsinthemultipletofsuperconformala nomalies,whichalsoincludesthetraceand supersymmetrycurrentanomalies.Wewilldiscussthisinsec.7.10. Weconsid ertheactionforscalar mult ipletscoupledtovectormultiplets: S = d4xd4 eV.(6.7 .1) Forsimp licity,weassumeanevennumberofscalarmultiplets,inpairsofopposite char gewithrespecttopolarvectorgauge“elds.ThetwoWeylspinorsinsuchapair formaDiracspinor,withtheusualtransformationunderparity.Thecolumnvector isthusinarealrepresentationofthesymmetrieswhichthepolarvectorsgauge.Wecan alsoconsiderthecouplingofaxialvectorgauge“elds,withrespecttowhichthetwo membersofapairhavethesamecharge.TheDiracspinorsofthepairscoupletothese axialvectorswitha 5.Toi ndicatethesetwotypesofvectors,andthecorresponding twotypesof v ectormult iplets,weseparate V intopolarandaxialparts: V = V++ VŠ; V+= Š V+*, VŠ= VŠ*.(6.7 .2) The*referstocomplexconjugationinthesenseof(3.1.9)( V *= Vt,since V = V), but canrefertomatrixcomplexconjugationifanappropriaterepresentationischosen.This isasp ecialcaseofthesituationdiscussedafter(6.5.59).Sinceisinarealrepresentationofthegroupof V+,thepolarv ectormult iplet,wecanuseimprovedruleswith respecttoit,butmustusetheunimprovedform(atoneloop)for VŠ,theaxialvector mult iplet. Weconsid ertheone-loopgraphswithtwoexternalpolarvectorsandoneexternal axialvector.Dependingonthegroupstruct ure,theanomalyinth e lineoftheaxialvectormayormaynotcancel.Iftheaxialanomalyisnonvanishing,axialgaugeinvariance

PAGE 418

4026.QUANTUMGLOBALSUPERFIELDSislost,andtheaxialvectorcannotbeconsi deredphysical.Tobegeneral,weconsider theaxialvectorasmerelyadeviceforde“ningtheappropriateaxialcurrent.Wethen evaluatethedivergenceofthatcurrentintermsofthepolarvector“eldsappearingat theothertwolegs. Varyingtheacti onwithrespecttoanyoftheaxialvectormultiplets VŠ,weobtain theaxialcurrentsuper“elds JA= TA,(6.7 .3) wherew ehavew ritten VŠ= VŠ ATA.Gau geinvarianceoftheactionrequirestheonshellconservationlaw 2JA=0(asfo llowsfromsubstitutingthetransformationlaw eVŠ= ei eVŠeŠ i intotheactionandvaryingwithrespecttothechiralgaugeparameter).Thereforewe de“netheanomaly AAby 2JA= AA.(6. 7.4) Wew ill“ndthatth eanomaly 2J isproportionalto W2.Thecompon ent(ax ial) currentisgivenby j€=1 2 [ €, ] J | .Itsdivergen ceisthereforegivenby €j€ [ 2, 2] J | ( 2W2Š 2 W2) | a b c df a bf c d,w hi ch isthefamiliarcomponent result. Theanomalycanbecalculatedbyevaluatingthematrixelement 2< TA > ,(6. 7.5) whereisnowcovariantlychiralwithrespectto V+.Inthecalc ulationsbelowweomit thegrouptheoryfactor.Wec omputethematrixelement < > andattheendwe musttake thetraceofitsproductwith TA. Wew illevaluatetheanomalybythreemetho ds:(1)theAdler-Rosenbergmethod, (2)withaPauli-Villarsregulator,and(3)withpoint-splittingregularization. IntheAdler-Rosenbergmethodweneedonlycomputeatrianglegraphwithone axialandtwopolarvectorsatthevertices.Other,self-energy-typegraphs,withonevectoratonevertexandtwoattheotheralsoco ntri bute.However,theircontribution merelycovariantizesthatfromthetrianglegraphandtherefore,byimposinggauge in va riance,thefullresultcanbeextractedfromthisgraph.Weusethebackground-“eld formalismofsec.6.5.Attheaxialvertexwehave,from < > itself,

PAGE 419

6.7.AnomaliesinYang-Millscurrents403D2 D2,(6. 7.6) whileattheothertwoweusethelinearizedexpression(6.5.65)with on-shell, Landaugaugepolarvectors( a a= DW=0): Š i ( a a+ WD).(6.7 .7) Thesupergraphisshownin“g.6.7.1: p Š qapaŠ iWDbpbŠ iWD D2p + q p D2Fig.6.7.1 Itiseasytocheck,byintegrationbyparts,thatthe WD, WDtermsdo not contributeonshell.Thereforethe D manipulationistrivialandwemustevaluatean ordinarygraph,asinscalarQED,with1atonevertex,and Š i a aattheothers.The Feynmanint egralis d4p (2 )4 p ap bp2( p Š q )2( p + q)2 a( q ) b( q).(6.7 .8) Thisdirectlygivesthecontributiontothematrixelement < > .(Ifconsid eredasan ordinarytrianglegraph,withexternalvectorsattachedafterwards,thefactorof2from functionallydierentiatingthetwo ascorrespondstothisgraphplusthatwithcrossed v ectorlines.) Accordingtooursupersymmetricdimensio nalregularizationprescriptiontherest oftheevaluationshouldbecarriedinDdimensionsandtheothergraphsshouldbe included.However,gaugeinvariancerequiresthat a( q )enterth eresulti ntheform F a b= Š iq[ a b ],andatermoft hisformc anonlybeobtainedfromthetrianglegraph,by extrac tingfromtheintegralthe( “nite)partproportionalto q aq b.Afterintr o ducing Feynmanpar ametersandshiftingtheloopmomentumthispartcanbeeasilyextracted,

PAGE 420

4046.QUANTUMGLOBALSUPERFIELDSandweobtainforthecompletecontributionto < > Š1 4 1 (4 )2 1 ( q + q)2 F a b( q ) F a b( q)(6. 7.9) or,in x -space < > =1 4 1 (4 )2 1 ( F a bF a b).(6.7 .10) Here F a b= [ a b ]=1 2 C (€ W€ )+1 2 C€€( W )(6.7.11) andhence F a bF a b=1 2 ( ( W ))( ( W ))+ h c = Š 4 2W2+ h c .,(6.7 .12) wherewehaveusedthe“eldequations W= 2W=0. Theanomalyisgivenby 2[1 4 1 (4 )2 1 F a bF a b]= Š1 (4 )2 1 22W2= Š1 (4 )2 W2.(6. 7.13) Thismustbemultipliedbythegroupgenerator TAandatracetaken(with W= WBTB). InthePauli-Villarsregularizationmethodwecomputem lim 2( < > Š < m m> )(6. 7.14) wheremisamassiveregulator“eld.Inthisre gularizedexpressionwecanusethe equationsofmotion 2 =0, 2 m= m m(6.7.15) sothattherelevantquantitytocomputeis Šm lim m < mm> (6.7.16)

PAGE 421

6.7.AnomaliesinYang-Millscurrents405Using(6.5.44,61)wehave < mm> = j j W ( j ) = 2Š m 2 +( +Š m2) 2= Š m 2 +Š m2 .(6. 7.17) Wemustther eforecomputealoopgraph,with 2replacedby D2(thisistheonly sourceof D s),and( +Š m2)Š 1expandedinpowersofthebackground“eld(weneed atleasttwo D s): 1 +Š m2 1 0Š m2 ( Š iWD) 1 0Š m2 ( Š iWD) 1 0Š m2 + ... (6.7.18) Theonlynonzerocontributioninthe m limitcomesfromthetermexplicitly written.Inmomentumspaceitco rrespondstoatrianglegraphwith D2atonevertex and WD, WDattheothertwo.Theanomalyisthereforegivenby Šm lim m2 d4p (2 )4 1 [ p2+ m2][( p Š q )2+ m2][( p + q)2+ m2] (2 W2) = Š 1 (4 )2 W2(6.7.19) asbefore. Inthepoint-splittingmethodwecompute D2< e > = < 2e > = < e (eŠ e )2 > .(6. 7.20) Usingthecommutationrelations(4.2.90)ofthecovariantderivatives,we“ndeŠ €e = €Š €W+1 2 €€( €W)+ O ( 3).(6.7 .21) Wethen expandtheremaininge =e [1 Š i + O ( 2)],expressev erythingin termsof < ( x )e ( x ) > = < ( x ) ( x + ) > ,andtake thelimit 0.

PAGE 422

4066.QUANTUMGLOBALSUPERFIELDSHowever,wecanlimitthenumberoft ermsweneedconsiderbyevaluating < ( x ) ( x + ) > “rst.Theonlytermswhicha redivergentinthelimit 0(from po we rc ounting)aregivenbythetadpoleandpropagatorgraphs,asshownin“g.6.7.2: Š i (aaŠ iWD) D2 D2D2D2 Fig.6.7.2 (Asbefor e,wehavetwofactors D2and D2fromand ,and +Š 0Š i ( a a+ WD).)The WDtermdoes notcontribute.Thegraphsare evaluatedas d4p (2 )4 eŠ i p1 p2 = 1 (4 )2 1 2 Š a d4p (2 )4 eŠ i pp ap2( p + k )2 = 1 (4 )2 1 2 i (6. 7.22) sothat < ( x ) ( x + ) > = 1 (4 )2 1 2 [1+ i + O ( 2)].(6.7.23) Wethusk eepfactorsmultiplying < > onlyto O ( 2),andalsod ropfactors O ( 2) whichhavea D€actingon < > .There sult is D2< e > ( Š €W D€Š1 2 €W€W) < e > ( Š €W D€+ 2W2) 1 (4 )2 1 2 (1+ i ) 1 (4 )2 ( Š i 1 2 €€W D€€+ W2).(6.7 .24) (Noteinparticu larthato nlythee partoftheremaininge contributes,andonly

PAGE 423

6.7.AnomaliesinYang-Millscurrents407upto O ( )ineŠ €e .) Usingthechiralrepresentationrelation D€€= iC€€W,(6. 7.25) weobtain thesameresultasb ythe previoustwomethods.(NotethatinordinaryQED thecalculationisslightlysimplerbecausethepoint-splitpropagatorgoesonlyas Š 1.) Athigherlo ops,andalsoatoneloopforrealrepresentations,ourcovariantFeynmanrulesapply.Consequentlythetrianglegraphcontributiontotheeectiveaction dependsontheconnectionsand“eldstrengt hsandn otonthegauge“eldsthemselves and,bysimplepowercounting,itisthereforesuper“ciallyconvergent.Wedrawtwo co nc lusions:Therearenoone-loopchiralanomaliesfortheYang-Millsmultipletitself (thechiralghostsareinarealreprese ntationoft hegroup),andthereare nohigher-loop chiralanomalies foranymultiplet:Forthechiralcurrentde“nedby(6.7.3)theAdlerBard eentheoremholds. TheAdler-Bardeentheoremisnotincon”ictwiththeexistenceofhigher-order contributionstothe -function.Aswementionedatthebeginningofthissection,the chiralcu rrentthatisinthesamemultipletwiththeenergy-momentumtensorisnotthe onewehavediscussedhere,buttheR-symmetryaxialcurrent.Itisamemberofthe superc urrent de“nedbycouplingtosupergravity,andingeneralitsanomalydoesreceive higher-orderradiativecorrectionsasdoth eanoma liesofitssupersymmetricpartners.

PAGE 424

Contentsof 7.QUANTUMN=1 SUPERGRAVITY 7.1.Introduction408 7. 2. Ba ck gr o und-quantumsplitting410 a.Formalism410 b.Expandingtheaction415 7.3.Ghosts 420 a.Ghostcounting420 b.Hiddenghosts424 c.Morecompensators426 d.Choice ofgaugeparameters429 7.4.Quantization431 7.5.Supergravitysupergraphs438 a.Feynmanrules438 b.Thetransve rsegauge440 c.Linearizedexpressions441 d.Exam ples 443 7.6.CovariantFeynmanrules446 7.7.Generalpropertiesoftheeectiveaction452 a.N=1 452 b.GeneralN455 7.8.Examples460 7.9.Locallysupersymmetricdimensionalregularization469 7.10Anomalies473 a.Introduction473 b.Conformalanomalies474 c.Classicalsupercurrents480 d.Superconformalanomalies484 e.Localsupersymmetryanomalies489 f.NottheAdler-Bardeentheorem495

PAGE 426

7.QUANTUMN=1 SUPERGRAVITY 7.1.Introduction Thequantizationofsuper“eldsupergravitypresentsanumberofnewfeatures andcomplicationsthatwedi scussinthischapter.Oncethegauge-“xingprocedureand ghoststructurehave b eendeterminedFeynmanrulescanbeobtained.Usefulassupergraphsareinglobalsupersymmetry,theirpowerisawesomewhenitcomestodoingperturbationtheorycalculationsinsupergravity.Muchofthesimplicityofsuper“eldcalculationsinsupergravity,ascomparedwithco mponentcalculations,occursbecause,asin globalsupersymmetry,wedealwithobjectshavingfewerLorentzindices.Thesupergravitysuper“eldisaLorentzvector,asc omparedtot heLorentzsecond-ranktensor andvector-spinorofcomponentsupergravity.ConsequentlytheinteractionLagrangians havefewerterms,andthetensoralgebraismu chsimple r.Forexample,thethree-gravitonvertexcontains171terms,whilethecorre spondingthree-vertex insupergravityconsistsofonly27.Asaresult,itispossibletodocalculationsinsuper“eldsupergravity thathavenotevenbeenattemptedinordin aryquantumgravityorcomponentsupergravity. Theinvestigationofthedivergencestructureofquantumsupergravityisalsovery mu ch fa c ilitatedbytheuseofsuper“elds.Manycancellationsduetosupersymmetry happenautomatically,anditismucheasiertolistandunderstandthepossiblecounterterms.Incomponentcalculations,withnon-supersymmetricgauge-“xingtermsforthe gravitonandgravitino,thecorrespondingcancellationsdonotoccurautomatically,and it is mu ch morediculttodeterminewhatin“nitiesmightbepresentorabsent. Background “eldmethodsplayacrucialrolehere.Sincethecalculationsarenever veryeasy,andthealg ebradoesgetcomplicated,itisessentialtokeepsomecontrolof thegaugeinvarianceofthetheory,andthisisbestaccomplishedbyworkinginthemanifestlygauge-invariantbackground“eld formalism.Inparticular,wehavetheusual propertyth atalldivergencesaregaugeinvariant:Theformalismavoidsthenoncovariantdivergencesofgravitationaltheoriesquantizedinnonbackgroundgauges.Weshall se et ha t,justasinYang-Millstheory,thebackground“eldquantizationhasthefurther virtueofsimplifyingsomeofthevertices.Italsoallowsustoworkwithonlybackgroundcovariantderivativesratherthanprepotentials,andconsequentlyleadstosome

PAGE 427

7.1.Introduction409improvem entinthepowercountingrulesforthetheory. Thenewfeatureofthequantizationprocedureistheappearanceoflargenumbers andnewtypesofghosts,besidestheFaddeev-PopovandNielsen-Kalloshghosts.They ariseeitherbecauseofcertainconstraintst hatthegauge-“xingfunctionssatisfy,orare introducedtoremovecertainnonlocalitie sthathaveb eenproducedbythegauge-“xing procedure.Inaddition,we“ndnumerousghosts-for-ghosts. Wedisc uss“rstthebackground“eldquantizationprocedure.OrdinaryquantizationFeynmanrulescaneasilybeobtainedfromtheoneswederivebysettingthebackground “eldstozero,butinpuresupergravitythereislittleadvantagetousingthem: Ingeneral, L -loopcalculationsinordinary“eldtheorypresentaboutthesamelevelof dicultyas L +1-loopcalculati onsinthe background“eldmethod.Inthefollowing sectionswediscussthebackground-quantumsplitting,whichwepatternaftertheonein Ya ng -M illstheory,thenumberandkindsofghostsonemayencounter,andthechoiceof gauge-“xingfunction.OncewehavetheLagrangian,wecandiscussgeneralproperties oftheeectiveaction,derivesuperg raphrules,anddoloopcalculations. Weconsid eronly n = Š1 3 supergravity.Insec.7.10.eweshallarguethat N =1, n = Š1 3 theoriesareinconsistentatthequantumlevelduetoanomaliesintheWard identitiesoflo calsupersymmetry(exceptwhenpartofanextendedtheory).

PAGE 428

4107.QUANTUMN=1SUPERGRAVITY7. 2. Ba ck gr o und-quantumsplitting Wedivi dethediscussionintwoparts:thesp littingitself,andtheexpansionof theaction.AsintheYang-Millscase,thesplittingintoquantumandbackground“elds is nonlinearanddoneintermsofexponentials.Thesimplestwaytounderstanditisas anexpansionofthe(constrained)covariantderivativesintermsofunconstrainedquantumprepotentials(neededforquantization) andconstrainedbackgroundderivatives; thesimplestwaytoobtainitisbyre-solvingtheconstraints,asinsec.5.3.,butusing backgroundcovariantinsteadof”atsupersp acederivatives.Exceptforsomesmallmodi“cationsexplainedbelow,theresultscanbewrittenalmostimmediately. Theexpansionoftheactionisalgebraicallylengthy,butstraightforward.Wegive th ep ar tq uadraticinthequantum“elds,buttheprocedurecanbeextendedfor“nding higher orderterms. a.Formalism Thequantum-backgroundsplittinginsupe rgravityfollowsapatternverysimilar tothatofYang-Mills,andwesimplyrepeatithere,referringthereaderbacktosec.6.5 formotiv ationandanexplanationoftheprocedure.Westartwiththeconventional derivati ves(with de gauged U (1);seesec.5.3.b.8) A= EA MDM+(A M +A€€ M€€), [ A, B} = TAB CC+( RAB M + RAB€€ M€€);(7.2 .1) covariantunderthe(vector-representation)transformations: A = eiKAeŠ iK, K = K ; K = KMiDM+( K iM + K€€i M€€).(7.2 .2) Thesolutiontotheconstraintsexpressesthederivativesintermsoftheunconstrained prepotentials H€and andordinary”at-spacederivatives DM,inthechiral representation,asinsec.5.3.Forthetimebeingweuse achiraldensi tycomp ensator.Weachieve ourquant um-backgroundsplittingbysubstitutingintothesolutionbackgroundcovariantderivativesforthe”atderivatives.The“elds H€and arethequantum“elds,

PAGE 429

7.2.Background-quantumsplitting411whilethebackground“eldsappearimplicitlyinthebackgroundderivatives. Wetake thebackgroundcovariantderivatives Ainthev ectorrepresentation, €=( ), i a=( i a),andrequiret hemtosatisfythes ameconstraintsas A. Thisimmediatelyallowsustosolvetherepresentationpreservingconstraints: {, } = T + R( M )(7. 2.3) anditshermitianconjugateareobviouslysatis“edby(cf.sec.5.3.b.2) €=( €+ € M + €€€ M€€),(7.2 .4a) = eŠ H ( + M + €€ M€€) eH,(7. 2.4b) where H = HAi Aintroducesth equant um“eld HAwhichalsoappearsinandthe quantumconnection .Wehavewri ttenthederivativesinachiralrepresentationwith respectto HA.Repla cing eH= ee andmulti plyingallquantitiesby e fromtheleft and eŠ fromtherightwouldtakeustoaquantumvectorrepresentation.However,for quantization,itissimplertoworkwith H Itisalsoconvenienttode“ne ba ckgroundcovariant ha ttedobjectsasin(5.2.23), butfrombackgroundcova riantderivativesand H : €= €, = eŠ H eH, €= Š i { €} ;(7. 2.5) A= EA B B+( A M + A€€ M€€)= Š ( Š 1)AeŠ H ( A) eH;(7. 2.6) andd e“ne TAB Cand RABintermsofthem.Wealsode“nethesuperdeterminants,with theirappropriatehe rmiticityconditions: E = sdetEA M, E E = sdet E EA M, E = sdet EA B; EŠ 1=( EŠ 1)eŠ H, E EŠ 1=( E EŠ 1), EŠ 1E EŠ 1=( EŠ 1)E EŠ 1eŠ H.(7. 2.7) Wehave usedtheidentity,foranyfunction f f E EŠ 1HE E = Hf + fi ( Š 1)A AHA(7.2.8) (dro ppingtheterm iHA( E EŠ 1 AE E )= Š iHA( Š 1)BT T T TAB B=0(see (5.3.42)).The

PAGE 430

4127.QUANTUMN=1SUPERGRAVITYbackgroundcovarianti zationofth eoperator eŠ Histhus eŠ E EŠ 1HE E= E EŠ 1eŠ HE E .Itresults inthehermiticityconditions(from(5.3.51b)and(7.2.9)) (1 eŠ E EŠ 1HE E)Š 1= eŠ H(1 eE EŠ 1HE E), EŠ 1=( EŠ 1)eŠ E EŠ 1HE E.(7. 2.9) Weuseaconv entionalconstrainttodeterminethevectorcovariantderivative €= Š i {, €} ,(7. 2.10) andconventionalconstraints T€€€=0(orequiv alently T€ (€ € )=0)and T€ ( € )€=0to determinethespinorconnections: €€€= Š € (€ € )ln Š ( ), € = Š1 2 T€ ,( € )€Š ( €€);(7. 2.11) (compareto(5.3.55)and(5.3.25)afterdegauging). Finally,weimposethe n = Š1 3 conformal-breakingconstraint,whichdetermines byaproce duresim ilartothatofsec5.3: = Š 1( eŠ H )1 2 (1 eŠ E EŠ 1HE E)1 6 EŠ1 6 ,(7. 2.12) where isbackgro undcovariantlychiral: € = =0.Forqu antumcalculations, wehaveto eitherexpress exp licitlyintermsofanordinarychiralsuper“eldandthe backgroundgauge“eldor,whatispreferablei ngen eral,derivecovariantFeynmanrules thatallowustoworkwithitdirectly. Thefullderivatives Atr an sformcovariantlyundertwosetsoftransformations: (a)Backgroundtransformations: A = eiK AeŠ iK, H= eiKHeŠ iK, = eiK eŠ iK, ( M )= eiK ( M ) eŠ iK, A = eiKAeŠ iK, K = K = KAi A+( K iM + K€€i M€€).(7.2 .13)

PAGE 431

7.2.Background-quantumsplitting413Thesetransformationsfollowfromthe requirementthatthefullderivative Atransform covariantlyand bebackgr oundcovariantlychiral:From(7.2.4a)or(7.2.6,12)itfollowsthatand transformcovariantly,andtherefore,from(7.2.4b) H transforms covariantly. (b)Quantumtransformations: A = A;( 7.2.14a) eH= ei eHeŠ i eX (), =ei Š1 3 ( a aŠ Š i G G a a) ,(7. 2.14b) A = LA B() ei BeŠ i ;(7. 2.14c) where =Ai A = ,[ €,] =0;(7. 2.14d) foranybackgroundcovariantlychiral ( € =0).Weha vetakenthest a ndardchiral representationtransformations(5.2.16,67)forthequantumsuper“elds H and except forcertainmodi“cationsrequiredbecauseofthe Asin H and.Since [ H ,]=( H AŠ HA) i AŠ BHA( T TAB C C+ R RAB( M )),(7.2.15) ei eHeŠ i generatesLorentztransformationterms.Weintroduce X ()= X M + X€€ M€€tocancelthem.Similarly,theusualtransformationlawof hastheadditional G G a atermb ecausewerequire tobebackgroundchiraland €( a aŠ Š i G G a a)=0.Thetrans formation (7.2.14c)of Afollowsfrom (7.2.14a,b).The-dependentLorentztransformation LA B()correspondstothe A Btermin(5.2.21)withadditionalcontributionsfrom X (). Inpractice,weneverneedtocompute X exp licitly.Using(7.2.15)andtheBakerHausdortheorem, ei eHeŠ i = eH+ Y ( M )(7.2.16) forsomeL orentztransformation Y ( M ).Si nce Hisascalaroperator,

PAGE 432

4147.QUANTUMN=1SUPERGRAVITY[ H, Y ( M )]= Y( M )( 7.2.17a) forsomeL orentzrotation Yandhence eH+ Y= eHeŠ X,(7. 2.17b) thus de“ning X .Intro ducing X into(7.2.14b)isequivalenttotheprescriptionofdroppingatanyst ageofthecalculationLorentzterms other thanthoseimplicitin A.For convenience,weintroducetheoperation <> thatremovesexplicitLo rentzgeneratorsas fo llows:Forany A = AA A+( A M + A€€ M€€), (7.2.18a) wede “ne < A > AA A, < eA> e< A >.(7. 2.18b) Thetransformationlawfor H canberewritten eH= < ei eHeŠ i > .(7. 2.19) Thequantum-backgroundsplittingwehavedescribedisequivalenttothefollowing spli ttingoftheprepotentialintermsofaquantumchiral H andbackgroundvector : eH ( split )= < e eHe > ,(7. 2.20) whichisanalogoustotheYang-Millscase(6.5.25).Theusualchiralrepresentation transformationlaw < e eHe >= ei 0< e eHe > eŠ i 0(7.2.21) canberewrittenaseitherbackground(7.2.13)orquantum(7.2.14)transformations analogoustothoseof(6.5.27). Asinthenonbackgroundcase ,thequantumt ransformationsmustpreservechirality(7.2.14d).Therefore,takesthefollowingform,expressingitintermsoftheunconstrainedsupergravitygaugeparameter L(cf.(5.2.14)): €= Š i €Š 3L,= 2Š 3L.(7. 2.22) (Wehav emadethere de“nition L Š 3Ltosimplifyquantization,aswillbe explainedinsec.7.4.)Furthermore,wechoosethe(quantum)€-gauge H= H€=0

PAGE 433

7.2.Background-quantumsplitting415(see(5.2.18)),whichdetermines€intermsof L: €= eŠ H 2 Š 3 L€+ O O ( R R G G W W ).(7.2 .23) Thissupersymmetricgaugechoicedoesnotintroduceanyghosts. Wenowtaket heclassicalsupergravityaction( 5.2.48)intermsoffullsuper“elds andexpressitintermsofthequantumgauge“eldsandthebackgroundcovariant derivati ves,using( 7.2.4-12): SC= Š32 d4xd4 EŠ 1, EŠ 1= E EŠ 1 EŠ1 3 (1 eŠ E EŠ 1HE E)1 3 eŠ H ,(7. 2.24) Thisexpressionisthedirectbackgroundcovari antizationof(5.2.72),includingafactor of E EŠ 1tomakeitadensity.Thequantum“eldsappearexplicitlyandin E ,wh ilethe background“eldsappearimplicit lyincovariantderivativesand E E b.Expandingtheaction Ournexttaskistoe xpandtheactioninpowersofthequantum“elds.Thisisa tedious buthealthyexerciseandweoutlinethestepsneededtogetthequadraticpart, whichweneedfordiscussinggauge-“xing,andfordoingone-loopcalculations.Cubic andhigher-ordertermsareneededforhighe r-loopcalculations,butwedonotderive themhere. Wemustexpa ndtheexponen tialsandthedeterminant EŠ1 3 in(7.2.24)inpowers of H .We“rstd e“neAby EA= < A> = EA B B ( A B+A B) B= A+A.(7. 2.25) Theexpansionof EŠ1 3 isthen,toq uadraticorderin(and H ): EŠ1 3 =[ sdet (1+)]Š1 3 = exp ( Š1 3 strln (1+)) =1 Š1 3 str +1 18 ( str )2+1 6 str (2)

PAGE 434

4167.QUANTUMN=1SUPERGRAVITY=1 Š1 3 ( Š 1)AA A+1 18 [( Š 1)AA A]2+1 6 ( Š 1)AA BB A.(7. 2.26) To “ndtheexplic itformofA B,wereturnt o(7. 2.5,6)andwriteAexp licitly.Since < €> = €= €, €=0 .( 7.2.27a) From < > = < eŠ H eH> = + < [ H ] > +1 2 < [[ H ], H ] > + < O ( H3) > (7.2.27b) weobtain B(+ B)fromtheco ecientof Bontherighthandsideof(7.2.27b). Since H isascalaroperator,wecandropLorentzrotationtermsproducedbythecommuta torsateachstage,becausetheycanpr o duceonlymoreLor entz terms(see (7.2.17a)).Weobtain a Bfromtherighthandsideof < a> = Š i < { €} > = aŠ i < { €,[ H ] } > Š i1 2 < { €,[[ H ], H ] } > .(7. 2.27c) AgainwecandropLorentztermsatintermedi atestagesofthecalc ulation:Theycontri buteonlyto a€,andsi nce€ B=0 (toallorders),theyd onotcontri butetothe determinant E .We“rs t “nd Btolowestorderin H : [ H ]=[ H bi b]= i [ H b] b+ iH b[ b] = i ( H b) bŠ H€( R R €Š G G€ )+ Lorentzterms andhence = Š H€G G€, b= i H b.(7. 2.28) (Againwecanignore€.) Pr o ceedinginthismannerwethen“nd,totheorderin H€necessaryforthe quadraticaction( H H€ €):

PAGE 435

7.2.Background-quantumsplitting417 = Š (1 Š1 2 iH ) H€G G€Š1 2 ( H€) H€[1 2 (€G G€ )+ €€( W Š1 2 ( )R R )] +1 2 H€G G€H€G G€Š1 2 R R R R H2, b= i H b, a = Š i ( Š €H€G G€+ R R H€), a b= € H b+ €€ .(7. 2.29) Wehave usedthefactthat,forthepartoftheactionquadraticin H ,wen eedonlythe linearpartsof €and€ ,andwecanalsodro panyto talderivativesofquadratic terms(but notifweweretocomputehigher-ordertermsintheaction).Aftersubstituting(7.2.29)into(7.2.26)we “nd,againdroppingirrelevantterms, EŠ1 3 =1 Š1 3 { € H€Š (1 Š1 2 iH ) G G H Š1 4 ( H€) H€[ (€G G€ )+ €€(2 W Š ( )R R )] +1 2 [( G G H )2Š 2 G G2H2] Š R R R R H2}+1 18 ( € H€Š G G H )2+1 6 {( € H€Š €€H€G G€)( € H€Š €€H€G G€) Š 2( €H€G G€Š R R H€) H€Š [( G G H )2Š 2 G G2H2]}.(7. 2.30) Wealsohavether elevanttermsof(using(7.2.10),andexpanding =1+ ): (1 eŠ E EŠ 1HE E)1 3 =1 Š1 3 i H +1 9 ( H )2, eŠ H =1+( + )+ Š iH .(7. 2.31) Finally,weobtainthequadraticpartoftheLagrangianbymultiplyingtogetherthe

PAGE 436

4187.QUANTUMN=1SUPERGRAVITYvariousc ontri butions,toobtain: E E EŠ 1=1+( + +1 3 G G H )+ Š1 3 i ( Š ) H +1 3 ( + ) G G H +1 3 R R R R H2+1 18 ( G G H )2+1 12 ( H )2Š1 36 ([ €, ] H€)2Š1 18 ( G G H )[ €, ] H€+1 3 R R ( H€)( H€) +1 12 ( H€) H€[ (€G G€ )+ €€(2 W Š ( )R R )] Š1 6 H€( € € + € € ) H€.(7. 2.32) Byusingtheidentity(with =1 2 € €) € € + € € = CC€€(Š + { 2, 2}Š1 2 [ Š R R €+ G G€ +( G G€) M+ W€€€ M€€, €])+2 R R R R M M€€Š ( ( R R ) ) M€€,(7. 2.33) wecanr ewrite(7.2.32)as E E EŠ 1=1+{ + +1 3 H€G G€}+{ +1 3 i ( Š ) €H€+1 6 H€ H€+1 12 ( €H€)2Š1 36 ([ €, ] H€)2Š1 3 [( 2+3 2 R R ) H€][( 2+3 2 R R ) H€]}+{1 3 ( + ) H€G G€+1 18 ( H€G G€)2+2 3 R R R R H€H€+1 12 ( 2R R + 2 R R ) H€H€+1 6 H€( R R 2+ R R 2) H€Š1 12 H€G G€[ €, ] H€+1 12 H€[( ( G G )€) €H€+( (€G G€ )) H€] Š1 18 ( H€G G€)[ €, ] H€+1 6 H€( W H€+ W€€€ €H€)},(7. 2.34)

PAGE 437

7.2.Background-quantumsplitting419wherewehavealsousedtheBianchiidentities(5.4.16,17,18).Theexpressioninthe “rstsetofbracesislinearinthequantum“elds.Ifsourcesarecoupledtothem,variationwithrespectto H and gives R R = J and G G€= J€,using d4xd4 E EŠ 1 = d4xd2 eŠ 3( 2+ R R ) = d4xd2 eŠ 3R R (7.2.35) (seesec.5.5.e;weareinthebackgroundvectorrepresentation).Theexpressioninthe secondsetofbracesisthedirectcovariantizationofthefreeLagrangian,whichis obtainedbysettingallbackground“eldstozero: Š1 3 IL(2)= +1 3 i ( Š ) €H€+1 6 H€[ Š D2 D2Š D2D2] H€+1 12 ( €H€)2Š1 36 ([ D€, D] H€)2.(7. 2.36)

PAGE 438

4207.QUANTUMN=1SUPERGRAVITY7.3.Ghosts Inthenextsectionweshalldiscussindetailthequantizationofsuper“eldsupergravity.Theprocedureisnotentirelystra ightforwarda ndrunsintoanumberofsubtletiesnotnormally encounteredinsimplertheories,soitisdesirabletoknowaheadof timethegeneralghoststructureofthequantizedtheory.Thisisthetopicofthepresent section.Wediscussthefollowingsubjects:(a)howthelinearizedghoststructurecanbe easilydeterminedbeforeperformingthequantization;(b)themodi“cationstotheFaddeev-Popovprocedurenecessarywhenusingconstrainedgauge-“xingfunctionsinthe pr esenceofbackground“elds;(c)howtoobtainonlypropagatorsthatgoas pŠ 2,and avoidinfra reddiculties,whilestillkeepingtheactionlocal,bytheintroductionof additional“elds;and(d)thenecessityforappropriateparametrizationofthegauge transformations(forwhich(density)compensatorsarecrucial)sothattheFaddeevPopovproce dureisapplicable,andsothatweonlyusethetypesofsuper“eldsallowed in an ar bitrarysupergravitybackground. a.Ghostc ounting We“rstgiveasim pleruleforcountingallthegh ostsinanygaugetheory.In ordinarygaugessomeoftheseghostsmaydecouple,butinbackground“eldgaugesall th eg hostscoupletothebackground.Webeginbyderivingtherulesforageneralcomponent-“eld gaugetheory.Tostreamlinenotation,wedropallindicesandindicate abnormal-statistics“eldsbyprimes.ThegeneralquadraticLagrangianforanygauge “eldcanbewri tteni ntheform A n A ,whereisa projectionoperatorand n isan integer(when A isatensor)orhalf -integer(when A isaspi nor: 1 2 /).Forphysical “elds n =1or1 2 .(Theope rators /,etc.maybecovariantw ithres p ecttobackgr o und“elds,andmayincludenonminimalcouplingstothebackground.)Thegauge invarianceis expressedas A = ,with =0.After gauge“xing,theLagrangian b ecomes A nA but,inordertocancelthe A = mode,whichdidnotoccurinthe originalLagrangian,wemustintroduceaghost B.ItsL agrangianisobtainedthrough thesubstitution A Š > A= Binthegauge-“xedLagrangian.Wethusobtain IL = A nA +( B) n( B)= A nA + B n +1B(7.3.1) whereisanewproj ectionoperator.Inthesimplestcases=1(e.g .,if A isthe

PAGE 439

7.3.Ghosts421photon“eld)andwearethrough.Moregenerally(e.g.,if A isanantisymmetrictensor gauge“eld) Bhasagaugeinvariance,andwemustcontinuetheprocedureuntilno gaugeinvarianceremains.(Thisisa“nitepro cedure,sinceeachghosthasonelessvector i ndexthanitspredecessor). The“nalLagrangianthushastheform IL = A nA + B n +1B+ C n +2C + ... .(7. 3.2) Fortenso r“elds,a “eldwithkine ticoperator mrepresents m “eldsofthattypewith kineticoperator ;forsp inors, mrepresents2 m “eldsw itha /.(If incl udesbackgroundinteractionstheircontributiontotheeectiveactionis lndet m= mlndet =2 mlndet /.)Thus,for physicaltensor“elds A ( n =1), the numberofsu ccessive“eld sgoesas1, Š 2,3, Š 4,...,whileforphysicalspinor“elds ( n =1 2 ),theygoas1, Š 3,5, Š 7,...,wheretheminussignsindicateabnormalstatistics. Thesenumbersrepresentthenetnumberofnormal-statisticsminusabnormal-statistics quantum“eldsinthelinearizedLagrangian,allcouplingtothebackground“elds.Furthermore,aswewillseebelow,allthe“elds inthiscounti ngdecoupleathigherloops (andatoneloopwheno neisquantizinginordinarygaugesratherthanbackground“eld gauges)exceptforthephysical“eldsand(fornonabeliantheories)theFaddeev-Popov ghostsofthephysical“elds.Theremayalsobeadditionalcompensating“eldscoming inpairsofoppositestatistics(catalysts:seebelow)whichcancelinthiscounting,and whichalsocancelinone-loopbackground“eldcalculations(butmaycontributefor higherloops).Exampleswherethiscountingincludesmorethanjustthephysicaland Fadd eev-Popov“eldsare:(1)thegravitino,whichhas1, Š 3inste adof1, Š 2,duetothe appearanceoftheNielsen-Kalloshghost(seebelow);(2) p -forms,whichhave 1, Š 2,3 ,...,( Š 1)p( p +1)inste adof1, Š 2,4,...,( Š 1)p2p, duetoNielsen-Kalloshghosts andhidde nghosts(seealsobelow). Generalizationofthecountingrulestosuper“eldsisstraightforward,thougheach casehastobetreatedseparatelybecauseofthegreatervarietyofsuper“eldgaugetransformations.We“rstgeneralizetosuper“eld stheresultof thepreviousparagraphfor obtainingthenumberofsuper“eldswithst a ndardkinetictermcorrespondingtoone withahigher-derivativekineticterm.When misthekineticoper atorforgeneral unconstrainedsuper“elds,itisequivalentto m generaltensorsuper“eldswithkinetic

PAGE 440

4227.QUANTUMN=1SUPERGRAVITYoperator ,or2 m generals pinorsuper“eldswithkineticoperator /.(Note that,asdiscussedinsec.3.8, sdet =1 unless hasnontr ivial -dependence.)However,forchiral super“eldsthesituati onisslightlymoresubtle(see(3.8.28-36)): d4xd2 m m i =1 d4xd2 i i 2 m i =1 d4xd4 ii,(7. 3.3a) d4xd4 m 2 m +1 i =1 d4xd4 ii,(7. 3.3b) d4xd4 € mi € m +1 i =1 [1 2 d4xd2 + h c .].(7. 3.3c) The formof(7.3 .3a)givestheresultforchirals calarsuper“elds,whereasthe formisapplicabletochiral(undotted)spinorsuper“elds.Thislatterresultcanalsobe relatedto(7.3.3c)bynotingthat(7.3.3a)for m =1and( 7.3.3c)for m =0arem erely dierent gaugechoicesforthegauge-“xedactionforthetensormultiplet(cf. (6.2.32-34)). Wenowconsi dersomeexamplesofsuper“eldgh ostcounting.Thesimplestisthe v ectormultiplet.TheclassicalLagrangianis V 1 2 V ,with V = i ( Š ),whereis chiral and1 2 isthesuperspin1 2 projectionoperator.Aftergauge“xing(whichremoves 1 2 )andth es ubsti tution V Š > i ( Š )withchira lghosttocan celthegauge modes,weobtain IL = V V + .(7. 3.4) (With d4 integrationthe and termsgivezero,modulononminimalcoup lingswhichweincorporateintothede“nitionof .)TheghostLagrangianis equivalenttothreeoftheusualterms (see(7.3.3b)).Wethusexpectthreechiral ghosts,whichagreeswithourresultsfr omexplicitquantizationinsec.6.5. As econdexampleisthatofthetensormultiplet,withclassicalactiond4xd2 1 2 +andgaugeinvariance = i D2DK K = K .After gauge“xing (whichremovest heprojector1 2 +),substitutionleadstoa“rstgenerationghost Lagrangian V 21 2 V,andits gaugeinvariance V= i ( Š )leadst oasecond

PAGE 441

7.3.Ghosts423generationghostLagrangian 2 .(The gaugeinvarianceof Visthesam easthe gaugeinvarianceofthe variation of : ( K )= ( K + i Š i ).)Ween dupwith one term,two V Vterms,a nd“ve terms. Athirdexam pleisthatofthegeneralspinorsuper“eld ,whicht ogetherwitha compensatingchiralscalardescribesthe(3 2 ,1)multi plet(seesec.4.6;weareusingthe secondformof(4.6.42),butwiththecompensator V gaugedtozero).TheLagrangian hastheform IL =1 2 €i €+ h c Š 2 + crossterms (7.3.5) whereisasumof projectionoperatorsand IL hasthegau geinvariance =+ iDK = Š D2K ,withchiralandreal K .Inthe gauge“xed Lagrangianisabsentandwehaveachiralspinorghost andarealscalarghost V(correspo ndingtoand K ,resp ectively)withoutanyfurthergaugeinvariance(the variation ,is not in va riantunderanychangesofor K ): IL = €i €+ + €i € + V V(7.3.6) ( crosstermscanbeeliminatedbyasuitablechoiceofgauge“xingfunction). TheotherformofthetheoryhastheLagrangian1 2 €i €+ h c .+ ... witha dierent,and gaugeinvariance = i D2DK1+ iDK2.Afterin cludingghosts,it b ecomes IL = €i €+ V 2 V 2+ V 1 2V 1+ 2.(7.3 .7) Thechiralscalar“eldisasecond-generationghost,arisingfromtheinvariance V 1= i ( Š )(duetotheinvarianceof under K1 K1+ i ( Š )).Wethus obtaintheequivalentofthreerealscalargh ostsand“vechiralscalarsecond-generation (normalstatistics)ghosts. Weconsid ernow n = Š1 3 supergravityitself,w ithkineticLagrangian H H + (+ H crossterms),whereisasumof projectionoperators.Wehave the(linearized)gaugeinvariance H€= D L€Š D€L, = D2DL,withg eneral spinorgaugeparameter L.After gauge“xingwehaveaLagrangian H H + + € i € ,wheretheg hosttermis obtainedbysubstitutioninthe

PAGE 442

4247.QUANTUMN=1SUPERGRAVITYgauge-“xedLagrangianwiththegaugeparameter Lreplacedbytheghost .We haveanew gaugeinvariance, =whereischiral.(Thisre”ectstheinvariance of th eo riginalgaugetransformationsunder L=.)Wethusintro duceasecondgenerationchiralghost andare“nallyledtotheform IL = H H + + /+ / .(7. 3.8) Weobtain theequivalentofthree“rst-generationgeneralspinorghosts,withLagrangian / ,andtwos econd-generationchiralspinorghosts.Inthenextsectionwewillderive theseresultsfromthegauge“xingprocedure,andgivetheresultsforavariantformof the n = Š1 3 compensatorwhichleadstoadierentsetofghosts. b.Hiddenghosts InadditiontotheNielsen-Kalloshghost,whichemergesfromacarefulapplicationofthegauge-averagingprocedure(see( 6.5.12,13)),thereisasecondsubtletythat mayoccur,andwhichmustbehandledcorrectlyinordertoarriveatthecorrectsetof ghosts.Thishastodowiththeoccurrenceofgauge-“xingfunctionswhichsatisfyconstraints.Inthenonsupersymmetriccasethesimplestexampleisgivenbythe2-form A a binabackgroundgravitational“eld.ThenaivetHooftgaugeaveraging IDf a ( bA a bŠ f a) exp ( Š d4xg1 2 f2)= exp [ Š d4xg1 2 ( bA a b)2](7. 3.9) wouldgivea nincorr ectresult,sinceth econstraintinthe functi onalimplies f =0, andintroducesextraneousde pe ndenceontheexternalgravitational“eld.Wewould ther eforeliketoputjustthetransversepartof f inthe functional,and inthegaugeaver agingfunctionas f21 2 f a( a bŠ1 2 a Š 1 b) f b.Howev er,sincethenonlythe transver separtof f appearsinthefunctionalintegral,theintegrandhasagaugeinvariance f a= a ,sowemustint roduceappropriategauge-“xingand(Faddeev-Popovand Nielsen-Kallo sh)ghostterms.Theintermediatestepsvarydependingonthechoiceof gauge-“xingfunction(e.g., f vs. Š 1 f ),butthenetresult isthatoneobtains Š 1 additionalscalar“elds,(a hiddenghost)andthusthetotalsetof“eldsconsistsof1 2-form, Š 21-fo rms,and+3scalars(vs.the+4expectedfromconsideringjusttheFaddeev-Popov ghostsofthevectorghostsofthe2-form),inagreementwithourgeneral countingargumentgivenabove.Similarargumentsapplytohigher-rankforms.

PAGE 443

7.3.Ghosts425Asimplerformofthea rgumentforthenecessityofthesehiddenghostscanbe giveninsupersymmetrictheories.Consideragainthechiralspinorgaugesuper“eld, withgaugeinvariance = i D2DK andgauge-“xingfunction F = F =1 2 i ( DŠ D€ €).Becauseofthechiralityof, F satis“estheconstraint D2F =0,sothat F isalinearsuper“eld.T heusualgauge-“xingprocedureinvolves introducingi nthef unction alintegral ( F Š f );however,thelinearnatureof F would implyt hat f isalsolinear,anunfortunatefeaturesinceitisimpossibletofunctionally integrateordierentiatewit hresp ecttolinearsuper“elds. Thisdicultycanbeavoidedbycompleting F t oagen eralsuper“eld,bythe additionofchiralandantichiralpiecestoit.Wedothisbyreplacing F inthe functionalwiththeexpression F = F +( D2 Š 1 + D2 Š 1 ).(7.3 .10) andfunctionallyintegratingover aswell.Thechi ralsuper“eld isthehidde nghost. Now D2 F = is unconstrained,andso f isalso. To understandtheprocedureweexamineitscomponentform.The -function ( F Š f )isapro ductof -functionsfortheindividualcomponentsof F Š f (see (3.8.17a)).Since F isthe1 2 partof F andtherestisthe0part,thecomponentsof thetwotermsin(7.3.10)appearindierentcomponent -functions.The D2f | D D2f | and D2 D2f | componentsaresetequaltocomponentsof withoutspacetimederivatives (whichiswhyw eincl udedthe D2 Š 1factor), andwitho utany F =1 2 i ( DŠ D€ €) contributions.Averagingwith expd4xd4 f2producesanac tionforthese componentsthatdoesnotcontributetothefunctionalintegralupon integration(trivial kineticterms).The f | Df | componentsproducestandardgauge-“xingtermsforthe gaugecomponentsofwhichareabsentintheWess-Zuminogauge(namely and B (4.5.30)),andwhosecontribut ionisthereforecan celedbycorrespondingghosts.Finally, the[ D€, D] f | componentgivesthegauge-“xingfunction bAa b+ a Š 1G ,where G = ImD2 | and A a bistheantisymmetrictensorcomponentof.Aver agingofthis componentof f givesaterm( bA a b)2Š G Š 1G :the usual A a bgauge-“xingtermasin (7.3.9)plusthehiddencomponentghost(+1scalarwith Š 1,countingas Š 1scalar with ),inagreementwiththeabovediscussioninthecomponenttheory.

PAGE 444

4267.QUANTUMN=1SUPERGRAVITYAsim ilarsituationoccursinsupergravity duetotheappearanceofthegauge-“xingf unction F= D€H€satisfying D2F=0.Thisw illbediscu ssedinmoredetailin thenexts ection. Notethatthesehiddengho stscoupleonlytobackground“elds,andthuscontri buteonlyatoneloop.Itwouldbedesirabletohaveageneralderivationofthese ghostsbasedonBRSTinvarianceofthegauge-“xedaction,fromwhichSlavnov-Taylor identitiescouldbederived.TheappropriateBRSTtransformationswouldbethose whoseSlavnov-Tayloridentitiesimpliedgauge-independenceoftheeectiveaction.At presentthisapproachhasnotbeenworkedout. c.Morecompensators UnwantedtermsintheLagrangian,suchasthoseleadingto pŠ 4termsinthe pr opagatorornonlocalvertices,cansometimesbecanceledbyintroducingadditional “eldsandgauge-“xingthemconveniently.Sinceonlytheghostsdiscussedintheprecedingsectionsaren eededtopreserveunitarity,contributionsofthesecatalyst“eldsmust themselv esbecanceledbytheirownghosts,andindeedthishappensattheone-loop level.Thecatalystsmayingeneralinteractwiththeotherquantum“elds,andhence contributeathigherloops,whereastheirghostsdont.Ifoneweretointegrateoutthe catalysts,theirhigher-loopcontributionswouldsimplyreproducetheunwantedterms thatthecatalystseliminatedinthe“rstplace. Catalystsarejustatypeof(tensor)compensator.Forexample,thecompensator intheStueckelbergformalism(sec.3.10.a)isintroducedsimplytoimproveultraviolet be ha viorofthepropagator,anddecouplesduetogaugeinvarianceoftheinteraction term.Inourcase,thesecompensatorsimproveinfraredbehavior,anddonotdecouple. Furthe rmore,catalystsgenerallyareintroducedbyghost“elds,whereaspreviouslywe discussedcompensat orsrelatedtoonlyclassical“elds. Asanexample, considerthelinearizedLagrangian IL = A [(1 Š )+ ] A ,with 2=and =0,1.Ifweread ierentialoperator,e.g., Š 1 ,theabove Lagrangianwouldleadto pŠ 4pr opagators.ToobtainthesimplerLagrangian IL = A A onecouldmakea“eldrede“nition A=[(1 Š )+ Š1 2 ] A butifwere nonlocalthiswouldintroducenonlo ca litiesintheinteractionterms.

PAGE 445

7.3.Ghosts427Insteadweintroduceacatalyst“eld B intheLagrangian,eitherwithatrivial kinetictermortogetherwithaghostwhichcancelsit.Wethenmakeshifts A A + OB B B + OA thatcanceltheunwantedterms,andgiveno AB cross terms.Forinstance,intheexampleabove,ifisamatrix,wechoose B sothat B = B (andintroducealsoaghost“eldwithoppositestatisticsand B= B)andwe addittotheLagrangiantoobtain IL= IL +(1 Š ) B B + B B.Firstmakingthe shift B B + A ,thent heshift A A Š (1 Š ) B ,weobtain IL= A A +(1 Š ) B B + B B.Forback groundinte ractions B and Bwillcancelatone loop, butclearlythe A shiftcanleadtoquantuminteractionsof B ( butnot B).An equivalentprocedureconsistsofmakingthesubstitution A A + B intheoriginal Lagrangian,andgoingthroughthegauge“xingprocedureforthenewgaugeinvariance thathasbeenintr o duced,namely A = B = Š ( = ).Ingeneral,thisisthe simplestprocedure. Asasuper “eldexample,weconsiderarealscalar V inthepresenceo fanon-shell backgroundsupergravi ty“eld,w ithLagrangian IL0= V ( Š 2+ a {2, 2} ) V (7.3.11) with a =0,1.Thesuper“eld V has pŠ 4termsinitspropagatorandcomplicatedvertices(couplingtot hebackgroundgravitational“eld),butitcanbeshownthattheresult fortheeectiveactionisindependentof a .Toshowt hisusingcatalyst“eldsweintroducethem,forexample,bymakingtheshift V V +( + ), € =0.(7. 3.12) Wehave nowthegau geinvariance V =+ = Š ,withachiralparameter.We choosethe gauge“xingfunction F = 2( V Š a 1 Š a )and gauge“xingterm 2(1 Š a ) F F ,andareledto theLagrangian IL = V V +2 a 1 Š a .(7. 3.13) The Lagrangianisequivalenttothatforthreeordinarychiral“elds i, i =1,2,3.The gauge-“xingpro cedurealsointroducesthreechiralghosts,justasfortheusual V super“eld(twoFaddeev-PopovandaNielsen-Kalloshghost).Theyexactlycancelthethree ordinarychiral“eldsattheone-looplevel,andleaveuswith V V .

PAGE 446

4287.QUANTUMN=1SUPERGRAVITYIfwehadconsideredasystemsimilartotheabove,butwhere V hadquantum interactions,the swouldalsohavesuchinterac tions.Thentheeectofthe swould betorepro duce,if integratedout,thenonlocalitiesthatwouldhavebeenintroducedif, insteadoffollowingthea boveproce dure,wehadmadeanonlocalrede“nitionof V to casttheoriginalLagrangianinthe V V form.Inthiscase,the(o-shell)Greenfunctionshav egen uine a -dependence. Thegeneralprocedureisthefollowing:ConsidertheLagrangianofanarbitrary super“eld oftheform n(0+i cii) ,(7. 3.14) where0+i i=1and0isaparticularsuperspinchos enforconvenience,e.g.,the highestsuperspinin orthesuperspinthatoccursmostfrequentlyininteractionterms. Ifsomeoftheconstants ciareequal,wemaycombinethe correspondingprojection operatorsiintoasingleone (incl uding0,whichisme relyawith c0=1).Also, some cimayvanish,whichimpliesacorrespondinggaugeinvariance.Wenowintroduce catalystsbytheshifts +i Oi i,iOj= ijOj,(7. 3.15) where Oiareoperatorsthatmaybenonlocal,butonlytotheextentthatallnonlocalitiesintheinteractiontermscaneventually bere moved.Thenwe“xthecorresponding gaugeinvariances =i Oii, i= Š i,(7. 3.16) insuchawaythattheLagrangianfor b ecomessimply n ,andallc rossterms betw een and iarecanceled:Thegauge-“xingfunctions Fi= Oi ( Š ci1 Š ci Oi i)(7. 3.17) withgauge-“xingterms (1 Š ci) Fi n( Oi Oi)Š 1Fi(7.3.18)

PAGE 447

7.3.Ghosts429givetheLagrangian n +i ci1 Š ci iOi  nOi i.(7. 3.19) ( Oi Oiisinvertibleon Fi.Also, itcanbeshownthat Oi( Oi Oi)Š 1Oi =i.)Notethat thisproce dureincludes“xingoftheordinarygaugeinvariance.WethenaddtheFaddeev-Popov andNielsen-Kalloshghostsasi [ 2 i( Oi Oi) 1 i+ 3 i n( Oi Oi)Š 1 3 i+ h c .].(7. 3.20) Intheinteractingcase, Oimustbechosen sothatanybackgrounddependencecomes outlocal,inc l udingextratermswhichmayresultfrommanipulationsofthebackground dependent and Oi.Itmaybe necessarytochoose Oisuchthat ihasitsow n gauge invariancei ndependentof ,ortocombi neseveraliinsuchawaythattheabove Lagrangianfor alsoneeds“xing,inwhichcasetheentireproceduremustberepeated forthose s.Howev er,the salwayshave fewercomponentsthantheir s(atleast for N =0or1supersy mmetry),sotheseriesmusteventuallyterminate. d.Choice ofgaugeparameters Inanygaugetheory,somecareisrequiredtoensurethattheFaddeev-Popov quantizationprocedurewillleadtocorrect,unitaryresults.Onewaytocheckunitarity istocompareresultswiththoseobtainedina ghost-free(e.g.,axial)gauge.InsupersymmetrictheoriessuchagaugeistheWess-Zuminogauge,andonewaytoinsureunitarityisbymakingcertainthatonecanpasssmoothlyfromcovariantgaugestothe physicalgaugewithoutintroducinganyextraunphysicaldegreesoffreedom.Thiswill certainlybethecaseifthesuper“eldgaugetransformationsaresuchthattheyallowthe gau gi ngtozerooftheunphysicalcomponentsbyalgebraic,non-derivativetransformations( A = andnot,e.g., or a a).Forexample,inordinarycomponentYangM illstheorythegaugetransformationcanbewritteneitheras A a= a oras A a= a .Howev er,thelatterchoicewouldgiveaFaddeev-PopovghostLagrangian c c (insteadofjust c c ),andtheextra wouldgiv eanontrivial contribution whichwoulddestroyunitarity.ItispossibletomodifytheFaddeev-Popovprescription tocorr ectlyhandlethesituation,butthesimplestprocedureistochoosethegauge parametersinsuchawayastoavoidtheproblem.

PAGE 448

4307.QUANTUMN=1SUPERGRAVITYInthecaseofsupergravity,itcanbeveri“edbytheprocedurejustdescribed(cf. 5.2.10)thatthe Lparametrizationistheonlycorrectone.Aswesawinsec.5.2.cthis parametrizationcanbeused onlyifwealsohavethecompensator(s)inthetheory. Eliminationofthecompensatorwouldintroduceconstraintsonthegaugeparameter,the solutionofwhichwouldexpress Lintermsofderivativesofoth ersuper“eldparameters. However,asintheexampleabove,thiswo uldintroducespuriousextraghostsinthe naiveFaddeev-Popovprocedureandunitaritywouldbelost,unlesstheprocedurewere modi“ed. Thereisonemorerestrictionwhich mustbeob servedinchoosinggauge parametrizations(andthusghosts):Ingeneral,notallsuper“eldswhicharerepresentationsofglobalsupersymmetryarealsorepres entationsoflocalsupersymmetry.Inparticular,for n = Š1 3 th eo nlytypeofchiralsuper“eldsallowedareoneswithonlyundottedspi norindices:Theexistenceofadottedchiralspinorwouldimply 0= { €, €} €= Š 2 RM€€€= Š RC€ (€€ ) =0.(7. 3.21) Generally,thechoiceofgaugeparametersmu stberestrictedtothosewhichcanexistin an ar bitrarybackground.

PAGE 449

7.4.Quantization4317.4.Quantization Inthissectionwepresentthedetailsofthequantizationprocedureforsupergravity.Thisinvolveschoosinggauge-“xingfunctionswhichallowallkinetictermstotake simpleforms,and“ndingtheresultingghosts(Faddeev-Popov,Nielsen-Kallosh,andhidden).Suchsimpli“cationsoftenrequiretheuseofappropriatecompensatorsand/orcatalysts.Thisproce dureis“rstappliedtothephysical“elds,thentotheresultingghosts, theghostsghosts,etc.(Theg hostsreduceinsizeateachstep,sotheprocedurequickly terminates.)Fornowweworkwit hon-shellbac kgro und“elds( R R = G G =0),so thepart oftheactionquadraticinthequantum“eldsbecomes(see(7.2.24,34)),inunits =1 (ormakingt heusualrescaling( H ) ( H )), S = d4xd4 E EŠ 1[ Š 3 + i ( Š ) H Š1 2 H H Š1 4 ( H )2+1 12 ([ €, ] H€)2+( 2H ) ( 2H ) Š1 2 H€( W H€+ W€€€ €H€)].(7.4.1) Thequantizationwithon-shellbackground“eldsissucientforcomputingphysical quantities(S-matrixelements)inpure N =1andexte ndedsupergravity.Wewilldiscussthegeneralsituationlater. We havethefollowing(o-shell)gaugeinvarianceunderthequantumtransformations(from(7.2.14)and(7.2.22,23)): H€=( L€Š €L)+ O ( H )+ O ( ), = Š ( 2+ R R ) Š 3L Š1 3 [( 2+ R R ) Š 3L] .(7. 4.2a) Notethatthesecondequationcanberewrittenas 3=( 2+ R R ) L.(7. 4.2b) (Onshell weca nset R R =0.) Tocan celthe H crossterms,wechoosethefollowinggauge-“xingfunction: F= €( H€+ i a € Š 1 Š 3)(7. 4.3)

PAGE 450

4327.QUANTUMN=1SUPERGRAVITY(forsomeconstant a tobedeterminedbelow).Thisisthemostconvenientgaugechoice. Itcorrespondstoamodi“cationofthetransversegauge €H€=0(sees ec.7.5.b).We havede“nedthechiraldAlembertians += + W M Š= + W€€€ € M€€,(7. 4.4a) on ar bitrarychiralsuper“elds(i.e.,withanynumberofundottedspinorindices)by + ... = 2 2 ... Š € ...€= 2 2 € ...€.(7. 4.4b) Wehave used 3=(1+ )3in(7.4.3)insteadofjust sothattheFaddeev-Popovprocedurewillcontributenonlocaltermsto onlythekineticterms oftheghosts,andnotto theirquant uminteractions,duetotheformof(7.4.2b).(Thisisthereasonforour introductionof Š 3intothetransformationlaws(7.2.22).)Nonlocalkinetictermscan bemadelocalbyuse ofcatalystghosts,sothisisaharmlessnonlocality,whereasnonlocalinteractiontermswouldbeaproblem. We“xthe gaugeby“rstcompletingthelinea rsuper “eldgauge“xingfunction Ftoagen eralsuper“eld,therebyintroducinghiddenghosts ,ands ubsequentlyaveragingovergaugesbyusingaweightingfunctionthatleadstosomeofthedesiredsimpli“cations:Wewishtocancelall H termsintheLagrangianwhichwouldcontributeto pr opagatorsexceptfor H H ,incl udingthe H crossterms.Thisgaugeischosenby introducinginthefunctionalintegralthefactor ID ID €ID ID € ( F+ 2 Š 1 +Š ) ( F€+ 2 Š 1 Š €Š €) exp{ d4xd4 E EŠ 1 [ Š1 4 ( Š € €)2Š1 12 ( + € €)2+( €)( €)]},(7. 4.5) andcarryingouttheintegralsover .Thisgivesthe gauge-“xingterms andthe hidden ghostaction SGF= d4xd4 E EŠ 1{[1 4 ( H )2Š1 12 ([ €, ] H€)2Š ( 2H ) ( 2H )]

PAGE 451

7.4.Quantization433+[ Š5 3 i ( H )( a 3Š a 3) Š4 3 ( a26+ a2 6)+10 3 a a 3 3] + €i € + Š 1}.(7. 4.6a) Uponlineari zation(andusingtheon-shellcondition R R =0), the termsbecome Š 5 i ( H )( a Š a )+30 a a ,sowechoose a =1 5 (7.4.6b) tocan celthe H crossterms in(7.4.1). Wecanshow, however,thatthehiddenghostLagr angiangivesnocontributions. (Notethatithasnoquantuminteractionsandcontributesatmostattheone-loop level.)Weperformtwosuccessiverotations(withunitJacobian) (1) + ib € 2 Š 1 Š €, € €;(7. 4.7a) (2) € €+ ic € 2 Š 1 +;(7. 4.7b) choose b and c tocan celthe €crossterms,andrewritethehidden-ghostactioninchiralform S = d2 eŠ 3+ h c .,(7.4 .8a) (recallthatthebackgroundisinvectorrepresentation)or,withthe“eldrede“nition Š3 2 S = d2 eŠ + h c .= d2 + h c .,(7.4 .8b) intermsofanordinarychiralsuper“eld = eŠ .Thusthe hiddenghostdecouples fromthebackground“eldandgivesnocontributiontotheeectiveaction.(Thisisjust thecovariantizationof(7.3.3c)for m = Š 1.) TheFaddeev-Popovghosts areobtainedins ta ndardfashionfromthegauge“xingfunctions.Forthetimebeingwewrite onlythekineticterms(arisingfromthe H and independentpartofthegaugetransformation;theremaindergivesrisetoghostquantum“eldinteractions).Aftersomealgebraweobtain

PAGE 452

4347.QUANTUMN=1SUPERGRAVITYŠ ( € Š €)( €Š €) Š1 5 [( 2 ) Š Š 1( 2 € €)+( 2 € €) + Š 1( 2 )].(7.4.9) Wenowwishtos implifytheghostLagrangianbyputtingitinthestandardform € €.Wetherefo reintroducecatalystswiththeshifts + ( V1+ iV2), + ( V 1+ iV 2).(7.4 .10) Inadditiontotheinvarianceduetotheseshifts,theLagrangianhasalsotheinvariance duetothefactthatthe“eldsappearonlyas €, € .Wethushav ethe gauge transformations =+ L ( V1+ iV2)= Š L €=0; = + L, ( V 1+ iV 2)= Š L, € =0;(7. 4.11) withthechiralspinorparameters, .These parameterswillintroducesecondgenerationchiralspinorghosts ,andwealso havetherealscalarghosts V 1,2,3,4associatedwiththeinvariancesparametrizedbythecomplex L L.After gauge“xingand somechangesofvariablesthekineticLagrangiancanbeputinstandardform(see (7.4.14a)).Theone-loopcontributionof Vi, V icancelsthatof V i, but Viand V ihavequantuminter actions(because and do,and Viand V ienterthroughthe shifts in(7.4.10)). Theaveragingin(7.4.5)hastobenormalizedbyintroducingaNielsen-Kallosh ghost3 ,tocompens atethecont ributionsfromthe “elds.ItsL agrangianis Š1 4 ( 3 Š € 3€)2Š1 12 ( 3 + € 3€)2+( 3€)( €3 ).(7.4 .12) Wecannowa pplytheusualprocedure(asdescribedinsec.7.3)ofthecatalyststoplace theLagrangianinthestandardform:Weshift 3 bytherepresentationswhosecoecientsin(7.4.12)arenot1,andthen“xthegaugetomakethem1(see(7.3.11-13)for anexample).Inthiscase,wemaketheshift 3 3 + 2 3+ 3, €3=0.(7. 4.13) Thisallowsusto“xthesuperspin0( 3)andtwooft he(four)superspin1 2 ( 3)par tsof

PAGE 453

7.4.Quantization4353 skineticterm.Wethenchoosethemostgeneralgauge-“xingfunctionsandweightingsforthenewinvariances(correspondingtoarbitraryvariationsof 3and 3,andthe correspondingvariationsof 3 ),introducetheappropriatenewghosts,makeshifts,etc. Thenetresultisthatallthecatalystscancelasintheexample(7.3.11),leavinguswith justtheNie lsen-KalloshghostwithconventionalLagrangian.(Thisghosthasnoquantuminteractions.) Infact,theformofthe(quantum-quadratic)Lagrangianwaspredictableforall “eldsexcept H ,sincebydime nsionalanalysisandLorentzinvariance(and,whenrelevant,chira lity)onlyitcouldhavenonminimalte rmsnotresultingfromdirectbackgroundcovariantization(i.e., Wterms). The“nalresultofthequantizationprocedureisthefollowing:Wewritethewhole eectiveLagrangianasasumofaquadraticpartandtherest,withthequadraticpart being S = d4xd4 E EŠ 1[ Š1 2 H€ H€Š9 5 +( €i €+ i € €+ 3€i €3 ) +(3 V 1 V1+ V 2 V2)+4 i =11 2 V i V i+2 i =1 (1 2 i 2i + h c .)], (7.4.14a) where = + W M + W€€€ € M€€.(7. 4.14b) Intheseformulae , 3 Viand V ihaveabnormalstatistics.Thisexpressionis sucientforone-loopcalculations.Notethatatone-loopthecontributionsfromthe various V scan celduetostatistics. Thehigher-loopcontributionscomefromquantuminteractiontermsoriginatingin threeplaces:(a)thehighero rder(cubic,quartic,etc.)termsintheexpansionofthe classicalaction(7.2.24);(b)thegauge-“xingterm;(c)thehigherordertermsintheFaddeev-PopovL agrangian.Thelatterhasthesymbolicform ( antighost ) g host( gaugefixingfunction ),wherethevariationi sthefu llnonlinear

PAGE 454

4367.QUANTUMN=1SUPERGRAVITYvariation( 7.2.19,22,23),withthegaugeparameter Lreplacedbytheghost .Since wehavema detheshifts(7. 4.10)fortheghosts,the“elds Vi, V iwillalsoappear.Thus, thequant umverticesareobtainedfromthehigherorderterms(beyondquadratic)in theexpansionof(see(7.2.24,7.4.6)) SC+ SGFŠ{ [ €+ €( V 1Š iV 2)] Š €[ + ( V 1+ iV 2)] } H€(+ ( V1+ iV2)),(7.4 .15) where H€( L)istheex pression obtainedbysubstituting(7.2.22,23)into(7.2.19).We haveperformedanintegrationbypartsinthesecondterm.Wenotethatwhileboth termsinthegauge“xingfunction(7.4.3)leadtointeractionsoftheghostswiththe background“elds,onlythe“rstterm €H€leadsto(local)interactionsbetweenthe ghostsandthequantum“elds.Thisistheendofthequantizationprocess. Wehave discussedthequantizationprocedureintheformulationwiththechiral compensator .Asdiscu ssedinsec.5.2.d,anotherpossi blechoiceforcompensatorisa realscalarsuper“eld V introducedthrougha variantrep resentation.Thetreatmentof thecorrespondingformulationcanbeobtainedbymakingthesubstitution(eveno she ll) 3 1+( 2+ R R ) V V = V .(7. 4.16) V hasthetransf ormationlaws Ba ck gr o und: V= eiKV Quantum: V= V +( L+ € L€).(7.4 .17) Weusenowth e gauge“xingfunction F= €H€Š1 5 V .(7. 4.18) Theshifts(7.4.10)areagainmade,andbyaproceduresimilartotheonedescribed aboveweobtainthefollowingresults:Thehigher-ordertermsintheactionareagain givenby(7.4.15),withthesubstitution(7.4.16)(butnow SGFdoesnotcontribute).

PAGE 455

7.4.Quantization437However,thequadratictermsarenow S = d4xd4 E EŠ 1[ Š1 2 H€ H€Š1 10 V V +( €i €+ i € €+ 3€i €3 ) +(3 V 1 V1+ V 2 V2)+7 i =11 2 V i V i+7 i =1 i i].(7.4 .19) Here , 3 Vi, Vi ,and ihaveabnormalstatistics,and iarechiral. Ingeneral,thetotal“eldcontentisthefollowing:(a)physical“elds H and (or V ),whichcontributeatallloops;(b)the“rst-generationFaddeev-Popovghosts and ,whichcontri buteatallloops;(c)the“rst-gen erationN ielsen-Kalloshghost 3 and allhigher-generationghosts,whichcontri buteonlyatoneloop;(d)thecatalystghosts Viand Vi ,whichcontri buteatonlymorethanoneloop(beingcanceledattheone-loop levelbythecont ributionfromthe Vs).Wewilldiscussinsec.7.10someofthedierencesbetweentheformulations(7.4.14)and(7.4.19).

PAGE 456

4387.QUANTUMN=1SUPERGRAVITY7.5.Supergravitysupergraphs a.Feynmanrules Inthenextsectionweshallconsiderfurtherthebackground“eldquantization anddiscussitsapplications.InthissectionweconsiderordinaryquantizationanddiscusstheFeynmanrulesforsupergravity-mattersystems.Thereisnoneedtogoagain throughthegauge-“xi ngprocedure.Wesimplytaketheresultsoftheprevioussection andsetthebackground“eldstozero.Thereforethesupergravityquantumactionis givenby(7.4.14,15),where E E =1,allthed erivativesare”atspacederivatives,andall chiral“e ldsordinarychiral.Furthermore,the“elds 3 Vi and i canbedropped sincetheyhavenointeractions(buttheFaddeev-Popov“elds ,andthec atalysts Vi, V ido).Equivalently,wecanworkwith(7.4.15,19),dropping 3 Vi ,and i. Matteractions,covariantizedwithrespectto H€and ,canbe a dded. Fr omthe”atspaceformof(7.4.14a)weobtainordinarypropagators.Inparticularwehave HHpropagator : Š €€p2 4( Š )(7. 5.1) propagator : p€p2 4( Š ),(7.5 .2) andtheusualpropagatorsfor and V .Verti cesareobtainedfromtheexpansionof (7.4.15),aswellasfrommatteractions.For example,considerthekineticactionofa scalarmu ltiplet cov: S = d4xd4 EŠ 1cov cov= d4xd4 ( E )Š1 3 (1 eŠ H)1 3 eŠ H .(7. 5.3) Wehaveex pressed covintermsofa”atspacechiralsuper“eld andweareworkingin thechiralrepresentatio n.Wemustexpandnowthevariousfactorsinpowersof H€and = Š 1.However,theexpansionswerecarriedoutin(7.2.30-32).Replacing backgroundcovariantderivativeswith”atspacederivatives,we“ndthecubic

PAGE 457

7.5.Supergravitysupergraphs439interactions S(3)= d4xd4 { ( + ) + [ Š H ai aŠ1 3 ( D€DH a) Š1 3 ( i aH a)] } = d4xd4 [( + ) + H a(1 2 i a Š1 6 [ D€, D] )].(7.5.4) Thesameexpansionscanbeusedto“ndthesupergravityvertices,butwithbackgr o undssettozerothealgebraismuchsimpler.Thus,from = eŠ HDeH, €= D€, weobtain € B= =€= a = a€=0, a b= Š i D€ b,(7. 5.5) where bisobtainedasthecoecientof bin(7.2.27b): b= i ( DH b) Š1 2 [( DH c) cH bŠ H c( cDH b)]+ ... .(7. 5.6) (To“ndthec ubicinteracti onswedonotneedthethirdordertermin bsinceitonly contributesatotal( D€)derivat ive.)Therefore EŠ1 3 =[ det ( a b+ a b)]Š1 3 =1 Š1 3 a a+1 6 a b b a+1 18 ( a a)2Š1 18 a a b c c bŠ1 9 a b b c c aŠ1 162 ( a a)3.(7. 5.7) We alsoexpand(7.2.31)oneorderhigherwhichgives,againdroppingatermwhichonly contributesatotalderivative, (1. eŠ H)1 3 =1 Š1 3 i ( H ) Š1 18 ( H )2Š1 6 H ( H )+1 162 i ( H )3,(7. 5.8a) eŠ H =1+ + + Š iH Š iH Š1 2 H ( H ).(7.5 .8b) Thecubicsupergravityactionisobtainedfromtheproductof(7.5.7)and(7.5.8).

PAGE 458

4407.QUANTUMN=1SUPERGRAVITYWeobtain thec ubicghost-antighost-quantum“eldverticesfrom(7.4.15)and (7.2.19,22,23).Weneed H€to“rstor derin H€and : H€=( D L€Š D€L)+3( D€LŠ D L€) Š1 2 [ Š i ( D L€+ D€L) €H a+( D2L) DH a+( D2 L€) D€H a+ iH ( D L€+ D€L)].(7.5.9) Thecubicghostactionisobtainedbysubstituting L= + ( V1+ iV2)(cf. (7.4.15)).Theseverticesaresucientfordoingsomeone-loopcalculations.However,as wehavealre adymentioned,atleastforon-shell“ elds,thebackground“eldmethodis muchsimple r.Inthismethod,theabove(covariantized)verticeswouldbeneededonly fortwo-loopc alculations. b.Thetransversegauge TheFeynmanruleswehavediscussedaboveusetheparticular(weighted)gauge of(7.4.3),whichisthemostconvenientforinternallines.However,whencomputing gaugeinvariantquantities,wecanuseanygaugefortheexternallines(thisistruein bothordi naryandbackground“eldmethods).Wediscussherethechoiceofaglobally supersymmetricgaugethatisconvenientformostcalculations. Thesuper“elds H€, containseveralirreduciblerepresentationsofsupersymmetry.Accordingto(3.9.40) thesuperspin contentof H€is(3 2 +1+1 2 +1 2 +0),while has superspin0.Accordingto(3.9.36,37),thespinorgaugeparameters L+L€contain superspins(1+1 2 +1 2 +1 2 +1 2 +0).(Theextrasuperspin1 2 representationsinthegauge parametercorrespondtosecond-generation ghosts.)Therefore,weshouldbeableto“nd a gaugewherewehaveeliminatedallsuperspinsbut3 2 and0.Wehavetwochoices,correspondingtoeliminatingthesuperspin0in (gauging to1),orin H€(inwhich case mustbekep t).Thesecondchoi ceismuchmoreuseful,andcanbeachievedby imposingthe transversegaugeconditionDH€=0.Notethatt hiscondition(andits complexconju gate)implies H =[ D, D€] H€= D2H€=0.

PAGE 459

7.5.Supergravitysupergraphs441c.Linearizedexpressions Forsomec omputations,weneedtheexplicitexpressionsforthegeometrical quantitiesintermsoftheprepotentials H€, .Hereweout linethe procedurefor obtainingthelinearizedexpressions;higher orderscanb eobtai nedinsim ilarfashion.We workinthec hiralrepresentation, andintheLorentzgauge N = .Afterweobtain theresultswewillconsiderotherLorentzgauges. We be ginwiththelinearizedexpressionsof(5.2.78a)(cf.also(7.5.5)) E€= D€ E= D+[ D, H ]= D+ i ( DH m) m(7.5.10) Weset =1+ and,usingforexample(7.5.5-7),wehaveatthelinearizedlevel E =1+ D€DH€(7.5.11) Fromthef ormofin(5.2.78c)weobtain =1+ X (7.5.12) where X 1 2 Š Š1 6 (2 D€D+ D D€) H€.(7. 5.13) IntheparticularLorentzgaugeweareusingweneednotdistinguishbetween”atand curvedindices.Weobtainthen E= E= D+ XD+ i ( DH b) bE€= E€= D€+ X D€.(7. 5.14) To “nd E awe write(againusing N = ) E a= E a+ i1 2 C (€ € ) E€+ i1 2 C€ ,( € )€ E(7.5.15) where E aŠ i { E, E€} .There fore E a= aŠ i ( DX ) D€Š i ( D€ X ) D

PAGE 460

4427.QUANTUMN=1SUPERGRAVITY+[ D€DH b+(1 6 [ D, D€] H dŠ1 2 ( + )) a b] b(7.5.16) Thelinearizedexpressionsforthe C scanbewor kedoutfromtheird e“nition.We“nd “nally E a= a+ i [1 2 D2D( H )€Š ( D€ X ) ] D+ i [ Š1 2 D2 D(€H€ )Š ( DX ) €€] D€+[( D€DH b)+( X + X ) a b] b.(7. 5.17) To “ndtheconnectionsweevaluate“rstthe CAB C,and usethetorsionconstraints. We “nd = Š C ( D ) X €€=1 2 D2 D(€H€ ),€= Š1 2 D2D( H )€, a = i1 2 D D2D( H )€+ iC ( D€D ) X .(7. 5.18) Theindependent“eldstrengthsare R = D2( Š i1 3 aH a), G a= Š2 3 D D2DH aŠ1 6 a b c d b[ D, D€] H dŠ1 3 a bH b+ i a( Š ), W=1 6 D2D( i €H )€.(7. 5.19) Theremaining“eldstrengthscanbereadfromthesolutionoftheBianchiidentities (5.4.16). Aswehavementionedseveraltimes,itis sometimesusefultochooseaLorentz gauge N = inwhich€ =0sothat,inthech iralrepr esentation,whenactingon a“ eldwithundottedindices, € ...= N€€ D€ ....(Thatsuch a gaugeispossible followsfrom R€€ =0,whichim pliesthattheaboveconnectionispuregauge.)We reachthisgaugebytheLorentztransformation A=[ L A], L = M + h c .(7. 5.20)

PAGE 461

7.5.Supergravitysupergraphs443sothat,inparticular, €€= €€Š E€€.(7. 5.21) Atthe linearizedlevel,setting €€=€€+ €€=0,w e “nd €€=1 2 D D(€H€ )(7.5.22) and = ( €€).Inthisgauge N€€= €€+ €€. Inthisgauge,thevariousquantitiesof(7.5.19)areshiftedaccordingtotheirindex structure.Inparticular,we“ndthatnow = Š C ( D ) X +1 2 D D€D( H )€.(7. 5.23) d.Examples Inthissubsectionweassumethataregularizationschemeexiststhatpreserves localsup ersymmetry.Suchaschemewillbediscussedinsec.7.9.We“rstcomputea masslesschiralloopcontributiontothesuperg ravityself-energy.Therelevantinteractionisgivenby(7.5.4),andthesupergraphisgiveninFig.7.5.1. k + p Hb( k ) Ha( Š k ) p Fig.7.5.1 Wenote thefollowingsimpli“cations:(a)The( + ) vertexleadstoo nlyatadpole contributiontothe or H self-energydiagram,andwesetthistozeroindimensional regularizationformassless s.Equivalently,weobservethatintheoriginalaction (7.5.3)thecompensator canbeabsor bedinto bya“eldr ede“nition.(b)Withsuitableregularizationtheresultshouldbegaugeinvariant,andwecanworkinthe

PAGE 462

4447.QUANTUMN=1SUPERGRAVITYtransversegaugewheretwoofthethreetermsinthe H vertexdo notcontribute.The Š H a i a ve rtexisthesameasintheYang-Millscase,ifwereplace VATAby H ai a. Theresultcanthenbereadfrom(6.3.31)(withtheadditionalmomentumfactorsfrom i a)1 2 d4k (2 )4 d4 H a( Š k ) d4p (2 )4 Š p2Š p c D€D+ D2D2p2( k + p )2 p a( k + p ) bH b( k ).(7.5 .24) Usingthegaugecondition thiscanber educedto Š1 8(D Š 1) d4k (2 )4 d4 H a( Š k ) k2k c D€DH a( k ) I ( k2),(7.5 .25) whereindimensionalregularization I ( k2)= dDp (2 )D 1 p2( k + p )2 =1 (4 )1 2 D (2 Š1 2 D)[(1 2 DŠ 1)]2(DŠ 2) ( k2)1 2 D Š 2=1 (4 )2 ( 1 Š lnk2+ const .).(7. 5.26) Whenac tingon H a( k ),againusingthegaugecondition,wecanrewrite k c D€D= k c1 2 { D€, D} = k2.Thefu llycovarian tresultcanbewri ttenasacontributiontotheeectiveactionoftheform[ c1 d2 ( W)2+ h c .+ c2 d4 ( G2+2 RR )] I However,thecoecients c1, c2cannotbedeterminedfromjustatwo-pointcalculation (ex ceptinthebackground“eldmethod:seesec.7.8).Onshellonlythe“rsttermsurvives.Althoughther esultisindependentof ,thecomp ensatorreappearsinthecourse ofseparatingoutthedivergentpart(seesec.7.10). Asas econdexamplewecomputesupergravitycorrectionstothechiralself-energy. ThegraphsarethoseofFig.7.5.2:

PAGE 463

7.5.Supergravitysupergraphs445 ( k ) ( Š k ) HH Fig.7.5.2 The“rstgraphgivesnocontribution(after D -algebra itisatadpole)whiletheothers addu pto d4k (2 )4 d4 ( Š k ) d4p (2 )4 1 p2( k + p )2 [ Š5 9 D2D2+1 9 p k Š2 9 ( p Š k )2] ( k ). (7.5.27) (The Š5 9 forthe pr opagatorfollowsfromitsnormalizationin(7.4.14).)Intheintegralwecanreplace p k = Š k2and p2=0.Thusthe totalresultvanishes.

PAGE 464

4467.QUANTUMN=1SUPERGRAVITY7.6.CovariantFeynmanrules Tothequanti zedsupergravityactionofsec.7.4wecanaddcovariantizedsupersymmetricmatteractionsandconsidergeneral matter-supergravitysystems.Ifthematteractionscontaincovariantderivatives,thesemustbesplitasinsec.7.2.Forconstrainedsuper“eldswemust“rstextractexplicitquantum“elddependence(e.g., eŠ H ,where ,arebackgr oundcovariantlychiral).Inprinciplewecanalsosplit mattersuper“eldsintoquantumandbackgroundpartsandconsiderageneralquantum systeminabackgroundofmatterandsupergravity.However,ingeneraltheprocedure ofsec.7.4isnotapplicable.Wecannotimposetheon-shellconditions R R = G G€=0. Theseconditionsmustbereplacedbytheequations R R = J ( matter ), G G€= J€( matter ) andthequantizationmustbecarriedoutwiththesupergravity“eldso-shell.Thisisa straightforwardbutalgebraicallycumberso meprocedure.Thereforeinthissectionwe willconsideronlypureon-shellsupergravitybackgrounds(noexternalmatter).General systemscanbehandledbyanextensionofourquantizationmethodsorbytheordinary (n on background)quantizationoftheprecedingsection. Giventhebackground“eldLagrangianwithquantummatterorsupergravity “elds,theFeynmanrulescanbederivedinexactlythesamewayasforglobalsupersymmetry.Ingeneral,forunconstrainedquantumsuper“elds,wecanreadtherulesdirectly fromtheL agrangian.Wehavetwotypesofvertices:thosearisingfromquantumselfinteractions,andthosecontainingalso(oronly)interactionswiththebackground“elds. Thebackground“eldsappearonlythrough“eldstrengthsandbackgroundcovariant deriva tives A= E EA MDM+ A,withthe ”atsuperspace DM.Therefo re,wewill encounterverticeswithallquantumlines,orwithamixtureof(atleasttwo)quantum linesandbackgroundlines.Forco nstrained,i.e.,backgroundcovariantlychiralsuper“elds,wemust“rstofallsolvethec hira lityconstraints,i.e.,write = e 0intermsof anordinarychiralsuper“eld.Thiswillintroduceinteractionsinvolvingexplicitlythe backgroundpotentials.Weshalldiscussbelowhowtoavoidthis,butatanyrateweend upwithanactiontowhichthemethodsofchapter6canbeapplied,withordinarypropagatorsandrulesforcalculation. Weobservethat attheone-looplevel, thecontributionfro mthege neralspinors canalsobeobtainedbysquaringtheirkineticoperatorandtakinghalfoftheresulting contributiontotheeectiveaction.Wehave

PAGE 465

7.6.CovariantFeynmanrules447( i €)( i €)=1 2 ( i €)( i €)+1 2 [ i €, i €] = + { W W W W M } ,(7. 6.1) wherewehaveused(5.4.16).Thespinoractionin(7.4.14a)thusbecomes d4xd4 E EŠ 1 3 i =11 2 i ( + { W W W W M } ) i + h c = d4xd4 E EŠ 1 i1 2 i i + h c .,(7.6 .2) andweobservethat allthe unconstrainedsuper“elds( H V )aredesc ribedbysimilar actions,withthesameoperator givenby(7.4.14b).Weshalldiscusslaterapplicationsofthi sresult. Wenowdes cribeamodi“cationoftheFeynmanr ulesforcovariant lychiralsuper“e lds,analogoustothemodi“cationfortheYang-Millscaseinsec.6.5.Theconsequencesofthemodi“cationare:ItguaranteesthattheFeynmanrulesforchiralsuper“eldswillnotintroduceexplicitbackground gaugepotentials,butonlythevielbeinand connections,anditactua llysimpli“essomeofthe D -algebra.Wefo llowapro cedure thatisidenticaltot hatofsec.6. 5.We“rst de“ne covariantfunctionaldierentiation forageneralsuper“eldby ( z ) ( z) E 8( z Š z),(7.6 .3) whichgives( ) d8zEŠ 1L = L .Wethende “necovariantfunctionaldierentiation foracovariantlychiralsuper“eld (w hichcouldcarryadditionalundottedspinor i ndices,butwedonotindicatetheseexplicitly)by ( z ) ( z) ( 2+ R ) E 8( z Š z)= Š 3 D28( z Š z),(7.6 .4) wherethesecondformisobtainedbyusingtheidentity(5.3.66b)andthechiralrepresentation( E€= N€€ D€)withthe particularLore ntz gaugewhere N = suchthat E€ = E€ €=€ =0.(7. 6.5)

PAGE 466

4487.QUANTUMN=1SUPERGRAVITYInthisrepresentationcovariantlychiralsuper“eldsarechiralintheusualsense: €...=0imp lies D€...=0 .J us ta si nt he Ya ng -M illscase,atthispointweneednot beexp licitastowhetherthechiralcovarianceiswithrespecttofullderivatives(containingbothbackgroundandquant um“elds)orjustbackground“elds,andtheobjects appearingin(7.6.4)canbefunctionsofboth,orjustbackground“elds(exceptwhenthe chiralsuper“elds aresupergravitysuper“elds,inwhichcasethecovariantderivativescan on ly be background).Wecanstayo-shell. Thecovariantizationoftheusualexpression D2D2 = b ecomesnow ( 2+ R )( 2+ R ) ... = + ... += + W M +1 2 i ( €G€) M Š1 2 iG€€Š R 2Š1 2 ( R ) + R R +( 2 R ),(7.6 .6) generalizing(7.4.4)oshell.Weobservethatonshell(recallingthatchiralsuper“elds canonlyhaveundottedindices) += ,wherethela tterquantityw asde “nedin (7.4.14b),aresultwhichweshalluselater. Asinsec.6.5cwestartwiththeaction S = S0+ Sint( ), S0= d4xd4 EŠ 1 .(7. 6.7) Sintalsocontainstheotherqu antum“elds butwehaveindicatede xplicitlyonlythe dependenceon .Wecon centrateonthefunctionalintegralover whichgives,using (7.6.3,4), Z ( J J )= ID ID exp [ S +( d4xd23J + h c .)] = [ expSint( J J )][ exp ( Š d4xd4 EŠ 1 J + Š 1J )],(7.6.8) whereisthefunctionaldeterminant = ID ID eS0.(7. 6.9) Ingeneraltheaboveexpressionfor Z dependson,andistobeintegratedover,theother

PAGE 467

7.6.CovariantFeynmanrules449quantum“elds.Weare consideringthemasslesscase,buttheresultsforthemassive casecanbeobtainedeasily.Exceptfor,theotherfactors,whichcontainquantum “eldself-interactions,contributeonlybeyon doneloop,ortod iagramscontainingexternalchirallines. Thedeterminantgivesthecompleteone-loopcontributionfromchiralsuper“eldsofdiagramswithonlyexternalsupergravitylines,andcouldbeevaluatedbyusing standardsuper“eldFeynmanrules,butwewishtoavoidthis.Instead,weshallusethe do ub lingtrickasinsec.6.5c.(Insupergravitywearealwaysdealingwithrealrepresentations).Wenowhave O O + J J =0, O O O O = 0 2+ R 2+ R 0 .(7. 6.10) Itssquare, O O2 Š J J =0, O O O O = ( 2+ R )( 2+ R ) 0 0 ( 2+ R )( 2+ R ) ,(7. 6.11) correspondstoanaction S 0= d4xd231 2 + = d4xd4 EŠ 11 2 ( 2+ R ) ,(7. 6.12) andintermsofitwecanwritethefunctionalintegral 2= ID ID exp [ S 0( )+ h c .]=( ID eS 0)2.(7. 6.13) Weinte grate S 0bysepara tingout 3D2from EŠ 1( 2+ R ),treating1 2 [ EŠ 1( 2+ R ) Š 3D2] asaninteractionterm.Theresultis = ID eS 0= { exp d4xd231 2 J [( 2+ R )( 2+ R ) Š D2D2] J } [ exp Š d4xd231 2 J 0 Š 1J ] |J =0.(7. 6.14) (Notethatwritinginstead( 2+ R )( 2+ R ) D2EŠ 1Š 3eŠ HD2EŠ 1 Š 3eHwouldgive

PAGE 468

4507.QUANTUMN=1SUPERGRAVITYtheusualru les,ex ceptfortheextra 3sfromthede“nition(7.6.4)).Therefore,acalculationoftheone-loopcontributionof totheeectiveaction(i.e., ln )consistsinevalua tinggraphswithpropagators pŠ 24( Š )andvert i ces[( 2... ]giv ingrisetoastring ... [( 2+ R )( 2+ R ) Š D2D2]i4( iŠ i +1)[( 2+ R )( 2+ R ) Š D2D2]i +1... (7.6.15) withd4iintegralsateachvertex.Weconcentrateonagivenvertexandatthenext onew erewrite( 2+ R )= D2Š 3EŠ 1.Wetempora rilytransferthe D2factor across the -functionandusetheidentity [( 2+ R )( 2+ R ) Š D2D2] D2=( +Š 0) D2.(7. 6.16) Wefurthersim plifytheexpressionbyusingtheanticommutationrelationstomovethe D sin +totherightuntiltheya reannihilatedbythe D2.Theresu ltingexpression, whichwecall +,containsno D s.Wenowreturnthe D2factortoitsoriginalplace, reexpressthevertexinitsoriginalform,andproceedtomanipulateitinthesameway. Wecancontinuea roundtheloopandtreatinthiswayallverticesbutthelast,andwe areledtothefollowingrules: onevertex : D2[ Š 3EŠ 1( 2+ R ) Š D2], othervertices : +Š 0.(7. 6.17) Themassivecaseisobtainedsimplybyaddingamassterminthedenominatorofthe pr opagator. Theserulesleadtoasimplerevaluationoftheone-loopcontribution,sincethere areno D sintheloopexcepttheone D2, butmoreimportantlythecontributionis manifestlyexpressibleonlyintermsofobjectswhichappearinthecovariantderivatives, andnotthegaugeprepotentials.Thisisevidentlytrueofthehigher-loopcontributions aswell.From Sintandthede“nitionofthecovariantfunctionalderivativein(7.6.4),the expression(7.6.8)leadstohigher-loopFeynmanruleswhichdonotexplicitlydependon thebackgroundprepotentials.Weobtainpropagators + Š 1forchirallines,wherethe full +canbeexpressedintermsofthequantum H€, ,andthebackgro undcovariant derivati ves.From Sint( J J )weobtainverti ceswithfactors( 2+ R ) E or( 2+ R ) E

PAGE 469

7.6.CovariantFeynmanrules451operatingoneachchiralorantichirallineleavingthevertices.(Thesegeneralizethe ordi nary”atspacerules.)Againwecanexpressthesequantitiesintermsofthequantum H€and ,andthebackgro undcovariantderivatives .For anactualmomentum spacecalculationthesehavetobefurtherexpressedintermsofordinaryderivativesand backgroundvielbeinandconnections.Wenowhavetheresultthatforallsuper“elds, whencalculationsarecarriedoutinthebackground“eldmethod,thecontributionsto thee ectiveactionfromindividualgraphsdonotinvolvethebackgroundsupergravity prepotentialsthemselves,butonlyvielbeinandconnections,whichdependon (multi)derivatives oftheprepotentials.Consequentlythereissomeimprovementinthe powercountingrul esforpotentiallydivergentgraphs.

PAGE 470

4527.QUANTUMN=1SUPERGRAVITY7.7.Generalpropertiesoftheeectiveaction Wean alyzeinthissectio nthege neralformoftheeectiveaction(background “eldfunctional),asconstrainedbytherequirementofbackground“eldinvariance. Thisanalysisisparticularlyimportantfordeterminingthedivergencestructureofsupergravity.Divergences,whichcouldbecanceledbycountertermsintheLagrangian,correspondto local termsintheeectiveaction,andtheirformislimitedbygaugeinvariance anddimensionality.Insomecasesweobtainstrongerresultsbyrestrictingourselvesto on-shellbackgro und“elds.Theon-shellrestrictionisnotserious:Thetheoryisnotperturbativelyrenormalizable,Greensfunctionsaregauge-dependentanddivergent,andat bestwecanhopethat gauge-indepe ndent,on-shellquantities(e.g.,theS-matrix)are “nite.Therefore,onlythedivergenceswhichdonotvanishon-shellaresigni“cant (divergenceswhichareproportionaltoth e“eldequatio nscanberemovedbya“eld rede“nitionwhichdoesnotaecttheS-matrix). Wew illdisc uss“rstthesituationin N =1superg ravity.Thediscussionisapplicablethentoextendedsupergravityexpressedintermsof N =1super“elds .Howev er, strongerstatementscanbemadeiftheextendedtheoriescanbeexpressedintermsof extendedsuper“elds.Sincethediscussiondoesnotdependondetailsoftheextended super“eldconstructions,butonlyonpropertiesthatgeneralizeour N =1back ground quantizationmethods,wedevoteasubsectiontothiscase. a.N=1 Ourbackground“eldsaresupergravity“elds,whilethequantum“eldscanbe supergravityormattersuper“eldsorboth.Sinceinthebackground“eldformalismthe eectiveactionisgaugeinvariant,itcanbeconstructedfromthe“eldstrengths R R G G€, W,andcovarian tderivat ives(withanoverallfactorof E EŠ 1,or 3forchiralintegrands).Furthermore,onshell R R = G G€=0.(Wecons ideronlyvanishingcosmological term:Otherwise, R R isanonvanishingdimensionalconstant,andthedimensionalanalysisischanged.)Thus,onlythechiral“eldstrength W(anditsc omplexconjugate) andcovariantderivativescanappear.Wealsohavetheon-shellconditions W= €W=0aswell asthecorrespondingequationsfor W .Indet ermining theformofthee ectiveactionwecanalsousethefollowingfacts:(a)Theeective actionisdimensio nless.Thedimensionsofthevariousquantitieswhichcanappearare

PAGE 471

7.7.Generalpropertiesoftheeectiveaction453[ d4x ]= Š 4,[ d2 ]=1,[ W ]=3 2 ,[ ]=1 2 .Ina ddition,withonlysupergravityinteractions,foran L -loopcontributio nwehaveafactor 2( L Š 1)withdimension Š 2( L Š 1).(b) F unctions G ( x1,....)arisefromloopintegralsafterallthe D -algebrahasbeencarried outa nd,iftheyhaveodddimension,mustcontainanoddnumberofspace-timederivatives(momentumfactors)wh ichhavea ni ndexstructure €.(c)Inte gralswiththechiralmeasure d2 musthavechiral integrands,i.e.,factorsof W or 2( W2)( 2 W =0 onshell),etc.(butinthelattercasetheycanberewrittenasfullintegralsanyway).(d) Dottedandundottedindicesmustbeseparatelysaturated. Anotherimportantfeatureisthefactthatallthe d4 termsinhaveanequal nu mb erof(spinor-)undierentiated W sand W s.Thisi sacons equenceoftheglobal chiralRinvarianceofthetheory(cf.(5.3.10);the Y transformationsareglobalinvariances;intermsofprepotentials,theyaresimplyphasetransformationsof ,leaving H invariant).Weshouldalsoremarkthat,apriori,asdiscussedintheprevioussection, pertur bationtheorydoesnotleadtoaforminvolvingonlythe W sandtheircovariant derivatives,butratherthequantitieswhichappearinthebackgroundcovariantderivatives,i.e.,backgroundvielbeinandconnectioncoecients,witha d4 integral.However, b ecauseofbackgroundinvariance,thesequantitiesmustarrangethemselvesintoaform thatismanifestlycovariantorcontainsonenoncovariantfactortimesacovariantobject thatsatis“esaBianchiidentity.(Thenon covariantterm,whenvaried,producesa derivativewhich,whenintegratedbyparts,giveszerouponuseoftheBianchiidentity.) Thus,thetermd4xd2 3W2reallyarisesfromanexpression(inthegauge(7.6.5))d4xd4 E EŠ 1 W,whichcant henberew rittenasachiralintegral. We“rst discussall local termsin,i.e.,allpossibleon-shelllocaldivergencesof thetheory.Ag enericlocaltermwillhavethestructure ( a)l( W W )m( W )n( € W )r,(7. 7.1) with W ( W )andtheindicescontractedinvariousways,andthespace-time derivativesdistributedinvariousways.WehaveusedtheinvarianceunderR-transformationstowriteonlytermswithequalpowersof W and W .Wenotethat unlesssome space-timederivativesactonthem, W and W cannotberaisedtoapowerhigherthan 4,becausetheyaresymmetricintheirthreespinorindicesandhencecontainonlyfour i ndependentLorentzcomponents.Thedimensionalityoftheabovetermis

PAGE 472

4547.QUANTUMN=1SUPERGRAVITYl +3 m +2 n +2 r .Ifitappearsat L loopsitismultipliedby( 2)L Š 1withdimension Š 2( L Š 1).Theoveralldimensionoftheactionmustbezero: d = l +3 m +2 n +2 r Š 2( L Š 1) Š 2=0.Theonly purelychiralterm,involvingjust W ,isthequadra ticexp ression d4xd2 3WW+ h c .= Š d4xeŠ 1ww+ h c .(7. 7.2) where w is th eW eyltensor.Ondimensionalgroundsitcanonlyappearattheone-loop level(withno factor,asf ollowsfromourdiscussionabove),andtheintegrandisa to ta ld erivativeonshell.Itis,infact,ondimensionalgrounds,theonlylocalterm(i.e., po ssibledivergence)whichcanoccuratthe one-looplevel,onshell.However,dueto theGauss-Bonnettheorem,itisjustatopologicalconstant,andvanishesintopologicallytrivialspaces.(Forafurtherdiscussion,seesec.7.10.) Atthetwo-loopl evelnolocaltermsarepossible.Wehaveafactor 2ofdi mension Š 2,anditiseasytocheckthatthereisnoway,fromamongthegenericexpression above,to“ndeitherachiralexpression(tobeintegratedwith d4xd2 ),orag eneral expression(tobe integratedwith d4xd4 ),whichcanleadtoatermwithdimensionzero. Thus,atthetwo-looplevelnoon-shelldivergencescanariseinsupergravity. Athigherlo opsthenumberandvarietyoflocaltermsincreases.Forexample,at threeloops ,thecomb ination W2 W2isapossiblelocaltermandthereforeapotentially divergentone(theonlyon-shellone,infact).Atanygivenloopthereareofcourselimitationsduetodimensionality,chirality,andindexcontraction.Inparticularitiseasyto ve rifythat,ondimensionalgroundsandinordertosaturateindices,allhigherloop termsmusthavefactorsofboth W and W andthereforevanishwheneither W =0or W =0.Thissit uationdescribesbackground“eldcon“gurationswhichareself-dualor antiself-dual.Weconcludethatsuchcon“gur ationsreceivenoradiativecorrections. Thenonlocalpartoftheeectiveactionhasastructureasin(7.7.1),including howeveranonlocalfunction G ( x1, x2,....)andwith“eldsevaluatedatdierentpoints ( x1, ),( x2, ), ... .W e “nd,usingsuperspaceperturbationtheoryanddimensionalanalysisthat,ifnomassive“eldsarepresent, theon-shelle ectiveactionhastheform d4x1d4x2d2 3W( x1, ) G ( x1, x2) W( x2, )+ h c .

PAGE 473

7.7.Generalpropertiesoftheeectiveaction455+ d4x1... d4x4d4 W( x1, ) W( x2, ) W€€€( x3, ) W€€€( x4, ) G ( x1... x4)+ ... ,(7. 7.3) wherethetermsnotexplicitlywrittencontainmorethanfour W sandtheircovariant derivatives.Thenonlocalfunctions G ( x1, x2,...)canbethoughtof aspolynomialsin thespace-time covariantderivativesactingonthe“elds,andfunctionsofthecovariant dAlembertian(e.g.,itsinverse),correspondingtotheresultofdoingvariousloopintegralsinmomentumspace.Theimportantpointisthatother,aprioripossibleterms withtwo,three,orfour W sarenotpresent.(Forexample, d4 WW W cannothaveits indicessaturatede venifderivat ivesareinc l udedwhile d2 ( W )4hasthewrongdimension,etc.)Insecs .7.8 ,10weshalldisc ussinmored etailtheformofthe G functions. b.GeneralN Weshalla ssumeinth issectionthatunconstrainedsuper“eldformalismsexistfor allsupersymmetricsystemsofinterest.Suchformalismshavenotyetbeendeveloped exceptfor N =2 ,a nd thereareindicationsthatiftheyexisttheyhaveanunfamiliar form.Weshallonlyassumethatthereexistco nstraintsonthecova riantderivatives thatallowthem,andtheaction,tobeexpressedintermsofordinaryderivativesand unconstrainedprepotentials.Wecanthenmimicthe N =1 background-quantumsplittingforg eneral N .Werepla cetheunconstrainedprepotentialsbyquantumprepotentials,andtheordinary derivativesbybackgroundcovariantderivatives.Furthermore,if covariantlyconstrained (e.g.,chiral)super“eldsarepresent,wecanderivecovariantrules forthe maswed idin N =1;the procedureisgeneral.Wewillnotrestrictourselvesto on -shellbackgrounds. Byanextensionofourfullycovariantbackground“eldmethodofsec.7.6,we obtainimprovedpower-counti ngrulesfordiscussinglocaldivergences.Theserulessimplyfollowfromthefactthat allquantumtermsintheeectiveactionareautomatically expresseddirectlyintermsoftheconstrainedbackgroundcovariantderivatives (andtheir “eldstrengths)andanexplicitexpansionintermsofunconstrainedbackgroundprepotentialsisunnecessary.(Onemightalsoexpect toneedthesup erspacegeneralizationof antisymmetrictensorgauge“elds,e.g.,thethree-formofD=11supergravity,butthe background-quantumsplitactioncanalwaysbewritteninaformwheresuch

PAGE 474

4567.QUANTUMN=1SUPERGRAVITYbackground“eldsappearonlyastheir“eldstrengths,andthese“eldstrengthsalready appearamongthe“eldstrengthsofthebackgroundcovariantderivatives.)Thisimplies thatalldivergenttermsmustbeexpressibleas localfunctionsofth ecov ariantderivatives. Thus,forsupersymmetricYang-Mills,wherethesameideasapply,allcountertermsmustbelocalfunctionsof ,andfors upergravityof andalso E M.(Conventionalconstraintsdetermineallof Afrom and E M.)Furthermore,becauseinthe derivation oftheFeynmanrulesverticesarealways integratedoverfullsuperspace, they willcarryafulld4xd4 N for N -extendedsupersymmetry. Forthecaseof extendedsupersymmetry,treatedwithextendedsuper“elds,there isatec hnicaldicultyinthebackground“eldmethodbecauseoftheappearanceofan i n“nitenumberofgenerationsofghostsuper“eldswithprogressivelyincreasingsuperspin.Forexample, N =2 Ya ng -M illstheoryisdescribedbyarealisovectorsuper“eld Va bwithgaugeinvariance Va b= Dc ( a bc ) + Dc€ ( ac b )€.Thistra nsformation implies thatthecorrespondingghosthasagaugeinvariance ( abc ) = Dd ( a bcd )( ),whichin turnimpliesaghostwithinvariance ( a bcd )( )= De ( a bcde )( ),etc.T he gaugesuper“e ldsunavoidablycontain“eldsofspinhigherthanthose(physicalandauxiliary)occurringinthegauge-invariantaction.Togaugetheseawaythegaugesuperparameters(and ther eforethecorrespondingghosts)mustcontainhigherspinsthanthegaugesuper“elds. However,onlya“nitenumberofghosts(i.e.,theusualFaddeev-Popovghosts,plusperhapscertaincatalystghosts)contributeatmorethanoneloop.Therefore,thehigherloopcontributionstotheeectiveactioncanbecalculatedinamanifestlybackground covariantformandwillobeythepower-countingrulesthatwederivebelow,whereasthe one-loopcontributionmayhavetobetreatedseparately.(Forexample,wecouldchoose backgroundnoncovariantgaugesforsomeoftheghostswhichcontributeonlyatone loopinsuchawaythatallbuta“nitenumb eroftheseghostsdecouple.Theeectof suchachoicewouldbetoproduceaone-loopeectiveactionwhichisnoncovariant,but thiswouldhavenoeectonphysicalquantities).Wediscussnowtheimplicationsof theser emarksandtheimprovedpower-countingrulestowhichtheylead.Forcompletenesswediscuss“rstthesituationinglobaltheories. The“rstexampleoftheimprovedpowercountingwasalreadygiveninsec.6.5for theFayet-IliopoulosD-termin N =1 Ya ng -M illstheory.Backgroundcovarianceimmediatelyimpliesthevanishingofsuchatermbeyondoneloop.For N > 1weobtain

PAGE 475

7.7.Generalpropertiesoftheeectiveaction457strongerresults: Si nc eY ang-Millstheoryisrenormalizable,theonlyalloweddivergenceinthebackground“eldmethodisproportionaltotheclassicalaction.However,beyondoneloopit musthavetheformd4xd4 N atthelinearedlevelbecauseofthecovariantFeynmanrules.Here hasdimension1 2 andisalocaloperator(nonnegativedimension). Sincetheactionisdimensionless,weobtaintheinequality Š 4+2 N +1 2 +1 2 0,which impliesthatonly N =0or1c anhavedivergencesbe yondonel oop. Thus N =2and N =4 su pe rsymmetricYang-Millstheorymustbe“nitebeyondoneloop.Furthermore, weknowfromex p licitone-loopcalculationsusing N =1super“elds(s eesec.6.4)that N =4isoneloop“niteaswell. Ontheotherhand, N =2does haveone-lo opdivergences.(Also,asfor N =1,loopcorr ections tothe N =2Fayet-I liopoulos termvanish.)Weem phasizethatwehadtomakeaseparate one-loopargumentbecauseofthe problemwithin“nitenumbersofghosts. Wecana pplysim ilarargumentsto N -extendedsupergravity.Thelocal(divergent)partoftheeectiveactionconsistsoftheintegralof EŠ 1timesa(covariant)productoffacto rsofvielbeinandconnections.At L loopsthelowestdime nsionalsuchterm is loc 2( L Š 1) d4xd4 N EŠ 1,(7. 7.4) mult ipliedperhapsbysomefunctionofadimensionlessscalar“eldstrengthfor N 4; suchafunctionmayhoweverbeforbiddenbyglobalon-shellinvariance(otheradditional factorswouldhavepositivedimension).Requiringthisexpression,possiblymultiplied byapolyn omialinthe “elds(withnonnegativedimension),tobedimensionless,we obtaintheinequality Š 2( L Š 1) Š 4+2 N 0,whichimplies L N Š 1.(SimilarargumentsforYang-Millsgivetheimproved Š 4+2 N +2 0inste adoftheabove Š 4+2 N +1 0.)Thus,fromtheseargumentsalone,we“ndthatin N -extended supergravitytheeectiveactioncanhavelocalterms,andthereforepossibledivergences, onlyat N Š 1loopsandbey o nd.(Thisissoeventhoughpossiblelower-loopinvariants canbeconstructed.TheimportantpointisthatourFeynmanrulesimplyintegration overfullsuperspa cewithintegrandsthatinvolvecovariantobjects.)Notethatthe divergencesexcludedbytheserulesareabsent bothon andoshell.

PAGE 476

4587.QUANTUMN=1SUPERGRAVITYAsim ilaranalysisinhigherdimensionsgivestheresultthathigher-loopdivergencesareabsentissupersymmetricYang-Millstheoryfor L < 2N Š 1 D Š 4 ,andinsu perg ravityfor L < 2N Š 1 D Š 2 (for L loopsinD-dimensions,where N referstothefour-dimensional value, i.e.thenumberofanticommutingcoordinatesis4 N ).Forlower dimensions (super-)Yang-Millsisrenormalizableanyway;forsupergravitytheaboveinequalityholds forD=3whileforD=2we“ndhigher-loop“nitenessfor N > 1. Ourbackground“eldapproachleadstoafurtherresultwhichisnotapparentin ordinaryquantizationornonsupersymmetricgauge“xing:Attheone-looplevel,in N =1lang uageandusingthebackground“eldformalism,theonlycontributionstothe (on-shell,topological)divergencesareproportionalto( W)2andcomefromchiral super“elds.Tounderstandthisweobservethatthedivergenceisjustacovariantization ofthedivergenceinthetwo-pointfunction,a nditscoecientcanbedeterminedbycalculatingaself-energydiagram.However,inourgauge,examinationofthequadratic actionin(7.4.14)(whichgivesthegeneralformforanytypeofsuper“eldinanon-shell supergravitybackground),revealsthatonlychiralsuper“eldverticeshaveenough D s and D stogivenonzerocontr i butions.Therefore inatheorywithanetzeronumber (physicalminusghost)ofchiralsuper“elds (any N 3theorywit ha ppropriatechoiceof auxiliary“elds(compensatingmultiplets)) thereareno(topological)one-loopdivergences. Atthetwo-l oopl evel no supergravitytheoryhason-shelldivergences. Wesu mmarizeourresultsinTable7.7.1,whichlistsallcaseswheredivergences mustbeab sentinpuresupersymmetricgaugetheories.Theresultscanbeclassi“ed intothreetypes:(A)absenceofdiverge ncesduetoone-loopcancellationsin N 3 supersymmetryofcontributionsof N =1chiralsuper“elds;( B)absenceoftwo-loop supergravitycountertermsbecauseinvaria ntsofa ppropriatedimensiondonotexist;(C) absenceofdivergencesathigherloopswhichisestablishedbyourargumentsabove. Theabsenceofhigher-loopdivergencescannotbeestablishedrigorouslyuntilthe correspondingsupergraphrule sareexp licitlyconstructed.Possibledicultieswithcarryingouttheprogramareinfraredproblemsduetolargenegativepowersofmomentain thesuper“eldpropagators,andtheexplicitconstructionoftheclassicalaction(whose formmaysurpriseus,ifthepropertiesofextendedsuperspacearenotasimpleextension ofthosefor N =1superspa ce).However,weemphasizethatoncetheactionhasbeen

PAGE 477

7.7.Generalpropertiesoftheeectiveaction459 loops N 123456 7 Ya ng -M ills0 1 2C CCCCC 4A C CCCCC supergravity0 1B 2B 3AB 4AB,C 5AB,CC 6AB,CCC 8A B, C CCCC Table7.7 .1.Absenceofdivergencesinsupersymmetrictheories written,thepowercountingrulesandourconclusionsimmediatelyfollow.Wenotethat inthe N =4 Ya ng -M illscasethe“nitenesscanalreadybeprovenwhenthetheoryis writteni nte rmsof N =2super“el ds, i.e., N =2 Ya ng -M illscoupledtoan N =2sc alar mult iplet;the N =2powercountingrul escanthenbeapplied.

PAGE 478

4607.QUANTUMN=1SUPERGRAVITY7.8.Examples Inthissectionweshallgivesomeexamplesandapplicationsofourcovariantformalismforcomputingsupergraphsinsupergravity.Werestrictourselvestoone-loop calculat ions.Higher-loopcalculationsarepossible,butthealgebraiscomplicatedif thereareinternalsupergravi tysuper“elds.W eshallco nsider“rstsomeone-loopcalculationswithmatter“eldsinsidetheloopand backgroundsupergravitysuper“elds.The algebrasimpli“esconsiderablyintheon-shellsituation. We be gi nb y “ndingone-loopchiral-“eldcontributionstothe(covariantized)onshelltwo-pointfunction,correspondingtothe“rsttermintheon-shelleectiveaction (7.7.3).Incontrasttothecalculationofsec.7.5.d,theseparatecoecientsofthe W2and G2+2 RR termsintheeectiveactioncanbede terminedfromthe two-poin tf unctionalonewhenthecovariantrulesareused(althoughherewe“ndonlytheformerterm sincethelattertermvanishesinouron-shellcalculation).However,wemustusedimensionalregularizationtokeeptrackoftermsthataretotalderivatives only infour dimensions,sinceinafour-dimensionalmomentum-spaceFeynman-graphcalculationtheyvanishbymomentumconservation.Weshallusetheon-shellconditionsonthebackground super“elds,butkeeptheexternalmomentum k oshell( k2 =0)intheloop integral. Also,weshallwriteourexpressionsinfourdimensions.However,thecalculationshould beca rriedoutinDdimensions,bothtoavoidultravioletdivergences,andtocircumvent thefactthat,whenD=4,thelinea rizedresultisatotaldivergence. Weconsid erachiralsuper“eld ...€€ ...with2 A undottedand2 B dottedindices, andaction1 2 d4xd2 3 + + h c ..(If B =0such“eldsc anexistonlyi non-shell backgrounds.)From(7.6.17),andintheLorentzgauge € =0,the lineari zedvertices are(onshell E EŠ 1= =1) One vertex : D2( 2Š D2) D2[ E E a aD+1 2 ( aE E a) D+ M D],(7.8 .1a) Other vertex : +Š 0 E E a aD+1 2 ( aE E a ) D+ W M D.(7. 8.1b) Thepropagatoris pŠ 2 ... €€... 4( Š ). The D -algebra istrivial.The M termsgiveacontributionproportionaltothe nu mb erofundottedindices,andwehaveafactorfromatraceoverallspinorindices.

PAGE 479

7.8.Examples461We “ndacontributiontotheeectiveaction 2=( Š 2)2( A + B ) d4k (2 )4 d4p (2 )4 1 p2( k + p )2 1 4 d2 [( p +1 2 k ) a( p +1 2 k ) bE E a ( Š k ) E E b ( k )+ A W( Š k ) W( k )]+ h c (7.8.2) Wehave usedthelinearized,on-shellrelations W= D2 E E a = D2E E a,and convertedthe d4 integraltoa d2 integral.Finally,doingthemomentumintegraland usingthe lineari zedrelation aE E b Š bE E a = C€€W,(7. 8.3) weobta intheresult 2=( Š 2)2( A + B )(1 12 Š A )1 2 d4xd21 2 W( x ) I ( Š ) W+ h c .(7. 8.4a) withthelogarithmicallydivergentintegral I of(7.5.26).Aftercovariantization2also containssomecontributionsfromgraphswith3,4,etc.externallines.Thechiralintegralabove,aftercovariantization,alsocontainsa 3factor. Separatingoutthedivergentpart,wehave 2= k11 (4 )2 1 2 d4xd2 31 2 W[ 1 Š ln 2 ] W+ h c .(7. 8.4b) where isarenormalizationmassand k1=( Š 2)2( A + B )(1 12 Š A )(7. 8.5) Forachir alscalar k1=1 24 .(Wehaveincl udedafa ctorof1 2 tocancelthe2dueto ourusingtheaction + h c .inste adof .)Forachiralspinor k1=20 24 .If thechirals pinorsuper“eldisthegauge“eldofthetensormultiplet,therewillbean additionalco ntri bution k1=5 24 fromthe“vesecond generationchiralscalarghosts discussedins ec.7.3.a.(The V ghostsdonotcontr i bute:seebelow.) Thenextcalculationwecouldimagineperformingisthatofatrianglediagram. However,sinceno WWW or WW W termispres entintheeectiveaction(cf.our

PAGE 480

4627.QUANTUMN=1SUPERGRAVITYdiscussionfollowing (7.7.3)),thecontributionfromsuchadiagrammustbecompletely containedinthethirdorder(in H€or )te rmsintheexpansionof2. Weobservethat, oncetheself-energycontributionfromachiralsuper“eldhas b eencomputed,thatfromavectormultiplet V istrivial,becausethewholecontribution comesfromthethreechiralghosts.Indeed,the V -backgroundinteractionsareextracted fromthec ovariant V V quadraticaction.However,ju stasinthebackgroundYangM illscalculation,each D or D containe dinthe operatorof(7.4.14b)bringswithit onefactoroftheexternal“eld,andforgra phswithlessthanfourexternallineswedo nothaveenough D s .T hisisanimportantfeatureofthebackground-“eldmethod:In o-shellYang-Millsoron-she llsupergravitybackground, general(nonchiral)super“elds donotcontributetoone-looptwo-andthree-pointfunctions;onlytheirchiralghostsdo. Thus,inthiscasethecontributiontothesupergravityself-energyfromavectormultipletis Š 3timesthatfr omaphysicalchiralscalarsuper“eld(3ghostswithwrongstatistics). Thecalculationoftheon-shellone-loopself-energycontributionsinself-interacting (q uantumandbackground)supergravityisnowtrivial.Nonewcalculationsneedtobe pe rformedbecause,fromtheactionin(7.4.14a)or(7.4.19),weseeagainthatonlychiral super“eldscontribute.Intheformwith V compensators(5.2.75a),theresultissimply Š 7timesthatfr omaphysicalchira l“eld(7chiralscalarghost swithwrongst atistics): k1= Š7 24 .Intheformwithac hiralcompensator,wehavethecontributionfromthe physical “eld,andcontributionsfromtwochiralspinorswiththeaction1 2 d4xd42+ h c ..The “nalansweris41timesthecontributionfromaphysical chiralsc alar: k1=41 24 .Fina lly,ifweusedthespinorcompensator(5.2.75b)wewould obtain k1= Š55 24 Thecalculationofachiral-orgeneral-“eldboxdiagramcontributingtothe WW W W termin(7.7.4)isinprinciplenomoredicult.Forachiral“eldwehave onevertex(7.8.1a)andthreevertices(7.8.1b)withone D2andfour D sintheloop,two ofwhichhavetobeintegratedbypartsontoexternallines.Forareal“eldwehaveone fact orof D or D ateachvertex(comingfromthelinearizationof ),sothe D -algebra istrivial.WhatisleftthenisaFeynmanintegralforaboxdiagram,withsome

PAGE 481

7.8.Examples463momentumfactorsinthenumerator. Ournextexample isthatofthecalculationofsomeone-loop,four-particleSmatrixelements,similartothecalculationwecarriedoutfor N =4 Ya ng -M ills.There wesawthatt herewerecancellationsandthewholecontributioncamefromthe V -loop. Thecorrespondingsituationthatwewill“ndhereisthatasimilarcancellationbetween ghosts,physical“elds,andcertain othercontr ibutionstakesplacein N =8superg ravity so thatinthatcasetheresultisessentiallyidenticaltotheYang-Millscase,andcanbe obtainedwithoutfurthercalculation.Wenowdiscussthesituationindetail. For N -extendedsupergravitydescribedby N =1super“elds,toc alculatecontributionstobackground N =1superg ravity,onesimplyaddscontributionsfromsuper“elds representingallthe N =1mult iplets.Thesuper“eldswhichenterinadditionto H are ofthesametypeasabove,namely V s, s,and s,andtheirLagrangianshavethe sameform(uptochoicesofcompensating“eldsanddualitytransformations,whichmay chan gethevaluesofcontributionstotopologica linvaria ntsandthec orrespondi ngsuperconformalanomalies(seesec.7.10)butdonotaecttheS-matrix).InTable7.8.1we givethenumberof“eldsofeachtype(themin ussignsindicateabnormalstatistics)for eachvalueof N N V H€ 1-74-31 2-21-21 30-1 -11 40-201 50-211 60021 80441 Table7.8 .1.Numberof“eldsofeachtypecontributingtotheone-loopeectiveaction Weuset heformof N =1supe rgravitywith V compensators.The(3 2 ,1)multi pletis describedbyageneralspinorsuper“eld(see sec.4.5.e)anditsquadraticLagrangian, incl udingghosts(whichisallthatisneeded)isgivenbythebackgroundcovariantization of(7.3.7).

PAGE 482

4647.QUANTUMN=1SUPERGRAVITYThefour-particleS-matrixisobtainedfromthediagramsinFig.7.8.1. Fig.7.8.1 Theexternalwavylinescorrespondto N =1superg ravity“elds,whilethesolidline loopcorrespondstovarious N =1mult iplets.Allbutthelastdiagramcorrespondto the“rstterm2of(7.7.3).However,bycovariancetheS-matrixmustcontainfourfactorsof W ,andby dimensionalitythecompletecontributionmustcomefromthesecond term4of(7.7.3),andthereforecanbeobtainedfromtheboxdiagram.(Onlythebox diagramhasenoughden ominatorfactorstobalancethedimensionsoffourfactorsof W .) Wewritethecontri butionfromtherelevantpartof4as =1 8 d4p1... d4p4(2 )16 d4 ( ( pi)) [ W( p1) W( p2) W€€€( p3) W€€€( p4)( C4G4+ C2G2+ C0G0)( pi) Š1 2 W( p1) W€€€( p2) W( p3) W€€€( p4)( C4G 4+ C2G 2+ C0G0)( pi)].(7.8.6) Here G0istheFeynmanintegralforascalarboxdiagram(6.5.68),while G2( G 2), G4( G 4)aresi milarcontributions extracted fromboxdi agramswithtwoandfourmomentumfactorsinthenumerator.Unlike N =4 Ya ng -M ills,theaboveexpressionisvalid onlyon-shell,whereaso-shelltheeectiveactiondiverges.(Becauseofcovariantization theexpressionabovecontainstermswithmo rethanfour“elds,butitdoesnotgivethe

PAGE 483

7.8.Examples465completecontributiontomore-th an-four-particleamplitudes.) Thecoecients Ciaredierentforeachvalueof N .Todet ermine Ciwemust computecontributionsfromeachofthesuper“eldslistedinTable7.8.1.Theonlydicultcalculationisofthecontributionfromthechiralsuper“elds i.Weshallth erefore restrictourselvesheretodiscussingresultsfor N 3where nochiralsuper“ eldsappear. Aswedisc ussedinsec.7.6,itisusefultosq uarethekin eticoperatorforthe spinors(andtakeonehalfofthecorrespondingcontributiontotheeectiveaction)in ordertomakeitsimilartotheotherkin eticterms.Thekine ticoperatorfor V ,and H€thentakesthe uni versal form = + W M + W€€€ € M€€,(7. 8.7) Therelativecoecientofthe and W termsis independent ofthechoic eof“eld. Therefore,inperformingone-loopcalculationsweneedonlykeeptrackoftheindex structure.Inparticular,thereisalwaysafactorfromatraceovertheLorentzindex:1 for V Š 1for Š 1for €and4for H€.( and €eachcountas Š 1 1 2 2= Š 1 duetoa Š 1forFe rmistatistics,a1 2 tocan celtheeectofhavingsquaredthekinetic operator ,and2forth etrace over .) Lookingatthekineticoperatorandagainrequiringthateachloopcontainatleast two Dsandtwo D€s,wediscovertwosourcesforsuchterms:Theexplicit W M (tothisorderwecanreplace by”ats uperspace D)andthos econtainedinthecovariantdAlembertian: =1 2 a a=1 2 ( E E a m m+ E E a D+ E E a€ D€+ a( M ))( E E a n n+ E E a D+ E E a€ D€+ a( M )), (7.8.8) where E E a mŠ a mandallotherquantitiescontainatleastonefactoroftheexternal “elds.(Theconnectiontermscanbedropped:Theycannotcontributetoanygraph withatmostfourexternallinesbecausetheydonotbringwiththemany D s.) Wenowmaket hefollowingobservations:

PAGE 484

4667.QUANTUMN=1SUPERGRAVITY(1)Since connectiontermscanbedropped, actsinthesamewayonall“elds. (2)Sincewe needtwo D sandtwo D s,andeachoneb ringswithitan E E ora W andan E E ora W ,ourresultw illcontainfoursuchfactors,twobarredandtwo unbarred.The E E verticeshavetheform E E a aDor E E a€ a D€. (3)Since [ aE E b ] W,andthev ectorindicesonthetwo E E sinatermwith onlytwosuchfactorsmustbecontracted(andthereforealsothespinorindices)inorder topro duceacovariantcontribution,thereare E E2 W2and E E2W2terms butno E E E EW W terms.(Similarlythereareno E E2 E E W termsor E EW W2terms.) (4)DuetothealgebraoftheLorentzgenerators,the W M W M factorsineither the E E2W2or W2 W2produceanextranumericalfactorof a (and b from W M W M ) relatedtothenumberofspinorindices. Therefore,therearethreetypesoftermstoconsider: (1)( E E )2( E E )2,(2)( E E )2( W M )2(and h c .),(3)( W M )2( W M )2.Eachterm takesthesameformforall N butwithacoecientdeterminedbysummingover V , ,and H thepro duct (numberofsuch“elds) (Lorentztracefactor) (M-factor). Thenumberof“eldsisgiveninTable7.8.1,thetracefactorwasdiscussedearlier,and the M -factoris,foreachofthethreetypesofterms,respectively:(1)1,(2) b ( a for theh.c.),(3) a b .Thevaluesoftheo verallnu mericalcoecientarepresentedinTable 7.8.2.Thesecoecientsarelabeled(1) C4,(2) C2,(3) C0,andappe arineq.(7.8.6). (Notetha tonly H contributestothelastcolumn,sinceonly H hasbothadottedand an undottedindex,andsohasboth W and W termsinitskineticoperator.)Our resultisthusthat:(1) N =1,2contain all typesofte rms,includingcontributionsfrom chiralsuper“elds ,whichwe havenotdiscussed;(2) N =3 ,4 receivenocontributions fromchir alsuper“elds;(3) N =5,6lack also the E E2 E E2typete rm;(4) N =8 receivesa contributionfrom only the W2 W2term,inanalogytothe N =4supers ymmetricYangM illscalculation.

PAGE 485

7.8.Examples467 NC4C2C0 3554 4244 5034 6024 8004 Table7.8 .2.Multiplicityofcontributionsofeachtypetotheone-loopeectiveaction Thecalculationoftheeectiveactionproceedsasfollows:For N =8superg ravity thereisnothi ngmoretodo.Onehasaboxgraph,withtwofactorsof W andtwofactorsof W andascalarloopintegraltoperform.Weobtainthe G0termsineq.(7.8.6) withafactor C0=4,w h ile C4= C2=0.For N =5,6oneh asaboxgraphwithtwo verti cesoftheform E E andtwowith W s,aswellasatrianglegraphwithonevertex containingtwo E E -factors.Theloopintegralfortheboxgraphcontainsnowtwoloopmomentumfactors,butgauge(localsupersymmetry)invariancecanbeusedtosplito apartwhi chgives [ aE E b ]soastoproduce W s,whiletherestmustcancelthetriangle graphcontributio n.(Theyactuallymaycontributeto2,whichis zerobymomentum conservation .)Finally,for N =3,4oneh asaboxgraphwithone E E factor ateach vertex,atria nglegraphwithonevertexcontainingtwo E E factors, andals oaselfenergytypegraphwithbothverticescontainingtwo E E factors.A gaingaugeinvariancecanbeusedtoextractthecompletecon tri butionfromtheboxgraph,theremainderaddinguptozero. TheS-matrixcanbeobtainedfromtheeectiveactionbydroppingthe piintegralsandta kingas umof G termsoverpermutationsofth eMa ndelstaminv ariants s =( p1+ p2)2, t =( p1+ p4)2, u =( p1+ p3)2.( IntheYang-Millscasethisalsoinvolves intercha ngeofinternalsymmetryindices).Thus,for N =8superg ravitywehave S ( s t u )=(2 )4 ( ( pi)) d4 W( p1, ) W( p2, ) W€€€( p3, ) W€€€( p4, ) [ G0( s t u )+ G0( s u t )+ G0( u t s )].(7. 8.9)

PAGE 486

4687.QUANTUMN=1SUPERGRAVITYThe -integrationsplitsuptheproductofthesuper“elds W intoasumoftermsinvolvingproductsofWeylte nsorsandgravitino“eldstren gthswhichcanbereplacedwith momentaandpolarizationvectorsforthevariousprocesses.Theactualvalueof G0is G0( s t u )= 2 Š (2 )4 (( Š ))2( ) ( Š 2 ) [sŠ 2 Š F (1,1,1 Š Š u s )+ tŠ 2 Š F (1,1,1 Š Š u t )],(7. 8.10) where =2 ŠD 2 and F isahypergeometricfunction.Itisultraviolet“nitebutinfrared di ve rgentforbothYang-Millsandsupergravity,butinthelattercasethedivergenceis milderbecauseofcancellationsinthe s t u permutatio ns.Theexpressionsfor G2, G 2, G4,and G 4aresomewhatmorecomplicatedandwillnotbegivenhere. Sofarourresultsarewithonly N =1superg ravityexternalparticles(gravitons andgravitini).However,theS-matrixcanbeextendedimmediatelytotheotherparticlesofan N > 1multi pleteitherbydirectglobalsupe rsymmetrytransformationsonthe S-matrixorbyre alizingthatthed4 ( W)2( W€€€)2canbeextende dt oasim ilar expressioninvolvingproductsoffouron-shell“eldstrengthsforextendedsupergravity. Wealsonote thatthe W2 W2formoftheresultimpliesthe he licityconservationpropertiesofthesupersy mmetricS-matrix. Theremarkablesimplicityofthecalculationsandresultsisdueinparttothe surprising decreaseinnumberofdiagramsonehastoconsiderasoneproceedsfrom N =1 to N =8.Inparticular, theabsen ceofchiralsuper“eldsfor N 3produ cesthecrucial simpli“cation,andtheensuingcancellationsbetweenvarious“eldsculminatesinthe absolutetrivialityofthecalculationfor N =8superg ravity.Attheotherextreme,the N =0theory (ordinaryEinsteingravity)wouldseemtorequireamajorcomputercalculation.

PAGE 487

7.9.Locallysupersymmetricdimensionalregularization4697.9.Locallysupersymmetricdi mensio nalregularization Iftheonlyinterestingsupergravitytheoriesarethosethatare“nite,theconstructionofaregularizationwhichmanifestlypreserveslocalsupersymmetryissomewhatofanacademicexercise.Nevertheless,weshalldiscusstheprocedurehereforcompleteness,andbecausethegeneralmethodisusefulfordiscussionofdimensionalreductiontointegraldimensions. Itispossibletoextendthesupersymmetricdimensionalregularizationmethodof sec.6.6forapplicationtosupergravity.Somemodi“cationsarerequiredbecause,unlike matter(scalarorvector)multiplets,supergravityisnolongeron-shellirreducibleafter dimensionalreduction.Thedimensionalreductionmustthereforebeperformedina mannerthatpicksouttheirreduciblepart.Insuper“eldlanguage,thedierenceoccurs b ecause( N =1)ma ttermultipletsare describedbyscalarsuper“elds,whereassupergravityisdescribedbyavectorsuper“eld.UponnaivedimensionalreductiontoDdimensions,thisvectorsuper“eldreducestoasuper“eldwhichisaD-dimensionalvector,describi ngpuresupergravity,plus4-Dscalarsuper“elds,describingvectormultiplets.Thedimensionalreductionmustthereforeberede“nedsothatonlytheD-dimensional-vectorsuper“eldappears.Wethereforeneedasuper“eldformulationthat describespuresupergravityforarbitraryD < 4.Forsimplicitywewilldescribetheconstructionfor N =1, butthemethodcanbeeasilygeneralizedtoanyfour-dimensional super“eldtheory.Forintegraldimensions,the N =1theory reducestopure N =2 supergravityforD=3or2,andpure N =4superg ravityforD=1. Webeginbyc onstructingcovariantderivatives.Sinceupondimensionalreduction theLorentzgroup SO (3,1)isbro kendow nto SO (D Š 1,1) SO (4 Š D),ourcovariant derivati vestaketheform A= EA+1 2 A b cM c b+1 2 Ab cMc b,(7. 9.1) wherethesupervectorindex A =( ,€ a ),andwehavere duceda4-dimensionalvector i ndex € intoaD-d imensionalvectorindex a pl usaninternalsymmetry ( SO (4 Š D))index a .The “eldstrengthsarede“nedasusual: [ A, B} = TAB CC+1 2 RAB c dM d c+1 2 RABc dMd c.(7. 9.2) Thederivativescontainedin EA= EA MDMrangeoverD(commuting)+4

PAGE 488

4707.QUANTUMN=1SUPERGRAVITY(anticommuting)coordinates.Thefact thatthespacetimecoo rdinateshavebeen reducedfrom4toDautomaticallytakescareofreducingthefundamentalsuper“eld H atoaD-componentvector ,asdis cussedabove( H = H ai a).However,ascomparedto theusual4-dimensionalcovariantderivatives,wehavelessgaugefreedomduetothe absenceofaLorentzgeneratorofthemixedtype Ma b,sothatana ddition alconstraintis neededtoaccountforthislostinvariance.Speci“cally,thismeansthattheobject N whichappearsuponsolvingtheconstraints,andwhichisgaugedawaybylocalLorentz transformationsinD=4,musthavethecomponentswhicharenotgaugedawayinD < 4 constrainedaway.Wethereforeimposethe followingsetofconstraintsforarbitrary D 4(forsim p licity,wechoosethecase n = Š1 3 ): Conventional : T = T [ b c ]= T€€= a €R€ c d= a €R€ c d=0, a €T€ c= i a c;(7. 9.3a) Chiral itypreserving : T c= T€=0;(7. 9.3b) Conformalbreaking : T b bŠ T€€=0;(7. 9.3c) Additionalconventional : a €T€ c=0;(7. 9.3d) wheretheadditionalconventionalconstraintistheonethatconstrainstheextracomponentsof N .( a €istheD-dimensionalPaulimatrix,whichprojectsouttheD-componentvectorindex a fromthe4-d imensional € ;sim ilarly, a €projectsoutthe4-D-component i ndex a .Our normaliz ationhereis a € b €= a b, a €b €= a b, a € a €+ a €a €= €€.)Theconventionalconstraintsaswrittenaresomewhat redundant,butitcanbeshownthat,inconjunctionwiththeremainingconstraints,they servetodetermine Aintermsof E,asusual.Thechir ality-preservingandconformalbreakingconstraintshaveasolutionsimilartothatofD=4,exceptthat HMin H = HMiDMisnowaD+4componentsupervector.Thesolutiontotheconstraintsis (cf.sec.5.3): E= N E,

PAGE 489

7.9.Locallysupersymmetricdimensionalregularization471E a= ( N a bŠ N a cAc b) E b+( f a E+ f a€ E€), E= eŠ De, E a= Š i a €{ E, E€} ,=MiDM, [ EA, EB} = CAB C EC, Aa b= i a € C€ b, N€ €= N N€€, detN =1, N = det ( N a bŠ N a cAc b), = 1 2 ŠD 2(D Š 2) [(1 eŠ )D(1 e )Š (D Š 2) E2 N2]Š1 4(D Š 1) E€ =0.(7. 9.4) Thematrix N€ €isdeterminedby(7.9.3d)totaketheform N€ €= N a bNa bN a bNa b = (1+ ATA )Š1 2 (1+ AAT)Š1 2 A Š (1+ ATA )Š1 2 AT(1+ AAT)Š1 2 (7.9.5) inanappropriateLorentz internalgauge,wheredoubles pinorindicesareconverted intovectorindicesandbackagainwith s(oftheappropriatetype),andthelastequationiswritteninmatrixnotationwith A = Aa b. N isdeterminedfromthisexpression for N€ €byusin gtherelation N€ €=( eX)€ €, X€ €= €€Y + Y€€ N =( eY) .(7. 9.6) Since N€ €isorthogonal, X isantisymmetric,andtheref orecanbeexpressedinterms ofthetraceless Y .Wehavenot giventheexplicitexpressionfor f a in(7.9.4),northe solutionsfortheconnections,buttheycanbeobtainedasinD=4withoutfurthercomplications,andwillnotbeneededhere. Thesupergravityactionis S = Š 2D Š 1 D Š 2 Š 2 dDxd4 EŠ 1

PAGE 490

4727.QUANTUMN=1SUPERGRAVITY= Š 2D Š 1 D Š 2 Š 2 dDxd4 EŠ1 (D Š 1) [ det ( a b+ Ac aAc b)]Š1 2(D Š 1) [(1 eŠ )(1 e )](D Š 2) 2(D Š 1) (7.9.7) Projectionoperatormethodscanbeusedtoshowthatthelinearizedactioncontains onlytheusualsuperspins3 2 +0.CouplingtomattercannowbeperformedasinD=4, withchiralLagrangiansintegratedbydDxd22(D Š 1) (D Š 2) Supergr aphcalculationscanbe perfo rmedwiththeusualfour-dimensional D -algebra.Wedomomentumintegrationas inconventionaldimensionalregularization,andminimallysubtractthedivergentpart using 1 timesalocal,covariant,D-dimensiona lcounterterm constructedfromtheDdimensionalcovariants. Thesameinconsistenciesthatoccurredi ngloba llysupersymmetricdimensional regularizationofcourseremaininthelocalcase.Nevertheless,asintheglobalcase,in actualcomputationstheinconsistenciesseemtodisappearaftertakingD 4.After minimalsubtraction,theremaining“nitequantitysatis“esthe4-dimensionallocal supersymmetryWard-Takahashiidentities(aftertakingD 4).Furthermore,the methodisperfectlyconsistentforreductiontointegraldimensions,andcanbeusedfor describingextendedsupergravityinlowerdimensions.However,weobservethatthe ab ov es uper“elddescriptioninnonintegraldimensionsde“esunderstandingintermsof components.(E.g.,sincetheD-beinin H ahasno ea bpart,what“eldgaugesthe internal Ma bsymmetry?) Sincethesubtractionprocedurepreserveslocalscaleinvariancewhenthecompensator isincluded,therenormalizedeectiveactionwillbesuperconformallyinvariant. However,D=4superconformalanomaliesare ingeneralpresentpr eciselybecausethe renormalizedeectiveactiondependson .Wediscuss thisinthenextsection.

PAGE 491

7.10Anomalies4737.10Anomalies a.Introduction Anomaliesinlocalconservationlawsareharmlessaslongasno“eldscoupleto thecorrespondingcurrent.Indivergentcomponenttheoriesthereisalwaysatleastone suchanomaly:thescaleanomaly.Thisanomaly,whichcanbeexpressedasanadditionalcontributiont othetraceoft heenergy-momentumtensor,occursbecauseanew massscaleisintroducedatthequantumlevel,therenormalizationmassparameter.For example,atheorythatisclassicallyconfo rmallyinvariant,andthushasaclassical energy-momentumtensorwithvanishingtra ce,getsquantumcontr i butionstothetrace. WhenEinsteingravityiscoupledtothequantumsystem,thisanomalyisharmless,as generalcoordinateinvariancemerelyrequire sconserv ationoftheenergy-momentumtensor( i.e.,thevanishingofitscovariantdivergence).However,itwouldbeharmfulin conformalgravity,sincelocal(Weyl)scaleinvariancedoesrequirevanishingofthetrace. Quantumc orrectionstotheenergy-moment umtensorar emostc onveniently de“nedbycouplingthequantumsystemtobackgroundgravityandde“ning T m n= R g m n (7.10.1) whereRistherenormalizedeectiveactionandisasuitablyde“nedvariation(see below).I tstraceisgivenby T m m= g m nR g m n (7.10.2) Alternatively,itcanbeobtainedby“rstint roducingacompensatingscalarintothetheory(5.1.34).Forconformaltheoriesthe compensatordecouplesfromtheclassical action,butingeneralitent ersintherenormalizedactionwhereitcouplestothetrace. Therefore,varyingRwithrespecttothecompensatordetermines T m m. Insupersymmetry,theenerg y-momentumtensoristhe componentofasuper“eld,the supercur rentJ€.(Morepr ecisely,1 8 [ D€, D] J€| + a b = T a bŠ1 3 a bT c c.) Thetraceoftheenergy-momentumtensorisacomponentofarelatedsuper“eld,the supertraceJ (1 2 ( D2J | + h c .)=1 3 T a a).Justastheenergy-momentumtensorcanbe de“nedfromthecouplingtogravity,thesupercurrent J€canbede“nedfromthe

PAGE 492

4747.QUANTUMN=1SUPERGRAVITYcouplingtothesupergravitysuper“eld H€: J€= H€ .(7. 10.3) Aswewilldisc ussbelow,thesupertracecanbede“nedbyfunctionaldierentiationwith respecttothecompensatingsuper“eld.Inclassicallocallysuperscaleinvarianttheories thecompensatordecouplesandthereforethesupertracevanishes.Ingeneral,itspresenceisameasureofthebreakingoflocalsuperscaleinvariance. Thesupercurrentalsocontainsthesupersymmetrycurrent(atlinearorderin ) andtheR-symmetryaxialcurrent(at =0); thesupertracealsocontainsthe -traceof thesupersymmetrycurrent(atlinearorderin )andthediver genceoft heaxialcurrent (theimaginarypartofthe 2component).Thus,inasupersymmetrictheorywhere scaleinvarianceisbroken,theaxialcurrenthasachiralanomalyandthesupersymmetry currenthasan S -supersymmetryanomaly,andthecoe cientsofallthreeanomaliesare equal.However,justastranslationalinvarianceisnotviolated(thetraceoftheenergymomentumtensorisanomalous,notitsdivergence),neitherisordinary Q -supersymmetry(the -traceofthesupersymmetrycurrentisanomalous,notitsdivergence). Inlocallysupersymmetrictheories,ina dditiontothesuperconformalanomalies describedbythesupertrace,theremayexistanomaliesintheWardidentitiesoflocal (Poincar e)supersymmetry.Forthecasesthathavebeenstudiedtheydonotoccurin N =1theoryfor n = Š1 3 (theminimalsetofauxiliary“elds)becausewecanregularize inama nnerconsistentwithlocalsupersymmetry;theydooccuringeneralfornonminimal(andnewminimal) N =1, n = Š1 3 theori es.Wewillusetheexistenceofsuperconfo rmalanomaliestoinfertheexistenceoftheseauxiliary-“eldanomaliesandconclude thatingeneralonly n = Š1 3 theoryisquantumconsistent. b.Conformalanomalies We“rstreviewon e-loopon-shellscaleanomaliesincomponenttheories.We areinterestedinquantumcorrectionstoma trixelementsoftheenergy-momentumtensorbetweenthevacuumandastatecontainingtwoormoregravitons.(Wecouldconsiderotherexternalparticle s,andalso o-shellanomalies,butwhengravityisquantizedonlytheonshellonesareunambiguous.)Equivalently,wecomputetheone-loop

PAGE 493

7.10Anomalies475eectiveactionforasysteminagravitationalbackground,functionallydierentiatewith respectto g m nand then setthebackground“eldonshell.AtthelevelofFeynmandiagrams,becauseofcovariance,weneedonlyconsiderthetwo-pointfunction(graviton pr opagatorcorrection),whichdeterminesalltheone-loopdivergences,andthethreepointf unction(trianglegraph),whichdeterminesthetraceoftheenergy-momentum tensor.Infact,iftheclassicaltheoryforthe“eldintheloopisconformallyinvariant, all therelevantinformationcanbeextractedfromthetwo-pointfunction. Inclassicaltheoriesconformalinvaria nceisbrokenintwoways:(a)bymassterms thatbreakitsoftlyandwhos ee ectcanbeseparatedout,astheycanbeforthedivergenceoft heaxialcurrent,and:(b)byhardterms,e.g.,derivativesof“elds,asfor Ya ng -M illsinD =4dime nsions,andforantisymmetrictensor“elds.Inthesubsequent discussion,whenwerefertononconformaltheorieswemeanclassicaltheorieswherethe breakingishard. Weconsid er“rstaclassicallyconformallyinvarianttheorysothatthetraceofthe classicalenergy-momentumtensoriszero.Wh encoupledtogravity,theclassicaltheory islocallyscaleinvariant.Byintroducinga ppropriateD-dependencethetheorycanbe dimensionallyregularizedsothatthisinvarianceispreservedintheregularizedeective actionnearD=4.(Theinvarianceisbrokenonlytoorder(D Š 4)2,andth ushasno eectevenin(D Š 4)Š 1divergentterms.)However,thecoecientofthe(D Š 4)Š 1factorisnotseparatelylocallyscaleinvariantexceptatD=4,andthereisnolocal“nite termthatcanbeaddedtoittomakeitso.Ther efore,therenormalizedeectiveaction, de“nedbysubtractingthisD-dimensional,local,covariantdivergenttermfromtheregulari zedeectiveaction(i.e.,byaddingacounterterm S)isnotloca llyscaleinvariant. Consequently,whenwecomputethetraceof T m nde“nedintermsoftherenormalized eectiveactionwe“ndanonzeroresult.Sincetheregularized,unrenormalizedeective actionUwasscale invariant,thescaleanomalyoft herenormalizedeectiveactionRisjustthetrace computedfromtheD-dimensionalcounterterm S. Wehave de“ned T m nin(7.10.1).Thevariationisde“nedintermsof by f g m n = gŠ1 2 f g m n (7.10.4) ( g = det ( g m n)),ordirectlyby

PAGE 494

4767.QUANTUMN=1SUPERGRAVITY g m n( x ) g p q( x) =1 2 ( m p n ) qgŠ1 2 4( x Š x).(7. 10.5) Thelocalscale(trace)anomalyisthen g m nT m n= g m nR g m n = g m n S g m n .(7. 10.6) Th el as te qu al it y hol ds on ly becauseweareconsideringclassicallyconformaltheories. Otherwise,theformertwoexpressions, thetotaltrace, donotequalthel astexpression, thetraceanomaly. (Ingen eral,the anomaly is understoodtobeacontributiontothe traceduetothedivergencesofthetheory.) SinceinclassicallyconformaltheoriesUislocallyscaleinvariantnearD=4, Stakestheformof(D Š 4)Š 1timesalocal(generalcoordinate)invariantthatisthe dimensionalcontinuationofa4-dimensiona lobj ectthatisloca llyscaleinvariant.From dimensionalandcovarianceconsiderationswe“ndtwoindependentfour-dimensional objectsofthisform:Intermsoftheirreduciblepartsofthecurvatureof(5.1.21),they aretheEule rnumber =1 (4 )2 1 2 d4xg1 2 [1 2 ( ww+ h c .) Š r€€r€€+3 r2],(7. 10.7) atop ologicalinvariantwhosefunctionalvariationvanishesandwhichitselfvanishesin topologicallytrivialspacetimeforD=4,andtheintegralofjusttheWeyltensor ( w2+ w2).Thedier encebetweenthetwovanishesonshell( r€€= r =0). Inquantumgravitythecoecientofanytermthatvanishesonshellisingeneral gauge-dependent,andinfactcanbemadetovanishbyanappropriategaugechoice,or canbeeliminatedbyalocal“eldrede“nitionofthemetric(sincesuchrede“nitionsof theactionareproportionaltothe“eldequat ions).Ther efore,weconsideronlytheonshellpartofthetraceanomaly,whichwewriteintermsof w2=1 2 ww.Wenote that w2hasthesimplescalingpropertyfor arbitrary D ( g m n g m n )( x )( g1 2 w2)( x)=1 2 (DŠ 4)( g1 2 w2) 4( x Š x).(7. 10.8) InD-dimensionstherelevantpartofUisgivenbyacovariantizationofagraviton self-energygraphandhastheform

PAGE 495

7.10Anomalies477U1 D Š 4 dDxg1 2 1 2 w ( `` 2 )D 2 Š 2w + h c .,(7. 10.9) where`` means +curvature termsnecessarytomakeUlocallyscaleinvariant inarbitraryD,and isarenormalizationmass.(Ualsocontains“nitelocallyscale invariantterms,anddivergenttermsthatvanishonshell.)Wethenhave SŠ1 D Š 4 dDxg1 2 w2+ h c .,(7. 10.10) R d4xg1 2 1 4 wln ( 2 ) w + h c ..(7. 10.11) Byintegrationbyparts(dropping“nitetermsproportionaltotheEulernumber,which canbeconsideredpartofthecorrespondingin“niteterm(7.10.10)),Rcanberewritten atD=4 R d4xg1 2 {1 2 r€€ln ( 2 ) r€€Š3 2 rln ( 2 ) r +( w2+ w2) ln [1 Š 1 + r r ] } (7.10.12) plusmore“nitetermsthatarelocallyscalei nvariant,andtermsof thirdorhigherorder in r and r€€.Sin ce[1 Š ( + r )Š 1r ]satis“est hescalec ovariant equation ( + r ) =0(withour conventionsofsec.5.1, + r isthekineticoperatorofalocally scale-covariantscalar),itcanbeshownthat ( g m n g m n )( x ) ln [1 Š 1 + r r ]( x)= Š1 2 4( x Š x),(7. 10.13) Therefore,using(7.10.8),weseethat(7.10.12)givesthesame(on-shell)trace(fromthe lastterm)asRin(7.10.11)(oras Sin(7.10.10)): g m nT m nŠ1 2 ( w2+ w2).(7. 10.14) We “nd(7.10. 12)amoreconvenientformofrepresentingR.The “rsttwotermsare covariantizedself-energycontributionsunambiguouslyexpressedintermsofthecurvaturescalarandRiccitensor,andareofnointerestforon-shelltracessincetheirvariationvanishesonshell.Thelastterm,whenthe ln isexpa ndedinpowersof r ,hasthe

PAGE 496

4787.QUANTUMN=1SUPERGRAVITYform d4xg1 2 w21 r + ... = d4xg1 2 r 1 w2+ ... ,(7. 10.15) sothatitreceivescontributionsonlyfromdiagramswithatleastthreeexternallines. Thistermisacovariantizedtrianglegraphcontributionandcouldbecomputedusing, forexample,the Adler-Rosenber gmethodwith r atthetopvertex(cf.also (6.7.10-13).Thetraceoperation g g actingon r in(7.10.15)isanalogoustothe 2in (6.7.13)). Inthemoregeneralcasewhenthequanti zedtheoryisnotcla ssicallyco nformally invariant,orgravityisalsoquantizedsothat g g U =0,localsc aleinvariancecannot beusedto determine g m nT m nfrom S.Itisthenn ecessarytocalculatethetotaltrace fromRdirectlyfromatrianglegraph.(Inthecaseofquantumgravityweuseabackground“eldgaugetomaintaincovarianceofR.)Thegen eralformoftheunrenormali zedeectiveaction near D=4is U= k11 D+1 (4 )2 d4xg1 2 [ k21 2 r ( 1 Š ln ( `` 2 )) r + k31 2 r€€( 1 Š ln ( `` 2 )) r€€Š k4( w2+ w2) ln (1 Š 1 + r r )](7.10.16) where =2 ŠD2 and DisthedimensionalcontinuationoftheEulernumberof(7.10.7): D=1 (4 )1 2 D 1 2 dDxg1 2 [1 2 ( ww+ h c .) Š r€€r€€+3 r2].(7. 10.17) Risobtainedbysubtractingoutthe Š 1terms. Therelevanttermfortheon-shelltraceisagainthelastonein(7.10.16),although nowitscoecientisnotrelatedtothoseoftheprecedingterms.Theon-shelltrace computedfromRisnotequaltothet racecomputedfrom S.I tr eceivesadditional contributionsfromtheclassicallynonconfo rmalpartofthetheory.Asmentionedabove,

PAGE 497

7.10Anomalies479wew illrefertot hepartattributableto Sasthetraceanomaly,whilecallingtheentire contributionfromRthetotaltrace. Thenumber k1hasbeencomputedinavarietyofways,anddeterminestheonshelltraceanomalyof“eldsintheloo p.Thisquan tityisusuallywritten ( T m m)anomalous= k11 (4 )2 1 4 [ ww+ h c .](7. 10.18) with360 k1=4,7, Š 52, Š 233,848,364,forascalar,Majoranaspinor,vector,RaritaSchwinger“eld,graviton,andsecond-rankan tisymmetrictensorgauge“eld,respectively, incl udingtheirghosts.Wenotethatalthoughthe“rstandlast“eldsbothdescribethe samespinzeroparticle(ifthescalarhasno improvementterm),theirtraceanomalies aredierent.Ontheotherhand,itcanbea rguedthattheyhavethesametotaltrace, whichisthephysicallyrelevantquantity,determinedbyR.(Forthe improved scalar “eld( T m m)tot=( T m m)anom, butfortheantisymmetrictensororunimprovedscalarthey aredierent:Thelattertheoriesarenotclassicallyconformallyinvariant.)Inlikefashion,third-andfou rth-rankantisymmetrictensor“elds,whichhavenophysicaldegrees offreedom,havezerototaltrace(infact,zer oreno rmalizedeectiveaction),although b ecauseofthequantizationprocedure,theyhaveanonzerodivergentcontributiontoUandthereforeanonzerotraceanomaly. Ausefulmethodforma kingscalebreakingpropertiesmanifestistointroducea compensatingscalarasin(5.1.34).Wethenhave ( g m n g m n f )( 2g p q)=1 2 Š 3 f 1 2 f ,(7. 10.19) sotheexistenceofanonzerotraceisequivalenttohavingdependenceon .Forex ample,(7. 10.10,11)becomes SŠ1 D Š 4 dDxg1 2 2D D Š 2 w2+ h c .,(7. 10.20) R d4xg1 2 1 4 wln ( `` 2 ) w + h c .;(7. 10.21) (notethatweh aveabsorbed into )andthelasttermi n(7. 10.12)becomes d4xg1 2 ( w2+ w2) { ln [1 Š 1 + r r ] Š ln } .(7. 10.22)

PAGE 498

4807.QUANTUMN=1SUPERGRAVITYIftheclassicaltheoryisconformallyinvariant, decouplesfromtheclassical action,andthus doesnotappearintheFeynmanrulesorinU:Itsonly appearancein Risthrough S.Itmus tbeintro ducedin StomakethistermscaleinvariantinDdimensions,andtocompensateforthisitmustalsoappearinR,sinceUisinde pendentof .Ontheot herhand,iftheclassicaltheoryisnotconformallyinvariant, is presentinU,andw illenterinRinama nnerwhichisnotrelatedtothewayitenters in S. c.Classicalsupercurrents Inthissubsectionwederivetheclassicalsupercurrentsforvariousmultiplets. Thesearethesuper“eldsthatcontainthesuperconformalcomponentcurrents.They canbeobtainedi nprincip lefromtheclassicalactionsbymeansofNoetherstheorem,or canbecalculatedasthevariationalderivativesofthecovariantizedactionswithrespect tothesupergravityprepotentials.Ingen eralwedonotimmediatelyobtainthesame results,unlessweperformsome“eldrede“nitions.Theserede“nitionshavenophysical eectsincetheyonlychangethecurrentsbytermsproportionaltothe“eldequations. Weconsid erminimalsupergravitywiththechiralcompensator. Theactionforascalarmultipletinthepresenceof(background)supergravityis (inthechiralrepresentation) S = d4xd4 EŠ 1 eŠ H +[ d4xd23(1 2 m 2+1 6 3)+ h c .]. (7.10.23) Ifwemakethe“eldrede“nition Š 1 andusethelinearizedequation(see(7.5.4)) EŠ 1Š 1( eŠ H )Š 1=1 Š1 3 D€DH€Š1 3 i aH a(7.10.24) weobta inthe superc urrent J€ J€= S H€ = Š1 6 [ D€, D] +1 2 i€ = Š1 3 ( D€ )( D )+1 3 i€ .(7. 10.25) The -independentcomponentof J€isthe(R-transformation)axialcurrent

PAGE 499

7.10Anomalies481J€| =1 3 Ai€A Š1 3 €,the linear -componentisthesupersymmetrycurrent,and atthe levelwe“ndthe(im proved)energy-momentumtensor. Wede “nethe supertrace J S 3 =1 6 m 2, D€J =0.(7. 10.26) Wecanv erify,usingtheequationsofmotion,theconservationequation D€J€= DJ .(7. 10.27) Quitegenerally,thisequationisadirectconsequenceoftheinvarianceofthe ac ti on under L-transformations(5.2.7,7.4.2b), H€= D L€Š D€L, 3= D2DL: LS = S H€ LH€+( S 3 L3+ h c .)=0.( 7.10.28) Iftheclassicaltheoryisconformallyinvariantthecovariantizedactionissuperscale invariant(i ndependentof ,po ssiblyafter“eldrede“nitions,e.g.,inthecaseaboveif m =0), thesupertracevanishes,and D€J€=0.(7. 10.29) Thisequationexpressestheconservationoftheaxialcurrent,andthevanishingofthe supersymmetrycurrent -trace,andoftheenergy-momentumtensortrace. Forthev ectormultipletthe”atsuperspacecomponentcurrentsarecontainedin thesuper current J€= W€W,where Wisthe”atsuper“eldstrength.However,to obtainthisexpressionfromcouplingtosupergravityrequiressome“eldrede“nitions whichwenowdescribe.Forsimplicityweconsidertheabeliancase. Inthesupergravitychiralrepresentationwehavetherealitycondition V= eHV Introducing V= e1 2 HV wehave now V = Vand W= i ( 2+ R ) ( eŠ1 2 HV )= i D22 N eŠ HDe1 2 HV = i Š3 2 D2 EŠ1 2 N eŠ HDe1 2 HV .(7. 10.30) Itisconvenienttocalculateinthegauge(7.6.5)where N = sothatspinorchiral “eldsarech iralintheusualsense.Inthisgauge,atthelinearizedlevel(seesec.7.5.c)

PAGE 500

4827.QUANTUMN=1SUPERGRAVITY EŠ1 2 N = Š D€DH€.(7. 10.31) However,ifwecalculatethesupercurrentby J€= H€ d4xd231 2 WW(7.10.32) itwillnotbe(Yang-Mills)gaugeinvariantbecausethegaugetransformationof V (or V)dep endson H€: V= i ( eŠ1 2 H Š e1 2 H), D€=0.(7 .10.33) Were medythisbymakingafurther“eldrede“nition V=( cosh1 2 H + sinh1 2 H1 2 H 1 2 H€[ D€, D]) V0.(7. 10.34) Thus V0hasthe H -independenttransformationlaw V0= i ( Š ),whichindicates thatthecomponentvector“eldin V0hasacurvedvectorindex,incontrasttothe”at i ndexonthatin V .Atthe linearizedlevel,we“nd 3 2 W= W0 Š D2H€ W0€,(7. 10.35) where W0 = i D2DV0(7.10.36) isthegauge-invariant“eldstrengthof V0(containingthecompon ent“eldstrengthwith curvedindices).Fromtheaction(7.10.32)we“ndthen J€= W0€W0 J =0; D€J€=0.(7. 10.37) Ifthe(covariantized)supersymmetricgauge-“xingterm(6.2.17)ispresent,we haveadditionalcontributions(for =1) J€ GF= Š1 6 [ D€, D][ V0{ D2, D2} V0+( D2V0)( D2V0)] +1 2 ( D2V0) i€( D2V0) Š V0i a[ D2, D2] V0

PAGE 501

7.10Anomalies483Š1 2 ([ D€, D] V0) { D2, D2} V0,( 7.10.38a) JGF=1 3 D2( V0[ D2, D2] V0).(7. 10.38b) Fo rt he tensormultiplet(chiralspinorsuper“eld)thereareanalogouscomplicationsduetothetransformationlaw = i ( 2+ R ) ( eŠ1 2 HK).(7. 10.39) Wehavein tro duced Kbyan alogyto V.Howev er,ifwerede“ne Kintermsof K0, and intermsof 0 ,byanal ogyto(7.10.34,35),we“ndthatthecovariant“eld strength G=1 2 e1 2 H+ h c .= Š1 2 e1 2 HE ( EŠ 1E )+ h c = Š1 2 e1 2 HE ( 3 2 EŠ1 2 N eHD eŠ H)+ h c .(7. 10.40) canbeexpressedas G= G0Š1 3 ([ D€, D] H€) G0Š1 2 H€[ D€, D] G0+[( DH€) D€Š ( D€H€) D] G0,(7. 10.41) where G0=1 2 D0 + h c .. Fromtheaction S = Š1 2 d4xd4 EŠ 1eŠ1 2 HG 2(7.10.42) weobtain J€= Š1 12 [ D€, D] G0 2+1 2 G0[ D€, D] G0= Š1 3 ( D€G0)( DG0)+1 3 G0[ D€, D] G0,( 7.10.43a) J =1 6 D2G0 2.(7. 10.43b) Wenote thatthesubstitution G0 + (cf.sec4.4.c.2)givesthesupercurrent J€for

PAGE 502

4847.QUANTUMN=1SUPERGRAVITYthenonconformalscalarmultipletwithLagrangian1 2 ( + )2.ThisL agrangianforthe scalarmultip let,identicalin”atspac etothe usualone,givesdisimprovementtermsto J€and J b ecausetheextraterms1 2 ( 2+ 2)leadtononmini malcouplingsd4xd23R1 2 2+ h c ..Ontheotherhand,theimprovedtensormultiplet(4.4.46) withaction Šd4xd4 GlnG doeshave J =0. Fromthe gauge“xingterm SGF= Š1 2 d4xd4 EŠ 1(1 2 Š h c .)2(7.10.44) weobtain additionalcontributions.Thecombinedcurrentfrom(7.10.42,44)canbewritten J€=1 6 ( D2) i €Š1 6 ( D) i €D( )Š1 2 i €D2+1 2 €+ h c ., (7.10.45a) J =1 12 D2[( D)2+ h c .](7. 10.45b) Asmentionedearlier,the“eldrede“niti onswehav eperfo rmedchangetheformof thesuper currents,butonlybyaddingtermsproportionaltothe“eldequations.Such termshave nophysicalconsequences. Thesupercurrentforthesupergravitymultipletitselfcanbeobtainedfromthe background-quantumsplittingofsec.7.2,byfunctionaldierentiationwithrespectto thebackgro und“eld.Wewillnotgiveithere. d.Superconformalanomalies Thediscussionofsec.7.10.bcanbetakenoverdirectlytothe N =1supersymmetriccase.Weconsiderquantumcorrectionstothesupercurrent J€,andin particular toitssupertrace J .Forcl assicallyconformallyinvariantsystemsthesupertracecanbe obtainedfromtheone-loopcounterterm,andwewillgenerallyrefertothiscontribution asthe superanomaly. Iftheclassicaltheoryisnotconf ormallyinvariantthesupertrace, computedfromtherenormalizedeectiveaction,doesnotequalthesuperanomaly.We

PAGE 503

7.10Anomalies485discussinthissectiontheminimal n = Š1 3 theorywithachiralcompensator. Wede “netherenormalizedcurrents J€= R H€ (7.10.46) J = R3 .(7. 10.47) (IntheversionofthetheorywithvariantmultipletcompensatorswehaveR V = J + J orR = J .)Wewillassumeforthetimebein gthatt heminimaltheoryhasno lo calsupersymmetryanomalies.Invarianceoftheeectiveactionunderlocalsupersymmetrytransformationsgivesthen €J€= J .(7. 10.48) ThesupertraceiszeroonlyifRisindepe ndentof .(Theoper ationisd e“nedin (5.5.44).) Thesuperanomalyisgivenby Jan= S3 (7.10.49) J = Janonlyiftheclassicaltheoryissuperconformal. Therelevantone-loopexpressionscorrespondingto(7.10.9,10,17)are U1 D Š 4 dDxd22(D Š 1) D Š 2 W( +2 )1 2 D Š 2W+ h c ., (7.10.50a) SŠ2 D Š 4 dDxd22(D Š 1) D Š 2 W2+ h c .= Š2 D Š 4 dDxg1 2 ( w2+ w2),(7. 10.50b) D=1 (4 )1 2 D [1 2 dDxd22(D Š 1) D Š 2 W2+ h c .+ dDxd4 EŠ 1( G2+2 RR )],(7.10.51) where W2=1 2 WW, isasuper covariantizeddAlembertian,and DisthesupersymmetricformoftheEuler numberof(7. 10.7).Theexpressi oncorresp o ndingto (7.10.16)is

PAGE 504

4867.QUANTUMN=1SUPERGRAVITYU= k11 D+1 (4 )2 d4xd4 EŠ 1[( k2Š k3)1 2 R ( 1 Š ln +2 ) R Š k31 2 G€( 1 Š ln 2 ) G€] Š k41 (4 )2 { d4xd23W2ln [1 Š ( 2+ R ) 1 Š R ]+ h c } ,(7. 10.52) andrepresent san unambiguouswayoforganizingtheo-shellcovariantizedcontributionsfromsupergraphswithtwoorthreeexternallines.Otherterms,withmorefactors of W,donotcontri butetoon-shellsupertraces.ThedAlembertian Šwasde “ned in(7.4.4). Iftheclassicaltheoryissuperconformal k1= k4.Otherwise,the yhavetobecomputedseparately,e.g.,fromaself-energyan dfromatriang lesupergraph,respectively. Forexample,t helastterm in(7.10.52)canbeexpandedas k41 (4 )2 d4xd4 EŠ 1W21 Š R + h c .+ ... ,(7. 10.53) andgives,atthelinearizedlevel, J€=1 3 k41 (4 )2 i €1 ( D2 W2Š D2W2).(7. 10.54) Thiscorrespondstothecontributionfromatrianglegraphwithtwolegsonshell.Its formisuniquelydeterminedbycovariancea ndpowercount ing,andtheactualvalueof k4canbedetermined,forexample,bytheAdler-Rosenbergmethod. Thesupertraceandsuperanomalyaregivenby J =1 3 k41 (4 )2 W2, Jan=1 3 k11 (4 )2 W2.(7. 10.55) Thesuperanomalycanbereadfromtheresultscontainedin(7.8.5)whichgivetheonshellvalueofthe“rsttermin(7.10.52).Forascalarmultiplet k1=1 24 ,wh ileforatensormultiplet,includingghosts,itis k1=25 24 .Inabackgr oundcovariantgaugeforthe v ectormult ipletthecontributionto Scomesentirelyfromtheth reechiralghostssince,

PAGE 505

7.10Anomalies487asdiscussedinsec.7.8,genera lsuper “eldsdonotcontributetothetwo-pointfunction. Thus,fortheYang-Millsmultipletwehave k1= Š3 24 .Forthegra vitinomattermultiplet,withtheeectiveLagrangianof(7.3.6)or(7.3.7)we“nd,byaddingcontributions fromthechiralghosts, k1= Š19 24 or5 24 ,resp ectively,forthetwodierentsetsofcompensat ors.Finally,forthesupergravitymultipletweobtainthevalues k1=41 24 Š7 24 Š55 24 dependingonwhetherweusea V ,or compensator.These numberscan alsobeobtainedfromacomponentanalysisofthetheories,usingthevalues k1of(7.10.18)forthecomponenttraceanomaly.(Changingfromonecompensator toanothercorrespondstoreplacingsomeofthespinzeroauxiliary“eldswithdivergencesofvectorauxiliary“elds.) Forthescal armultiplet,whichisclassicallysuperconformallyinvariant, k4= k1=1 24 ,andfort hesamereason ,forthev ectormult iplet k4= Š3 24 .Sincethe tensormultipletisphysicallyequivalenttothescalarmultiplet,ithasthesamevalue k4=1 24 ( = k1sincetheclassicaltheoryisnotsuperconformal).Thisresulthasbeen checkedbya nexp licitcalculation. Forthesuperg ravitymultiplettheexplicitcalculationshavenotbeencompletely carriedout.Ifweconjecturethatthecontributionstothesupertraceagaincomecompletelyfromthechiral“eldsinthequantumaction,wecandeterminethecoecients k4. Sincewearediscussingthesupertrace,chiralspinorsareequivalenttochiralscalarsor, whatamountstothesamething,theresultisindependentofthetypeofcompensator weuse.Thisgi vesthevalue k4= Š7 24 forthesupergravitymu ltipletand,byasimilar reasoning, k4=5 24 forthe(3 2 ,1)gravitino mattermultiplet.(Forexample,inthe(2,3 2 ) mult iplet,replacingthechiralcompensatorwitha V compensatorreplacestwo sand two €switheight swithoppositestatistics.Forthe(3 2 ,1)multi plettheequivalent ofone andone €in(7.3.6)isfourmore s,asin(7.3.7).) Forthescal arandvectormultiplets,thesupertraceresultsarealsoconsistentwith thecalculatedvaluesofthecomponentaxialcurrentanomalies(providedweassignthe correctR-weights1 3 Š 1forthefe rmionsofthescalarandvectormultiplet,respectively). However,theconventionallyquotedvalueforthegravitinoaxialanomaly ( mj5 m=21 24 (4 )Š 2r r )doesnotmatchthe energy-momentumtraceforeitherthe

PAGE 506

4887.QUANTUMN=1SUPERGRAVITY Nsmaxtotaltrace( k4) 00 8/360* 1/27/360 152/360 3/2127/360 2232/360 =( Š 1)2 smax+1(15 smax 2Š 2)/90 1 1/21/24 13/24 3/25/24 27/24 =( Š 1)2 smax+1(4 smax+1) /24 2 1/21/12 11/12 3/21/12 21/12 =( Š 1)2 smax+1/12 3all0 Table7. 10.1.Valuesofthetotaltracecoecients(*Complexconformalscalar) (2,3 2 )or(3 2 ,1)multi plet.Thisisaconsequenceofthefactthatthecomponentanomaly wascalculatedf oraclassically conser ved gravitinoaxialcurrent,whereasthecomponent currentcontainedin J aisnotclassicallycons erved:Itcontainsadditionaltermswhich givenonvan ishi ngcontributionsto mj5 m.(Itsenerg y-momentumpartnerisnottraceless:e.g.,thetraceofthe quadratic partoftheEinsteintensor,representingtheenergymomentumtensorofthegraviton“eld,isclassicallynonvanishingevenonshell.Thisis duetotheconformalnoninvari anceofEinsteingravity.) Thevaluesofthe k4coecientscalculatedonthebasisofourconjecturearepresentedinTable7.10.1,whic hgivesthesup ertracein N =0, N =1,andexte ndedsupersymmetry.That k4=0for N 3r e”ectsagaintheabsenceofanetnumberofchiral super“elds. Theveri“cationofourstatementsawaitsanexplicitcalculationoftherelevanttrian gl es upergravitysupergraph,andabetterunderstandingofsomeofthecomponent

PAGE 507

7.10Anomalies489calculations.Ifourconjectureiscorrect,itisrathercurious,andnotunderstood,that thesupergravit ytheorywiththe V compensatorbehavesasifitweresuperconformal ( k1= k4)or,equivalent ly,thatthesuperanomalyandsupertracedieronlyifchiral spinorsarepresent. e.Localsupersy mmetryanomalies Wecanusetheexi stenceofsuperconformalanomaliestoinfertheexistenceof anomaliesintheWardidentitiesoflocalsupersymmetryfor n = Š1 3 .Wedemonstrate thisexplicitlyforthecaseofquantummatte rmulti pletscoupledtobackgroundsupergravity,butexpectsimilarresultswhensupergravityitselfisquantized.We“rstconsider N =1superg ravity. Atthe linearizedlevel,theWardidentitiesr e”ecttheinvarianceoftheeective ac ti on underthe(linearized)localsupersymmetrytransformations( L= L) H€= D L€Š D€L,(7. 10.56) n = Š1 3 : 3= D2DL, n =0: = Š 2 D2L+ i D2DK n = Š1 3 ,0: H= i ( Š1 3 D2L+1 3 D€D L€+1 2 n +1 3 n +1 D D€ L€+ DL).(7. 10.57) Wehave usedth e gauge H=0for n = Š1 3 ;the gauge H= Š i D€H€( 1 H=0, 1 eŠ H=1;sees ec.5.2.b)for n =0,sothat EŠ 1canbelinearizedas 1+1 2 ( D+ D€ €);andthegauge=1forother n .For n = Š1 3 ,0weha vemade theshift H HŠ1 3 i D€H€sothat J€isthesuper conformalcurrent(couplingto co nf ormalsupergravitysaxialvector,andnottheotherauxiliaryaxialvector).Wehave

PAGE 508

4907.QUANTUMN=1SUPERGRAVITY0= R= d4xd4 ( H€) J€+ d4xd2 ( 3) J + h c d4xd2 ( ) + h c d4xd4 ( iH) + h c ., (7.10.58) where J€ R H€ J R3 R R ( iH) .(7. 10.59) Ifwerequirethat(7.10.58)besatis“ed,weobtainthe(linearized)conservationlaws n = Š1 3 : D€J€= DJ D€J =0; n =0: D€J€= Š 2 D€= DŠ D€€=0; othern : D€J€=1 3 D2+1 3 D€D €+1 2 n +1 3 n +1 D D€ €, D( )=0.(7. 10.60) TheinvariancesusedtoderivetheseconservationlawsarethoseofPoincar esupergravity,andtheirviolationwouldimplythatthemultipletcontributingtoRcannotbe coupledconsistentlytothecorrespondingformofsupergravity.Ontheotherhand,the violationofthesuperconformalconservationlaw D€J€=0imp liesonlythatthemultipletcannotbecoupledconsistentlytoconformalsupergravity. Weevaluatema trixelementsoftheconservationequations(7.10.60)betweenthe vac uumandanon-shellsupergravitystate.Inparticular,ifweconsiderone-looptrianglegraphsweknowthepreciseformofthel eft-handside.Asdiscussedintheprevious subsection,powercountingandcovariancedeterminesuniquelythematrixelementof thesuper current:

PAGE 509

7.10Anomalies491< ( H H ) | J€| 0 > i €1 [ D2( W)2Š D2( W€€€)2].(7. 10.61) Thenwehaveforthematrixelementof D€J€< D€J€> DW2.(7. 10.62) Itisnotzero(exceptwhenthesupertracevanishes),andisindependentoftheformof thecompensator. Wenowexami nethematrixelementoftheright handsideof(7. 10.60).Webegin byconsid eringcontributionstotheone-loopeect iveactionfromaclassicallysuperconformalmultiplet.Its(locallysupersymmetr ic,covariantized)actionisindependentof thecompensator.However,asdiscussedint heprevioussubsection,thecompensator entersthe(one-loop)renormalizedeectiveactionafterthedivergenceshavebeensubtractedout.Wecanasknowiftheform(7.10.62)iscompatiblewiththerighthandside of(7.10.60).SincethecompensatorentersRonlybecausewehavesubtractedoutthe covariant, local, counterterm S( H ,compensator),thecorrespondingcurrentmustalsobe local.For n = Š1 3 asolutiono f(7. 10.60)is J W2, but forn = Š1 3 thereexistsno local or thatsatis“estheconservationequation. Weconcl udethatanysuperconformal N =1mult ipletthathasanonzeroo ne-loopsupertracegivesacontributionto RthatviolatesthePoincar esupergravit yconserv ationlawsfor n = Š1 3 i.e.hasa local supersymmetryanomaly. Therefore,ingeneral,superconformalmultipletscanbecoupledconsiste ntlyon lyto n = Š1 3 supergravity.(Theanalysi saboveis inconclusive, however,ifthesupertracevanishes,e.g.fo rasy stemofonevectorandthreescalarmultiplets,whichhasnoone-loop divergenceorsupertrace.) Inthecasewheretheclassicaltheoryisnonsuperconformal,thecompensatorsmay coupletononlocaltermsintheeectiveaction.Thus,for n = Š1 3 ,0, <> D1 D2 W2(7.10.63) cansatisfy(7.10.60)andwecannotconclude,withoutfurtheranalysis,thatPoincar e supergravityanomaliesarepresent.However,wecanstillconcludethatananomalyis presentfor n =0since( 7.10.60)implies D2J€= €J€=0,whereas D€J€ DW2implies D2J€ i €W2and €J€ i ( D2W2Š D2 W2),neit herofwhichvanisheven

PAGE 510

4927.QUANTUMN=1SUPERGRAVITYonshell.Thisoccursbecausethesupertraceisanirreduciblemultipletofsuperspin0 ( W2isachiralscalar),whereasthecompensatormultipletfor n =0hass uperspin1 2 Fornonmini malsupergravity( n = Š1 3 ,0)wew illseebelowthatanomaliesare ab sentonlyunderveryspecialcircumstances.Ingeneral,theirpresenceisrelatedtothe nonexistenceofachiralmeasure.Aninterestingwaytounderstandtheoriginofthe anomalyistousethefactthat(inappropriatesupersymmetricgauges)only(physicalor ghost)chiralsuper“eldscont ributetothedivergencesandreq uireregularization.Inparticular,wecanaskifPauli-Villarsregularizationispossibleforchiralscalarsuper“elds withthevariousnonconformalcouplingsofsec.5.5.Sinceonly n = Š1 3 hasachiral measurethatallowsmasstermsforchiralsuper“eldswith conformalkin eticterms, itis theonly n thatallowsPauli-Villarsregularizationforthosesuper“elds.(Inotherregularizationschemes,thesamedicultywithchiralmeasuresshowsupinotherways:e.g., in dimensionalregularization,“ndinganalogstothechiralintegrandsinthelasttermof (7.10.52).)Infact,wewillshowbelowthattheonlyquantum-consistentcouplingsto su pe rgravityarethosewhich:(1)allowPauli-Villarsregularization,(2)havevanishing supertrace,or(3)havecouplingsthatco rrespondtoextendedsupersymmetry.For n = Š1 3 allchirals uper“eldscanhavemassterms,soallcouplingsarepossible.For other n couplingtothevector mult ipletaloneisimpossible(itisclassicallysuperconformalandhasclassicallysuperconformalchiralghosts),couplingtoascalarmultiplet aloneispossibleonlyforthenonconformalc ouplingthatallowsmass ( butnotself-interaction)terms,andcouplingt othecombinat ionofthetwor equiresacancellationthat o ccursinextendedmultiplets(andprobablynowhereelse,ifthecancellationistobe exactlymaintainedathigherloops). Todisc ussthesituationquantitatively,we performanexp licitveri“cationofthe conservationlaw(7.10.60)forcontributionsfromchiralscalarswithnonsuperconformal couplings.Fromtheactionsofsec.5.5(withthede“nitionsin(7.10.59))we“ndthe classicalcurrents J€= Š1 6 [ D€, D] +1 2 i€ forn =0, Š2 n+1 2 [ D€, D] +1 2 i€ forn =0,

PAGE 511

7.10Anomalies493 J =1 3 D2 n = Š1 3 =3 n+1 2 D2D n =0 =3 n+1 2 D othern (7.10.64) Wenowim aginecom puting reno rmalized one-loopmatrixelementsofthesecurrents betw eenthevacuumandanon-shellbackgroundsupergravitystate.Thematrixelementof J€musthavethefor m(7. 10.61)and,inparticular,its =0compon enthas theform i € Š 1( w2Š w2).Weobservethat i€ givesnocontributiontothiscomponentandt hereforenocontributionatall,sinceanycovariantsuper“eldthatvanishes at =0vanishesid entically.(Thetopvertexofthegraphcontainsonlycrossterms A Bof | =A+ i B,whereasthegravitationalco uplingsareproportionaltoAAand BB.)Therefore,tocomputematrixelementsof any ofthecurrentsin(7.10.63)itissuf“cienttocomputematrixelements < ( H H ) | | 0 > fortwo-particleon-shellgraviton states( H H ),andthe na pplyappropriateoperators(e.g., < J€> < [ D€, D] > =[ D€, D] < > ,etc.). Bypowercountingandcovarianceargume nts,thereno rmalizedmatrixelementhas the uniqueform < > = c 1 ( D2W2+ D2 W2)(7. 10.65) wherecisanumericalfactor.Wenowsubstitutethecorrespondingexpressionsof (7.10.64)intotheconservationlaws(7.3.59).Since <> =1 2 (3 n+1) D2D< > =0(7. 10.66) always,we“ndthattheconservationlawsare never satis“edfor n =0( unless c =0). For n =0s ubsti tuting(7.10.64)into(7.10.60)gives Š1 2 D D2< > =3 n+1 3 n +1 D D2< > ,( 7.10.67a) whichissatis“edonlyfor n= Š1 2 ( n +1).(7. 10.67b) For n = Š1 3 ,thisistheonlyvalueof nde“ned,evenclassically(seesec.5.5.f.2).For

PAGE 512

4947.QUANTUMN=1SUPERGRAVITYother n ,thisval ueisexactlytheonethatallowsamassterm. Toinvest igateanomalycancelingm echanisms,weconsidercontributionsfromone v ectormult ipletand l identicalscalarmultipletswitharbitraryweight n.Theco ntri butionofthevectormultiplettothenonlocalpartof <> and <> mustva nish b ecauseofclassicalsuperconformalinvariance(furthermore,theghostsmusthave ng host= Š1 3 ).Thecontributionto < D€J€> is Š 3timesth atofaphysicalscalarmultiplet .The l scalarmultipletscontributetoboththeleftandrighthandsidesof (7.10.60).Theconservationlawnowbecomes Š1 2 D D2< > ( l Š 3)=3 n+1 3 n +1 D D2< > l ,( 7.10.68a) whichgivesthecondition nl=1 2 (3 l Š 1) n +1 2 (1 l Š 1).(7.10.68b) Inparticular,for l =3,which correspo ndsto the N =4v ectormult iplet,wheredivergencescancel,thesuperconformalcoupling n= Š1 3 isrequired.For l =1,the N =2 v ectormu ltiplet,we“nd nl= n .R ecallthatfor n =0theconserv ationlawsrequirethe supertrace D€J€tovanish identically eventhoughthetheorymaystillhavedivergences (i.e.,thesuperanomalymaybenonzero).Wethushave Š2 n+1 2 [ D€, D] < > l Š1 6 [ D€, D] < > ( Š 3)=0,agreeingwith(7.10.68)for n =0. Wehaveth usfoundthatfor N =1only n = Š1 3 isgenerallyquantumconsistent, whileforother n onlyveryspecial nonsuperconformalcouplingsareallowed.These argumentscanbeappliedtoextendedsup ergravity.Inparticular,thestandard N =2 theory,which(intermsof N =2super“elds)h asanisovectorcompensator Va b,isq uantumconsistent,basicallybecauseithaschiralmeasure.Thusan N =2v ectormu ltipletwillgivecontributionstotheeectiveactionwhichareanomaly-free.Whenanalyzedintermsof N =1super“elds, N =2superg ravitydecomposesintoa(3 2 ,1)multipletcoupledto N =1, n = Š 1supergravity.The N =2v ectormultipletdecomposes intoa N =1v ectormultipletandanonconformalscalarmultiplet,butwith n= n = Š 1whichisco nsistentwiththenoanomalyconditionwederivedabove.Itis

PAGE 513

7.10Anomalies495likelythatt hisextendedsupersymmetryisnecessaryfor n = Š1 3 forthiscan cellationof theanomaliesinthesupergravitationalco nservationlawtooccurathigherloops. f.NottheAdler-Bardeentheorem Insec.6.7weconsideredtheanomalyinthe(axial)Yang-Millscurrentand,on thebasisofthecovariantrules,concludedthatit(anditscomponentaxialcurrent)satis“estheAdler-Bardeentheorem.Ontheoth erhand,thesupertrace(anomaly)ingeneralreceiveshigher-ordercorrections(the -functionisnotzero),andthereforethecomponentR-cu rrentdoesnotsatisfytheAdler-Bardeentheorem.(Weareconsideringhere matrixelementsofthecurrentbetweenthevacuumandon-shellYang-Millsstates, ratherthansupergravitystates.)Although thecurrentslookthesameclassically(fora scalarmultip letinanexternalvectormultipletorsupergravitybackgroundthe A a A termdoes notcontribute),thedierencearisesbecauseofdierentrenormalizationprescriptions. Inthe“rstcase,whentheaxial-vectorgaugesuper“eldisexternal(otherwise,in thepresenceofo ne-loopanomaliesthequantumtheorymakesnosense),itispossibleto renormalizethehigher-loopeectiveaction( V+, VŠ( ext ))andd e“ne Jrenormsothatit isnota nomalous.Ontheotherhand,if VŠ( ext )isrepla cedwith H€( ext ),thehigherloope ectiveaction( V+, H€( ext ))isusuallyrenormalized inama nnerwhichisconsistentwithPoincar esupergravity gaugeinvariance.Inthatcase,wedonothavethe freedomtorede“ne J€ renormsoastoremoveitshigher-loopsupertrace(anomaly).Ifwe giveupsuper-Poincar einvarian ce,wecanrenormalizesothat €J€ renorm=0athi gher loops.However, thisJ€ renormwillnotcontainaconserved (symmetric)energy-momentumtensoratthe level.Therefor e,therenormalizedchiralR-currentwhichisinthe samemultipletwiththerenormalizedconservedenergy-momentumtensordoesnotsatisfytheAdler-B ardeentheorem.

PAGE 514

Contents of8.BREAKDOWN 8.1.Introduction496 8.2.Explicitbreakingofglobalsupersymmetry500 8.3.Spontaneousbreakingofglobalsupersymmetry506 a.Renormalizabletheories506 a.1.Classicaleects506 a.2.Loopcorrections509 b.Nonrenormalizabletheories511 c.Globalgaugesystems513 8.4.Traceformulaefromsuperspace518 a.Explicitbreaking518 b.Spontaneou sbreak ing520 8.5.Nonlinearrealizations522 8.6.SuperHiggsmechanism527 8.7.Supergravityandsymmetrybreaking529 a.Massmatrices532 a.1.Vacuumconditions532 a.2.Gravitinomass533 a.3.Waveequations534 a.4.Bosemasses535 a.5.Fermimasses536 a.6.Supertrace538 b.Super“eldcomputationofthesupertrace539 c.Examples540

PAGE 516

8.BREAKDOWN 8.1.Introduction Themoststrikingfeatureoftherelationbetweensupersymmetryandthe observedworldistheabsenceofanyexperi mentalevidencefortheformerinthelatter. Theparticlesweseedonotfallintosupersymmetricmultiplets,nordotheyshoweven anapproximatemassequalitythatwouldindicatetheywereinmultipletsbeforesymmetr yb reaking.Thusifsupersymmetryisanunderlyingsymmetryofthephysicalworld, itmustbeba dlybroken,orotherwisehiddenfromdirectexperimentalveri“cation. At af undamentallevel,itisdiculttoaccepttheideaofaglobalsupersymmetry withoutbelievingthatthereexistsanunderlyinglocalsupersymmetry:Sincewebelieve thatgravitymustbequanti zed,andsinceevenglobalsupersymmetryimpliesthatthe gravitonrequiresaspin3 2 gravitinopartner,thenthegravitinomustbethegaugeparticleoflocalsupersymmetry,howeverbadlybrokenglobalsupersymmetrymaybe.Then, asinanygaugetheory,thesupersymmetrybreakingmustbespontaneous(i.e.,bythe vac uum)andnotexplicit( i.e.,intheactionitself).Ifwebelieveinlocalsupersymmetry withsymmetrybreaking,wemustunderstandmechanismsforthisbreaking.Itcanbe th roughtheHiggsmechanism,orduetocosmologicalfactorssuchasboundaryconditionsorhightemperaturee ectsintheearlyuniverse,ornonperturbativedynamical eects,orviadimensionalcompacti“cation.Itisalsoreasonabletobelievethatthe breakinghappensatalargeenergyscale.Ifthisisso,wemayhopethatthedynamical eectsofthesupergravity“eldscanbeignoredatalowerenergyscale,andthatthe eectivelowenergytheoryisabrokengloballysupersymmetrictheory.Wecanstart withanexactgloballysupersymmetrictheory,atsomescalewheresupergravity“elds havedecoupled,andinvestigateitsspontaneousbreakingabinitio,orwecanputthe breakinginbyhand,asanexplicitmanifestat ionoftheoriginallocalbreaking.(Ingeneral,ifwestartwithalocallysupersymmetr ictheorythatexhibitssymmetrybreaking andsetgravitational“eldsandcouplingstozero,softbreakingtermsareinduced). Unlikeothersymmetries,therearesomein terestingand unexpectedrestrictionson thepossiblebreakingofglobalsupersymmetry.Someofthesehavetheirorigininthe supersymmetryalgebraitself ,wh ileothersaremosteasilyobtainableinthecontextof super“eldperturbationtheory.The“rstres triction,whichfollowsfromthealgebra,is

PAGE 517

8.1.Introduction497thefollowingtheorem: Ifsupersymmetryisnotspontaneouslybroken,i.e.,ifthevacuumisinvariant undersupersymmetrytransformations,thenitsenergyiszero;conversely,ifthere existsastateforwhichtheexpectationvalueoftheHamiltonianiszero,supersymmetryisnotspontaneouslybroken.Furthermore,ifsupersymmetryisspontaneouslybrokenwithoutanattendantmodi“cationofthesupersymmetryalgebra, thenthevac uumenergyispositive. Asdiscussedinsec.3.2,thisresultfollowsdirectlyfromthecommutationrelations { Q Q } = P ,whichgivein particular Evac= Š 1 2 N €< 0 |{ Qa Qa€}| 0 > = 1 2 N a < 0 || Qa |2| 0 > (8.1.1) Thus: Ifallthecomponentsofthesupersymmetrycharge(generators)annihilatethevacuum,itsenergyiszero.Ifanyoneofthemdoesnotannihilatethevacuum,then itsenergyispositive. Weem phasizethat thistheoremassumest hatthesupersymmetryalgebraisnot chan ged.Withanappropriateinterpretationofthetotalenergyandcharge,thetheoremalsoholdsinsupergravity.Ontheotherhand,explicitbreakingdoeschangethe algebraandthennegativeo r zeroenergyispossible. Thesecondimportantresult,provedforafairlylargeclassofrenormalizablemodels,isthatinspontaneouslybrokenglobaltheories,therearemasssumrulesrelating fermion andbosonmasses,whichtaketheformstates mB 2Šstates mF 2=0.(8. 1.2) Thesesumrulesareextremelyrestrictive,andmaketheconstructionofrealisticmodels dicult;however,forlocallysupersymmetr icandexplicitly(softly)brokenglobally supersymmetrictheories,t hegeneralizationsofthisformulaarephenomenologically acceptable. Athird resultisthefollowingtheorem,whichcanbeproveninperturbationtheory forfour-dimensionaltheories:

PAGE 518

4988.BREAKDOWNIfsupersymmetryisnotspontaneouslybro kenatthetreelevel, thenitisnotbrokenbyradiativeco rrections.AColeman-Weinbergmechanismisnotpossible. Thistheoremisnotvalidintwo-dimensionalsupersymmetry. Ifsupersymmetryisspontaneouslybrok en,amasslessGoldstonefermionmustbe present.Therefore,ifonecanprovethatnomasslessfermionstatescanexist,supersymmetrycannotbebrokenspontaneously.Usingthisfact,Wittenhasgivencertaincriteria (indextheorems)thatallowonetoruleoutinasimplemanner,incertaincases,the po ssi b ilityofspontaneoussupersymmetrybreaking.Inalocallysupersymmetrictheory, theGoldstonefermionisabsorbedbythespin3 2 gravitinoviaaconventionalHiggs mechanism.Indextheoremshavenotbeeninvestigatedinsupergravity. Globalsupersymmetrybreakingismosteas ilydiscu ssedinsuper“eldlanguageasa breakingof Q -translationalinvarianceinsupers pace.Thiscanhappeneitherbecause thevac uumisnot Q -translationallyinvariant(spontaneousbreaking),orbecauseone hasexp licit -dependenceintheeectiveaction(eitheratthetreelevelornonperturbatively,forexampleviainstantoneects,whichcouldintroducesuchexplicitdependence).Ifthebreakingisspontaneous,it meansingeneralthatsomesuper“eldhasa nonzerovacuumexpectationvalue(ifLorentzandinternalsymmetryinvariancearenot tobebroken,ithastobeaneutralscalarsu per“eld).F urthermore,thenonzeroexpectationvaluemustresideinotherthanthe -independentcomponentofthe“eld,sothat someex p licit -dependenceisintroduced.For N =1ma ttersuper “elds, itmeansthat oneoftheauxiliary“eldsmusthaveanonzerovacuumexpectationvalue.(Unlessgauge invarianceisbroken,vacuumexpectationvaluesforthegaugecomponentscannotbe physica llyrelevant.) Supersymmetrybreakinginalocalconte xtandthesuperHiggsmechanismcan alsobedescribeddirectlyinsuperspace.Allthestandardmethods,suchasthetheory ofnonlinearrealizations,canbeappliedand allthestandardresults,suchastheconversionoftheGoldstinointohelicitymodesofamassivegravitinoandtheexistenceofUgauge,canbegeneralizedtothesuper“elddiscussionofspontaneouslybrokensupersymmetry;theresultingformalismisconsiderablysimplerthanacomponentapproach. However,some i ssues(atthepresenttime)canbesettledonlybyconsideringcomponentsdirect ly,e.g.,whatarecomponent“eldmasses,whataretheconditionsforspontaneousbre akingtooccur,whatistheWittenindex,etc.Therefore,althoughmostofthe

PAGE 519

8.1.Introduction499materialinthischapterisatthesuper“eldl evel,wecannotavoidsomecomponentcalculations,andwealsoomitsometopicsthathavenot,asyet,receivedanadequate superspacetreatment. We“rst discusssoftexplicitbreakingofglob alsupersymmetry(sec.8.2).OurcriterionforsoftnessistheanalogofSymanzikscriterioninordinary“eldtheory:In renormalizablegloballysupersymmetrictheo ries,theonlyrelevantdivergencesarelogarithmic.Weaskwhatnonsupersymmetric termscanbeadde dtothecl assicalaction withoutspoilingthedelicatecancellationsbetweenbosonandfermioncontributionsthat areresponsibleforthe absenceofq uadraticdivergences.Sincewecancasttheproblem insuper“eldlanguage,weareabletotakeadvantageofthesuper“eldpowercounting rulesofchapter6. Wenexttreatspontan eousbreakingofglobalsupersymmetryforbothrenormalizableandnonrenormalizabletheories(sec.8.3).(Nonrenormalizabletheoriesarerelevant toourdiscussionofbreakinginthecontextoflocalsupersymmetry.)Wedonotdiscuss Wittensindextheorem,orbreakingofsupersymmetrybyinstantons;asnotedabove, withourpresenttechniquestheseissuescanbehandledonlyatthecomponentlevel. Wedo,howev er,giveasuperspacederivationofthesupertracemassformulae(sec.8.4). Thisderivationismuchsimplerthanthecomponentcalculation(whichwealsogive, partlyforcomparison,butalsobecauseitprovidessomeextrainformation,e.g.,the massesoftheindividualcomponents). Finally,wediscussthesuperHiggseect.WeshowhowtheGoldstinocanbe describedbyanonlinear(super“eld)realizationofsupersymmetry,andhowstandard radialandanglevari ablescanbeintroducedinmodelswithspontaneouslybroken supersymmetry(sec.8.5).W eexhib itthesuperHiggsmechanism(sec.8.6)andgivea deta ileddiscussionofthecaseofarbitrarysupe rsymmetricmattersystemscoupledto supergravity(sec.8.7).

PAGE 520

5008.BREAKDOWN8.2.Explicitbreakingofglobalsupersymmetry Oneoftheimportantfeaturesofsupersy mmetrictheoriesistheperturbativenorenormalizationtheorem(sec.6.3.c):ThesuperspacepotentialP()forchiralsuper“eldsreceivesnoradiativecorrections,sotha ts calarmultipletmassesandcouplingconstantsarenotrenormalized(asidefromtheeectofwavefunctionrenormalizations). Fu rthermore,forrenormalizablemodels,onlyl ogarithmicdivergencesarepresent(asdiscussedinsec.6.5,quadraticallydivergentD-termsarenotgeneratedifgaugeinvariant regularizationisused).Whensupersymmetryisexplicitlybrokenthisisnolongerthe case,and,ingeneral,quadrati callydivergentcorrectionscanbeinduced.Equivalently, theparametersofane ectivelowenergytheorycandependquadraticallyonmasses associatedwiththetheoryde “nedathighenergies,andsomeofthenaturalnessof supersymmetrictheoriesisdestroyed.Ho wever,thereexistsa setofsupersymmetry breakingtermswhoseeectis soft :Whena ddedtoasupersymmetr icLagrangian,any ne wd ivergencesthatthesetermsgeneratearel ogarithmic.Moreprecisely,ifweintroducecountertermsintheclassicalLagrangiantocancelthenewdivergences,afterrenormalizationtheirdependenceontherenormalizationmass(orhighenergycuto)isonly logarithmic.Inthissection,wedescribethesetofsoftbreakingterms,andtheadditionaltermsthattheyinduce. Breakingsupersymmetryisbreaking Q -translationalinva riance.Thisisdoneby introducingexplicit -dependenceintotheLagrangian.Equivalently,wecanintroducea super“eld( x )witha“xed -dependentvalue.Thissuggeststhefollowingprocedure: Givenasupersymmetricaction,wegeneratenewtermsbycoupling,inamanifestly supersymmetricfashion,someexternal(spurion)super“eld(s)tothequantum“elds. Supersymmetrybreakingisachievedbygivingthese“eldssuitable( -dependent)“xed values.Atthe componentlevel,thisintroducessomenonsupersymmetricterms.Soft breakingisachievedbyonlyallowingnewco up lingsthatareconsistentwiththe(power co un ting)renormalizabilitycriteriaofsuper“eldperturbationtheory,sothatnodivergencesworsethanlogarithmicareintroduced .T heinducedin“nities correspondtoconventionald ivergenttermsintheeectiveactioni nvolving productsofthequantumand spurion“elds.Whenthespurion“eldsaregiventheir“xedvalueswecandeterminethe correspondingnewcomponentin“nities.Generally,wewill“ndthatinacomponent languagesymmetrybreakingtermsofdime nsiontwoaresoft,buttermsofdimension

PAGE 521

8.2.Explicitbreakingofglobalsupersymmetry501threearenot(withsomeexceptions).Thesetermscorrespondtosplittingthemassesof particlesinmultipletsbyhand,oraddingsomenew,nonsupersymmetricinteractionsof averysp ecialform.Wewill“ndthattherearee ssentia lly“vedistincttypesofsoft breakingtermsthatcanoccursingly,orincombinations.Ingeneral,onesuchterm inducestheothers,sothatwe shoulddiscussthemallatthesametime.However,since theirphysi calsigni“canceisdierent,wep refertotreatthemoneatatime. Weconsid erconventionalrenormalizableLagrangians(cf.sec.4.3)oftheform S = d4xd4 [ ieVi+ trV ]+ d4xd2 [1 2 WW+P(i)]+ h c .(8. 2.1) wherePisapolynomialofdegreethreeorless.Bypowercountingweknowthatthe onlydivergencesofthetheorycorrespondtotermsintheeectiveactionoftheform d4xd4 d4xd4 Vn d4xd4 V ( D )2( D )2Vn d4xd4 V (8.2.2) wherethe D -derivat ivesaresui tablydistributedandtermswith n > 1are relatedto termswith n =1by gaugeinvariance(weincludeghostsamongthechiral“eldsin (8.2.2)).Webreaksupersymmetrysoftlybycouplingadditionalexternalsuper“eldsina mannerconsistentwiththepowercountingcriteria(seesec.6.3):Nomorethanfour D sactingontheinternallinesshouldappearatanyvertexwheretheexternalspurion “eldisinserted.Inadditiontotheoriginald ivergen cesofthetheory,wemaygenerate newones,involvingthespurion“eldsaswell,andtheyaretheonesthatinterestus.In thissectionwedonotconsiderdivergencesinvolvingspurion“eldsonly,whichcorrespondtoinsertionsintovacuumdiagramsandcontributeonlytothevacuumenergy (cosmologicalconstant);see,however,sec.8.4. Sincethespurion“eldscanneverintroduceanyadditionalspinorderivativesintoa loop,ifinanysoftbr eakingtermthespurion“eldissetto1theresultingtermmustbe eitheraconventionalrenorma lizablesupersymmetrictermor atot al(spi nor)derivative. Thepossibleadditionalcouplingsthatintroduceexplicit -dependenceintotheaction correspondtomultiplyingaspurionfactorinto ,2, W2,3,or D( W).Indetail, wehave:

PAGE 522

5028.BREAKDOWN(a) Sbreak= d4xd4 U d4x 2A A (8.2.3) where U = 22 2isaneutraldime nsion-zero“xedgeneralscalarsuper“eld.Atthe classicallevel,whenaddedto(8.2.1),suchatermbreakstheequalityofthebosonand fermionmassesofascal armultipletbyadding Š 2tothemassesofA=2Š1 2 ReA and B=2Š1 2 ImA .Toinvesti gatethedivergencesitintroduces ,weconsiderloo pswithordinaryverticesandexternal U verti ces.Welookforlocaltermsintheeectiveaction, involvinga d4 integralandfactorsof U andthequantum“elds,ofdimensionnogreater than2.(Thisisourstandardpowercountingofsec.6.3.)Since U isdime nsionless, suchtermsare: U ,correspondingtoalogarithmicrenormalizationof(8.2.3),i.e.,of 2; U (+ )(butonlyifsomechiral“eldismassive);and UD D2DV (butonlyifthe gaugegrouphasa U (1)factor;the D -factorsarerequiredbygaugeinvariance).Therefore,theactionmayreceiveadditionall ogarithmicallydivergentcorrections: d4xd4 U + d4xd4 UDW+ h c d4x [ 2m A+ 2D](8. 2.4) (b) Sbreak= d4xd2 2+ h c d4x 2(A2Š B2)(8. 2.5) where = 22isaneutraldi mension-onechiralsuper“eld.Thisadditioncorresponds toanotherwayofsplittingthemassesofscalarsandpseudoscalarsawayfromthemass ofaspinorinachiralmultiplet.New,logar ithmicallydivergenttermsaregivenby d4xd4 + h c d4x F(8. 2.6) whereF= ReF .Since is neutralunderwhateverinternalsymmetrygroupsmaybe present,noin“nitiesinvolvinggauge“elds(asmightarisefroma eV term)canbe i nduced.

PAGE 523

8.2.Explicitbreakingofglobalsupersymmetry503(c) Sbreak=1 2 d4xd2 WW+ h c 1 2 d4x + h c .(8. 2.7) where = 2isaneutral,dimensionzerochiralsuper“eld.Againthisinvolvesvertices withonlyfour D sandisthereforesoft ,and providesamechanismforgivingmassesto fermionsin gaugemultiplets.Thefollowingdivergenttermsmaybegeneratedinadditiontocorrectionsto Sbreakitself: d4xd4 + h c d4x F d4xd4 ( + h c .) d4x A d4xd4 + h c d4x [FA Š GB] d4xd4 d4x [A2+B2](8. 2.8) Foragiventhe ory,notallofthesetermsneedappear;forexample,thethirdtermwill onlybegeneratedatthetwolooplevel,andonlyifamassivechiralsuper“eldispresent. (d) Sbreak= d4xd2 3+ h c d4x Re ( A3Š 3 AB2)(8. 2.9) with asin(c).Unlikethepreviouscases,thisintroducesanallowednonsupersymmetricinteractionterm.Ingeneralweinducethesamedivergencesasincase(c). Thebreakingtermd4 ( + ) canber educedbya“eldrede“nition (1+ )to thepreviouscases. Anotherpossibility,whichgivesagaugeinvariantmassmixingbetweenthe fermionsofa gaugemultipletandofascalarmultipletintheadjointrepresentation,is (e) Sbreak= d4xd4 DU W+ h c .

PAGE 524

5048.BREAKDOWN= d4xd4 D2DU DV + h c d4x Re [ + A D] (8.2.10) withadimension Š 1“eld U = 2 2,or,equivale ntly,w ithadime nsion1 2 chiral spinor super“eld = D2DU .L ogarithmiccorrectionsareinducedfor Sbreakitselfand for: d4xd4 + h c d4x 2F.(8.2 .11) Theabovepossibilitiesforsoftbreakingare”exibleenoughtocoverallinterestingphysicalsituationswithoutintroducingalargenumberofarbitraryparameters.(Withseveralmultiplets,becausecancellationsarepossible,othertypesoftermscanbesoft,e.g.,d4xd4 DU 1D2d4x ( F1A2Š F2A1).) Itisalsointerestingtoexaminesomeca sesofbreakingthatarenotsoft.Wementiontwo: (a) Sbreak= d4xd4 U ( D)( D)+ h c .(8. 2.12) with U asin(e),shiftsthemassofthespinorinascalarmultiplet.Butitleadstoverticeswithsix D s,asdoes (b) Sbreak= d4xd4 U (+ )3 d4x A3.(8. 2.13) Bothwillpro ducequadraticallydivergentterms,forexample d4xd4 U + h c d4x A(8. 2.14) We canunderstandthedierencebetweencases(a)and(b)ononehand,and(a) ontheotherasfollows:Theybothleadtofermion-bosonmasssplittingsforthescalar mult iplet.However,theformer,inadditiontosplittingmasses,alsoaectssomeofthe componentinteractionterms,anditis thedelicatebalanceofmasstermsand

PAGE 525

8.2.Explicitbreakingofglobalsupersymmetry505in te ractionsthatkeepsthedivergencesundercontrol.Ontheotherhand,thereisno dicultygivingmasstothefermionofavect ormultiplet,orintroducingmassmixing betw eenthefermionsofthetwomultiplets. Itisausefulandsimpleexercisetochecksomeoftheaboveconclusionsbyexaminingsupergraphsinvolvingthespurion“ elds.Wenotethatsomeoftheinducedterms wehave listedmaybemissing b ecauseofgrouptheoryrestri ctions,or,insomecases, b ecauseoftheabsenceofmasses,e.g.,thethirdtermincase(c).Incertaincasespossibletermsaremissingbecausethecorrespondinggraphsrequireor p ropagators andthese bringwiththemnumeratormassfactorsthatreducethedegreeofdivergence ofthediagrams.Forexample,incases(c)and(d)aterm 2cannotbeproduced b ecausethecorrespondingdiagramsmustcontaintwomassfactorsandhenceareconvergent.

PAGE 526

5068.BREAKDOWN8.3.Spontaneousbreakingofglobalsupersymmetry Ifglobalsupersymmetryisspontaneouslybroken,amassless(Goldstone)fermion mustbepres ent.ThiscanbeestablishedbytheusualreasoningthatprovestheGoldstonetheorem:Ifthesupersymmetrychargedoesnotannihilatethevacuum,thereexist operatorswhose(anti)commutatorwiththesupersymmetrychargehasnonzerovacuum expectationvalue,and,inparticular,wecanwrite < 0 |{ Q, S a }| 0 > = d4x x b < 0 | T ( S b ( x ) S a (0)) | 0 > (8.3.1) where S a isthesupersymmetrycurrent,satisfying aS a =0,and Q= d3xS 0 .The leftha ndsidenotbeingzero(itisactuallyproportionaltothevacuumenergydensity (8 .1 .1 )) im pliesthattherighthandsidereceivesacontributionfromasurfaceterm;this is th ec as eo nlyifthematrixelementvanishesatin“nitynotfasterthan | x |Š 3,whichis po ssibleonlyifamasslessfermion intermediatestateispresent. Thespontaneousbreakingofsupersymme tryingloballysupersymmetrictheories canbeinvestigatedbyexaminingtheeectivepotentialatitsminimum,whereitequals thevac uumenergy.Theeectivepotential U isobtainedfromminu sthee ectiveaction byse ttingallmomentaandallcomponent“eldsthatarenotscalarstozero.Wemust thenminimize U withrespecttoalltheremainingcomponent“eldsandaskifitvanishesa tthemi nimum. a.Renormalizabletheories a.1.Classicaleects We“rstco nsiderasystemwithonlychiralscalarsuper“elds,andarenormalizableclassicalactiongivenby(4.1.11): S = d4xd4 ii+ d4xd2 P(i)+ h c .(8. 3.2) wherePisapolynomialofdegreenohigherthanthree.Toinvestigatetheclassicalvacuumwesetallmomentaandfermion“eldstozero;wethenhaveeectively = A Š 2F ,andsi nceeach integrationrequiresa 2 2factor,weobtaintheclassical potential

PAGE 527

8.3.Spontaneousbreakingofglobalsupersymmetry507U = Š FiFiŠ [ FiPi( A )+ h c .](8.3 .3) wherePi= P Ai asin(4.1.13).Theclassicalvac uumisdescribedbytheconstant( x i ndependent)classical(expectation)valuesofthescalar“eldsobtainedbysolvingthe classical“eldequationsforconstant“elds,i.e.,byextremizingtheclassicalpotential. Extremizingwithrespecttothe Fi“rst,we“nd Fi= Š Pi( A );substitutinginto U we obtain U =i | Pi( A ) |2.(8. 3.4) Werequire U Ai = Pij( A ) Pi( A )=0.(8 .3.5) Thepotentialwillvanishattheextremuma ndsupersymmetrywillnotbebrokenonlyif thesimultaneou sequationsPi( A )=0haveasolution.Thisreq uirementisequivalentto thatofrequiringthatallthe F shave zerovacuumexpectationvalue.Wecanwork directlyinsuperspace,byde“ningP()asthe supers pacepotential. Theconditionfor supersymmetrynottobebrokenisformallythatthesuperspacepotentialhavean extremumwithrespecttothesuper“elds: P i =0. Weconsid ertwoexamples: (a)TheWe ss-Zuminomodel,withaction d4xd4 + d4xd2 [ a +1 2 m 2+1 6 3]+ h c .(8. 3.6) Thesuperspacepotential,whendierentiatedwithrespecttogives a + m +1 2 2andsettingthistozeroalwaysgivesusasol ution.Hencethevacuumenergyiszeroand thereisnosupersymmetrybreaking.Thisis thecaseevenifweconsideranarbitrary (nonrenormalizable)polynomialpotential. (b)Ontheotherhandwecanconsider theORaiferteaighmodel,givenby d4xd4 [ 00+ 11+ 22]+ d4xd2 [01 2+ m 12+ 0]+ h c .(8. 3.7) forwhichweobta intheequations

PAGE 528

5088.BREAKDOWN1 2+ =0 201+ m 2=0 m 1=0,(8. 3.8) whichhavenosolution.Inthiscasesupersymmetryisbrokenattheclassicallevel. Inthegeneralcase,thesituationdependsonthetopologicalstructureofP.In particularthepresenceofanextremum,andi tsstabilityundervariationsofparameters inP,canbestudiedinrigorousfashionandgivesrisetoindextheoremstodetermine whethersupersymmetrycanorcannotbespontaneouslybroken.Weremarkthatsupersymmetrybreakinginthesenseaboveimp liesthatthefermionmassmatrix,whichis givenbythematrixof secondderivativesPijevaluatedattheminimumof U (see (4.1.12)),hasazeroeigenvaluecorrespondingtothezeromassoftheGoldstonefermion. Indeed,if(8.3.5)issatis“edwithPj =0,Pijmustbe singular. Includinggaugeinvariantinteractionswitharealgaugesuper“elddoesnotfundamentallychangethediscussion.Gaugeinvarianttermsthatcanbeaddedto(8.3.2) havethegeneralformd4 [ eV+ V ] (thelasttermonlyif V isa U (1)gauge“eld), andtheonlycomponentof V thatcanhaveanonzeroexpectationvalueistheauxiliary “eldD;thislea dstoadditional termsintheclassicalpotentialoftheform Š1 2 (D)2Š DŠ AA D.Extre mizingwithrespecttoDandtheneliminatingitgives anadditionalcontribution1 2 | + AA |2totheclassicalpotenti al,whichmustbeseparatelyzeroforsupersymmetrynottobebroken.Theexpressionfortheclassicalpotentialcanbereadfrom(4.3.7).Wenotethatif =0,ifitcanbe arrangedfor + AA to equalzero,thensome(charged)scalar“eldmustacquirea(nonvanishing)vacuum expectationvalue,andthegaugegroupwillbespontaneouslybroken.Thus,tohave both gaugeinvarianceandsupersymmetry,itisnecessarythat =0. We observethatifsomeexpectationvalueofanauxiliary“eldisnonzero,i.e., f = < F > or d = < D> ,thesupers ymmetrytransformation ofthespi nor“eldofthe mult ipletbecomes(see(3.6.5,6)) = f + ... or

PAGE 529

8.3.Spontaneousbreakingofglobalsupersymmetry509= id + ... ,(8. 3.9) whichistypicalbehaviorforaspontaneouslybrokensymmetry.Thespinor“eld describestheGoldstino,and f or d sets thescaleofsupers ymmetrybreaking. a.2.Lo opcorrections Wenowestab lishthefollowingresult:Infourdimensions,ifsupersymmetryisnot spontaneouslybrokenattheclassicallevel,itisnotbrokenbyradiativecorrections. Thistheoremcanbeprovenmostreadilybyusingresultsofsuper“eldperturbationtheory,anditmightbeviolatedbynonperturbativeeects,althoughnoexampleisknown infour dimensions.We“rstconsiderthesituationwithonlychiralsuper“elds. Abasicfe atureofperturbationtheoryisthatt heeectiveactionisobtainedwith a d4 integral.Ifweconsiderclassicalconstant“eldsoftheform= A Š 2F Dactingonthemissimply ,sothatt hederivativesdonotintroduceany factors;consequently,inthe d4 integration,wemustget and factorsfromthe s,andtheseare accompaniedbyan F andan F factor.Therefore,addingth ecla ssicalpoten tialtothe quantumcorrections,wehaveatotalpotentialoftheform Ueff= Ši [ FiFi+ FiPi( A )+ Fi Pi( A )]+ij FiFjGi j( A A F F )(8. 3.10) Dierentiatingwithrespectto Aiand Fiweobtain Š U Ai = FjPij+ FjFk Ai Gk j(8.3.11a) Š U Fi = Fi+ Pi+ FjGj i+ FjFk Fi Gk j(8.3.11b) Now,ifattheclassi callevelthereexistvaluesofthe A ssuchthatPi=0,sothat Fi=0sati sfytheextremumequationsandmaketheclassical U vanish,iti sclear from theaboveformthatthisresultisnotcha ngedbythequantumcorrectionssincethe additionaltermsalsovanishesfor Fi=0.Theimport antingre dientisthatthe quantum correctionsarebilinear intheauxiliary“elds. Thereforetheclassicalminimumoftheclassicalpotentialisstillanextremumof thequantumcorrectedpotential,anditisstillsuchthatthetotalpotentialvanishes

PAGE 530

5108.BREAKDOWNthere.Furthermore,ifthesupersymmetrya lgebrastillholds,itmustbeanabsolute minimum(nonegativeenergy)andthereforesupersymmetrycannotbebroken.Conversely,ifsupe rsymmetryisbrokenattheclassicallevel(somePi =0forany A s),the aboveequationhasno Fi=0solution s,andhenceradiativecorrectionscannotrestore thesymmetry. Were markthatinlowerdimensionsthesituationisslightlydierent.Therethe superspaceintegrationsare d2 ,wh ilesuper“eldsstillhavetheform A Š 2F .Therefore,thequantumcorrectionstotheeectivepotentialcanhavetermsoftheform FG ( A F )witha single F .Whentakingd erivativeswithrespectto F ,thefactorin frontc andisappear,andwe“ndthattheclassicalextremumnolongerneedbean extremumofthequantumpotential. Inthepresenceofgaugesuper“eldswecanhaveadditionalcontributionstothe eectivepotential.TermsproportionaltoD 2orDF arequadraticinauxiliary“elds anddonotchangeourconclusions:If F =D=0ares olutionsoftheclassicalequations,theywillalsobesolutionsofthequantumcorrectedequations.However,itispossibletogeneratetermsoftheformDf ( A A ),andsuchterms,nolongerquadraticin theauxiliary“elds,couldchangeourconclusions(recallthatapureDtermisnotgenerated).Nevertheless,aslongasgaugeinvarianceandsupersymmetryareunbrokenat thetreelevel(whichimpliest hatthetheorydoesnothaveaFayet-Iliopoulosterm),even aterm linearinDisharmless.Thisisbecausesuchatermarisesonlyfromthecovariantizationoftermsintheeectiveactionoftheformd4xd4 g ( ,) d4xd4 g ( eV,) d4x D A g ( A A ) A ;t hu st hi st er mi sa tl ea st b ilinearintheD, A “elds,andhencewecanusethesameargumen tsasabovetoconcludethattheclassical solutionD= A =0(whichmustbe thecaseifgaugeinvarianceandsupersymmetryare unbrokenclassically)isstillasolutionatthequantumlevel.(Alinear A termwould spoilthisargument,butsuchatermcannotbewrittenasa d4 integral.) Ifclassicalgaugeinvarianceisbroken,andaDf ( A A )isgen erated,ithasbeen shownthatforaspeci“cclassofmodelsasupersymmetricsolution( F =D=0)ofthe quantumcorrect edequationsexistswiththe A sshiftedfromtheircl assicalvalues;thus, eveninthiscase,supersymmetryisnotbrokenbyradiativecorrections.However,the generalsituationisinneedoffurtherclari “cation.Alloftheseresultsholdforthe

PAGE 531

8.3.Spontaneousbreakingofglobalsupersymmetry511nonrenormalizab lesy stemsthatwediscussbelow. b.Nonrenormalizabletheories Wenowconsi dermoregeneralsitu ations.Forglobalmodelscoupledtosupergravity(seesec.5.5.h)therenormalizabilitycriterionistoorestrictive:Thecombined systemsarenotpower-countingrenormalizable,andsincewecanmake“elddependent Weylresca lings,thereisnoreasontoinsistonpolynomialityofthematteractions. However,no nderivativesuper“elddependentrescalingsdonotchangethenumberof derivativesintheactionandthereforewere strictourselvestoactionsthatleadtocomponentL agrangianswithnomorethantwospacetimederivativesinthepurelybosonic termsoftheaction,andnomorethanonespa cetimederivativeinthetermscontaining fermions;thisispreservedbysuper“elddependentrescalingsthatdonotinvolvespinor derivatives.Inthissubsectio nwediscuss interactingchiralscalarsuper“elds;weextend thediscussiontogaugesystemsinthefollowi ngsubsection.Thereadershouldreview ourdiscussionofK¨ ahlermanifoldsinsec.4.1.b. Weconsid erasystemof N chiralsuper“eldsidescribedbythesuperspaceaction S = d4xd4 IK (i, j)+ d4xd2 P(i)+ h c ., =( x , ), D€=0, =(), D =0.(8 .3.12) Asdiscussedinsec.4.1.b,the“rsttermoftheaction S canbegivenageometricalinterpretation:i, jcanbethoughtofascoordinatesofacomplexmanifoldwithK¨ ahler potential IK Wer ecallthatthe(complex)component“eldsofiarede“nedbyprojection Ai=i| i= Di| Fi= D2i| .(8. 3.13) Wede notevacuumexpectationvaluesofthecomponent“eldsby ai= < Ai> fi= < Fi> =0.(8. 3.14) Thevacuumexpectationvaluesareobtainedbysolvingtheclassical“eldequationsfor x -independent“elds.

PAGE 532

5128.BREAKDOWNTheaction S leadstothesuper“eldequations D2IKi+Pi=0(8.3 .15) andtheirhermitianconjugates(weusethenotationof(4.1.13,25a)).Takingvacuum expectationvaluesandevaluatingat =0usingt hede“nitions(8.3. 13,14),weobtainin particular IKi j( a ) fj+Pi( a )=0.(8 .3.16) Asintherenormalizablecase(sec.8.3.a),spo ntaneoussupe rsymmetrybreakingoccursif fi=0is not asolutiontothese equations.Furthercomponentequationsareobtained bydierentiati ng(8.3 .15)with D2andevaluatingat =0.W e “nd [ IKij k( a ) fk+Pij( a )] fj=0.(8. 3.17) Af te r “ndingthevacuumsolution(s),wecanchoosetoworkinnormalgauge(4.1.27)at thevacuumpoint.Inthatcasethevacuumequations(8.3.16,17)reduceto fi+Pi=0,Pijfj=0.(8. 3.18) IfPij( a )isnonsing ular all fj=0andsupersy mmetryisnotbroken.Conversely,if fj=0isnotaso lutionof(8.3.16,17)thensupersymmetryisbrokenandPij( a )mustbe singular. Returningtotheac tion(8.3.12),weshiftthe“eldsi i+ < i> andinvestigate”uctuationsaboutthevacuumstate. Inparticularwecanreadothemassesof thevariousparticlesfromtheresultingaction;alternatively,wecan“ndthemassmatricesofthecomponent“eldsbyexpandingthesuper“eldequations(8.3.15)tolinearized orderint he”uctuations: D2[ < IKi j> j+ < IKij> j]+ < Pij> j=0.(8. 3.19) A pplying Dand D2andevaluatingat =0asin(4 .1.21),we“nd Fi+PijAj=0 i € i€+Pijj =0 Ai+ IKik jl flfk Aj+PijkfkAj+PijFj=0(8.3 .20)

PAGE 533

8.3.Spontaneousbreakingofglobalsupersymmetry513wherew ehavedroppedthe <> onP...and IK... ...andassumedthatweareinnormal gauge.Eliminatingtheauxiliary“eldsweidentifythefermionandbosonmassmatrices: MF=Pij( a ) MB 2= Š IKik jl flfkŠ Pik Pkj Pijk fkPijkfkIKjl ik fkflŠ PikPkj .(8. 3.21) Again,asabove,ifsupersymmetryisbrokenPijhasatleastonezeroeigenvalueandone ofthecorrespondi ngmasslessfermionsistheGoldstino. Weevaluate thegradedtraceofthemassmatrixsquared.Thissupertracegives themassrelation strM2=J ( Š 1)2 J(2 J +1) MJ 2= trMB 2Š 2 trMFMF *= Š 2 IKik il flfk(8.3.22) Sinceweareinnormalgaugewecanrewritethisas strM2= Š 2 Rk l flfk(8.3.23) where Rk l= Ri l k iistheRiccitensorofthemanifoldevaluatedati= < i> (see (4.1.28)).Theresultismanifestlycovariant,andthus(8.3.23)holdsinanarbitrary gauge.Inparticular,formodelswithconventionalactions iitheK¨ ahlermanifoldis ”at andweobtainthesimplemassformulaJ ( Š 1)2 J(2 J +1) MJ 2=0.(8. 3.24) Wealso observethatincontrasttorenormalizab lemodels,spontaneoussupersymmetry breakingcanoccurinamodelwithasing lechiralmultiplet,forexamplewith IK = cos (+ ),P()=,where Š < A < c.Globalgaugesystems Inthissection,werepeatthepreviousanalysisbutincludegaugesuper“elds V = VATA.B ecausegaugesymmetriesareusua llydescribedbyexplicitmatrix

PAGE 534

5148.BREAKDOWNrepresentations( TA)i jofthegenerators,webeginwiththisformulation;wethenchange overt oamoregen eralformulationwherewedescribetheactionofthegeneratorsby K illingvectors.Asdiscussedinsec.4.1.b,thisallowsustochoosenormal coor dinates in whichthecomputationofthemassmatricessimpli“es(however,wecannotusenormal gauge ).Werestrictourselvestomodelswherethegaugedgroupisunbrokenor isotropic atoneormorepointsofthemanifoldofscalar“elds.(Theusualmatrixrepresentation assumesisotropyattheorigin,i.e.,theoriginiskept“xedbygaugetransformations.)A fo rmulationintermsofKillingvectorsshouldallowonetogaugegroupsthatarerealizednonlinearlyateverypointonthemanifoldofscalar“elds,i.e., hasac onstant termeverywhere(theconstan ttermca nnotbeelim inatedbyshiftingthescalar“elds); however,thesuper“elddescriptionofthemoregeneralcasehasnotbeenworkedout. Weconsid ertheaction S = d4xd4 [ IK (i,j)+ trV ] + d4xd2 [P(i)+1 4 QAB(i) WAWB]+ h c .(8. 3.25) withcovariantlychiral: j= k( eV)k j, WA= i D2( eŠ VDeV)A.(8. 3.26) Thechiralquantities QAB= AB+ O ()can generatemassesforthegaugefermionscontainedin V .Wehaveincl udedthegl obalFayet-Iliopoulosterm trV (4.3.3). WechoseaK¨ ahlergauge(see(4.1.26))where IK itselfisinvariant;wecanalways dothis ifthegaugegroupisunbrokensomewhereonthemanifoldasdiscussedabove. Thengaugeinvarianceof S requires IKj( TA)j iiŠ j( TA)j iIKi=0, Pj( TA)j ii=0, QDE, j( TC)j ii+( TC)D AQAE+( TC)E AQAD=0.(8. 3.27) Thematrices( TC)E Aformtheadjointrepres entationofthegenerators,andare,uptoan overallf actor,thestructureconstants;thustheyareindependentofanyspecialchoiceof

PAGE 535

8.3.Spontaneousbreakingofglobalsupersymmetry515coordinatesonthescalarmanifold. Webeginbyderiv ingtheeq uationsforthevacuumexpectationvalues.Wede“ne (Yang-Mills)covariantcomponent“eldsbycovariantprojection(see(4.3.4,5)) Ai=i| i= i| Fi= 2i| ,( 8.3.28a) = W| f=1 2 ( W )| ,(8. 3.28b) i € €=1 2 [ {, W} ] | ,D= Ši 2 {, W}| ,(8. 3.28c) where fBisthecomponentgauge“eldstrength(see(4.2.85)).Wealsoneedtheidentity( dA= < DA> ) < 2 2i> | = dA aj( TA)j i.(8. 3.29) Thesuper“eldequationsthatfollowfrom(8.3.25)are: 2IKi+Pi+1 4 QAB, iWAWB=0 j( TA)j iIKiŠi 2 ( QABWB)+1 2 i €( QAB W€B)+ trTA=0.(8. 3.30) Theequationsforthevacuumexpectationvaluesareobtainedbyevaluatingat =0 theaboveequations,andtheequationobtainedbydierentiatingthe“rstonewith 2. We “nd IKi j fj+Pi=0 IKij k fkfj+ ak( TA)k jdAIKi j+Pijfj+1 2 QAB, idAdB=0 aj( TA)j iIKi+( QAB+ QAB) dB+ trTA=0(8.3 .31) where IK QAB,Pareevalu atedwithi ai. Wenowgen eralizetoarbitrarycoordinatesbyrewritingtheaboveintermsof ho lomorphicKillingvectors kAi.W er ep lacethespeci“cformoftheYang-Millsgauge transformation(4.1.35) i= i A( TA)i jj, i= Š i jA( TA)j i(8.3.32)

PAGE 536

5168.BREAKDOWNwiththemoregeneralform(4.1.31): i=AkAi, i=AkAi(8.3.33) whereiisde “nedbyanalogywi th(4.1.34b): i exp ( iVAkAj j ) i.(8. 3.34) Theconditions(8.3.27)thatensuregaugeinvarianceoftheactionbecome IKikAi+ IKikAi=0 PikAi=0 QDE, ikCi+ i ( TC)D AQAE+ i ( TC)E AQAD=0.(8. 3.35) Asdiscussedabove,aformulationintermsofKillingvectorsenablesustousenormalcoordinatesandthustosimplifyourcomputations.Thus,forexample,wecancomputethemassmatricesofthevariouscomponent“eldsand“ndasupertracerelation thatgeneralizes(8.3.22).We“ndthelinearized“eldequationsforthecomponent“elds byexpa ndingthecovariantizedformofthesuper “eldequation s(8. 3.30)aroundthevacuumandapplyingtheoperators1, 2tothe“rstand1, ,[ ]tothes econdof theequationsandevaluatingat =0.Theresult,inno rmalcoordinates,is Fi+PijAj=0 i €€ i+ AkAi+Pij jŠi 2 QAB, idAA=0, Ai+[ idAkAi j+ IKik ljfk flŠ Pik Pkj] Aj+[Pijkfk+1 2 QAB, ijdAdB] Aj+[ QAB, idB+ ikAi]DA=0, ( QAB+ QAB)DB+( QAB, idB+ ikAi) Ai+( QAB, idBŠ ikAi) Ai=0,i 2 ( QAB+ QAB) €B€+( kAiŠi 2 QAB, idB) i+1 2 QAB, ifiB=0,

PAGE 537

8.3.Spontaneousbreakingofglobalsupersymmetry517( QAB+ QAB) €fBŠ ( kAikBi+ kBikAi) AB€=0.(8. 3.36) wherewehav edropped <> onP..., IK... ..., QAB....Inour normalizat ionthev ectorwave equationis €fŠ m2 VA€=0.(8. 3.37) Eliminatingtheauxiliary“eldswe“ndthemassmatricesfromwhichweobtainthe supertrace(innormalcoordinates) strM2= Š 2[ idAkAi i+ IKik lifk fl+ tr ( Qi1 Q + Q Qj1 Q + Q ) fi fj+ i ( Q + Q )Š 1AB( kAi QBC, idCŠ kAiQBC, idC)].(8.3.38) Acovariantform ula,va lidinanycoordinatesystem,isobtainedbyreplacing kAi iwith kAi ; iand IKik liwith Rk lstrM2= Š 2[ idAkAi ; i+ Rk lfk fl+ tr ( Qi1 Q + Q Qj1 Q + Q ) fi fjŠ itr ( Qi1 Q + Q ) kAidA](8. 3.39) wherewehaverewrittenthelasttermusingthegaugeinvariancerelations(8.3.35).In thecoordinatesystemwheretheYang-Millsgaugetransformationsaregivenby(8.3.32) wehave kAi ; i= Š i ( TA)i iŠ i ( TA)j i aji(cf.(4.1.29b,31,32d)).

PAGE 538

5188.BREAKDOWN8.4Traceformulaefromsuperspace Inthelasttwosections,wefoundthesupertraceusingessentiallyacomponent approach,andnottakingadvantageofthesuper“eldformalism.Thereisamucheasier waytoevaluatethesu pert raceexpressionwithoutevercomputingcomponentmass matrices:Iftheactionisexpandedincomponentsandtheone-loopeectivepotentialis evaluated,itsquadraticallydivergentpartisproportionaltothesupertrace strM2; moreover,wecaneasilyreadothisquadra ticallydiverg enttermfromtheclassical super“eldactionifweimagineperformingasuper“eldone-loopcalculation. a.Explicitbreaking Wecand evelopthemethod(andderivesomenewmassformulae)by“rstconsideringthecaseofexplicitsoftbreakingof supersymmetry.Forexample,weconsidera masslessscalarmultipletandaddtoittheexplicitsoftbreakingterm(8.2.3) Sbreak=d4xd4 U .Wenowcalculatet hequadraticallydivergentpartoftheoneloope ectivepotential;thecoecientisthecontributionoftheterm(8.2.3)tothe supertrace.Recallthatsoftbreakingtermsarede“nedbythepropertythattheygiveat mostlogarithmicallydivergentcontributionstotheeectiveaction,andyethereweare calculatingquadraticdivergences;however,insec.8.2weignoredvacuumdiagrams (which haveonlyspurion“eldsexternally),whereasherethatisallweareinterestedin. Inthecalculation,wehavetoconsiderthesumofone-loopdiagramswith n masslesschiralpropagatorsand nU -spurionvertices( U = 22 2;altho ughthecalculation simp li“esifweusetheexplicitformof U ,wew illkeep U general,si ncethentheresults canbeappliedtoothercases).Ateachvertexwehavefactors D2, D2actingonthe pr opagators;however,eachpropagatorisproportionalto pŠ 2,andth us,togeta quadraticdivergence,wemustcancelallbutonepropagatorwithanumeratorfactor. Thisrequires n Š 1facto rsof D2 D2Š p2;there mainingfactorisneededforthe loop (seesec.6.3,e.g.,(6.3.28)).Hencewe“nd =n d4 d4p (2 )4p2 Š 1 n ( Š U )n= d4 ln (1+ U ) d4p (2 )4p2 .(8. 4.1) Thereforethesupertraceis

PAGE 539

8.4Traceformulaefromsuperspace519strM2= Š 2 d4 ln (1+ U )= Š 2[ D2 D2ln (1+ U )] | = Š 2[ D2 D2U ] | = Š 2 2.(8. 4.2) Comparingtothecomponentexpressionin(8.2.3),weseethatthisisindeedthecorrect result:themassofthescalar A hasbeenloweredby 2(thefactorof2arisesbecause A iscomplex).Itisclearthatnodiagramcontai ningchiralself-int eractionscanchange theresult:Wen eededafactorof D2 D2ateachvertex,andachiralvertexcomeswith onlyafactor D2. Suc ha di agramcanbeonlylogarithmicallydivergent.Also,since supersymmetricmasstermscanbetreatedas interactions,ourresultsholdinthemassivecase.Thissameargumen talsoimp liesthatexplicitbreakingtermsofthetypes consideredin(8 .2.5,6,9)cannotcontributetothesupertrace. Nextweconsiderexplicitbreakingterms(8.2.7)foranabelianvector“eld: Sbreak=1 2 d4xd4 WW+ h c .=1 2 d4xd4 ( + ) VD D2DV .Thecalc ulation isalmostidenticaltotheabove:Eachpropagatorisstill pŠ 2,ex ceptthatavector pr opagatorhasanextra Š 1r elativetoachiralpropagator,andeach + vertex ( = 2)comeswith afactor D D2D,whichacts precisely inthesamewayasafactor D2 D2,ex ceptfora Š 1thatcan celstheextra Š 1f romthepropagator.Thuswe“nd strM2= Š 2 d4 [ Š ln (1+ + )]=2[ D2 D2ln (1+ + )] | = Š 2[ D2 D2 ] | = Š 2 2.(8. 4.3) Asbefore,thisagreeswiththecomponentexpression(8.2.7)(thefactor2comesfrom thetwohelicitycomponent softhefe rmion).Wecancombinetheexplicitbreaking terms(8.2.3-7)withthepreviousonesand“ndsimplythesumof(8.4.2,3). Finally,weconsider(8.2.10);sincethishasonlyafactor D D2insidetheloopat eachvertex,itcannotcontributetothequadraticdivergenceorthesupertrace.

PAGE 540

5208.BREAKDOWNb.Spontaneousbreaking Intheexamplesoftheprecedingsubsection,weusedratherelaboratemethodsto deriveresultsthatcanbefoundmoreeasilybyexplicitcomputationofthemasses;here wew illapplyt hesemethodstoderiveresultsthatrequiredthesomewhatlengthycalculationsofsec.8.3.We“rstconsidertheaction(8.3.12).Weexpand S tos econdorder inquantum“eldsi,withtheco ecientsevaluatedatthebackgroundclassicalvalues < i> : S(2)= d4xd4 jIKi ji+ d4xd2 Xijij+ h c .(8. 4.4) where Xij= Pij+ D2IKij(8.4.5) Incompleteanalogywith(8.4.1),thequadraticallydivergenttermintheone-loopeectiveactionis = d4p (2 )4p2 d4 tr [ ln ( IKi j)](8.4.6) where IKi jŠ i jplaystheroleof U and,asabove,thec hiralvertex(here Xij)doesnot contribute.Thesupertraceistherefore strM2= Š 2 d4 tr [ ln ( IKi j)]= Š 2 { D2 D2[ trln ( IKi j)] }| = Š 2[ trln ( IKi j)]k l flfk(8.4.7a) andhence,using(4.1.30b), strM2= Š 2 Rk l flfk,(8. 4.7b) inagreementwith(8.3.23). Fo rt he casewithgaugeinteractions(sec.8.3. c),weagainobtainthesupertraceby examiningthequadraticdivergenceintheone-loopeectiveaction.Tosecondorderin quantum“el dswehave S(2)= d4xd4 [ IKi j k( eV)k ji+1 4 ( QAB+ QAB) VAD D2DVB],(8.4 .8)

PAGE 541

8.4Traceformulaefromsuperspace521wherewehavedroppedtermsthatdonotcontributetothequadraticdivergence(that is,chiralinteractionsortermscorrespondingto(8.2.10)).Now( eV)k jIKi jŠ i kplays theroleof U above,and1 2 ( QABŠ AB)playstheroleof .The “nalresultis strM2= Š 2 d4 { tr [( VATA)i j+ ln ( IKi j)] Š trln (1 2 [ QAB+ QAB]) } = Š 2[ dA( TA)i i+ dA( TA)i j aji+ Rk l flfk+ tr ( Qk1 Q + Q Ql1 Q + Q ) flfkŠ tr ( Ql1 Q + Q )( TA)l i aidA](8. 4.9) wherewehavereplacedd4 2 2andused[ €, { €, } ]= Š 2 iW,and (8.3.29).Wethusrecovertheresult(8.3.39).(Wehavechosentoworkinthecoordina tesystemde“nedby(8.3.32)simplybecauseitismorefamiliar;thecomputationis eq ua llystraightforwardintermsofKillingvectors.)

PAGE 542

5228.BREAKDOWN8.5.Nonlinearrealizations Experiencewithspontaneouslybrokeni nternalsy mmetrieshasshownthatmuch usefulinsightcanbegainedbystudyingthegeneraltheoryofnonlinearrealizations. Themethodsthathavebeendevelopedcanbeappliedquitesuccessfullytosupersymmetry. Onewaytoformulateanonlinearrealizationofsupersymmetryistoconsider(nonlinearly) constrainedsuper“elds;however,itisfarfromobvioushowtochoosesuchconstraints,andsowewillreturntothisapproachafterwehavestudiednonlinearrealizationsdirectly. ThesimplestnonlinearrealizationistheVolkov-Akulovmodel.Itisfoundbyconsideringacovariantlytransformingsetofhypersurfacesinsuperspace.Let ( x )= (8.5.1) de“neahypersurfa ce;ittransformsas ( x)= (8.5.2) wherewerecallthat x= x Ši 2 ( + ), = + .Thisi mplies ( x)= + = ( x )+ =0(8.5 .3) andhence ( x Ši 2 [ ( x )+ ( x )])= ( x )+ (8.5.4) or ( x )= +i 2 ( €+ €) €.(8. 5.5) Thisgivesanonlinearrealizationofthealgebracarriedbythespinor“eld ( x );itisby nomeansunique,butothernonlinearrealizationsarerelatedtoitby“eldrede“nitions. Notethat (oranyotherequivalentnonlinearrealization)containsaconstanttermin itstransformationlaw,andisthereforeasui table“eldfor describingtheGoldstino. To “ndaninvariantaction,wereca llthattheone-form(3.3.31) s€( x , )= dx€+i 2 ( d €+ €d )(8. 5.6)

PAGE 543

8.5.Nonlinearrealizations523is in va riantundersupersymmetrytransformations.Ifweconstrainthisone-formtolie onthehypersurface ( x )= ,we “nd s€( x )= dx€+1 2 ( i m €) dx m dx mv m a,(8. 5.7) wherew ehavede“nedaninversevierbein v m a.Sincetheone-for misinvar iant,the vierbeinmusttransformcovariantly,i.e.,supersymmetrytransformationsof must i nducecoordinatetransformationsof v a m.Nowit iseasytowritedownaninvariant actionintermsofthedeterminant v = det ( v a m): S= d4xd4 vŠ 14( Š ( x ))= d4xvŠ 1.(8. 5.8) Wenote that cvŠ 14( Š ( x ))isascalarsuper“eldwhosecomponentsare functionsof ( c isanarbitrarydimensionalconstantthatsetsthescaleofsupersymmetrybreaking;seebelow).Wecanals oconstructachi ralsuper“eld= D2outof Since[ 4( )]2=0thesesup er“eldssat isfythenonlinearconstraints 2=2=0(8.5 .9a) = cŠ 1 D2 = cŠ m( D2 )m(8.5.9b) = cŠ 1D2 D2=1 2 cŠ 1 D D2D(8. 5.9c) etc.Thesolutiontotheconstraints(8.5. 9a,b)or(8.5.9a,c)ispreciselyorrespectively. Theexpectationvalues < > < > of,,thatfollowfrom <> =0are typicaloft heexpectationvalueofamultipletwithspontaneouslybrokensupersymmetry(asinsec.8.3):theauxiliarycomponents( 2or 2 2forandrespectively)get nonvanishingexpectationvalues c ,andallotherco mponentscanbetakentohavevanishingexpectationvalues.Thefermioncomponentsatone Š levellowerthantheauxiliary“elds,e.g., = D | or = D2D | ,havethisconst anttermintheirtransformationsas = c + ... ,con“ rmingouridenti“cationof c asthesupersymmetrybreakingscale. Thevierbein v canbeusedtowritedownotherinvariantactions; any expression covariantizedwith v issupersymmetric.For example,wecancouple toascala r“eld

PAGE 544

5248.BREAKDOWNA asfollows: SA= Š1 2 d4xvŠ 1( v a m mA )2(8.5.10a) where A transformsas A ( x )= A( x)= A( x Ši 2 ( + )).(8.5.10b) Wenowdiscussthepre scriptionfordescribingspontaneouslybrokentheoriesin termsof .Inaspontane ouslybrokentheory,theGoldstinoisone,oringenerala uniquelinearcombination,ofthefermioniccomponentsoftheordinarysuper“elds.We introducestandardvariablesbyreplaci ngtheGoldstinowiththeVolkov-Akulov“eld ,andthes uper“eldsbynewsuper“eldswhosecomponentstransformhomogeneouslyas in(8.5.10b),andinparticular,withnomixingofdierent -components. Webeginby constructingahomogeneouslytransformingsuper“eld outof andanordinarysuper“eld.Consider ( x , ) ( x +i 2 ( + ), Š Š )(8. 5.11) where( x , )is any super“eld.Undersupersy mmetrytransformations( x, ) ( x Ši 2 ( + ), + + )=( x , ),we“ndforthetransformationof : (x Ši 2 ( + ), )=(x Ši 2 ( + Š Š ), Š Š )=(x Ši 2 ( + Š Š ) Ši 2 [ ( Š )+ ( Š )], Š + Š + ); (8.5.12a) using(8. 5.4),wehave (x Ši 2 ( + ), )=(x +i 2 ( + ), Š Š )= ( x , ).(8.5 .12b)

PAGE 545

8.5.Nonlinearrealizations525Thusweseethatundersupersymmetrytransformations transfo rmshomogeneously (dier ent -componentsdonotmix)butnonlinearlywithrespectto ;the x -coordinate undergoesatranslation.Therefore,thewholesupersymmetrygroupisrealizedon by elementsofthePoincar egro up.Thisisageneralfeatureofn onlinearrealizations:Given agroup G (herethesupersymmetrygroup) andalinearlyrealizedsubgroup H (herethe Poincar egro up),thenonlinearrealizationsonsuitablyde“ned“eldsisperformedbyelementsof H .Consequently, wecanimpose anytranslationallyinvariantconstraint on withoutbreakingsupersymmetry.Forexample,ifweconstrainitentirelybysettingit equalto c 2 2(or c 2inthechiralcase),weexpressallthecomponentsofintermsof andrecoverthepreviousresult(8.5.9). Wenowconsi deramodelwithspontaneousbreakdownofsupersymmetry.Wecan describethemodelintermsofstandardcompon ents,i.e.,componentstransformingasin (8.5.4,10b),asfollows: (1)Foreachsuper“ eldsweconstructtheassociated (2)Weidentifythefermioniccomponent oftheappropriatelinearcombinationof sthatistheGoldstino,andisthereforeasuitablecandidatetobereplacedby andconstrainthecorrespondingcomponent inthecorrespondi nglinearcombinationof stozero.Th isgivesusthecombinationofthecomponentsofthes thattransformsas(8.5.4). (3)Weexpresstheremainingcomponentsofthesintermsoftheremainingcomponentsofthe s. Thisprocedureistheanalogofgoingtoradialandanglevariablesfornonlinearsigma models:theremainingcomponentsofthe scorrespondtotheradialvariables, whereas correspondstotheanglevariable. Asanexample,we considertheORaiferteaighmodelofsec.8.3.a,withthreechiralsuper“eldsi, i =0 ,1 ,2 .U si ng (8.3.18),we“ndthattheauxiliary“eld F0ofthe mult iplet0getsanonvanishingexpectationvalue F0= c .Todes cribethesystemin termsofstandardcomponents,we“rstde“ne homogeneouslytransformingsuper“elds ibyintro ducingtheVolkov -Akulov“eld asanextravariable;thenwerestorethe numberofde greesoffreedombyconstraining 0=0 (thefermioniccomponentof 0) andeliminating 0(thefermioniccomponentof0)infav orof .Wecandothis

PAGE 546

5268.BREAKDOWNb ecause,examining 0,we “nd 0= 0Š c + ... ,where c = < F0> .Ifspontaneous symmetrybreakingd idnotoccur(i.e.,if c =0),wecoulds tillde “nehomogeneouscomponents, butwecouldnotremovetheextradegreeoffreedom(thechangeofvariables from 0to wouldbesingu lar).Havingeliminated 0infavorof (thestandardangle va riable,whichtransformsas(8.5.4)),wecanproceedtoexpresstheremainingcomponentsofiintermsof andthecomponentsofthehomogeneoussuper“elds i(the standardradialvariables,wh ichtransformas(8.5.10b)).

PAGE 547

8.6.SuperHiggsmechanism5278.6.SuperHiggsmechanism Whensupersymmetrybreaksinasy stemcoupledtosupergravity,a superHiggs mechanismeliminatestheGoldstinoandgivesmasstothegravitino(theGoldstino b ecomesitslongitudinalcomponent).Toexam inethesuperHiggsmechanismindetail, west udythelocallysupersymmetr icanalogoftheconstrainedsuper“eldsoftheprevioussection(8.5.9). ThebasicingredientofthesuperHiggsmechanismisthetransformationlawofthe Goldst ino, = c + ... (seepr evioussection);whentheGoldstinoiscoupledto supergravity,thesupersymmetryparameter b ecomeslocal: = c ( x )+ ... (8.6.1) Consequently,theGoldstinocanbecompletelygaugedaway;sincethenumberof dynamicalmodesofthetheoryshouldnotchange,weexpecttheGoldstinotore-emerge somehow,anditdoessobygivingthegravitinoamassandbecomingitslongitudinal mode.Toseethisdirectly,wedescribetheGoldstinobyalocalconstrainedchiral“eld obeyingthelocalsuperspaceversionof(8.5.9a-b).Thuswetake2=0, = cŠ 1( 2+ R ) (hereweco nsider onlyminimal( n = Š1 3 )supergrav ity).These constraintshaveaconsistentsolutionintermsofasinglefermicomponent“eld(the Goldstino).Becauseoftheconstraintson,a nyloca llysupersymmetricaction(without explicitderivatives,see(5.5.15))reducesto Shi ggs= Š 2 d4xd23( + )+ h c .(8. 6.2) Theconstrainedsuper“eldisanonlinearfunctionoftheGoldstino;however,whenwe gaugetheGoldstinoaway(goto U-gauge) itsimpli“estobecome= Š 2c (alternativelyandequivalently,wecaneliminatetheGoldstinobyarede“nitionofthegravitino. Insuperspace,thiscorrespondstorescaling by(1+ )Š1 3 ;howev er,itissimplerto chooseUgauge).Theaction(8.6.2)becomes,using(5.6.60,64) Shi ggs= Š 2 d4x eŠ 1[ (3 S +1 2 (€ |€ |€ )€)+ c + h c .](8.6 .3) where S isthecomplexscalarauxiliary“eldofthesupergravitymultiplet;wealsohave the Š 3 Š 2| S |2termfromthesupergravityaction(5.6.63).Eliminating S byitsequ ation

PAGE 548

5288.BREAKDOWNofmotion,we“ndacosmologicalconstantandgravitinomassterms: Shi ggs= Š 2 d4x eŠ 1[ 1 2 (€ |€ |€ )€+ h c .+3 | |2+ ( c + c )](8.6 .4) Asdiscussedinsec.5.7,grav itinomasstermswhena ccompaniedbyacosmological constantdonotingeneralmeanthatthegravitinoismassive.However,if ( c + c )= Š 3 | |2(8.6.5) thenthecosmologicaltermcancels,andwecanunambiguouslyidentifythe termsas massterms. Sinceanyspontaneouslybrokentheorycanbedescribedintermsofstandardvariables,andinparticular,theGoldstinocanbedescribedintermsof,inanyspontaneouslybrokentheoryinwhich thecosmologicalconstantvanishesthegravitinomassis Re whenthe(superspace)kinetictermhastheusualnormalization Š 3 Š 2.Itisf o und byse ttingallmatter“eldstotheirvacuumexp ectationvalues.Moregenerally,when (8.6.5)isnotsatis“ed,wecanstill“ndtheapparent mass Re andthecosmological constantfromthetransforma tionoftheGoldstinoandbycomparingtothesuperspace action(8.6.2).

PAGE 549

8.7.Supergravityandsymmetrybreaking5298.7.Supergravityandsymmetrybreaking Supersymmetrybreakinginalocalcontextcanbestudieddirectly,usingthe componenttoolsofsection5.6.Wecandete rmineconditionsforsu persy mmetrybreakingandderiveamassformulaanalogousto(8.3.35).However,itismuchmoreecient torecastthep roblemasa global supersymmetryproblemthatcanbestudiedusingthe tec hniquesofsecs.8.3b-cand8.4.Weconsiderageneralsystemofinteractingscalar andvectormultipletscoupledto N =1superg ravity.Themattermultipletsare describedbychiraland(rea l)gaugescalarsuper“eldsi, VA,resp ectively;thesupergravitymulti pletisdescribedbytherealaxial-vectorsuper“eld H mand(for n = Š1 3 ) thechiralcompensator .Howev er, H mplaysnodirect roleinthesupersymmetry breakingmechanismorinthederivationofmassformulae.Thereforealltherelevant informationcanbeext ractedfromaglobal nonrenormalizable systemdescribedby ,i, and VA. Webeginbyre ducingthecoupledmatter-supergravitysystem.Theaxial-vector realgaugesuper“eldofsupergravity H mcontainsthegra vitonandgravitinophysical de greesoffreedom,aswellastheaxialvectorauxiliary“eld A m(5.2.8).Inthepresence ofthecompensator thesupergravitygaugegroupconsistsofthefullsuperconformal group,andwehaveatourdisposalallofthecomponentgaugetransformationsof (5.2.10).Consequently,wecangototheWe ss-Zuminogaugediscussedafter(5.2.10), andfurther,usetheremainingsuperconformaltransformationstoremovethegraviton trace,thegravitino -t race,andthelongitudinalpartoftheaxialvectorauxiliary“eld. Inthisgauge H mcontainsonlythetracelesscomponentsofthegravitonandthegravitino,andthetransversepartof A m;t he sp inzerocomplexauxiliary“eld S ,the -trace ofthe(left-handed)gravitino( )L= € €,thetraceo fthevierbein,o requivalently, itsdet erminant e = dete a mand Š 1 mA marecontainedin (theselasttwoarethereal andimaginary partsofthe -independentcomponentof ).Si nceonlythesequantities arerelevantforstudyingspontaneoussupersymmetrybreaking(e.g.,thespinzero bosonscang etvacuumexpectationvaluesandthe -tracecanmixwiththematter fermions),wecanignorethe H mdependenttermsintheLagran gianandworkentirely with andthemattersuper“eldsinaglobalse tting.Thissimpli“esthediscussion enormously;however,because Im | replacesthedivergenceof A m,therear esomesubtletiesassociatedwithitscontributiont othema ssesofthematter“elds(seesubsec.

PAGE 550

5308.BREAKDOWN8.7.a.4).Wetreatonly n = Š1 3 supergravity;analogousmethodscanbeusedforother n butsince n = Š1 3 allowsthemostgeneralcoupling,itisthemostinterestingcase. Thesupergravitymultipletitself(through )a ectsthepatternofsymmetry breaking.At“rstsight,thisseemsstrange:IntheusualHiggsmechanism,wedonot expectthepatternofsymmetrybreakingtodependonthecouplingstogauge“elds(at thetreelevel!).However,ananalogoussituationarisesinanonsupersymmetriccontext, whenscalar“eldsarecoupledtogravity.Weconsidertheaction S = d4x g [ Š 3 Š 2r ( g ) Š1 4 g m nGij( A ) mAi nAj+ rV1( A )+ V2( A )].(8.7.1) To “ndthevacuumexpectationvaluesofthesc alar“eldswecannotignorethegravitational“eld.IngeneraltheRicciscalar r willhaveanonzeroexpectationvaluethat aectsthemassesandscalarpotential.Howe ver,wen eednotconsiderthefullEinstein system;itissucienttolookforsolutionsoftheform g m n= 2 m nandtotreatthesystemofsca lar“elds Ai(subjecttothecondition =0at allpoints).(Thisisanalogoustokeeping andignoring H m.)Twopossiblesituationscanarise:If < V2> =0 wehaveanon zerocosmologicalconstant, Š 1Š 1 < V2>1 2 x2,andthevac uumvalues andthemassesof Aiareshiftedfromtheir”atspacevalues.If < V2> =0,thecosmologicalconstantvanishesandaconsistentsolutionis =const ant.Inthiscasethe gravitational“elddoesnotmo dify”atspaceresults.(Thisisnotthecaseinsupergravity:Evenifthecosmologicalconstantvanishes,thesupergravityauxiliary“eldsmodify globalresults.) Returningtothematter-supergravitysy stem,weconsidertheaction(5.5.32) S = d4xd4 EŠ 1( H ) {Š3 2 eŠ1 3 2( trV + G )+[ 1 R ( g +1 4 QABWAcovWB cov)+ h c .] } ,(8. 7.2) where EŠ 1isthesuper determinantofthevielbeinand R isthescalarcurvaturesuper“eld(seee.g.,( 5.2.74-6)).Thesupergravityactionisgivenbythe“rstterminthe expansionoftheexponential.Here G (i,j)isanarbit rarygaugeinvariantfunctionof chiralsuper“ elds,withde“nedby( 8.3.26), g (i)isachiralf unction,and WAcovisthe

PAGE 551

8.7.Supergravityandsymmetrybreaking531(supergravitycovariant)Yang-Mills“eldstrength.Thefunction G hasanaturalinterpretationasaK¨ ahlerpotentialwithgaugetransformations G G +(i)+ ( j) compensatedbyscalingsof : exp [1 3 2()] The exp [ Š1 3 2 trV ]factoristhelocalf ormoftheFayet-Iliopoulosterm(4.3.3). Itisgaugeinvariantbyvirtueofacombinedgaugetransformationof V andsuperscale transformationsof EŠ 1(5.3.8-10).Itspres enceseverelyrestrictstheformofthe g terms; theymustbe R -invariant(see(3.6.14)and(4.1.15))sothatthewholeactionisinvariant underthesuperscaletransformationsof EŠ 1(seebelow).Inthe 0 limittheaction (8.7.2)becomes(8.3.25),withtheidenti“cation G IK g P Asdisc ussedabove,wecansplitothetermsindependentof H m.Furthe rmore, accordingtothediscussionfollowing(5.5.28)the dependenceof Wcovcanbefactored out( Wcov Š3 2 W )sothattherele vantpartof(8 .7.2)becomes S = d4xd4 [ Š eŠ ( trV + G )] + d4xd2 [ 3g +1 4 QABWAWB]+ h c .(8. 7.3) Wehavesett hegravit ationalconstant1 3 2=1.Wew illrestoreitw hennecessary. Underthegaugetransformation trV tr [ V + i ( Š )], G G ,( D€=0),the actionisinvariantifwerescale eŠ i tr .Thustheloca lFayet-I liopoulostermacts asaconventionalgaugetermfor .If =0,asnot edabovetheformof g (i)is extremelyrestricted: 3g mustbe gaugeinvariant. Wenowa nalyzetheg lobalsystem(8.7.3)subjecttotheconditionthatthecosmologicalconstantvanishes.Wecanthenchoose Re <> | = =const ant.Withthe identi“cation Š eŠ trV eŠ G (i,j)= IK 3g (i)=P(8. 7.4) wehavetheactiono f(8. 3.25)withouta global Fayet-I liopoulosterm.We labelc omponentsof as

PAGE 552

5328.BREAKDOWN | = A D | = , D2 | = S (8.7.5) withexpectationvalues < A > =1, < S > = s .(8. 7.6) a.Massmatrices Webeginbyexp licitlycomputingthemassmatricesforthevarious“eldsinthe system.Thiscalculationis a littlelengthy,sowewillsimplifyitasmuchaspossible withoutlossofgenerality.Thus,werescale toremove g fromthechira lpartofthe action: 3g 3.Wealsor ede“ne G : G G +1 3 ln ( g ge3 trV).Thismakes inert undergaugetransformations,andabsorbstheFayet-Iliopoulosterminto G .Inthecase when < g > =0weca nnotperformthisresca ling.However,thiscaseisnotinteresting, sincethensupersymmetryisnotbrokeneveninthepresenceofaFayet-Iliopoulosterm (ifthecosmologicalconstantvanishes). a.1.Vacuumconditions Thesuper“eldequations(8.3.30)fortheaction(8.7.3)are: D2( eŠ G) Š 3 2=0(8.7 .7a) eŠ G 2Gi+3 3Gi+1 4 QAB, iWAWB=0(8.7 .7b) Š eŠ GkAiGi+1 2 ( QABWB) Š1 2 €( QABW€B)=0(8. 7.7c) wherewehaveused(8.7.7a)tosimplify(8.7.7b),andthrownawaysometermsthatlead tohigherorde rspi norand/orderivativeinteractionsthatdonotenterbelow.Wehave written(8.7.7)intermsofKillingvectorsbymakingthesubstitutions ( TA)i jjŠ ikAi,j( TA)j i ikAi.Thevac uumconditions(8.3.31)foundbyapplying spinorderivativesto(8.7.7)become s Š 3 eGŠ Gi fi=0, Gi j fj+3 eGGi=0,

PAGE 553

8.7.Supergravityandsymmetrybreaking5336 s 3=( QAB+ QAB) dAdB, 2eŠ GGi( ikAi)+( QAB+ QAB) dB=0, 3 eGGijfj+ Gij k fkfj+ Gi j( ikAj) dA+9 GieG( s Š eG)+1 2 Š 2eGQAB, idAdB=0.(8. 7.8) Theassumptionthatthecosmologicalconstantvanishesisequivalenttothecondition thattheseequationshaveasolutionforconstant .Wealsohav ethe gaugeinvariance conditions(8.3.c13)(theseholdforgeneralvaluesofthe“elds,notjustatthevacuum poin t): GikAi+ GikAi=0, Š ikCiQAB, i+( TC)B DQDA+( TC)A DQDB=0.(8. 7.9) Togivethegravi tationalactionthecorrectnormalization(5.2.72),weidentify 2=3 Š 2eG.(8. 7.10) a.2.Gravitinomass Asdisc ussedinsec.8.6,wecan“ndthespin3 2 massbysettingall matter “eldsto theirvacuumexpectationvalues,an dcompari ngthecoecientsofthed4xd4 andd4xd23terms.From(8.7.3),thekinetictermhasacoecient Š 2eŠ G= Š1 3 2, andthechiraltermhasacoecient 3(thefactorsof comefromthede“nitionsofthe dynamical“elds(8.7.5));hence,using(8.6.2),we“ndthatthespin3 2 massis m =3 eG(8.7.11) Wesimp lifyourcom putationfurtherbychoosingnormalcoordinates Gi j= i j, Gi j1...= Gi j1...=0.Using(8.7 .10,11),andnormalcoordinates,werewritethevacuumconditions (8.7.8)as s Š m Š Gi fi=0

PAGE 554

5348.BREAKDOWN fi+ mGi=0 6 m Š 2s Š ( Q + Q )ABdAdB=0 3 i Š 2kAiGi+( Q + Q )ABdB=0 mGijfj+ ikAidA+ mGi(3 s Š m )+1 6 2QAB, idAdB=0.(8. 7.12) a.3.Waveeq uations Wenow “ndthelinearizedwaveequations(8. 3.36)thatfollowfrom(8.7.7).Asin sec.8.3,weexpandthe“eldsinsmall”uctuationsabouttheirvacuumvalues.Forthe remainderofthissubsection,allquantities G... ...and QAB...areevaluatedati= aiand i= ai.We “nditusefultointroduceshiftedvariables A A Š GiAi, Š Gii, S S Š GiFi.(8. 7.13) From(8 .7.7a)wehave SŠ A( s Š m ) Š 2 m AŠ 2 mGiAiŠ ( Gijfj+ sGi) Ai=0 (8.7.14a) i € €Š kAiGiAŠ 2 m Š 2 mGi i=0.(8. 7.14b) From(8 .7.7b) ,we “nd Fi+ m [(2 AŠ A) Gi+(3 GiGj+ Gij) Aj+ Ai]=0(8. 7.14c) i €€ i+2 mGi + m (3 GiGj+ Gij) j+ kAiAŠi 6 2QAB, idAB=0.(8. 7.14d) From(8 .7.7c),we“nd ( Q + Q )ABDB+ QAB, idBAi+ QAB, idBAi+3 i Š 2[ kAiAiŠ kAiAi+ kAiGi( A+ A)]=0(8.7.14e)

PAGE 555

8.7.Supergravityandsymmetrybreaking535 1 2 ( Q + Q )ABi € B€+1 2 QAB, i( fiBŠ idB i) +3 Š 2kAi( i+ Gi )=0.(8.7 .14f) Weareleftwiththe equationsofthephysicalboson“elds.Thesesimplifygreatlyifwe us e( 8.7.14a,c,e);we“nd A=0 (8.7.14g) Ai+{3 Š 2( Q + Q )Š 1AB( kAiŠ1 3 i 2QAC, idC)( kBj+1 3 i 2 QBE, jdE) + ikAi jdAŠ m2( GikGkj+3 GiGkGkj+3 GikGkGjŠ Gik jlGkGl) + m [(3 s Š m ) i j+3(3 s Š 2 m ) GiGj]}Aj+{3 Š 2( Q + Q )Š 1AB( kAiŠ1 3 i 2QAC, idC)( kBj+1 3 i 2QBE, jdE) +1 6 2QAB, ijdAdBŠ m2(3 GikGkGj+3 GiGjkGk+ GijkGk) + m (3 s Š 2 m )( Gij+3 GiGj)}Aj=0(8.7 .14h) ( Q + Q )AB€fBŠ 3 Š 2( kAikBi+ kBikAi) AB€+3 Š 2kAiGiX€=0(8.7 .14i) where X€= €ImA Š ImGj€Aj= €ImA+( GjkBjŠ GjkBj) AB€.Here €is theYang-Millscovariantderivative. a.4.Bosemasses We no wd iscusstheseresults.From(8.7.14g)weseethatthecomplexscalar Ais massless.Fortherealpart,thisisnosurprise: ReAisthetraceofthegraviton,which ismasslessbecausethecosmologicaltermwasassumedtovanish.However,theimaginarypartrequiressomecare.Thepseudoscalar ImA isnotrecognizableasoneof the“eldsofthesupergravit ymulti plet;itstandsfor Š 1 A where A€isthe

PAGE 556

5368.BREAKDOWNdi ve rgenceoftheaxialvectorauxiliary“eld.Thustheequation ImA= ImA Š ImGiAi=0shouldbe replacedby A€Š ImGi€Ai=0(8.7 .15) Formany purposes,itmakeslittledierencewhetherweuse ImA orreplaceitwith Š 1 A .Bydimension alanalysisandLorentzinvariance, A€canenterthewave equationofthescalar“elds Aionlythroughitsdivergence: Ai+ cGi€A€+ ... =0.(8. 7.16) Substitutingin(8.7.15),we“nd ( Ai+ cGiImGjAj)+ ... =0.(8. 7.17) Whenwehave A insteadof A€,wegetthesameresult,si nceinsteadof (8.7.16)we have Ai+ cGi ImA + ... =0,(8. 7.18) andusingthe A waveequation,wer eobtain(8.7.17). However,ifgaugeinvarianceisbrokenthegauge“eldwaveequationcangetaspuriouscontributionfrom ImA thatisnotpresentwhen Š 1 A isusedinstea d.Indeed, substi tuting(8.7.15)intothe“rstformof X€(with €ImA replacedby A€)givesa zerocontributiontothespin1mass.Whengaugeinvarianceisunbroken,wegetno contributionfromtheformwith A aswell: €ImAdoesnotaectthespin1mass, andthevacuumexpectationvalueof kBjiszero.However,ifgaugeinvarianceisbroken, theexpectationvalueof kBjisnotzero(equivalently, €( GiAi) = Gi€Ai)and X givesas puriouscontributionthatmustberemovedbyhand. a.5.Fermimasses Thecomponent correspo ndsto the -traceofthegravitino;wede“netheGoldstinoasthatcombinationofmatter“eldsthatcouplesto (itmakesno essentia ldierencewhetherweuse or ,sincewearea lwaysfreetoaddtermstothegravitino). Thuswede“ne Gi i+1 2 m GikAiA.(8. 7.19)

PAGE 557

8.7.Supergravityandsymmetrybreaking537We alsode“netransverse“eldsthatareorthogonalto : iT iŠ Gi AT A+i m dA .(8. 7.20) (These satisfy GiiT+1 2 m GikAiAT=0.)Inte rmsofthese,thespinorwaveequations b ecome: i € €Š 2 m ( + )= 0( 8.7.21a) i € €+2 m ( + )=0(8. 7.21b) i €€ i T+ m ( Gij+ GiGj) jT+( kAiŠ kAjGjGiŠi 6 2QAB, idB) A T=0(8.7 .21c) i € B€ T+[6 Š 2( Q + Q )Š 1BA( kAjŠi 6 2QAC, jdC) Š 2 idB] jTŠ [ m ( Q + Q )Š 1BCQCA, iGi+i m dBkAlGl] A T=0.(8. 7.21d) Caremustbetakentoensurethatthemassoperatoron T, Tisrestrictedtothe  transversesubspace,i.e.,preservestheorthogonalityto Observethatsincethetraceofthegravitinoisa negativenorm state, i.e.,aghost, itskinetictermhasaminussignrelativetophysicalspinors(thesameistrueforthe traceofthegraviton;thewhole mult iplethasnegativenorm,ascanbeseenfromthe action(8.7.3)).Consequently,thoughthemassmatrixinthe system(whichis decoupledfromtheotherspinors)doesnotvanish,botheigenvaluesarezero(themass matrixisnothermitian).Actually,wedidnothavetoexplicitly“ndthewaveequation toarriveatthisresult:TheconditionthattheGoldstinocanbegaugedaway(thatwe cangotoaU-gauge)impliesthatboththeGoldstino andthe -traceofthegravitino musthave zeromassinthegaugethatweuse.

PAGE 558

5388.BREAKDOWNa.6.Supertrace Havingfoundthewaveequations(8.7.14gi,a8),andunderstood theirsigni“cance, wecanevaluatethesu pert race.Thespin0contributionis(recallthatwearestillin normalcoordinates): Š 2[ikAi idAŠ 3 Š 2( Q + Q )Š 1AB( kAiŠ1 3 i 2QAC, idC)( kBi+1 3 i 2 QBE, idE) Š m2( GijGij+3 GiGjGij+3 GijGiGjŠ Gik ijGkGj) + m (3 s Š m )N+3(3 s Š 2 m )( m Š s )](8.7.22a) whereN i iisthenumberofc hiralmultiplets.Thecombinedcontributionofthespin 0andspin1 2 “eldsis: Š 2[9 Š 2( Q + Q )Š 1ABkAikBi+ i ( Q + Q )Š 1ABdC( kAi QBC, iŠ kAiQBC, i) + ikAi idA+ Gij ik fkfjŠ ( N +1) m2+( N Š 1)3 ms + tr(1 Q + Q Qi1 Q + Q Qj) fjfi].(8. 7.22b) Thespin1contribution,omittingthe X€termisgivenbyt heexpression 3 3 Š 2( Q + Q )Š 1AB( kAikBi+ kBikAi)andca ncelsthe“rsttermof(8.7.22b).(Thenormalizationcomesfromthe3statesofaspin1particleandfromtheform(8.3.37)ofthe spin1waveequ ation.)Finally,thespin3 2 contributionisjust Š 4 m2.Thusweget (usingthegauge-invariancerelations(8.7.9)tosimplifysomeexpressions) strM2= Š 2[ikAi idA+ Gij ik fkfjŠ ( N Š 1)( m2Š1 2 2( Q + Q )ABdAdB) Š itr(1 Q + Q Qi)kAidA+ tr(1 Q + Q Qi1 Q + Q Qj) fjfi],(8. 7.23) innormalcoordinates,or,ingeneral,usingcoordinateinvariance,wehave strM2= Š 2[ikAi ; idA+ Ri j fjfiŠ ( N Š 1)( m2Š1 2 2( Q + Q )ABdAdB) Š itr(1 Q + Q Qi)kAidA+ tr(1 Q + Q Qi1 Q + Q Qj) fjfi].(8. 7.24)

PAGE 559

8.7.Supergravityandsymmetrybreaking539Were mindthereaderthathere strM23/2 J =0 ( Š 1)2 J(2 J +1) MJ 2.(8. 7.25) b.Super“eldcomputationofthesupertrace Ifouronlyinterestisthesupertraceformula(8.7.71),wecanobtainitwithfar lessworkusingthete c hniquedevelopedinsec.8.4.b.(Ofcourse,ingeneralweare interestedinthemassmatri cesthemselves,andnotjustthesupertrace).Westartwith lndet ( IKi j)= Š ( N +1) G + lndet ( Gi j)+ Nln ( Š eŠ trV )(8. 7.26) wheredierentiationiswithrespectto = eŠ trVandnot Beforea ddingcontributionsfromthegravitinomassandcorrectingfortheaxial v ect orauxiliary“eld(seebelow),thesupertracereadfrom(8.4.9)is strM2= Š 2[ikAi ; idAŠ ( N +1) trd + Rk l flfkŠ ( N +1)( Gi j fjfi+ iGikAidA) + tr ( Qk1 Q + Q Ql1 Q + Q ) flfkŠ itr ( Qi1 Q + Q ) kAi].(8. 7.27) whereweuse(4.1.29,30): Rk l=[ lndet ( Gi j)]k l,l=[ lndet ( Gi j)]l.(8. 7.28) Theexpression(8.7.27)hasnotmadeuseofthevacuumconditions(8.7.8)or (8.7.12),a nddoesnotincludeeitherthespin3 2 co nt ributionortheaxialvectorauxiliary “eldcorrectiontothe spin1massmatrixdiscussedinsubsec.8.7.a.4.Aswesawinthe previoussection,thespin3 2 contributionmustbeincludedseparately,sincethe -trace ofthegravitinocannotcontri butedirectly:theconditionforthesuperHiggsmechanism tooccurandforthegravitinotoabsorbtheGoldstinoinU-gaugerequirestheGoldstino-gr avit ino -tracesystemtobemassless.Thespin1correction,thoughsomewhat subtle,canalsobefoundwithoutextensive computation.Asdescribedinsec.8.7.a.4, wesimply subtract Š 2( Q + Q )Š 1ABkAiGi( kBjGjŠ kBjGj)= Š 2 2( Q + Q )ABdAdB(see

PAGE 560

5408.BREAKDOWNdiscussionfollowing(8 .7.22b)foranexplanationofthefactors). Onefurtherpointdeservescomment:Whenwerescaled toremovethepotential g (seethebeginningofsec.8.7.a),welostsightofthecontributionoftheFayet-Iliopoulosterm.Whenwemaketheshift G G +1 3 ln ( g ge3 trV),the trd terminequation (8.7.27)isabsorbedintothe iGikAidAtermasaconsequenceofR-invarianceof g ;itis moststraightforwardtoworkinthecoordinatesystemwheretheKillingvectorstakethe formofusual gaugetransformations: 3 gtr ( TA) Š gi( TA)i j aj=0(8.7 .29) andhence tr ( TA) Š1 3 [ ln ( g ge3 V)]i( TA)i j aj=0(8.7 .30) Usingthevacuumequations,wecansubstituteintothesupertrace(8.7.27). Includingthegravitinoandthespin1correctionterm,werecover(8.7.24). c.Examples Wecanusethesup ertraceformulaetostudymanycasesofinterest.Inparticular,inextende dsupergravit ytheorieswee ncounternonminimal G and Q terms.For example,in N =4superg ravity,whichcontainsonephysicalchiralmultiplet,threevectormultiplets,three(3 2 ,1)multi pletsandthesupergravitymultiplet, G Š ln (1 Š ), Q 1 Š 1+ .(8. 7.31) Weca nnottreattheactual N =4theorysincea descriptionoftheinteracting(3 2 ,1) mult ipletisnotavailable,but(8.7.31)suggest slookingatas ystemwithonescalarmultipletand n v ectormult iplets VA,co upledto N =1superg ravity,with G asaboveand QAB= 1 Š 1+ AB.(8. 7.32) We “nd,with G= 2 G = 1 (1 Š a a )2 (8.7.33)

PAGE 561

8.7.Supergravityandsymmetrybreaking541thesupertrace3/2 J =0 ( Š 1)2 J(2 J +1) MJ 2= Š 2( n +2) Gf f = Š 2( n +2) m2.(8. 7.34) Notethatthe Q and R termsin(8.7.17)combinebecause ( Q + Q )2= Š 4( G)Š 1[(1+)(1+ )]Š 2.Unlessascal arpotential g ()isint roduced, nosupersymmetrybreakingwilloccur.Ho wever,it ispo ssibletoaddsuchatermin N =1superg ravity,andthereexistmechanismstogeneratetermsthatactlikeapotentialev enin N =4superg ravity. For N > 4thean alogsof G and Q areexpressedintermsofanovercompletesetof “elds.Wemayexpecthoweverthat Q and G arerelatedsuchthat det ( Gi j) det ( Q + Q ) h (i) h ( i)where h (i)isaholomo rphicfunction.Inthatcase wemayalsoexp ectasimpleresultforthesupertrace. Wecanalsoconst ructmodelswithaFayet-Iliopoulostermandvanishingcosmologicalconstant.Forexample,consider G = eV+ 23 ln [ eV]+ +1 3 ln [( + )( + )](8.7.35) whereand ar ec hiral“elds,transformingunderthegaugetransformationwhile isinert,and ischosensoastomakethecosmologicalconstantvanish(thepotential andtheFayet-I liopoulostermar eincl udedin G asthe -term).We“ndasolutionto (8.7.8)with d =0forsom e “niterangeof (ascanbeveri“edbya pertur bationexpansionabout =0).

PAGE 562

542INDEXINDEX Action,component15,150,331 scalarmultiplet15,150,302 superconformal245,303,312 supergravity255,259,309 v ectormultiplet23,26,162,168,306 Actions,ingravity238 insupergravity299 Adler-Bardeentheorem407,495 Adler-Rosenbergmethod402,478,486 Algebra,superconformal65 super-deSitter67 super-Lie63 super-Poincar e63 supersymmetry9 Anholonomycoecients236,249 Anomalies,inYang-Millscurrents401 localsup ersymmetry489 (super)conformal474 Anomalycancellation494 Anomaly,chiral407 trace473,476,479 Antisymmetrictensor186 Auxiliary“eld16,151,162,252,326 Axial( n =0 )s up ergravity257,274,288 Axial-vectorauxiliary“eld246 Background“eldmethod373 Ba ck ground-quantumsplitting373,377,379,382,410.414 Backgroundtransformations379,412 Beta-function,vanishingof369 Bianchiidentities22,25,29,39,140,174,181,204,292 Bianchiidentities,solutionof25,40,176,184,294,296 Bisection120,123,126 Breakingandauxiliary“elds508 Breaking,radiative509 soft500 spontaneous496,506 BRSTtransformations342,345

PAGE 563

INDE X 543Casimiroperator72,87 Catalystghost426 Centralcharge64,72  Checkobjects39,251,277 Chiralspinorsuper“eld95,123,159,188 Chiralsuper“eld89 Cliordvacuum69 Commutatoralgebra320 Commutator,graded56 Compensator,conformal240,480 de ns ity242,250,255,259,267,286 tensor242,274 Compensators112,267 Compensators,gravitinomultiplet208 Components,auxiliary13,108 byexpa nsion 10,92 byproj ection11,94 covariant24,178 gauge108 ofscalarmultiplet94 ofsupergravitymultiplet38,245,261,322 ofvectormultiplet160 physic al108 Conformalinvariance65,80,240 Conjugation,hermitian57 rest-frame123 Connection,centralcharge86 gauge18,165,170 isospin86 Lorentz36,86,235,252 Constraints,conformalbreaking265,274,470 conformalsupergravity270 conventional21,35,171,237,270,276,410,470 Poincar esupergr avit y 274 representation-preserving172,270,278,470 solutionof172,276,279,470 Contortion41,115,273,289,298 Converter163 Cosetspace,and -models117 andsuperspace74

PAGE 564

544INDEXCosmologicalconstant528 Cosmologicalterm44,312,333 CovariantFeynmanrules382,446 Covariantfunctionalderivative384,447 Covariantization,ofactions43,300 Covariantlychiral172 CP(n)models113,179 CPT77 Curvature38,236,264 Degauged U (1)289,298 Degreeofdivergence393 Delta-function8,97 Densitycompensator250,267,269 Deriva tive, D -9,83 spinor8,56 superfunctional101,168 Derivatives,covariant18,24,35,165,170,235,249,269 DeSittersupersymmetry67,335 Determinant,vierbein238 Dilatationgenerator65,81,275 Divergences358,452 D -manipulation48,50,360 Doublingtrick386,449 Duality,forthegravitinomultiplet211 ofminimaland n = Š1 3 supergravity310 ofnonminimalandchi ralmultiplets200 oftensorandchiralmultiplets190 transformation190,204 Eectiveaction47,357,373,452 Energy,positivity64,497 Energy-momentumtensor473,481 Eulernumber476 Fa dd eev-Popovghost52,340,344,381,420,432 Fa ye t-Iliopoulosterm178,218,308,389,514 Fe rmi-Feynmangauge342,345 Fe ynmanrules46,53,348,438 Fieldequations153,169,313 Fieldstrength25,40,122,167

PAGE 565

INDE X 545Fieldstrength,conformal124 gravitinomultiplet206 supergravity244,266 Ya ng -Mills156,167,176 Fieldstrengths,o-shell147 -trace474,481 Gauge,normal156 supersymmetric37,338,415,440 Gaugeaveraging52,341,344 Gauge“xing52,341,343,428 Gauge-restoringtransformation115,161,164,173 Gaugetransformations159 GaugeWZmodel198 Generalcoordinatetransformations233 Ghostcounting420 Goldstino498,509,513,522,525,527 Gravitinomass333,533 Gravit inomulti plet 206  Hatobjects250,282,411 Hiddenghost424,432 HyperK¨ ahlermanifold158,222 Hypermul tiplet218 Improvedtensormultiplet191 Indexconventions7,54,542 Indices,”at35,234,252 isospin55 wo rld35,234,252 Integral,Berezin8,97 superfunctional103 Jacobiidentities22 K gaugegroup34,170,172,270 K¨ ahler,manifold155,511 po tential155,511,531 K illingvectors157,514 Lagrangemultiplier203

PAGE 566

546INDEX gaugegroup159,162,173,247,279 Legendretransform191 Liederivative232 Light-cone,basis55,108 fo rmalism108,142 Linearsuper“eld91 Localscaletransformations240 Loca lityin 48,357 Lorentzgenerators35,76,235,249 Lorentztransformations,local35,234 orbital233 Mass,gau geinvariant26 Massmatrices532 Measure, chir al301 general300 Minimal( n = Š1 3 )s up ergravity256,287 Mult iplet,gravitino206 N =2scal ar218 N =2tens or223 N =2v ector216 N =4 Ya ng -Mills228,369 nonminimal scalar199 scalar15,70,149 tensor186 3-form193 variantt ensor203 variantv ector201 v ector18,159,185 Nielsen-Kalloshghost53,376,381,434 Nonlinearrealizations117,522 Nonlinear models117,154,219 Nonminimal( n =0, Š1 3 )s up ergravity256,287 No-renormalizationtheorem358 Normalcoordinates157,533 ORaiferteai ghmodel507 Po we r-counting358,393,454,455

PAGE 567

INDE X 547Prepotential147,173 Prepotential,gravitinomultiplet206 supergravity244 Ya ng -Mills159,173 Projectionoperators120 Qu antumtransformations378,413,431 Rarita-Schwinger“eld246 Recursionrelations547 Reduction,productof D s85 Regulari zation393 Regulariza tion,bydimensionalreduction394 inconsistenciesin397,472 localdimensional469 Pa u li-Villars398,404 po int-splitting399,405 Re pr esentation,chiral79,165,174,284 irreducible120 o-shell13,108,143 on-shell13,69,138,143 superconformal80 super-deSitter82 super-Poincar e75 v ector79 Riccite nsor 237 R-transf ormations96,153 Rwe ight96,153,169 Scal arpotential153 Scalei nvaria nce 240 Se lf-energy49,390,443,460 Smatrix391,463 Softbreakingterms502 Spurion500 S -supersymmetry66,246 Stueckelbergformalism112 Superan omaly484 Supercoordinatetransformations34 Supercovarian tization324 Sup ercurrent473,480

PAGE 568

548INDEXSuperdeterminant99,254 Super“eld9,75 Super“eldstrength140 Superfo rm28,181 Superhelicity13,73 Sup erHiggseect498,527 Superpotential507 Sup erscaletransformations250,271,275 Sup ertrace100,513,518,538 Sup ertracemultiplet473,481,486 Supervector34 Symmetrizat ion7,56 Tangents pace 35,86 Tangent-sp acebasis183 Tensorca lcul us 326 Timebeing,the250,357,384,410,433,485 To rsion38,236,264 Torsion, ”atsupersp ace36,87 Transf ormationsuper“eld96 Transver segauge440 U (1)covariantderivatives269 U-gauge527 Variantrep resentation31,201 Variation,co varian t 168 Vielbeindeterminant42,254,255 Vielbein,”at28,86 supergravity34 Vierbein232,246 Volkov-Aku lovmodel522 We ss-Zuminogauge,supergravity38,246,261,317 Ya ng -Mills20,161,163 We ss-Zuminomodel150 Weylte nsor 237