
Citation 
 Permanent Link:
 http://ufdc.ufl.edu/AA00022267/00001
Material Information
 Title:
 Decays of the [gamma](1S) into a photon and two charged hadrons
 Creator:
 BrevaNewell, Luis
 Publication Date:
 2004
 Language:
 English
 Physical Description:
 vi, 121 leaves : ill. ; 29 cm.
Subjects
 Subjects / Keywords:
 Angular distribution ( jstor )
Average linear density ( jstor ) Gluons ( jstor ) Hadrons ( jstor ) Luminosity ( jstor ) Momentum ( jstor ) Photons ( jstor ) Pions ( jstor ) Quarks ( jstor ) Signals ( jstor ) Dissertations, Academic  Physics  UF ( lcsh ) Physics thesis, Ph. D ( lcsh )
 Genre:
 bibliography ( marcgt )
theses ( marcgt ) nonfiction ( marcgt )
Notes
 Thesis:
 Thesis (Ph. D.)University of Florida, 2004.
 Bibliography:
 Includes bibliographical references.
 General Note:
 Printout.
 General Note:
 Vita.
 Statement of Responsibility:
 by Luis BrevaNewell.
Record Information
 Source Institution:
 University of Florida
 Holding Location:
 University of Florida
 Rights Management:
 Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for nonprofit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
 Resource Identifier:
 024493469 ( ALEPH )
847495562 ( OCLC )

Downloads 
This item has the following downloads:

Full Text 
DECAYS OF FIlE T(1S) INTO A PHOTON AND TWO C(tARGED IIADI()NS
By
LUIS BREVANEWELL
A DISSEtITATION PRESENTED TO TttE GRADUATE SCHOOL
OF T1HE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF 'HE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2004
To my Parents.
ACKNOWLEDGMENTS
I owe my gTatitude to many people without whom this work would not have
been possible. First of all, I would like to thank my supervisor, Dr. Yelton. His
office door was always open for me, and he spent long hours of his time listening
to my ideas even though only a few of them ever worked. He gave me freedom to
explore and branch off in different directions while steadily guiding me forward at
the same time. After these years of working together I consider him more than a
mentor, and a think of him as a close friend.
During my stay at the University of Florida I have met many people inside
and outside the physics world. I would like to thank my first year graduate teachers
Dr. Sikivie and Dr. Woodard who were a source of inspiration for me. My thanks
also go to my fellow graduate students Vijay Potlia, Rukshana Patel, Necula
Valentin, Jennifer Sippel, Suzette Atienza, G. Suhas, and many more for all our fun
physics discussions and late homework sessions. I would also like to thank the good
friends ]I made outside the physics building, James Power, Yaseen Afzal (Paki),
Ramji Kamakoti (Ramjizzle), Dan DeKee (Double Down), and Fernando Zamit
(Fernizzle) who always reminded me that there is more to life than Physics.
The most important person I have met during my graduate career is my wife,
Jennifer. I thank her for enduring all those endless days when I would answer her
questions with only grunts and nods while my attention remained fixed on the
computer screen. Her love and support axe invaluable to me.
Finally, I would like to thank my parents Manuel and Charlene, my sister
Teresa, and my brother Gaizka. They have always been there for me.
TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ..... .......................
ABSTRA CT . . .. .. .. . .
CHAPTER
1 THEO RY . . . .
1.1 Particle Physics ...................
1.1.1 The Standard Model ............
1.1.2 Quantum Chromodynamics .........
1.1.3 Introduction to the Radiative Decays of the
1.2 Radiative Decays of Quarkonia Overview .....
2 EXPERIMENTAL APPARATUS .............
2.1 The Cornell Electron Storage Ring ........
2.2 The CLEO III Detector ...............
2.2.1 Superconducting Coil ............
2.2.2 Tracking System ...............
2.2.3 Crystal Calorimeter ...............
2.2.4 Ring Imaging Cherenkov Detector.....
2.2.5 Muon Detectors ..................
3 ANALYSIS OF THE DATA ...................
3
T(lS) .... 6
. . 14
. . 17
17
20
21
21
24
25
28
.. 31
3.1 Data Sample ..................................... 31
3.1.1 Continuum Subtracted Distributions ................. 32
3.2 Event Selection ....... ............................ 34
3.2.1 Skim Cuts ....... ........................... 35
3.2.2 Analysis Cuts ....... ......................... 37
3.2.3 Cut Summary, Efficiencies, and Fake Rates ............ 43
4 EXCLUSIVE RADIATIVE DECAY T(1S) 'tr,+r .......... 48
4.1 Robustness of The Mass Distribution ..................... 48
4.2 Statistical Fit of the Invariant Mass Distribution: Signal Areas
and Their Significance ..........................
4.3 Angular Distribution of The Signal ................ .
4.3.1 Optimum Mass Interval ......................
4.3.2 Background Subtraction ......................
49
50
51
53
4.3.3 Statistical Fit of the Helicity Angular Distribution: J As
signments and their Probability Distribution ........
5 EXCLUSIVE RADIATIVE DECAY T(1S) "K+K .........
5.1 Statistical Fit of the Invariant Mass Distribution: Signal Areas
and Their Significance ...................
5.2 Angular Distribution of The Signal ............
6 EXCLUSIVE RADIATIVE DECAY T(1S) .ypp ......
7 SYSTEMATIC UNCERTAINTIES .................
7.1 Cuts ................................
7.1.1 Justification of the DPTHMU Cut ........
7.2 Angular Distribution of Signal ................
7.3 Different Hadronic Fake Rates Between IS and 4S .
7.4 Other Systematic Sources ..................
7.5 Overall Systematic Uncertainties ..............
8 RESULTS AND CONCLUSION ..................
* 66
. 67
80
. 86
. 86
. 87
. 88
. 97
. 97
99
.. 100
APPENDIX HELICITY FORMALISM FOR TWO BODY DECAYS. ...
REFERENCES ..........................................
BIOGRAPHICAL SKETCH .................................
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DECAYS OF THE T(1S) INTO A PHOTON AND TWO CHARGED
HADRONS
By
Luis BrevaNewell
December 2004
Chair: John Yelton
Major Department: Physics
Using the CLEO III detector we report on a new study of exclusive radiative
T(1S) decays into the final states yir 7r, 'IK+K and ypp. We present branching
ratios for the decay modes T(1S) + yf2(1270), T(1S) + yf(1525), T(1S) 
yfo(1710) with fo(1710) * K+K and T(1S) > yf4(2050).
CHAPTER 1
THEORY
1.1 Particle Physics
Particle physics is the branch of physics dedicated to the study of matter
and energy at the most fundamental level. This means that the job of a particle
physicist is to identify the smallest constituents of matter and describe how they
interact with each other.
Humankind has been interested in this subject since ancient times. Two of the
first particle physicists in recorded history are the Greek thinkers Empedocles and
Democritus from the fifth century BC. Empedocles stated that our complex world
was made from combining four fundamental elements (earth, air, fire and water) in
different proportions. Democritus on the other hand, believed that the apparently
continuous objects in the natural world were not really continuous, but made from
voids and indivisible particles called atoms.
Over the last half century Particle physics has advanced tremendously and we
now have a beautiful, but incomplete, theory firmly grounded on experiment that
describes the fundamental constituents of matter and how they interact with each
other. This theory is called "The Standard Model."
1.1.1 The Standard Model
According to the Standard Mode], the fundamental building blocks of matter
are point like particles which interact with each other in as many as three different
ways. Each type of interaction, or force, is itself carried by point like particles
called force carriers. The particles which are force carriers are bosons I and are
collectively called gauge bosoms because they are needed for the theory to be gauge
invariant. The nonforce carrying particles are fermions. 2
The three interactions described by the Standard Model are called the
electromagnetic force, the weak force, and the strong force. The force of gravity is
not included in the Standard Model, and this is one of the reasons the Standard
Model is not yet complete. The electromagnetic force carriers are photons, the
weak force carriers are the Z', W+, and W particles, and finally, the strong force
is carried by eight kinds of gluons. Table 11 summarizes the situation.
Table 11: Gauge bosons and the force they carry.
Symbol I Name Force Carried
ly Photon Electromagnetic
Z Z Weak
W+ W+ Weak
W W Weak
g gluon Strong
Particles which interact through a particular force are said to couple to it
and to carry an associated charge. The nomenclature is as follows, particles that
interact through the electromagnetic force have an electromagnetic charge called
electric charge, those which interact through the weak force have a weak charge
called weak isospin, and those which interact strongly have a strong charge called
color.
The rest of the particles in the Standard Model which are not force carriers,
the fermions, are subdivided into different groups depending on their properties
(See table 12) reminiscent of the way chemists organized the elements into
1 Bosons are 'defined as particles with integer spin in quantum mechanics.
2 Fermions are defined as particles with half odd fractional spin.
3
the periodic table during the second half of the nineteenth century. Fermions are
divided into quarks (generically represented by the symbol q) and leptons. The
main difference between these two groups is that quarks interact through the strong
force while leptons do not. There are six types of leptons and six types of quarks
(also called the six quark flavors) which are grouped into three generations. Each
generation consist of two quarks and two leptons. All three generations replicate
the same set of force charges, the main difference between generations is the mass
of the particles (for example, the ratio of the masses of e : 1 : is 1 : 200 : 3500).
For each fermion there is an antifermion with equal mass and spin and opposite
charge.
Table 12: Fermion symbols classified into quarks, leptons, and the three genera
tions along with the generational common charges.
First Second Third if Electric Weak Has
Generation Generation Generationi Charge Charge Color
u c t +2/3 +1/2 Yes
Quarks d s b 1/3 1/2 Yes
Leptons e VT 0 +1/2 No
e [A T 1 1/2 No
1.1.2 Quantum Chromodynamics
Quantum chromodynamics (QCD) is the part of the Standard Model that
describes the strong force. QCD is based on local gauge invariance and color
symmetry. There are three possible color charges for quarks called r (red), b (blue),
and g (green). Antiquarks have opposite colors called f, b, p. The strong force
between quarks only depends on their colors and is independent of their flavor.
A very important characteristic of the strong force is that gluons themselves
carry a color charge and an anticolor charge so they can interact with other quarks
through the strong force and change their color. Since there are three colors and
three anticolors one might think that there are nine different gluons. However,
there is one linear combination of color anticolor states that has no net color and
leaves a quark unchanged. There are therefore 9 1 8 gluons. The 8 individual
gluon color states can be written as follows,
1
(1 > = (rb+br)
V2
i(rb bf)
12 >=1
13>=1 r bb)
1
14 >= (rg+gr)
(1.1)
5> = (r gf)

1
1
8 >z= (rr + g 0 2bb)
and represent the eight different gluons that exist in nature. The single and unique
color state left out is called a color singlet,
j9 >= I (r+b +g )
(1.2)
which is invariant under a redefinition of the color (a rotation in color space). In
group theory this decomposition of the color states into an octet and a singlet is
denoted by 3 o = 8 & 1. It is worth noting here that a colorless sate, such as 13 >
or 18 >, is not necessarily a color singlet state.
This situation is analogous to the perhaps more familiar example of two spin
1/2 particles. Each particle can have their spin up (T) or down (1) corresponding to
four possible combinations, each giving a total spin S = 0 or 1. The S = 1 states
form a triplet,
>LL
12 > = (Tl> +i1 IT>) (1.3)
13 > I 1>
and there is a singlet state with S = 0,
1
14 >= /I( Tl> 1I T>). (1.4)
Since gluons themselves carry color they can interact with each other through
the strong force. This interaction among the force carriers is unique to the strong
force and, when included in perturbative QCD calculations, leads to two important
properties observed in nature called "asymptotic freedom" and "color confinement."
Asymptotic freedom means that the interaction gets weaker at short distances.
Color confinement is the requirement that observed states have neutral color, or in
other words, they must be in a color singlet state.
Confinement explains why free quarks or free gluons, which have a net color
charge, have never been observed. It also explains why no fractional charged
particles made from a qq bound state have never been observed since it is not
possible to construct a color singlet for such a state (in group theory terms
3 9 3 = 6 D 3 where we have a sextet and a triplet, but no singlet). On the other
hand, color singlets can be constructed for a qq or qqq system. The color singlet for
qq is simply the state shown on the right side of Equation 1.2. The color singlet for
qqq can be obtained from the decomposition 3 0 3 0 3 = 10 D 8 (D 8 G) I and is,
qqq > r singlet (rgb grb + brg bgr + gbr rbg). (1.5)
Particles that are bound states of qq are abundant in nature and are called
mesons, those that are bound states of qqq are called baryons and also abound in
nature. Both groups are collectively called hadrons. There have been hundreds
of different hadrons observed, confirming the validity of QCD and the quark
flavors. Nevertheless, QCD leaves room for more possibilities. A bound state of two
gluons gg (sometimes called a glueball) can be in a color singlet and in principle
could be observed. Although glueballs are allowed by QCD there is no convincing
experimental observation of one. Another possibility are bound states that are
a mixture of the previous states, such as aqq + Jgg with arbitrary a, i3. These
states are called "hybrid mesons" or simply "hybrids". It is believed [1], [2], [3] that
hybrids are necessary to explain the spectrum of light mesons between 12 GeV.
As with any quantum mechanical theory, the spectrum of bound states is
a fundamental test. Glueballs are allowed by QCD, yet there is no conclu
sive experimental evidence of their observation, despite intense experimental
searches [4], [5], [6] complemented by lattice QCD calculations [7], [8] and other
theoretical contributions like bag models [9], fluxtube models [10], QCD sum
rules [11], weakly bound boundstate models [12], and QCD factorization formalism
models [13]. Physicists cannot be sure they understand QCD until such states are
observed, or until they can explain why we cannot observe them.
1.1.3 Introduction to the Radiative Decays of the T(1S)
The T(1S) is a meson composed of bb quarks. Mesons of this kind, composed
of a quark and an antiquark of the same flavor, are in general called quarkonia.
There is a convention behind the name of the T(1S) The T symbol is
reserved for particles composed of bb where the combined spin of the quark and
antiquark is 1. The "1S" symbol is borrowed from atomic spectroscopy with the
"1" meaning that the bb pair are in the lowestenergy boundstate, and the "S"
3 A good example of this is the successful description of th6 hydrogen atom's
spectrum by quantum mechanics.
meaning that the bb have a relative angular momentum L 0. A description of
particle naming conventions can be found in [14].
The T(1S) is unstable, existing for only about 1024 seconds after which it
decays into daughter particles (which in turn decay themselves if they are not
stable). The term "radiative decay" is reserved to any T(1S) decay where one of
the stable daughters is a photon.
The different ways the T(1S) can decay must obey the symmetries in nature.
Examples of such symmetries are "parity" and "charge conjugation", both of which
will be described soon. Symmetries are very important in the standard model. The
usefulness of symmetries can be seen in Noethers' theorem, which states that for
every symmetry there is an associated conserved quantity. Examples of Noethers'
theorem are the conservation of momentum connected to translational invariance
and the conservation of angular momentum associated with rotational invariance.
In the next sections we will use the parity and charge conjugation symmetries
to find T(1S) decays which conserve the associated symmetry constants and are
allowed by nature. In particular, we will show that the radiative decay of the
T(1S) through a photon and two gluons (see Figure 11) is allowed. The key
observation is that the two gluons must be in a color singlet since both the T(1S)
and the radiated photon have no color and color must be conserved. This means
that the two gluons satisfy color confinement and could form a glueball, although
more conventional meson states, or hadrons in no bound state at all, are also
possible outcomes. Regardless of what the gluons do, their energy will eventually
manifests itself as hadrons. Sometimes, two charged hadrons of opposite charge will
emerge. This work examines those two hadrons from a radiative T(1S) decay to
experimentally probe the gg spectrum.
Parity
The parity operator, P, reverses the sign of an object's spatial coordinates.
Consider a particle Ja > with a wave function T,a( t). By the definition of the
parity operator,
PXJ(F, t) = Pa Ta(Y, t) (1.6)
where Pa is a constant phase factor. If we consider an eigenfunction of momentum
0~(,t = ei('t (1.7)
then
pqy(Z t) = Pa9(X,t) = PA@pX(, t), (1.8)
so that any particle at rest, with = 0, remains unchanged up to a multiplicative
number, Pa, under the parity operator. States with this property are called
eigenstates with eigenvalue Pa. Pa is also called the intrinsic parity of particle a,
or more usually just the parity of particle a, with the words at rest left implicit.
Since two successive parity transformations leave the system unchanged, Pa = 1,
implying that the possible values for the parity eigenvalue are Pa = 1.
In addition to a particle at rest, a particle with definite orbital angular
momentum is also en eigenstate of parity. The wave function for such a particle in
spherical coordinates is,
IF i (Y, t) = RP1 (r) Y'(0, )(19
where (r, 0, 0) are spherical polar coordinates, R1(r) is a function of the radial
variable r only, and the Y (0, 0) is a spherical harmonic.
The spherical harmonics are well known functions which have the following
property,
Y1m (9, i = (1)lY m(r 0, 7r + ) (1
(1.10)
Hence
Pi nim( ,t) Pailnim(,t) = Pa(1)'Fnim(X, t) (1.11)
proving that a particle with a definite orbital angular momentum I is indeed an
eigenstate of the parity operator with eigenvalue Pa(1)'.
The parities of the fundamental fermions cannot be measured or derived. All
that nature requires is that the parity of a fermion be opposite to that of an anti
fermion. As a matter of convention fermions are assigned P = +1 and antifermions
are assigned P = 1. In contrast, the parities of the photon and gluon can be
derived by applying P to the field equations resulting in P = 1 and P. = 1.
The T(1S) hasP = PbP(1)L = 1 since L = 0.
Parity is a good quantum number because it is a symmetry of the strong and
electromagnetic force. This means, that in any reaction involving these forces,
parity must be conserved.
Charge Conjugation
Charge conjugation is simply the operation which replaces all particles by
their antiparticles. In quantum mechanics the charge conjugation operator is
represented by C. For any particle Ja > we can write,
Cla >= coa > (1.12)
where ca is a phase factor. If we let the C operator act twice to recover the original
state Ja >,
Ia >= 2a >= C(ca, >) = c2aCla >= cacala > (1.13)
which shows that CaCc = 1. If (and only if) a is its own antiparticle, it is an
eigenstate of C. The possible eigenvalues are limited to C = ca ca 1.
All systems composed of a the same fermion and an antifermion pair are
eigenstates of C with eigenvalue C =()(L+s) This factor can be understood
because of the need to exchange both particles' position and spin to recover the
original state after the charge conjugation operator is applied. Exchanging the
particles' position gives a factor of (I)L as was shown in the previous section,
exchanging the particles spin gives a factor of (1)s+l as can be verified by
inspecting Equations 1.3 and 1.4, and a factor of (1) which arises in quantum field
theory whenever fermions and antifermions are interchanged. With this result we
can calculate the charge conjugation eigenvalue for the T(IS) and obtain C = 1
since L + S = 1.
The photon is an eigenstate of C since it is its own antiparticle. The C
eigenvalue for the photon can be derived by inserting C into the field equations and
is C'= 1.
Finally, we consider a system composed of two gluons that are in a color
singlet. The two gluons are bosons and they must have a symmetric wave function,
T1G under a g, i g2 exchange. Under this exchange, the orbital angular momentum
part of the wave function contributes with a factor ( 1)L', the spin part of the
wave function for two spin 1 particles contributes with a factor of (I)S, and the
color singlet part of the wave function contributes with a factor of +1 since it is
symmetric. To ensure that T'G is symmetric we need L + S to be even. This implies
that C = ()L+S = 1. The L = 0 and L = 1 possible gg bound states are shown
in Table 13. The Jpc = 1+ is peculiar because it is impossible for a qq system to
have these quantum numbers. If this state is ever observed, it must be a glueball.
Experimental searches for such a state have been done, for example, in [15].
Table 13: Possible gg bound states with L = 0 or L = 1. The possible quantum
numbers are limited by the condition that J L + S be even, which is needed to
ensure a symmetric wave function for the two gauge bosons.
0 ++
0 2 2++
1 1 0+, 1+, 2k
Charge conjugation is a symmetry of the strong and electromagnetic force. For
those particles that are eigenstates of C, C is a good quantum number because in
any reaction involving these forces C must be conserved.
The P and C values of various particles used in this analysis are shown in
Table 14 along with their quark composition, orbital angular momentum, and
internal spin.
Table 14: Symbol, name, quark composition, angular momentum (L), internal spin
(S), parity (P), and charge conjugation eigenvalues (C) for a few of the particles
used in this analysis.
Symbol[ Name ]Quark Composition ILI S fP IC
T(1S) Upsilon(1S) bb 0 1 1 1
1+ Pion ud 0 0 1 x
Ir Pion dii 0 0 1 x
K+ Kaon us 0 0 1 x
K Kaon sf 0 0 1 x
p Proton uud 0 1/2 +1 x
p Antiproton ia 0 1/2 1 x
Photon x x 1 l 1 1
Possible Decays of the T(1S)
At this point we can understand the different possible ways the T(1S) can
decay. The Possible decays are limited because the strong and electromagnetic
force must conserve color, C and P.
The simplest possibility is for the bb pair to interact electromagnetically and
annihilate into one virtual photon This is allowed by parity, charge conjugation
symmetry and color conservation. The decay of the T(1S) to one gluon is not
allowed by color conservation and is therefore forbidden. T(1S) decays to two
photons are forbidden by charge conjugation. T(1S) decays to two gluons are
4 Such a photon is called virtual because it cannot conserve the 4momentum of
bb and is unstable, only existing for a brief period of time, as allowed by the uncer
tainty principle, after which it decays.
also forbidden by charge conjugation. T(1S) decays to 3 gluons are allowed (3
gluons can form a color singlet). T(1S) decays to 3 photons are also allowed, but
are largely suppressed by decays to one photon since 3 successive electromagnetic
interactions are much less likely to occur than a single one. Finally, T(1S) decays
two one photon and two gluons are also allowed under the condition that the two
gluons be in a color singlet state.
The three different possible T(1S) decays with least amount of interactions
(also called lowest order decays) are shown in Figure 11.
g
(a) 7 Z
b
(b)
9
(c)
1. +
Figure 11: Lowest order decays of the T(1S) allowed by color conservation, charge
conjugation symmetry, and parity. (a) Shows the decay into three gluons, (b) shows
a radiative decay, and (c) shows the electromagnetic decay through a virtual pho
ton that in turn decays electromagnetically into a pair of charged fundamental
particles, such as quarks or charged leptons (the charged leptons are represented by
the symbol 1).
Observable Resonances
In this work we search for a resonance X produced in a radiative T(IS) decay,
T(1S) *X.
(,1.14)
13
Using charge conjugation symmetry on both sides,
1 lCx. (1.15)
Therefore, Cx = +1.
In order for us to observe X it must decay into two charged hadrons,
X * hh, (1.16)
where h = 7r, K, p are the hadrons whose momentum we are going to measure.
Applying charge conjugation to this last decay,
+1 = IL+ S (1.17)
where L and S axe respectively the angular momentum and spin of the h+h
system. This last equation implies that L + S must be even. This has consequences
for the possible X parities we can observe. By parity conservation in 1.16,
Px = ( ) ( ) 1 .. ,1_)L. (1.18)
For h =7r, K, S = 0 and L J must be even, which implies that Px = +1. For
h = p, S can be 0 or 1 and particles with both positive and negative parities can be
detected.
Table 15: Possible S, L, J, P and C values for X from the radiative decay
T(1S) y X reconstructed in different decay modes.
Decay Mode_ P C
Xr + 0 even even +1 +1
X* K+K 0 even even +1 +1
XPP 0 even even + +1
x p I odd even and odd 1 +1
14
1.2 Radiative Decays of Quarkonia Overview
Theoretical models exist for glueball production in quarkonia decay [12]
and for the glueball spectrum. For example, a quenched lattice calculation [71
predicts a jPC = 214 glueball in the 2.2 GeV/c2 mass region (see Figure 12).
According to Table 15, in the charged pion and kaon modes, we are limited to
detect glueballs in the leftmost column where P = C = +1 of Figure 12, while in
the proton mode we are restricted to the two left most columns.
12
10 2+_ 3L
2% ...... .... 1 4
8
3
E 6 2++~
4 04
2
0 0
++ +
PC
Figure 12: Quenched lattice calculation result for the glueball spectrum for differ
ent P and C values. The mass scale is shown in terms of a scaling parameter from
the QCD lattice calculation named r0 on the left and in absolute terms on the right
by taking r1 = 410 MeV.
5 Here J stands for the internal angular momentum (spin) of the glueball,
J= L+S.
For some time bound states not considered to be pure glueballs, such as
f2(1270), f4(2050), r7, and r' have been observed in J/I radiative decays at
the 10 production level [16] 6 In 1996, the BES collaboration claimed the
observation of a resonance they called the fj(2220) particle in the radiative Ji4
system at the 10' level [171 (see Figure 13). A lot of excitement was generated at
the time because it is possible to interpret the fj(2220) as a glueball. However, this
result has not been confirmed.
2.0 2.2 2.4 2.6
Mass (GeV)
Figure 13: The mass spectrum obtained by the BES collaboration in radiative
J/bP decays into different hadronic modes.
Key to identifying a particle as a glueball are (a) suppressed production in
twophoton collisions (unlike quarks, gluons don't carry electric charge and do not
6 The J/ particle is a cI2 bound state. Since the strong force is flavor blind the
situation in J/I decays is in principle similar to that in T(1S) decays.
couple to photons), and (b) flavor symmetric decays, since a pure glueball has no
valence quarks.
CLEO has already done several studies of radiative decays of the T(1S).
Naively, one expects these types of decays to be suppressed by a factor
[(qb/q,)(m/mb)2 0.025 (1.19)
with respect to J'0 radiative decays. This comes from noticing that the quark
photon coupling is proportional to the electric charge and the quark propagator
is approximately 1/m for low momentum quarks. In 1999, CLEO made the first
observation of a twobody T(IS) radiative decay [18]. The spin of the observed
resonance could not be measured, ut its mass and width where consistent with the
f2(1270) particle. Under this assumption, comparing the measured branching ratio
of T(IS)  f2(2220) to the measured branching fraction of the J1,0 7 f2(1270),
a suppression factor of 0.06 0.03 was obtained. After the BES result for the
fj(2220) in radiative .JA)v decays, a corresponding search was performed by CLEO
in the radiative T(IS) system 1191. This analysis put limits on the fj(2220)
production in radiative T(IS) decays.
In this work we are privileged to have available the largest collection of
radiative T(1S) decays in the world. With it, we can study the structure of color
singlet gg hadronization, and shed more light on the fj(2220) result from BES.
CHAPTER 2
EXPERIMENTAL APPARATUS
To carry out, our study of the di.hadron spectrum we need to first produce
the T(1S) resonance and secondly observe its daughter particles flying away at
relativistic speeds. These two tasks are respectively accomplished by the Cornell
Electron Storage Ring and the CLEO III detector.
2.1 The Cornell Electron Storage Ring
The Cornell Electron Storage Ring (CESR), located at Cornell University. is a
circular particle accelerator that produces e+e collisions.
In order to produce such collisions electrons and positrons need to be created,
accelerated and stored. CESR's different components, shown in Figure 21, have
been carrying out this task since 1979.
A typical CESR run begins at the linear accelerator (LINAC) where electrons
and positrons are produced. To create positrons, electrons are evaporated off' a
filament and linearly accelerated by electromagnetic fields towards a tungsten
target. The collision creates a spray of electrons, positrons and photons. The
electrons are cleared away with magnetic fields and the positrons are introduced
into the synchrotron. The filling procedure is identical, except that the tungsten
target is removed.
Once the electron and positron beams are introduced into the synchrotron,
they are accelerated to the operating energy. in hour case they are accelerated to
the point where their combined energy is the T(1S) mass, 9.46 GeV.
Once the beams are at the desired energy they are transfered to the storage
ring, where they will remain for about an our. At one point in the storage ring the
beams are forced to cross paths. This is the point where e+e collisions occur 1
and where the center of the CLEO III detector is located. If the accelerator is
0 BUNCH OF POSITRONS
o BUNCH OF ELECTRONS
Figure 21: The Wilson Laboratory accelerator located about 40 feet beneath
Cornell University's Alumni Fields.
performing well a high collision rate results. A high collision rate is crucial for the
success of an accelerator and the experiments it serves. The important figure is the
number of possible collisions per second per unit area; this is called the luminosity.
In order to maximize the luminosity, the beams are focused as small as possible at
This point is not fixed in space, but varies from event to event inside a small
volume of space called the interaction region (IR).
19
the IR. During the CLEO III installation several magnetic quadrupoles were added
to CESR to improve the beam focus. CESR has consistently outdone itself while
collecting luminosity over the years (see Figure 22).
223100100
8000
7000
6000
.E
00.
14000
01
00
'C 0
Figure 22: CESR yearly luminosity. The gaps in 1995 and 1999 correspond to
down times when the CLEO II.V and CLEO III detectors where being installed.
The CLEO III detector measures the time integrated luminosity over a period
of time by counting how many times a benchmark process occurs during a certain
time interval at the IR. For redundancy, there are two benchmark processes that
are used, one where the e e particles interact to produce a new e+e pair,
and the other one where the e+e annihilate and produce two photons. Using
the known crosssection for each process, the number of events is converted to a
luminosity called the Bhabha integrated luminosity for the first process, and the y',
integrated luminosity for the second one. The term "integrated" is sometimes left
out and the total luminosity is referredto as simply the Bhabha or y luminosity,
with the time integration left implicit.
2.2 The CLEO III Detector
When the e+e collision occurs, the two particles are annihilated we enter
the world of particle physics. Nature decides what to do with the energy from
the annihilation. We have no chance of directly observing what is happening at
the annihilation point, but eventually long lived semistable particles are created
that fly off at relativistic speeds. These particles carry information about what
happened after the e~e collision, and can tell. us what nature did. Ehe CLEO III
detector has the important mission of detecting and measuring such particles.
As one can see in Figure 23, the CLEO III detector is a composite of many
detector elements. These subdetectors are typically arranged as concentric cylin
ders. The entire detector is approximately cube shaped, with one side measuring
about 6 meters, and weighs over 500 thousand kilograms.
C L E O IIIS l no d C l
Barrel
RiCH Drift
Silicon I
SC Ouadrupole
Pylon
CEndcap
SC Calorimeter
Iron Chambers
Figure 23: The CLEO III detector.
As implied by the name. CLEO III is not the only CLEO detector. CLEO
III was preceded by CLEO II.V, CLEQ.I1, CLEO I.V and the original CLEO
detectors. Fhe CLEO III detector was a major upgrade compared with the
previous version of CLEO [201, ,21], and has an improved particle identification
system together with a new drift chamber and a new silicon vertex detector.
2.2.1 Superconducting Coil
All the detector subsystems except, for the muon chambers are located inside
a superconduction coil. The coil remains unchanged since CLEO II. It is kept in
a superconducting state by liquid helium. The purpose of the coil is to provide a
1.5 Tesla magnetic field, which is mforin to 0.2%, to bend the paths of charged
particles in the detector. By measuring how much much a charged particle bends,
experimenters can measure the moment am of the particle.
The coil inner radius is 1.45m and its outer radius is 1.55 m, with a radial
thickness of 0.10 in. The total length of the coil in z is 3.50 m. It is wound from
a 5mm x 16 tnmi superconducting cable (Al surrounding CuNbTi strands). It
is wound in 2 layers, with 650 turns per layer, on an aluminum shell. When in
operation a current of 3300 amps flows through the coil.
2.2.2 TlYacking System
After particles from the interaction point pass through the beam pipe, they
begin to encounter the active detector elements of the tracking system. There are
two subdetectors responsible for tracking the curving path of charged particles.
The first one encountered by particle, is the silicon vertex detector, and the second
one is the central drift chamber. The CLEO li tracking system is responsible
for tracking a charged particle's path and measuring its momentum. Typical
momentum resolution is 0.3% for 1 GeV tracks. The tracking system also measures
ionization energy losses with an accuracy of about 6%.
Silicon Vertex Detector
The silicon vertex detector in CLEO III [22], also called SVD III, is a four
layer barrelonly structure with no endcaps that surrounds the beam pipe. This
detector (see Figure 24) provides four 0 and four z measurements covering over
over 93% of the solid angle. The average radius of inner surface of the four layers is
25 mm, 37.5 mn, 72 mm, 102 mm. The detector is constructed from 447 identical
doublesided silicon wafers, each 27.0 mm in 52.6 mm in z and 0.3 mnm thick.
The wafers are instrumented and read out on both sides. The instrumentation on
each side consists of an array of aluminum strips oil the wafer surface. These strips
are connected to preamplifiers at the end of the detector. The inner side has 512
strips in thc z direction and the outer side has 512 in the 6 direction. Therefore
each wafer contains 512+512 sensors. The 447 wafers are arranged in the 4 layers,
as follows: 7 sections in 0, each with 3 wafers in z, total = 21 wafers in the first
layer; 10 sections in each with 4 wafers in z, total = 40 wafers in the second
layer; 18 sections in each with 7 wafers in z, total = 126 wafers in the third
layer; 26 sections in phi, each with 10 in Z, total = 260 wafers in the fourth layer.
Charged particles traversing the wafer lose energy and create electron hole
pairs. Approximately 3.6 eVis required to create a single electronhole pair. The
electrons and holes then travel in opposite directions in the electric field applied
to the surfaces of the wafers until they end up on the aluminum strips, and the
detector registers a "hit'". When combined together, the hit on the inner side of a
wafer and the hit onl the outer side give a measurement of the (z, Q@). Tho wafer
position itself determines r.
The Central Drift Chamber
The CLEO III central drift chamber (DR 111) is full of a gas mixture with 60%
Helium and 40% propane held at about 270 K and at a pressure slightly above one
atmosphere. The drift chamber is strung with array of anode wires of goldplated
23
1 cm
tLJ
Figure 24: View of the SVD III along the beampipe.
tungsten of 20 rn in diameter and cathode wires of goldplated aluminum tubes
of 130 ,m in diameter. The anodes are kept at a positive voltage (about 2000 V),
and the cathodes are kept grounded, which provides an electric field between the
anode and the cathode wires. Anode and cathode wires are often called "sense"
and "field" wires respectively.
As a charged particle passes through the DR 1II, it interacts electromagneti
cally with the gas molecules giving energy to the outer electrons which become free
in a process called ionization. The free electrons from the ionized gas molecules
drift in the electric field toward the nearest anode wire. As the electrons get close
to the anode, the electric field becomes very strong which causes an avalanche
as further ionization is induced. The result of the avalanche is a large number of
electrons collapsing upon the sense wire in a very short amount of time (less than
one nanosecond). When this happens to a sense wire, we say that there is a "hit".
The current on the anode wire from the avalanche is amplified and collected at
the end of the anode wire. Both the amount of charge and the time it takes it to
move to the end of the detector are measured. A calibration of the drift chamber
is used to convert the amount of charge to a specific ionization measurement of the
incident particle. A calibrated drift chamber can also convert the time to roughly
measure the position along the sense wire where the charge was deposited.
The wires are strung along the z direction. About 2/3 of the outer part of
the drift chamber (the farthest part from the interaction point) is strung in with a
slight angle (about 25 miliradians) with respect to the z direction to help with the
z measurement. Wires strung in the z direction are called "axial" wires, while those
that are strung at a slight angle are called "stereo" wires.
The DR III consists of an inner stepped section with 16 axial layers, and an
outer part with conical endplates and 31 small angle stereo layers. There are 3
field wires per sense wire and they approximately form a 1.4 cm side square. The
drift resolution is around 150 Im in r and about 6 mm in z. All wires are held
at sufficient tension to have only a 50 /im gravitational sag at the center (z = 0).
There are 1696 axial sense wires and 8100 stereo sense wires, a 9796 total.
2.2.3 Crystal Calorimeter
The CLEO Crystal Calorimeter (CC) is composed of 7784 thalliumdoped
CsI crystals. Each crystal is 30 cm long (16.2 radiation lengths) with 5cm x 5cm
square front face. The crystals absorb any incoming electron or photon which
cascades into a series of electromagnetic showers. The electronic system composed
of 4 photodiodes present at the back of each crystal are calibrated to measure the
energy deposited by the incoming particle. Other incoming particles other than
photons and electrons are partially, and sometimes fully, absorbed by the crystal
giving an energy reading.
The CC is arranged into a barrel section and two endcaps, together covering
95% of the solid angle. The CC barrel section is unchanged since CLEO II; the
endcaps have been rebuilt for CLEO II to accommodate the new CESR interaction
region quadrupoles. The barrel detector consists of an array of 6144 crystals, 128
in 0 and 48 in z, arranged in an almostprojective barrel geometry. That is, the
crystals are tilted in z to point to a few cm away from the interaction point, and
there is also a small tilt in 0. The CC barrel inner radius is 1.02 m, outer radius
is 1.32 m, and the length in z at the inner radius is 3.26 m. It covers the polar
angle range from 32 to 148 degrees. The barrel crystals are tapered towards the
front face (there are 24 slightly different tapered shapes), the endcap crystals are
rectangular, but shaved near the outer.radius to fit in the container. The CC
endcaps consist of two identical end plugs, each containing 820 crystals of square
crosssection, aligned parallel to the beam line (not projective). There are 60
crystals in the "fixed" portion of the "keystone" piece of the endcap. 760 in the
part that slides. The keystone is made up of two parts, one on top that has 12
crystals that for mechanical removal reasons is separate from a container holding 48
crystals. The endcap extends from 0.434 m to 0.958 m in r. The front faces are z =
1.308 m from the interaction point. It covers the polar angle region from 18 to 34
degrees in +z, and 146 to 162 in z.
The photon energy resolution in the barrel (endcap) is 1.5% (2.5%) for 5
GeVphotons, and goes down to 3.8% (5.0%) for 0.1 GeVphotons.
2.2.4 Ring Imaging Cherenkov Detector
The Ring Imaging Cherenkov (RICH) detector [23] is a new detector subsys
tem for CLEO III. It replaces the CLEO I.V time of flight system designed to
measure particles' velocities.
Cherenkov radiation occurs when a particle travels faster than the speed of
light in a certain medium,
v > c/n. (2.1)
Where v is the velocity of the particle, c is the speed of light in vacuum, and n
is the index of refraction of the medium the particle is traveling in. The charged
particle polarizes the molecules of the medium, which then turn back rapidly
to their ground state, emitting radiation. The emitted light forms a coherent
wavefront if v > c/n and Cherenkov light is emitted under a constant Cherenkov
angle, J, with the particle trajectory forming a cone of light. The cone halfangle is
given by the Cherenkov angle which is,
cos 6 = 1. (2.2)
vn On*
If the radiation angle, 6, is measured, the speed of the incident particle is known.
This measurement, combined with the momentum measurement from the tracking
system, gives a measurement of the particles mass, and can be used in particle
identification.
The threshold velocity at which Cherenkov radiation is emitted is vmin n
When a particle traveling at the threshold velocity transverses the medium a very
small cone with 6 0 is produced. The maximum emission angle occurs when
v,,. = c and is given by
1
cos 6max (2.3)
n
The RICH (see Figure 2 5) consists of 30 modules in phi, 0.192 m wide and
2.5 m long. The detector starts at a radius of 0.80 m and extends to 0.90 m. Each
module has 14 panes of solid crystal LiF radiator at approximately 0.82 m radius,
0.192 m wide, 0.17 m long, 1 cm thick. Inner separation between radiators is
typically 50/um. The LiF index of refraction is n = 1.5. The radiators closest to z
= 0 in each module have a 45 degree sawtooth outer face, to reduce total internal
reflection of the Cherenkov light for normal incident particles (see Figure 26). The
radiators axe followed by a 15.7 cm (radial) drift space filled with pure N2. The
drift space is followed by the photodetector, a thingap multiwire photosensitive
proportional chamber.
Figure 25: The RICH detector subsystem.
track
I
10 MM
track
Y7
10m MM
Figure 26: The two kinds of RICH LiF radiators. For normal incidence particles
(z 0) a sawtooth radiator is necessary to avoid internal reflection.
With this index, particles in the LiF radiator with beta = 1 produce
Cherenkov cones of halfangle cos'(1/n) = 0.84 radians. With a 16 cm drift
space, this produces a circle of radius 13 cm. The RICH is capable of measuring
the Cherenkov angle with a resolution of a few miliradians (see Figure 27). This
great resolution allows for good separation between pions and kaons up to about 3
GeV as Figure 28 shows.
tz
L
2
0
1 2 3 4 5
Radiator row
6 7
Figure 27: Cherenkov angle resolutions per track as a function of radiator row for
Bhabha events. Row 1 corresponds to the two rows closest to z = 0, etc.
2.2.5 Muon Detectors
The muon detectors (MU) are the most external subsystem of the CLEO III
detector. They remain unchanged from CLEO II, and are composed of plastic
proportional tubes embedded in the magnet iron return yoke. They cover 85%
of the 47r solid angle (roughly 30150 degrees in polar angle). If a series of hits is
detected in the muon chamber layers they most likely correspond to muons because
other particles are blocked by the iron. Besides detecting muons, the heavy iron
I I I I I *
*Data
o Monte Carlo
0
I I I o I I I ,
0.4
: : :
" 0 .3 ...................... ..................... ...................... ..... ................... ..............: ........7 ''
0 0.2 ........
0 .1 ................. ........................ .... ..................... ..................... r.... ...........
0.5 1.0 1.5 2.0 2.5 3.0
Momentum (GeV/c)
Figure 28: Pion fake rate as a function of particle momentum for kaon efficiency of
80% (circles), 85% (squares) and 90% (triangles).
of the return yoke protects the inner subsystems of the CLEO III detector from
cosmic ray background (except for cosmic ray muons of course).
There are three planes of chambers in the barrel section, arranged in 8 octants
in 0. The plastic barrel planes lie at depths of 36, 72, and 108 cm of iron (at
normal incidence), corresponding to roughly 3, 5, and 7 hadronic interaction
lengths (t6.8 cm in iron) referred to as DPTHMU. There is one plane of chambers
in each of the two endcap regions, arranged in 4 rough quadrants in 6. They lie
at z = 2.7 m, roughly covering the region 0.80 < I cos(O)I < 0.85. The planar
tracking chambers use plastic proportional counters at about 2500 V with drift gas
of 60% He, 40% propane, identical to (and supplied by the same system as) the
drift chamber gas. Individual counters are 5 m long and 8.3 m wide, with a space
30
resolution (along the wire, using charge division) of 2.4 cm. The tracking chambers
are made of extruded plastic, 8cm wide by 1 cm thick by 5 m long, containing eight
tubes, coated on 3 sides with graphite to form a cathode, with 50 Pm silverplated
CuBe anode wires held at 2500 V. The orthogonal coordinate is provided by
8cm copper strips running perpendicular to the tubes on the side not covered by
graphite.
CHAPTER 3
ANALYSIS OF THE DATA
This work builds on the techniques developed by previous CLEO radiative
T(1S) analyses [18], [19], modified to be used with CLEO III data. A new
technique based on kinematic fitting is developed to, together with the new
RICH detector, improve efficiency and particle identification.
We search for radiative T(1S) decays in the modes T(1S) . yr+Tr, yK+K,
and ypp. The e+e collision data, has both resonant events, where the e+e
annihilate to give a T(1S) and continuum events, where the e+e collision does
not give a T(1S) To be sure we are observing T(1S) and not a continuum process,
the continuum must be subtracted by using a pure source of correctly scaled
continuum events. Pure continuum data can be obtained by operating CESR at an
energy different form the T(1S) mass. After subtracting the underlying continuum,
we examine the dihadron invaxriant mass spectrum in search of resonances. We
determine the spin and production helicity (the projection of the spin on the
momentum vector at production time) of any found resonances by examining the
photon and hadron angular distributions.
3.1 Data Sample
The analysis presented here is based on CLEO III data. Throughout this
document, and unless otherwise stated we use 7y luminosity (the luminosity types
used by CLEO are defined at the end of Section 2.1). We prefer to use the 7y over
the Bhabha luminosity because the resonant process T(1S) e+e artificially
increases the reported Bhabha luminosity by about 3% in T(1S) data. This extra
contribution would need to be accounted for when doing a continuum subtraction.
Choosing the yy luminosity avoids this complication.
The CLEO III data is divided into numbered sets. Sets 18 and 19 have a
luminosity of 1.13 0.02fb1 in the beam energy range 4.7274.734 GeV. This
data, which we call the T(1S) data (or simply the IS data), has both resonance
e+e , T(1S) and continuum events. We take the number of resonant events
from [24], NT(1s) = (2.1 0.1) x 107. This number contrasts with the previous
generation measurement of CLEO II, where Nr(ls) 0.15 x 10' were available.
In datasets and 18 and 19 there are also 0.192 0.004fb1 taken below the
T(1S) beam energy (4.7144.724 GeV). This data, which we call the T(1S)off data
(or simply the 1Soff data), has relatively low statistics and corresponds to purely
continuum events.
'To improve our continuum statistics we use 3.49 0.07f b' from datasets
9, 10, 12, 13, and 14 of data taken near the centerofmass energy of the T(4S),
which for our purposes is defined as data with beam energy in the 5.2705.300 GeV
range. This set of data, which we call the T(4S) data (or simply the 4S data), is a
source of pure continuum because no T(4S) * BB resonant event can survive our
"cuts" (the cuts are presented in Section 3.3).
3.1.1 Continuum Subtracted Distributions
We use the continuum data taken at the T(4S) energy to subtract the under
lying continuum present in the T(1S) data. This is important because continuum
background processes like e+e  'yp with p , 7r+7., e4e * y/ with
S K+K, and direct e+e + .yh (we will use the convention h = 7, K, p
from now on), look like the signal events we are searching for. To first order, the
cross section of these continuum process scales like 11s, where s is the square of
1 Cuts are simply conditions that an event must satisfy to be considered in the
,analysis. Cuts are necessary to eliminate background that would otherwise make a
measurement difficult or even impossible.
the center of mass energy of the c+e system. Taking the luminosityweighted
average beamenergies of each interval,, and the y'f luminosities (see Table 31), we
calculate that the T(4S) data scaled down by a factor of 0.404 2 represents the un
derlying continuum in the T (IS) data. This is true up to differences in momentum
distributions and phase space. The error in the continuum scale factor is unknown
because the luminosity ratio is expected to have a small but undetermined sys
tematic error. We make the somewhat arbitrary decision to retain three significant
digits in the continuum scale factor because it is sufficient for our purposes and
there are 0.5% effects from second order terms in the cross section formulae.
To eliminate the contribution of continuum events from a T(1S) data variable
distribution (e.g., the invariant mass of two tracks, the photon angular distribution)
we proceed as follows,
Obtain the T(4S) data distribution for the same variable.
Efficiency correct both the T(1S) and T(4S) distributions, using a
GEANT [25] based Monte Carlo (MC) simulation of the detector. Examples of
MC efficiency distributions are shown in Figures 33 and 73.
2 If we are to be mathematically strict we should calculate the scale factor as
E lSrun / I 4S run
1S runs S lun 4S runs S4S run
with obvious notation. This is equivalent to redefining the average energy as
1 _1 ELrun
E2 L EE2
runs run
However, our energy intervals are sufficiently narrow and this calculation does 4
not change the last significant digit of the original scale factor. Similarly, taking
into account second order terms in the energy dependence of the cross section, like
the one that appears as mn/s in the explicit formula for the e+e + 7p cross sec
tion, also has an insignificant effect on the scale factor.
Subtract the T(4S) distribution from the T(1S) distribution using a T(4S)
scale factor of 0.404.
We call this set of steps "continuum subtraction" by definition. For com
pactness, we call such a distribution the "continuum subtracted
distribution/plot" or " continuum subtracted distribution/plot". Except
for statistical fluctuations and phase space effects, the resulting 1S distribution
should not have any contribution from continuum processes that scale as 1/s.
It is important to notice that any continuum subtracted distribution is
efficiency corrected. This means that a fit to a continuum subtracted distribution
(for example, the continuum subtracted invariant mass distribution) gives the
efficiency corrected number of events directly. Strictly speaking, this number of
events is only correct if all the other variables we cut on have the same initial
distribution in data and MC, or if the efficiency does not depend oil them. In this
note we use "flat MC", defined as MC that is generated with a flat distribution in
the mass and the helicity angles 0, 0, (these angles are defined in the appendix).
For example, the number of events obtained from the continuum subtracted
invariant mass fit needs to be corrected to account for the fact that the helicity
angle distributions are not flat in data (see for example Figure 49, and the
efficiency is highly dependent on these variables (see Figure 73). This correction is
done in Section 7.2.
A summary of the results from this section is shown in Table 31.
3.2 Event Selection
Event, selection for T(1S) * yh+h is straightforward and can be thought of
in terms of three major stages.
First we skim the data, keeping only those events that contain exactly one
highenergy photon and two tracks. Next, we require that the total 4momentum
of these three elements be consistent with the colliding e+e 4momentum. Finally,
35
Table 31: Summary of the data used in this analysis. The continuum scale factor
is obtained using yy luminosities because the Bhabha luminosity is artificially high
during T(lS) running due to the process T(1S) ee.
T(1S) T(4S) T(1S)off
Dataset 18, 19 9, 10, 12, 13, 14 18, 19
Average Ebeam (GeV) 4,730 5.286 4.717
Range of Ebam(GeV) 4.7274.734 5.270 5.300 4.714 4.724
Z(re)(fb') 1.20 0.02 3.56 : 0.07 0.201 0.004
L(y7)(fb1) 1.13 0.02 3.49 0.07 0.192 0.004
T(1S) continuum scale factor 1 0.404 5.84
in the third stage, we project the surviving data onto the three different hadronic
modes via hadron separation and QED suppression cuts.
However, checking the 4momentuml involves using the tracks masses. This
means that the information from stage 2 should somehow be useful in stage 3. This
is indeed the case, and the details of how we do it are revealed in this section.
3.2.1 Skim Cuts
We skim over the data in the "hardGam" subcollection. The hardGam
subcollection was developed with this type of analysis in mind. For an event that
passes the triggers 3 to be classified as hardGam it must pass the following cuts,
eGaml > 0.5
eSh2 < 0.7
eOverP1 < 0.85
eVis > 0.4
aCosTh < 0.95 where "eGaml" is the highest isolated shower energy
relative to the beam energy, "eSh2" is the energy of the second highest shower
relative to the beam energy, "eOverPl" is the matched shower energy relative to
3 Triggers are basic criteria that an event must satisfy to recorded during the
data collection processes. Triggers are designed to get rid of trash and noise and
reduce the size of the data sample while keeping all of the important information.
36
the momentum of the track with highest momentum, "eVis" is the total energy
detected (charged tracks are assumed to be pions) relative to the center of mass
energy, and "aCosTh" is the z component of the unitary total momentum vector.
MonteCarlo predicts that about 75% of the generated T(1S) 4 yh+h signal
passes the hardware and software triggers and gets classified as hardGam.
As mtrentioned above, we use this data to make our skim. To write an event
from the hardGam subcollection into our skim we require the following topological
cuts,
There are exactly two "good tracks"; there can be any number of tracks that
are not "good tracks" but these are not used in the analysis. We define a "good
track" as a track that satisfies the following cuts; drift chamber track ionization
energy loss (dE/dX) information is available, the ratio of number of wire hits to
those expected is between 0.5 and 1.5, the pion fit has x2/d.o.f. < 20 (here d.o.f.
stands for degrees of freedom), and the distance of closest approach to the beam
spot in the xy plane (called DBCD) is less than 5 3.8P (nun) if P < 1 GeV/c
(where P is the tracks momentum in GeV/c) and less than 1.2 mm for tracks with
P > 1 GeV/c. This DBCD cut is common in the more sophisticated CLEO 11/11.5
analyses. It performs better than a simple DBCD < 5 mm cut, because it takes
into account the fact that traks with higher momentum have a better measurement
of DBCD since they scatter less.
There is exactly one "good shower", there can be any number of showers
that are not "good showers" but these are not used in the analysis. We define a
"good shower" as an unmatched shower with energy > 4 GeV.
These topological cuts are about 85% efficient for generated signal events that
have passed the triggers and have been classified as hardGam.
The overall skim efficiency is between 60,65%, depending on the mode (see
Table 34).
3.2.2 Nnalysis Cuts
After our skim we call any cuts we make "analysis cuts". These cuts are done
at analysis time and are mode dependent. As a convention, and unless otherwise
stated, efficiencies for individual analysis cuts are reported relative to the events in
the skim (not relative to the events generated).
4momentum Cut
All fully reconstructed events should have the 4momentum of the 6 e
system. This constraint is usually implemented with a simple twodimensional
AEp box cut, where AE is the difference between the reconstructed energy for
the event and the colliding e c energy (EcMA). and p is the magnituiide of the
reconstructed total momentum for the event. Typical values for these cuts are
0.03 < AEiEcM < 0.02 and p < 150 MeV/Ic (taken from [19)).
The traditional AEp box cut is somewhat useful. However, it does not take
into account the correlation between the measured energy and momentum. Indeed,
the signal lies in diagonal bands in the AEp plane, making a boxsbaped cut not
optimal (see Figures 31a and 31b).
We use an alternative approach to the 4momentum cut. After a simple
substitution, Ey = Ay = ECM Eh+h (where E,. is the photon's energy, p, is
the magnitude of the photons momentum, ECM is the energy of the e6 system.
and Eh+h is the energy of the hadron pair), we can write the Ep conservation
equations as:
Ph +h + (EC  Eh+h )P, =P AM (3.1)
where, Pgh+h is the dihadron momentum, fr is the photon's momentum unit
vector, and i5CAM is the momentum of the &~c' systen (which is a few MeV because
of the crossing angle). Equation 3.1 is a 3constraint subset of the 4momentum
constraint and has the convenient property of avoiding the use of the measured
photon's energy, which has nonGaussian asymmetric errors. It is important to
notice that Equation 3.1 contains the dihadron energy, therefore, it can help
discriminate between the various particle hypotheses.
We proceed as follows. After vertexing the hadron pair using the beam spot
with its error matrix, we calculate the photon's direction from the hadron pair's
vertex and the shower position. We then fit the event to the 3 constraints expressed
in Equation (1) using the techniques outlined in [26] and cut on the X2 of the
3constraint fit, yX p(h) < 100. To complete the 4momentum requirement, we
calculate AE(h) = Eh+h + E,  ECM, where Eh+h is the updated dihadron
energy after the constraint, and E, is the measured photon's energy, and require
0.050 ECM < AE(h) < 0.025 FrM. Furthermore, we now have available
x2_P/h) differences between different particle hypotheses, which help in particle
identification (ID). This is discussed in more detail in the next section. Figure 31
compares the performance of the old and new approaches to the Ep cut.
At this point it is a good idea to check that the 4momentum cut, rejects
background events that make it through our skim cuts (see Section 3.1). These
events typically have one high energy shower, two tracks, and (an) additional
element(s). One such background is T(IS) , ,yr+r7r0. Out of 25000 T(1S)
with 71 y 7r+r MC events 4 only 4 survive our 4momentum cut. We conclude
that our 4momentum cut is good at rejecting background events that pass the
skim cuts but have additional elements such as an extra photon, 7r, pair of tracks,
etc.
Hadron Separation
We define AX D(h,  h2) for the particle hypotheses h, and h2 of our charged
track pair (e.g. AX ( K)) as follows,
4 Thanks to Vijay Potlia for generating these events
0.
0.
0.'
0.1
0.50 0.40
0.30 0.20 0.10 0.00
AE (GeV)
Figure 31: Distributions for different 4momentum cuts. Signal MC events
(T(1 S) , y7r+r) are represented by the black dots, and "background" M( events
(e+e   K+K) appear as (red) triangles. Plot a) has no cuts. Plot b)
has the old 4momentum boxcut. Plot c) has the new 4momentum cuts. Plot d)
has the new 4momentum cuts, and also a cut defined as X9 __(7r) X 2_K) < 0.
The particle ID potential of the newly available X 2_P(h) is evident.
ID(h, h2) adE/dX(h+) E/dX(h2) + dX(h ,dx(h
I d 2 dd; 1 d,
2 log(LRIcH (h)) + 2 log(LRIcH(h4))
(3.2)
2 tog(LRIC1 (h)) 4 2 log(C R CH(h2)).
where the idea is to combine the dE/dX and RICH information into one number.
Pairs of tracks with AXID(hi h2) < 0 are more likely to be of type hb than of typc
h2,
In practice, we only add a tracks RICH information if its momentum is above
the Cherenkov radiation threshold for both mass hypotheses and there are at least
3 photons within 3a of the Cherenkov angle for at least one of the mass hypothesis.
50Q I ... f ... I i. i
25 a)
25 b)
00 AA 1
25 0
A
25 d) .
Jil 1 1>f ALL:
0.
0.
0.
n3
0.10 0.20
We also require that both hypotheses were actually analyzed by RICH. during
pass2.
In addition to RICH and dE/dX, and as hinted in the previous section, the
difference in XEP(h) from the constraint expressed in Equation 3.1 can help the
particle ID (see Figures 31c and 31d). We define,
A(E_ hj  h2) X,, P(hi) (3.3)
Events with AXEP(hl h2) < 0 are more likely to be of type h., than of type h2.
In this analysis, to select h, and reject h2 the default cut is simply AXID(hl 
h2) < 0. This simple cut is highly efficient, has low fake rates., and is sometimes
sufficient. However, out of the six possible cases when one hadron fakes another,
there are three important cases where it pays off to also use AX 2(h h2)
together with AXID(h h2) in an optimal way:
1. 7r background to K. This background comes trom the continuum process
e+W +ip, p 7"f.
2. r background to p. Again, this background comes fiom the continuum
process ee ',p, p  iri
3. K background to p. This background comes from the continuum process
e yp K K
In other words, the important cases occur when the lighter mass hypothesis
fakes the heavier mass hypothesis.
Mathematically, one would expect that simply adding both AXD and AXE
together (like we just did when combining RICH and dE/dX), and cutting on the
grand AX 2 is the way to go. Unfortunately. this simple approach fails because of
large nonmathematical tails in the individual XEP distributions.
Instead, for each of these three cases we define the best cut values (cl. c2) in
A D < cl and AX _P < c2 as those that maximize
T'CC2 R (34)
where Re (Rf) is the efficiency (fake rate) of the particle ID cuts and W is the
rough ratio of the background to signal in the data sample for each case. Each W
can't be known a prwri, but a rough idea of its value can be obtained by doing
a first iteration of the analysis with, for example, c1 = c2 = 0. We use W = 20,
W = 60, and W = 30 for cases 13 respectively. Figure 32 shows F(c, c2) for each
case. Table 32 shows the optimized cut values and their effect on particle ID.
10 =
10
10 r
10 5 0 5 10
Cl
Figure 32: Contour plots of F(cl, c2) as defined in Equation 3.4 for different sig
nals and backgrounds from flat MC. Top has a K signal and a 7r background. Cen
ter has a p signal and a ir background, Bottom has a p signal and a K background.
Quantum Electrodynamic Suppression
Quantum Electrodynamic (QED) background in this analysis comes from the
abundant processes e+e* Ie+e and e+e'  y .
42
Table 32: Cut values, efficiencies, and fake rates for AXAD < cl and AX2_ < c2 in
fiat MC. The different c1, c2 values axe chosen so F from Equation 3.4 is maximun.
Efficiency and fake rates of each cut are reported relative to events in the skim.
Statistical errors in the efficiencies are 0.1% or less. Errors in the fake rates are
statisticall and are shown for completeness only.
CutValue Signal efficiency (%) Fake rate (%)
cl c2 AXeD AX Both AXeD AXEZP Both
w faking K 3 2 94.2 95.2 89.9 1.73 0.09 14.5 + 0.3 0.31 fr 0.04
7r faking p 1 2 99.0 99.8 98.9 0.89 0.06 1.03 + 0.07 0.03 0.01
K faking p 1 3 98.8 99.3 98.1 1.89 0.09 1.09 + 0.07 0.10 0.02
To reject e+e + e+e we require both tracks to have a matched shower with
energy E such that IE/p(7r)  0.951 > 0.1, and to have 'ZWD(h  e) < 0.
To reject e e ",/lA+p we simply require DPTIIMU < 5 (DPTHA LU
was defined in Section 2.2.5) for both tracks in the K and p modes because particle
ID cuts make the pion (and therefore the muon) fake rate small (see Tables 32
and 36). For the 7r mode we cannot use particle ID in a practical way because
muons and pions have similar masses. Instead, to separate pions and muons we
use a much stronger cut requiring that both tracks be within the barrel part of
the muon chamber (I cos(O)I < 0.7), both have P > 1 GeV/c and both have
DPTHMU < 5
To improve the overall muon suppression cut efficiency with virtually no
increase in muon fakes, we flag an event as "not muonic" if any of the tracks
deposit more than 600 MeV in the CC., This increases the cut efficiency by about
90% in the ir mode and makes the detector more hermetic.
5 Other analyses (for example [18]) typically use DPTHMU < 3, our CLEO III
MC has too many pion tracks with 3 < DPTHIMU < 5 and two few tracks with
DPTHMU = 0.
43
3.2.3 Cut Summary, Efficiencies, and Fake Rates
Table 33 summarizes the cuts used in this analysis. Figure 33 shows the
overall Monte Carlo efficiency after all cuts. Figure 34 shows the fake rates
according to the MC and the data for different particle ID cuts. The data fake
rates and limit fake rates, which are measured for pions and kaons faking other
hadrons, are calculated from the p and peaks in the continuum (see Figures 42
and 51). Tables 34 through 36 sununarize the results of Figures 33 and 34.
Table 33: Cuts used in this analysis.
Motivation Definition for T(1S)  h+h (where h = 7r, K, or p)
Skim cuts
Data acquisition Event must pass hardware (Levell) and software triggers
t (pass2) and be of type hardGarn
Topological There are only two good tracks and only one isolated shower
with E > 4 GeV
Analysis cuts
Reconstructed event must X2_p(h) < 100.0 and 0.050 < E(h)/EcM <0.025
'have 4momentum of the
centerofmass system
Hadron separation Default is, AXID < 0. The three cases with a large fake pop
ulation because of continuum processes use a simultaneously
optimized cut on AXID and AXEp, and are summarized in
Table 32
QED background Both tracks have a matched shower energy E that satisfies
e+e 7e+e IE/p(r) 0.951 > 0.1, and AxeD(h e) < 0
QED background For h K, p both tracks have DPTHMU < 5. For
e e yy / h = 7r (At least one track has a matched shower energy >
600 MeV) or (( Both tracks have cos(O) < 0.7 and P >
1 GeV/c) and (both tracks have DPTHMU < 5))
Table 34: Efficiencies in % for cuts (as outlined in Table 33) for flat signal MC.
Efficiencies in the second group are reported relative to the number of candidates
that make it to the skim. The third part of the table shows the overall reconstruc
tion efficiency. Statisticall errors are 0.1% or less.
Skim cuts 7r 1K _
Hardware Trigger 89.7 89.0 90.9
Software Trigger 96.5 96.2 96.6
hardGam 78.4 79.9 76.5
Topological (ntrarck, = 2 & Er > 4GeV) 73.3 67.6 72.3
Overall Skim efficiency 64.3 60.362.4
Analysis cuts 7r K p
4momentum 98.6 98.5 99.0
QED e+e  suppresion 93.9 87.4 93.1
QED p+p A+p suppresion 74.7 93.0 98.3
Hadron separation 97.1 89.0 97.8
Overall analysis efficiency 66;.9 79.1 89.1
Overall reconstruction efficiency 43 4 7.6] 55.6
'Table 35: Efficiencies in % for cuts (as outlined in Table 33) for flat 4S MC. Effi
ciencies in the second group are reported relative to the number of candidates that
make it to the skim. The third part of the table shows the overall reconstruction
efficiency. Statisticall errors are 0.1% or less.
Skim cuts 7I K[ p
Hardware Trigger 88.4 88.0 89.0
Software Trigger 94.6 94.6 94.8
hardGam 79.2 80.5 77.9
Topological (ntraks = 2 & E > 4GeV) 75.2 69.5 75.8
Overall Skim efficiency 66.9_ 6.7 66.6
Analysis cuts 7 r K p
4momentum { 97.1 97.0 97.4
QED e+e + ye e suppresion 93.4 89.1 .93.2
QED p~t 4 y p+p suppresion 75.8 92.5 98.1
Hadron separation 91.4 96.6
Overall analysis efficiency ]67.4] 80.2 87.3
Overall reconstruction efficiency i45.0] 5.3 58.2,
Table 36: Final efficiencies and fake rates after all cuts in %. Statistical errors
for efficiencies are 0.1% or less. Statistical errors for fake MC rates are shown for
completeness only. The IS DATA corresponds to the IS off resonance data sample.
The MC was generated flat.
cuts= K cuts p cuts
IS 7r MC 43.0 0.14 0.01 < 0.007
4S 7r MC 45.0 0.29 0.02 < 0.01
1S p DATA  < 1.2 < 0.2
4S p DATA 0.20 0.06 < 0.06
IS K MC 1,27 0.03 47.6 < 0.02
4S K MC 0.92 0.04 50.3 < 0.06
1S DATA < 3.8 < 2.0
4S DATA 4.14 0.69 < 0.4
IS p MC 0.08 0.02 0.34 0.04 55.6
4S p MC 0.05 0.01 0.37 0.01 58.2
1 2
Invariant mass (GeV/c2)
Figure 33: Final efficiencies for each mode as a function of invariant mass for the
iS (solid) and 4S (dashed) Monte Carlo data. The MC was generated with a flat
angular distribution.
0.50
0.25
0.00
W
o 0.025
0.000
0.050
0.000
. .. I
1TIT f kin K'K
I I
n'nt faking pop
K K fokin pop
II I
Invariant Mass of Fakes (GeV/c2)
Figure 34: Fake rates for IS (hollow circles) and 4S (hollow squares) according to
flat MC. The (red) downward pointing triangle is obtained using the p peak in 4S
data.
CHAPTER 4
EXCLUSIVE RADIATIVE DECAY T(1S) ,T+7r
In Figure 42 the 7r+r invariant mass plot is shown for both 1S and 4S data.
Figure 43 shows the continuum subtracted 7r+ r invariant mass distribution (as
defined in Section 3.2.1) with the most likely statistical fit overlayed (which is
described in the next section). The number of events within 11 of the p region
(0.62 0.92 GeV/c2) left after the continuum subtraction is 200 300, and 50 of
these belong to the fo(980) lowmass tail.
4.1. Robustness of The Mass Distribution
In [27] the decay T(1S)  77r' is analyzed, and it is shown how the analysis
stream warps the shape of the reconstructed resonance. This effect, which arises
because of the particular 7r' behavior, raises problems when fitting the invariant
mass distribution.
In Section 3.2.1 we claimed that if the data and MC had the same O.y and
0, distributions, the fit to the continuum subtracted invariant mass distribution
automatically gives the correct efficiency corrected number of events.
Here we test this claim. To this end, we generate 10000 T(1S) y f2(1270),
with f2(1270)  7r+ with flat 0, and 0,. distributions. We treat this MC
as data and carry out the first two steps of the continuum subtraction process.
The resulting peak has a mass of 1.278 0.002 GeV/c2 and a width of 0.193
0.006 GeV/c2, consistent with the generated mass and width of 1.275 GeV/c2 and
0.185 GeV/c2. More importantly, the number of reconstructed events from the fit is
10040 180, which is consistent with the number of generated events.
Figure 41 shows the reconstructed and efficiency corrected events, a fit to
them, and the generated events.
49
We conclude that there is no warping of the mass distribution, and that the
analysis stream behaves like we expect when obtaining the efficiency corrected
number of events from data.
N Reconstructed and Elf. Corrected
MC Generated
2400
2000 "
"u 1600 1
>
~12O
N
> 800
W
400
1.00 1.10 1.20 1.30 1.40 1.50
i*7T Invariant Mass
Figure 41: Reconstructed events and efficiency corrected events, a fit to them, and
the generated events for T(1S)  'yf2(1270), with f2(1270) 4 7r+r with flat 01,
and 0, distributions.
4.2 Statistical Fit of the Invariant Mass Distribution: Signal Areas and
Their Significance
The results of this section are summarized in Table 41.
Figure 43 shows possible signals for for T(1S) * yfo(980), T(1S)
yf2(1270) and T(1S) 'yf4(2050). To determine the number of events in each
signal we fit the invariant i+Tr mass continuum subtracted distribution with three
spindependent, relativistic BreitWigner line shapes. The masses and widths are
allowed to float, except for the width of the f4(2050) which has a very large error
if allowed to float and is set to its PDG value. The PDG values[16] for the mass
and width of the three resonances are mtfo(980) = 980 10 MeV/c2, fo(980)
70 + 15 MeV/c2, mf2(127,) = 1275.4 + 1.2 MeV/c2. rF2(1270) = 185.1+3.4 MeV/c2,
mf4(2050) = 2034 11 MeV/c2, and Pf4(2050) = 222 + 19 MeV/c2.
To measure the statistical significance of each signal we do multiple X2 fits
fixing the signal area to different values while letting the mass and width of
the signal whose significance is being measured to float within 2a of the PDG
values. At the same time the rest of the fit parameters are fixed to the values that
originally minimized the X2. We assign each of these multiple fits a probability
proportional to eX/2 and then normalize. We calculate the chance of the signal
being due to a random fluctuation by adding the normalized probabilities for the
fits with a negative or 0 signal. This method fails for the highly significant f2(1270)
signal because the e2/2 value of fits with negative or 0 f2(1270) signal is of the
order of 1054 and our computing software can only handle numbers as small as
1041. For completeness we state that the significance of this signal is < 10'5.
To measure the upper limit for the fj(2220) we also do multiple X2 fits
for different fixed signal values, while keeping its mass and width constant at
myfj(2220) = 2.234 GeV/c2 and Ffj(2220) 17 MeV/c2 as in [17]. The resulting
probability plot is shown in Figure 45.
We find clear evidence for the f2(1270), evidence for the fo(980) and weak
evidence for the f4(2050). We also put a 90% confidence level upper limit on
fj(2220) production. Fit results are shown in Table 41.
Table 41: Results for T(1S) * rtr. The branching fractions of f2(1270) and
f4(2050) are taken from the PDG [16]. Errors shown are satistical only.
Mode Area B.F. (105) Significance
yfo(980), fo(980) 7r+r 340, T6 8.3 x 106 (4.30
,yf2(1270) 1230 100 10.4 0.8 < 1045 (> 14a)
f4(2050) 85 30 3.6 1.3 5.2 x 101 (2.6u)
yfj(2220), fj(2220) * 7r+7r < 13 < 6.2 x 102 _
4.3 Angular Distribution of The Signal
The helicity angle distributions of 0, and Oy are defined and described in the
appendix. In this section we first obtain the helicity angular distributions of the
51
fo(980), f2(980). and f4(2050) and then fit them to the predictions of the helicity
formalism.
In practical terms, obtaining the helicity angle distribution of a particular
resonance from data consists of two steps. First, we choose an invariant mass
interval around the resonance peak to select events from the resonance and obtain
a helicity angular distribution which has both signal and background events.
Second, we subtract the contribution to the helicity angular distribution of the
background events in the chosen mass interval to obtain what we want; the helicity
angle distribution of the resonance.
Choosing the mass interval is not a trivial thing. If its too wide there will
be too much background, and if its to narrow there will not be enough signal.
To choose the optimum mass interval we need to know how much signal and
background we are selecting. Therefore, the two steps described in the previous
paragraph are related. How we deal with this is revealed in the next two sections.
4.3.1 Optimum Mass Interval
As described above, the first step in obtaining the helicity angle distribution
for a resonance is choosing an invariant mass interval to select events from such a
resonance. A standard 11' (which corresponds to 1.6a for a spin 0 BreitWigner)
cut around the mean mass of the resonance can be chosen as a "standard" interval.
We could proceed this way, but in our case because of the large subtractions
involved when obtaining the angular distribution, a considerable increase statistical
significance of each bin in the helicity angular distribution can be achieved by
choosing the mass interval carefully (see the last column of Table 42).
Let's consider for example the f0(980) angular distribution. We begin with
the T(1S) and the T(4S) efficiency corrected distributions. Before the continuum
subtraction each bin in the fo(980) angular distribution has contributions from
the high end mass tail of the p and the low mass tail of the f2(1270). In order to
52
get the final angular distribution, both of these contributions are taken away by
first doing a continuum subtraction using the scale factor a = 0.404, and then
by subtracting the f2(1270) distribution outside the mass interval scaled by an
appropriate scale factor /3 equal to the ratio of f2(1270) inside the mass interval
and outside of it 1 After all this, the contributions to the bins in the final angular
distribution are,
Nfo(980) := nfo(98o) + np(T(1S)) + 'ff2(1270) anp(T(4S)) /On'2(1270) hfo (98o) (4.1)
where nfo(980) (nf2(1270)) is the number of fo(980)(f2(1270)) obtained by integrating
the fitted spindependent, relativistic BreitWigner function inside the mass
interval, np(T(lS))(np(T(4S))) is the number of p's inside the mass interval from
continuum events at the T(1S)(T(4S)) energy, and n'f2 (270) is the number of
f2(1270) outside the mass interval being used to subtract the contribution of
f2(1270) to the fo(980) angular distribution. Following this last definition, =
nf2(1270)/nf2(1270).
Each of these terms has an associated error. Assuming that An v, that
the efficiency correction has infinite statistics, and ignoring the errors on the
continuum scale factor a, the overall error is,
ANf0(980) = ./nfo(980) + (a + 2)np(T(4S)) + (2 + 0 + /32)nf2(1270)
We arrive to the conclusion that the mass interval (Fn Am, fn + Am) which
produces the helicity angle distribution with smallest relative binerrors is the one
1 Strictly speaking this is correct only to first order. The f2(1270) distribution
itself has a small contamination from fo(980) and f4(2050). This effect, which we
call crosscontamination, is ignored in Equation 4.1. Later, in section 4.2.2, we will
show how to eliminate this small crosscontamination using all the resonances. The
difference in how we actually get the heicity distributions and Equation 4.1 has an
insignificant effect when calculating the optimal mass interval
that maximizes,
Ff(90)(, A) nf (980) (4.2)
F/')fo(980) + (a + a2)np(T(4S)) + (2 + 3 T f2)n (127o)
The plot of Ff0(980) (in, Am) is shown in figure 46.
This same technique can be applied to the f2(1270) and f4(2050) resonances.
Results are shown in Table 42.
Table 42: Mean masses, widths in GeV/ic2 and inverse of the average bin relative
error (F) from background subtractions for the angular distribution of different
resonances. Standard mean masses and widths, corresponding to 1f, are taken
from the fit in Figure 43, and are labeled with the subscript "s", while those that
maximize F are labeled with the subscript "m". The last column shows the factor
by which the effective statistics increase.
Resonance (f, Am,) F(m5, Am5) (fn., Amm) F(m, F m:A
fo(980) (0.970,0.070) 5.7 (0.985,0.060) 6.0 1.11
f2 (1270) (1.270,0.120) 18.3. (1.590,0.420) 20.6 1.27
f4(2050) (2.120, 0.220) 3.2 (2.240, 0.250) 3.4 1.13
4.3.2 Background Subtraction
The continuum subtraction of the helicity angle distribution is defined in
Section 2.1. The subtraction of the tails from other resonances requires a closer
look. Let us call the continuum subtracted helicity angle distribution of the events
in the fo(980), f2(1270), and f4(2050) mass intervals cf0(9O), Cf2(127o), and cf4(2o50)
respectively. Let us call the felicity angle distribution of the events that come
exclusively from the resonance we are trying to select in the same mass intervals
(that is, the true helicity angle distribution of the resonance) tfo(980), tf2(127o), and
tf4(205o) respectively. The continuum subtracted helicity angle distribution of a
resonance is being contaminated by the tails of the other resonances. Keeping this
in mind we write 2
Cfo(980) tfo(980) + Otf2(1270)
Cf2(1270) tfo(1270) + ytf0(980) + btf4(2050) (4.3)
Cf4(2050) ifo(2050) + dtf2(1270)
Where the small numbers /3, y 6, and E are the ratios of the number of events
from a resonance in the mass interval where the contamination is taking place to
the number of events from the same resonance in the mass interval used to select
it. Using the (mm, Am,,) values in Table 42 and the fit in Figure 43 we obtain
/3 8.5 x 103, = 6.0 x 102, 6 = 9.7 x 102, and e = 0.13.
To obtain the background subtracted helicity angular distributions we simply
invert the system of equations expressed in Equation 4.3. The solution can be
conveniently expressed as,
tf2(l27O) = (Cfo(127o) "YCfo(980) 6Cf4(2050))
tfo(980) = Cfo(980) Otf2(1270) (4.4)
tf4(2050) = Cfo(2050) Ftf2(1270)
As a check, if we ignore the second order terms the previous solution becomes,
2 Here the cross contamination between the fo(980) and the f4(2050) is ignored.
There is no mathematical problem in including these contamination terms, but the
system of equations would be more complicated than it needs to be since this type
of cross contamination is negligible.
tf2(1270) Cfo(1270) 'YCfo(980) bCf4(2050)
tf0(980) CfO(980) OCf2(1270)
tf4(2O50) =Cf0(2o50)  f(20
which is indeed is the solution when crosscontamination is ignored.
4.3.3 Statistical Fit of the Helicity Angular Distribution: J Assign
ments and their Probability Distribution
For each resonance, we fit the data to the simultaneous cos 01, and cos 0,
helicity angle distribution obtained from data using Equation 4.4 to the helicity
formalism prediction, Equations 2327. projected on each angle and folded in
opposite directions around their syrmnetry axis in order to show both distributions
on the same plot. The corresponding J value of the best fit for each resonance,
shown in Figures 48 through 410, is defined as the J assignment (J,). We obtain
J = 1 for the fo(980) which is inconsistent with the known spin of the fo(980)
which is J = 0. For the f2(1270) and the f4(2050) we obtain J = 2 and Ja = 4
respectively, which is consistent with. their known spins.
To have an idea of how well the angular distribution determines J among the
hypotheses J = 0, 1, 2, 3, 4, we do a statistical fit for each hypothesis and assign
each one a probability proportional to e(x'++2d.f)12 where d.o.f. are the degrees
of freedom in the fit. The resulting normalized probability distributions give an
idea of the assigned J significance and are shown in Figure 411. In particular,
this figure shows that the J, = I for the fo(980) inconsistency can be due to a
statistical fluctuation.
File r/userluiuLdeo3/ntp/data_50.ntp
ID IDB Symb Datefflme
1 1 1 040807/0140
1 2 1 040807/0140
98765 981 1 040807/0203
Area
9112.
2.1195E+04
569.1
Meau RM.S
0.8080 0.2895
0.7647 0.2072
1.227 0.7650
0 1 2 3
Mass (GeV/c )
Figure 42: Invariant mass of 7r+r for IS (top) and 4S (bottom) data. For the 4S
data we show a blowup of the mass region 1.5 3 GeV/c2 where the p* can be
seen.
57
MINUIT X' Fit to Plot 9&3
UIS*Gamma h h. masspipi axis
File: Generated internally
Plot Area Total/Fit 1544.1 / 1542.1
Func Area TotaliFit 1359.7 / 1359.7
2
x = 44.7 for 54 8 d.o.f.,
Errors Par.
Function 1:f0
MOD 435.70 1l
MEAN 0.96672 1.5
WIDTH 7.05219E02 3.2
* SPIN 0.00000E+00 .0
* THRESHOLD 0.28000 0
Function 2: f2
MOD 998.59 77
MEAN 1.2678 7,5
WIDTH 0.12353 1.4
*SPIN 2.0000 00
* THRESHOLD 0,28000 0.0
Function 3:A_4
MOD 10437 38
MEAN 2.1191 5.4
* WIDTH 022200 0.0
* SPIN 4,0000 0.0
* THRESHOLD 0.28000 00
abolic
5.4
714E02
912E 02
DOOEi 00
OOE+00
'96
905E .03
273E02
000E+00
OOOE400
.84
51 5E.02
)00E 00
OOOE+00
O00E00
Minos
167.0
1.2374E02
4.3455E02
0.0000E+00
0.0000E+00
78.12
7.1319E03
1.3459E02
0.0000E+00
0 0000E+00
38.76
5.8898E02
00000E+00
0.0000E+00
0.0000E+00
22JUL2004 11:51
Fit Status 3
E.D.M. 8.848E06
C,L.= 52.7%
177.3
1.81 60E02
4.4879E02
0.3000E+00
0.OOOOE+00
77.92
8.0978E03
1 5382E02
0,0000.E00
0.OOOOE+00
38.91
4 9096E02
0.OOOOE+00
0.0000E+00
0.OOOOE+00
300 1 1 1 1 1 1 1 f. f. I
0 1 2 3
Mass (GeV/c2)
Figure 43: Continuum subtracted (as defined in Section 3.2.1) invariant mass of
7+7r from T(1S)  '+.r.
I
cite: Gen rolt iyntenally
ID iD8 Symb
9 3 32
98765 981 1
967 5 982 2
DOeFTime
040806/1839
040806/1845
040806/18A5
mean RM S.
1.408 0.4996
1.416 0.4800
1.422 0.4836
1.5 2.0 2.5 3.0
Mass (GeVic )
Figure 44: Blow up of the invariant mass of 7r+7w in the f4(2050) mass region,
including the upper limit on the fj(2220) shown as a dashed line.
File: tmp mntbohrhserlluicleo3mnJitulsgammahthtemp.m"p
go IoB Symb DateTime Area
1 29 1 04080811845 1.000
1 25 1 04086/1845 o0,M01
Mean RMS
6. 94 5008
5,132 3650
0 25
Area
Figure 45: Normalized probability distribution for different fj(2220)  +7 r sig
nal areas. The shaded area spans 90% of the probability.
1010 Fr
50 Lk W 1, L 1i., .
950 975
600 F
400'
200
1290 1490 1690
1000 1025 1050
1890
1990 2090 2190 2290 2390
Mean Mass (MeV/c)
Figure 46: Contour plot of the inverse of the average relative binerror from back
ground subtractions in the fo(980) (top), f2(1270) (middle), and f4(2050) (bottom)
angular distribution.
MINUIT )2 Fit to Plot 9&82
Ul S*Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 606.18/60618
Func Area Tota/Fit 436.31 / 436.31
X 2= 25.6 for 20 1 d.o.f.,
Errors Parabolic
Function 1: JO
AREA 218.15 42.27
300 ,
200
100
z
Minos
 42.27
5AUG2004 23:06
Fit Status 3
E.O.M. 6.301E13
C.L.= 14.1%
+ 42.27
1.0 0.5 0.0 0.5 1.0
Icosel IcoseI
Figure 47: Angular distribution for the excess events in the fo(980) mass region.
The fit corresponds to J = 0.
MINUIT X2 Fit to Plot 9&82
UlS4Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 606.18/606.18
Func Area Total/Fit 478.89 / 478.89
2
6=,rs 15.2 for 20 2 d.o.f.,
rmrs Parabolic
Function 1:J1
AREA 239.45 39.31
e 1.32483E03 0,3852
Minos
39.33
0.OOOOE+00
5AUG2004 23:06
Fit Status 3
E.D.M 1.115E05
C.L.= 65.1%
+ 39.29
+ 0.3537
1.0 0.5 0.0 0.5 1.0
ICOS0t IcosO, l
Figure 48: Angular distribution for the excess events in the fo(980) mass region.
The fit corresponds to J = 1.
MINUIT X2 Fit to Plot 9&72
Ul S*Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 1673.4/ 1673.4
Func Area Total/Fit 17338/1733.8
X2= 27.3 for 20 3 d.o.f.,
Errors Parabolic
Function 1:J2
AREA 866.97 43.10
E 4.25267E04 + 7.7843E02
0.29838 0.1098
Minos
45.85
0.OOOOE+00
0.1478
5AUG2004 23:06
Fit Status 3
E.D.M. 9.701 E06
C.L.= 5.4%
+ 46.07
+ 0.1238
+ 9.9862E02
, 200
0
0
o
_%
0
1.0 0.5 0.0 0.5 1.0
IcosO J IcoseI
Figure 49: Angular distribution for the excess events in the f2(1270) mass region.
The fit corresponds to J = 2.
MINUIT X,2 Fit to Plot 9&92
U1 SGamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 15993 / 159.93
Func Area Total/Fit 127.08 / 127.08
2
X2= 19.6 for 20 3 d.o.f.,
Errors Parabolic
Function 1:J4
AREA 63.528 1.408
0 1.5697 5.8174E02
4) 0.16919 0.6887.
Minos
14.91
0.3398
0,OOOOE+00
5AUG2004 23:06
Fit Status 3
E.D.M. 5.300E07
C.L.= 29.3%
+ 0.OOOOE+00
+ 0.0000E+00
+ 0.O00E+00
. I i I I I I I I I I 1 6i
1* TTiF
0.5
Icos0eI
0.5
Icosl tI
Figure 410: Angular distribution for the excess events in the f4(2250) mass region.
The fit corresponds to J = 4.
40 F
dI
40
1.0
.... I
0
0.8
20.4
I 0.0
0.8
0.4
0.0
0 2 4
Figure 411.: J, probability distribution for resonances in the fo (980) (top),
f2(1270) middlele, and f4(2050) (bottom) invariant mass region when only the
hypotheses Ja = 0, 1, 2, 3, 4 are considered.
CHAPTER 5
EXCLUSIVE RADIATIVE DECAY T(1S) K+K
In figure 51 the K+K invariant mass plot is shown for both resonance and
continuum running. This figure also has an inset showing the UdE/dX(K) for both
tracks for events in the 1.1 3 GeV/c2 region. This inset indicates that most
of the events in this mass region have indeed two kaons and that the amount of
p reflection is small. Furthermore, when the 7r+7 invariant mass is plotted for
these events only 40 out of 700 can be fit under a p peak (this fit is what we use to
calculate the pion faking kaon rate in Table 36).
Figure 52 shows our fit to the K+K invariant mass continuum subtracted
plot as defined is Section 3.2.1 with the most likely statistical fit overlayed (which
is described in the next section). The number of events near the 0 region (1.01 
1.03 GeV/c2) left after the continuum subtraction is 50 70.
5.1 Statistical Fit of the Invariant Mass Distribution: Signal Areas and
Their Significance
The results of this section are summarized in Table 51.
From the measurement of the previous section we expect a small contribution
( 50 events assuming no interference) from f2(1270) + K+K. We do find
some evidence for f2(1270)  K+K events (110 40) in the fit. We also find
strong evidence for the resonance f2(1525) and weak evidence for the fo(1710)
resonance. The f2(1270), and fo(1710) are fitted with their widths fixed to their
PDG values[16] because they have large errors if allowed to float. The rest of
the resonances parameters are consistent with their PDG values[16], which are
mf2(1525) = 1525 5 MeV/c,, Ff2(1525) = 76 110 MeV/c2, nif,(1710) = 1715 +
67
6 MeV/c2, and Ff(1710) = 125 10 MeV/c2. We also observe an excess of events in
the 2 3 GeV/c2 region which we can't attribute to any known resonances.
Significances of the signals of the identified resonances in the fit are calculated
as described in the previous section. The T(1S)  'fj(2220), fj(2220) K+K
upper limit is also calculated using multiple fits, except that this time the events
under the fj(2220) are of unknown origin, so we use a first order polynomial
allowed to float. The significance of the excess of events in the 2 3 GeV/c2
invariant mass region is calculated assuming a normal distribution; we simply add
up the number of events in each bin along with its error. Results are shown in
Table 51.
Table 51: Results for T(1S) + K+K
Mode Area Branching Fraction (105) Significance
7f2(1270) 110 40 23 8 5.4 x 104(3.3y)
f(1525) 360+8o7 3'.9 +09 < 1045(> 1a
yfo(1710), f0(1710) K+K 75 30 0.35 0.14 7.5 X 104(3.2,7)
,K+K (2 3 GeV/c2) 220 20 1.03 0.12 8.8a
yfj(2220), fj(2220) K+K < 10 < 5 x 102
5.2 Angular Distribution of The Signal
In this section we adapt the ideas presented in Section 3.4.3 to the T(1S) 
yK+K situation.
The derived statistical errors from the signal and background subtractions can
be used to calculate the mass interval which best represents the helicity angular
distribution. The inverse of the expected average relative bin error as a function of
the mass interval is shown in Figure 55, and the mass interval that maximizes it
are tabulated in Table 52.
The tails from the resonances contribute to the continuum subtracted helicity
distributions,
68
Table 52: Mean masses, widths inmass and inverse of the average bin relative
error (F) from background subtractions for the angular distribution of different
resonances. Standard mean masses and widths, corresponding to if, are taken
from the fit in Figure 52, and are labeled with the subscript "s", while those that
maximize F are labeled with the subscript "m". The last column shows the factor
by which the effective statistics increase.
Resonance (in,, Am,) F(m,, Am,) I (rnh, Amm) F(mm, Amm) F(m'ZmL
f2(1270) (1.276,0.185) 3.5 (1.300,0.100) 4.1 1.37
f2(1525) (1.540,0.085) 9.6 (1.565,0.100) 9.8 1.04
fo(1710) (1.760,0. 125) 3.1 (1.780,0.095) 3.3 1.13
Cf2(1270) tf2(1270) + Otf2(1525)
Cf2(1525) tf2'(1525) d Ytf2(1270) + btfo(1710) (5.1)
Cfo(1710) tf]o(110)  tfI(i525)
Where again the small numbers 3, y, 3, and c are the ratios of the number
of events from a resonance in the mass interval where the contamination is taking
place to the number of events from the same resonance in the mass interval used
to select it. Using the (mm, Am,,) values in Table 52 and the fit in Figure 52 we
obtain /3 = 1.3 x 102, y = 0.14, 3 = 0.11, and E = 0.19.
The background subtracted helicity angular distributions are,
tf2(1525) yb_(cf2(1525) 'YCf2(1270) 6Cfo(1710))
tf"(1270) =C(1270) Otf2'(1525) (5.2)
tf(=7(0) C,(17) t(1525)
The the best fit for each resonance and the excess of events in the 23 GeV/c2
are shown in Figures 56 through 59. The best spin assignment for the f2(1270) is
Ja = 2, for the f (1525) it is Ja = 2, for the it is fo(1710), and it is J = 1 for the
excess of events in the 2 3 GeV/c2 mass region. The J = 2 value for the fo(1710)
69
is inconsistent with its known spin. Also, examination of the normalized helicity
amplitudes for the f2(1270) reveals that they are inconsistent with those obtained
for the f2(1270) in the 7r+7r mode.
The assigned J probability distributions are shown in Figure 511. They reveal
that the inconsistencies in the f2(1270) and fo(1710) are not significant and can be
attributed to the statistical uncertainty.
File: "/bohr/userl fluis/de3/ntp/data_50.ntp
ID IDB Symb Date/Time
1 1 1 040807/0219
1 2 1 040807/02 19
1 10 1 040807/0235
Area Mean R.M S.
1274. 1.320 04283
2031. 1.156 02585
705.0 0.2244 1.208
01480 1.170
600 I
400
200
1200 _
6
800 .2
6 2 2 6
a~d(K )
400
0
0.9 1.4 1.9 2.4 2.9
Mass (GeV/c )
Figure 51: Invariant mass of K+K for IS (top) and 4S (bottom) data. For
the 4S data the inset shows the O'dE/dX (K) for both tracks for events in the
1.1 3 GeV/c2 mass region. This inset indicates that most of the events are con
stant with having two kaons.
MINUIT )? Fit to Plot 9&3
U1S4Garnmah h. masskk axis
File: Generated internally
Plot Area Total/Fit 794.78 / 575.02
Func Area Total/Fit 494.50 / 429.38
2
x = 37.5 for 40 7 d.o.f.,
Errors Para
Function 1: f_2(1270)
AREA 197.37 64.
MEAN 1.2757 3.31
*WIDTH 0.18510 + 00(
* SPIN 2.0000 + 0.
* THRESHOLD 0.98000 0.0
Function 2: f2'(1510)
MOD 318.76 64.
MEAN 1.5384 7.8
WIDTH 8.54208E02 2.7
* SPIN 2.0000 + 0.04
*THRESHOLD 0.98800 + 00
Function 3: f0(1710)
MOD 96.117 + 38.
MEAN 1.7622 2.2
* WIDTH 0.12500 0.04
* SPIN 0.00000E+00 004
* THRESHOLD 0.98800 0.04
15
bolic
92
I89EQ2
)00E00
)OOE+00
61
)67E 03
982E02
)OOE+00
00E+00
27
476E02
000E+00
)OOE+00
)OOE+00
Minos
64.93
8.9556E02
0.OOOOE+00
0.0000E+00
0.0000E+00
59.53
7.5447E03
2.2981 E02
0.0000E+00
0.0000E+00
38.70
2.0985E02
0.0000E+00
0.0000E+00
0.0000E+00
7AUG2004 14:23
Fit Status 3
E.D.M. 1.627E06
C.L.= 26.9%
64.91
3.0045E02
0.OOOOE+00
0.0000E+00
0.0000E+00
7290
8.8799E03
3.6400E02
0.0000E+00
0.0000E+00
37.77
2.6333E02
0.0000E+00
0.OOOOE+00
0.OOOOE+00
0.9 1.4 1.9 2.4 2.9
Mass (GeVic2)
Figure 52: Continuum subtracted invariant mass of K+K.
01
10
0 El
T
File: Generated internally
ID IDB Symb
9 3 32
98765 981 1
98765 982 2
1.80
Date/Time
040807/1422
040807/1426
040807/1426
2.00 2.20 2.40
Mass (GeV/c2)
Figure 53: Blow up of the fj(2220) region, with the 90% CL upper limit over
laid. The mass and width are taken to be mfj(2220) = 2.234 GeV/c2 and
Ffj'(2220) = 17 MeV/c2 as in [17].
Mean R.MS.
1.739 0.4861
2.086 0.5922
2.091 0.5835
2.60
File: /tmp mntvbohr/userlAuis/cdeo mn ftiulsgammahh'tern'.tmu
ID IDB Symb DatefTme Area
1 29 1 040807/1426 1.000
1 25 1 040807/1426 0.9054
R.M.S,
3.582
2.723
0 25
Area
Figure 54: Normalized probability distribution for different fj(2220) K+K
signal areas. The shaded area spans 90% of the probability.
1200 1250 1300 1350 1400
150 1
100 lo
1500 1550 1600 1650 1700
100
50 L 1 i i
1700 1750 1800 1850 1900
Mean Mass (MeV/c2)
Figure 55: Contour plot of the inverse of the average relative binerror from back
ground subtractions in the f2(1270) (top), f2(1525) (middle), and fo(1710) (bot
tom) angular distributions.
MINUIT X2 Fit to Plot 9&82
UlSGamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area TotaVFit 517.49 / 517.49
Func Area Total/Fit 129 01/129.01
X2= 17.8 for 20 3 d.o.f.,
Errors Parabo!lic
Function 1:J2
AREA 64505 17.78
e 7.95385E03 1.347
D 1.4108 0.4602
Minos
17.78
0.OOE+00
 0.2689
5AUG2004 23.16
Fit Status 3
E.D.M 5.475E05
C.L.= 40.4%
17.78
0.5745
0.0000E400
. I  I I I I I I I I
II ~
I I I, ,
I I I I I I
0.5
IcosO)
. I
. I I I
0.5
IcosOOeI
Figure 56: Background subtracted K+K angular distribution in the f2(1270)
mass region as defined in the text. The fit corresponds to J = 2.
. I
I
1IT
I I
MINUIT X2 Fit to Plot 9&72
U1S4Garnma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area TotaV/Fit 284.44 284.44
Func Area Total/Fit 367.87 / 36787
X2= 20.9 for 20 3 o.o.f.,
Errors Parabolic
Function 1:J2
AREA 183.94 t 18.70
0 6.76443E04 + 0.1885
0.50343 + 0.1298
o 50
"0
Minos
19.29
0.OOOOE+00
0.1593
5AUG2004 23:16
Fit Status 3
E.D.M. 3.771 E06
C.L.= 23.3%
+ 19.25
+ 0.2757
+ 0.1248
1.0 0.5 0.0 0.5
IcosO) tcosO"t
Figure 57: Background subtracted K K angular distribution
mass region as defined in the text. The fit corresponds to J = 2.
a>
I
1.0
in the .f (1525)
MINUIT X2 Fit to Plot 9&92
UlS4Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 129.47 / 129.47
Func Area Total/Fit 79.031 /79.031
X2= 28.0 for 20 3 d.o.f.,
Errors Parabolic
Function I: J2
AREA 39.516 10.80
8 2.99860E04 0.3847
1.5696 0.2633
Minos
10.81
0.0000E+00
0 3060
5AUG2004 23:16
Fit Status 3
E.D.M. 1 721E05
C.L.= 4.5%
10.80
+ 0.4645
+ .O.0E+00
25
0
1.0 0.5 0.0 0.5 1.0
IcoseO4 Icosel
Figure 58: Background subtracted K+K angular distribution in the fo(1710)
mass region as defined in the text. The fit corresponds to J = 2.
MINUIT X' Fit to Plot 9& 102
U1SGamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area TotaVFit 411.73 i 411.73
Func Area 7otal/Fit 370.86 / 370.86
2
x = 18.9 for 20 2 d.o.f.,
Errors Parabolic
Function 1:J1
AREA 165b43 17.19
G 015542 + 0.3003
100
0
Figure 59: K+K
Minos
17.19
0.0000E00
5AUG2004 23:16
Fit Status 3
E.D.M. 4.735E09
C.L.= 40.0%
+ 17.18
+ 0.1921
0.5 0.0 0.5 1.0
Icosel lcosej
angular distribution for events within the 2 3 GeV/c2.
MINUIT X2 Fit to Plot 9&92
U1 S*Garnma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 129.47 / 129.47
Func Area Total/Fit 62.194 / 62.194
2
m, 33.3 for 20 1 d.o.f., Parabolic
Function 1: JO
AREA 31.097 + 0.92
50 ,
5AUG2004 23:16
Fit Status 3
E.D.M. 3.484E15
C.L.= 2.2%
Minos
10.92 + 10.92
1.0 0.5 0.0 0.5 1.0
IcosO7 lcosOI
Figure 510: Background subtracted K+K angular distribution in the fo(1710)
mass region as defined in the text. The fit corresponds to J = 0.
79
1.2 I
0.8
0.4 
1 0.0
0.8
0.4
1 0.0 _
0.8
0.4I
1 0.0
0.8
0.4
0.0 
0 2 4
Ja
Figure 511: Ja probability distribution for resonances in the f2(1270)(top),
f2(1525) (middletop), fo(1710) (middlebottom) invariant mass region and the
excess of events in the 2 3 GeV/c2 (bottom). region when only the hypotheses Ja
= 0, 1, 2, 3, 4 are considered.
CHAPTER 6
EXCLUSIVE RADIATIVE DECAY T(1S) pp
In figure 61 the pp5 invariant mass plot is shown for both IS and 4S data, with
an inset showing that most of the events indeed have a proton and an antiproton
and that the p and 6 reflections are small. The enhancement at 3.1 GeV/c2 in
the 4S plot corresponds to the process e+e  7yJ/0 with J/ * pp. This
enhancement is not as pronounced in the IS plot because the IS data only has
32.4% of the luminosity of the 4S data. The number of events in the J/ invariant
mass region after continuum subtraction is 6 3 and is consistent with 0.
Figure 62 shows the continuum subtracted invariant ppi mass distribution
(as defined in Section 3.2.1) with a 90% confidence level upper limit for fj(2220)
overlaid. A direct fit to the fj(2220) yields 12 5 events.
There is an excess of events in the continuum subtracted invariant mass plot.
We measure this excess and the upper limit of e+e 4 yfj(2220), fj(2220)  pp
the same way we measured the excess of events inside 2 GeV/c2 < m(K+K) <
3 GeV/c2 region and the fj(2220) upper limit in Section 5.1. Results are shown in
Table 61.
The pp angular distribution for the mass range 2 GeV/c2< m(pp) < 3 GeV/c2
is shown in Figure 64 with the most likely J assignment, J = 1, fit overlaved.
The probability distribution for Ja is shown in Figure 65.
Table 61: Results for T(1S) 'p!.
Mode Area B.F. or 90 % U.L. (105) Significance
ypp (2 3 GeV/c2) 85 18 0.41 0.08 4.85u
yfj(2220), fj(2220) + pp < 20 < 9 X 102
Rle: *Abohr/user1Auis/deo3/ntp/data_50ntp
ID IDB Symb Date,'Time
1 1 1 040807/0303
1 2 1 040807/0303
1 10 1 040807/0304
2.0 2.2 2.4 2.6
Mass (GeV/c2)
Area Mean R.M.S.
104.0 2.334 0.3628
141.0 2.187 0.3419
139.0 0.2583 1.259
0.1224 1.119
2.8 3.0 3.2
Figure 61: Invariant mass of p3 for IS (top) and 4S (bottom) running. For the 4S
running the inset is consistent with the events having a proton and an antiproton.
Fle: Generated internally
ID IDB Symb
9 3 32
98765 981 2
98765 982 1
Date/Time
040807/1446
040807/1446
040807/1446
1.8 2.0 2.2 2.4 2.6
Mass (GeV/c2)
Mean R.M.S.
2.521 0.3093
2.685 0.3291
2.813 0.2553
2.8 3.0 3.2
Figure 62: Invariant mass of pp. The plot is continuum subtracted and efficiency
corrected. An overlay with the 90% confidence level upper limit for fj(2220)
is shown. The mass and width are taken to be mrf,(2220) 2.234 GeV/c2 and
Ffj(222o) = 17 MeV/c2 as in [17].
File: tmpmntroohr/userlAuis/ldeo3/mnjitlulsgammahh/temp.tmp
ID IDB Symb Date/Time Area
1 29 1 040807/1446 1.000
1 25 1 040807/1446 0.9110
Mean R.M.S.
12.78 5.430
11 80 4.590
0 25
Area
Figure 63: Normalized probability distribution for different fj(2220) * pp signal
areas. The shaded area spans 90% of the probability.
MINUIT X2 Fit to Plot 9&22
UlS,Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 256.69 / 256.69
Func Area Total/Fit 134.57 / 134.57
X2= 19.5 for 20 2 d.o.f.,
Errors Parabolic
Function 1:J1
AREA 67.287 10.91
0 1.20611E03 0.4148
25
0
C,
Minos
10.93
0,0000E+00
2AUG2004 18:12
Fit Status 3
E.D.M. 1.845E05
C.L.= 36.4%
+ 10.88
+ 0.3544
1.0 0.5 0.0 0.5
IcoseY Icos0"I
Figure 64: pp angular distribution
2 GeV/c2< m(pp) < 3 GeV/c2.
for the excess of events in the mass range
1.0
0.8
0.6
0.4
0.2
1 0.0
0
2 4
Assigned J
Figure 65: J, probability distribution for the excess of events in the mass range
2 GeV/c2< m(pp) < 3 GeV/c2 when only the hypotheses Ja = 0, 1. 2, 3, 4 are
considered.
CHAPTER 7
SYSTEMATIC UNCERTAINTIES
Systematic uncertainties are any sources of experimental uncertainty other
than the statistical ones. Limits on the accuracy of our detector simulation and any
physical processes that interferes with the experimental measurement are typical
examples of systematic uncertainties. These uncertainties need to be identified,
quantified, and if possible, corrected.
7.1 Cuts
From now on we report individual cut efficiencies relative to the events that
survive all the other cuts.
According to our MC, most of the skim cuts, except for the hardGam re
quirement, are nearly 100% efficient. Therefore, such cuts should not be a source
of systematic uncertainty. The only skim cut worth taking a closer look at is the
hardGam cut, which is about 93% efficient in MC. Measuring this efficiency in our
data is not possible because events not classified as hardGam are not in the data
to begin with. Closer examination of our MC reveals that that most of the 7%
inefficiency in hardGam comes from the eOverP1 cut (4%), some from the Sh2 cut
(2%), and the rest (1%) from the other cuts present in hardGam. We can measure
the efficiency of eOverP1 in data by looking at the eOverP2 distribution of data
tracks from p and 0 decay. We are satisfied by this check on the MC modeling (see
Tables 71 and 72), and we won't measure how well MC models the rest of the
hardGam cuts, which are 97% efficient in MC.
To quantify the quality in the MC modeling of our analysis cuts in the 7r
and K modes we use the p and signals present in our data' In this case the
process is straightforward. We measure the p and 0 signal signals in data and MC
over a floating background function with all cuts in place, and with all in place
cuts except the one under consideration. From these numbers we calculate the
effective efficiency of the cut. The differences between data and MC are taken as
the systematic errors which are added in quadrature. Results are shown in Tables.
71 72. The 4momentum cut does not appear because its efficiency is close to
100%.
Cuts MC eff. Data eff. Systematic Error
eOverP2 96.2 96.6 0.4
QED e+e  ye+e suppresion 93.1 94.2 1.2
QED M+p + ,p+p suppresion 75.3 77.7 3.2
Hadron separation 96.9 96.8 0.1
Overall analysis cut systematic error [ I 3
Table 71: Efficiencies for T(1S)  yr+r for flat signal MC, efficiencies from
data (p), and the derived systematic error in %. Efficiencies are reported as the
number of signal events after all cuts divided by the number of signal events with
all cuts except the one under consideration. Statistically errors are 0.1% or less.
For the proton case we don't have a clean sample with high statistics of
e+e + yop5 events in data. By extension we take the systematic error in this mode
to be 10%.
7.1.1 Justification of the DPTHMU Cut
The reason we prefer to use DPTHMU < 5, instead of the more traditional
(see [18]) DPTHMU < 3 used in CLEO II, is that, for some unknown reason, our
CLEO III MC has too many 7r tracks with 3 < DPTHMU < 5.
1 A study using K or A signals from hadronic environments would have larger
statistics, but is problematic because the large number of tracks and showers artifi
cially decrease the cut efficiency. See [28] Appendix A.3 for an example.
Cuts MC Eff. Data Eff. Systematic Error
eOverP2 98.5 100 1.5
QED e+e  e+e suppresion 98.1 99.3 1.3
QED p+ly * / suppresion 93.5 96.4 3.1
Hadron separation 88.4 82.2 7.0
Overall analysis cut systematic error 8
Table 72: Efficiencies for T(1S) * yK+K for flat signal MC, efficiencies from
data (0), and the derived systematic error in %. Efficiencies are reported as the
number of signal events after all cuts divided by the number of signal events with
all cuts except the one under consideration. Statisticall errors are 0.1% or less.
To observe this fact we first select a relatively clean sample of pion tracks
by requiring the event to have a 7r+ invariant mass consistent with the p mass,
and to pass all our analysis cuts except for the cut on DPTHMU on one track.
For such events we plot the rate as a function of momentum, at which the 7r track
whose DPTHMU cut we released has DPTHMU > 3 and DPTHMU > 5. To
increase our statistics we do this procedure twice, once for each track, and average
the fake rate. Figure 71 shows the results. Clearly the CLEO III MC we are using
has some problem modeling the DPTHMU < 3 cut.
To keep the systematic error low we choose a cut at DPTHMU < 5
(DPTHMU < 3 gives a systematic error of about 20%). This does not change the
efficiency in data very much, but it increases the efficiency r,, ted by MC, bring
ing it closer to reality. The increase in IL fakes after loosening the cut is estimated
to be low using QED MC (see Figure 72).
7.2 Angular Distribution of Signal
The photon and tracks from the process T(1S)  'X with X  h h have
a different angular distribution than that of flat MC. Examples of possible angular
distributions are shown in the appendix.
89
File: /bohr/userlAuis/deo3Mtp/data50.ntp
ID IDB Symb Date/Time Area Mean R.M.S.
1 4 31 040525/1408 0.2350 2.758 0.8547
1 3 31 040525/1408 0.4317 2.684 0.8088
1 6 31 040525/1408 9.679E02 2.996 0.7400
1 5 31 040525/1408 6.0516E02 2.792 0.6995
0.20
0.10
0.00
0.040
0.020 
0.000
I I I
0 1.6 3.2
Track Momenta (GeV)
Figure 71: Pion faking muon fake rates for a cut on DPTHMU < 3 (top) and
DPTHMU < 5 (bottom). MC is shown as solid circles while p from data is shown
as hollow circles. Fake rates are reported relative to all events that pass all cuts,
except for the DPTHMU cut for one of the tracks.
Figure 73 shows the efficiency in flat MC as a function of cos 0. and
cos Oh+' Note that the K and p modes are nearly insensitive to the track an
gular distribution, while the 7 mode is more sensitive to Oh+. This happens because
of the stronger muon rejection cut in the 7r mode.
We measure the systematic effects of flat MC efficiency by convoluting each
plot in Figure 73 with different possible angular distributions calculated in the
2 0, and Oh+ are the helicity angles of the sequential decay, defined in the ap
pendix (see Figure 2).
File: r/u erluisde 3 tp/mc s2 mumu ga ma new. n p
ID IDB Symb Date/Time Area
1 3 31 040518/1402 3.676
1 30 32 040518/1402 3.022
Mean R.M.S.
2.361 0.9088
2.549 0.8590
Track Momenta (GeV)
Figure 72: Muon faking pion fake rate (in %) for DPTHMU
and DPTHMU < 5 (solid circles). There is a complete overlap
< 3 (solid squares)
in the last bin.
appendix. Tables 7374 show the necessary correction factors relative to flat MC
efficiency for decays with definite 7 and X helicities A. and Ax, due to the nonflat
photon and hadron distributions. We call these factors E'X~x(,) and Ex (h)
respectively.
The efficiency of a decay with definite A,, Ax can be obtained using the flat
MC efficiency corrected by a factor JX X= E 'x (h). In general, the
=\'A~X (7) X E,\Yx () ngnrl h
final state is a mixture of all possible Aly, Ax pairs, and the efficiency correction
factor is,
iJx = cos2 9 cos2 (DE J + sin2 E)EJ< + cos2 E) sin2 (D(.1
I I I
(7.1)
U U9
0 0.40
LL]
0.0001
0.00 L
1 0 1
cosOe
0.40 hi i 0
0.80
UOS L 1 7
0.60 r
0.40K
0.20 !
0.00 0 1
cosO,"
0.60
0.40
0.20
1 0 1
cosO.
Figure 73: Flat MC efficiency as a function of cos9, (left column) and as a func
tion of CoSOh+ (right column). Top row corresponds to the pion mode, middle row
to the kaon mode and bottom row to the proton mode.
The fits in Figures 48 and 410 and 5658 measure the pair ((3, 4)). These
values are summarized in Table 75, where they are used to obtain Ex for each
mode.
The pair ((, 4)) carry an error which is a source of systematic uncertainty.
We calculate this systematic uncertainty by inspecting the differences in efficiency
when (6, 45) move away from the value which gives the minimum chisquared, 2
alo or a12
E=0
all
7 t 0.910 1.18
K 0.925 1.15
p 0.922 1.16
Table 73: ExJX (7), efficienciy correction with respect to flat MC factors due to
the nonfiat angular distribution of the photon in T(1S) + yX for different X spin
values (Jx) and y, X helicities (Ay, Ax).
Mode Jx alo al a12
0=0,4)=0 19=1 0
0 1.00 
1 0.785 1.11 
X > 7+7 2 0.829 0.934 1.15
3 0.898 0.853 1.03
4 0.899 0.863 0.922
0 1.00 
1 0.944 1.03 
X 4 K+K 2 0.950 0.988 1.04
3 0.957 0.971 1.01
4 0.965 0.962 0.992
Table 74: & xx (h), efficienciy correction factors with respect to nonfiat MC due
to the track angular distribution in T(1S) * yX, with X * h+h for different X
spin values (Jx) and y, X helicities (Ay, Ax).
under the condition X 2 < X2 + 1 For Jx > 1 both (, (D) are free to move,
defining a surface in the (E, 4) plane. These surfaces are shown in Figure 74 for
the f2(1270), f4(2050), and f2(1525).
At this point we can check whether (0, 1) depend on the mass of the de
cay. We split the f2(1270) and the f2(1525) into a high mass and a low mass
region. The plots of the error surfaces of the measured (), 1') show no significant
separation for the different mass regions (see Figures 75 and 76).
3 In a two dimensional linear problem such a set of points defines the surface of
the standard error ellipse.
Results for the correction factor and its systematic error are shown in Table 7
5.
Upper limits on T(1S) * yf2(2220), f2(2220) * h+h are changed to
include the angular distribution's effect on efficiency. Since we can't measure
the helicity amplitudes in this case, we choose the worst possible case where the
corrected efficiency is lowest. This always corresponds to 8 =D 0. Results are in
Table 76.
1.5
0E f2(1270)
E f4(2050)
[O f2'(1525)
1.0 
0.5
0.0 I I
0.0 0.5 1.0 1.5
9 (radians)
Figure 74: Surfaces in the (8, D) plane used to determine the systematic uncer
tainty in the efficiency correction factor for the.modes with f2(1270), f4(2050),
and f2(1525). The f4(2050) surface may seem large, but when drawn in spherical
coordinates it is a small "north pole cap".
0.0 0.5 1.0 1.5
0 (radians)
Figure 75: Measurecd (9, D) surfaces for the f2(1270) high and low mass regions.

Full Text 
xml version 1.0 encoding UTF8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchemainstance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E8U957T5A_HFSDD0 INGEST_TIME 20140724T23:50:48Z PACKAGE AA00022267_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
PAGE 1
'(&$<6 2) 7+( 76f ,172 $ 3+2721 $1' 7:2 &+$5*(' +$'5216 %\ /8,6 %5(9$1(:(// $ ',66(57$7,21 35(6(17(' 72 7+( *5$'8$7( 6&+22/ 2) 7 +( 81,9(56,7< 2) )/25,'$ ,1 3$57,$/ )8/),//0(17 2) 7+( 5(48,5(0(176 )25 7+( '(*5(( 2) '2&725 2) 3+,/2623+< 81,9(56,7< 2) )/25,'$
PAGE 2
7R P\ 3DUHQWV
PAGE 3
$&.12:/('*0(176 RZH P\ JUDWLWXGH WR PDQ\ SHRSOH ZLWKRXW ZKRP WKLV ZRUN ZRXOG QRW KDYH EHHQ SRVVLEOH )LUVW RI DOO ZRXOG OLNH WR WKDQN P\ VXSHUYLVRU 'U
PAGE 4
7$%/( 2) &217(176 SDJH $&.12:/('*0(176 L $%675$&7 YL &+$37(5 7+(25< 3DUWLFOH 3K\VLFV 7KH 6WDQGDUG 0RGHO 4XDQWXP &KURPRG\QDPLFV ,QWURGXFWLRQ WR WKH 5DGLDWLYH 'HFD\V RI WKH 76f 5DGLDWLYH 'HFD\V RI 4XDUNRQLD 2YHUYLHZ (;3(5,0(17$/ $33$5$786 7KH &RUQHOO (OHFWURQ 6WRUDJH 5LQJ 7KH &/(2 ,,, 'HWHFWRU 6XSHUFRQGXFWLQJ &RLO 7UDFNLQJ 6\VWHP &U\VWDO &DORULPHWHU 5LQJ ,PDJLQJ &KHUHQNRY 'HWHFWRU 0XRQ 'HWHFWRUV $1$/<6,6 2) 7+( '$7$ 'DWD 6DPSOH &RQWLQXXP 6XEWUDFWHG 'LVWULEXWLRQV (YHQW 6HOHFWLRQ 6NLP &XWV $QDO\VLV &XWV &XW 6XPPDU\ (IILFLHQFLHV DQG )DNH 5DWHV (;&/86,9( 5$',$7,9( '(&$< 76f WWWW7 5REXVWQHVV RI 7KH 0DVV 'LVWULEXWLRQ 6WDWLVWLFDO )LW RI WKH ,QYDULDQW 0DVV 'LVWULEXWLRQ 6LJQDO $UHDV DQG 7KHLU 6LJQLILFDQFH $QJXODU 'LVWULEXWLRQ RI 7KH 6LJQDO n 2SWLPXP 0DVV ,QWHUYDO %DFNJURXQG 6XEWUDFWLRQ ,9
PAGE 5
6WDWLVWLFDO )LW RI WKH +HOLFLW\ $QJXODU 'LVWULEXWLRQ $Vn VLJQPHQWV DQG WKHLU 3UREDELOLW\ 'LVWULEXWLRQ (;&/86,9( 5$',$7,9( '(&$< 76f ..a 6WDWLVWLFDO )LW RI WKH ,QYDULDQW 0DVV 'LVWULEXWLRQ 6LJQDO $UHDV DQG 7KHLU 6LJQLILFDQFH $QJXODU 'LVWULEXWLRQ RI 7KH 6LJQDO (;&/86,9( 5$',$7,9( '(&$< 76f A\SS 6<67(0$7,& 81&(57$,17,(6 &XWV XVWLILFDWLRQ RI WKH '37+08 &XW $QJXODU 'LVWULEXWLRQ RI 6LJQDO 'LIIHUHQW +DGURQLF )DNH 5DWHV %HWZHHQ ,6 DQG 6 2WKHU 6\VWHPDWLF 6RXUFHV 2YHUDOO 6\VWHPDWLF 8QFHUWDLQWLHV 5(68/76 $1' &21&/86,21 $33(1',; +(/,&,7< )250$/,60 )25 7:2 %2'< '(&$<6 5()(5(1&(6 %,2*5$3+,&$/ 6.(7&+ Y
PAGE 6
$EVWUDFW RI 'LVVHUWDWLRQ 3UHVHQWHG WR WKH *UDGXDWH 6FKRRO RI WKH 8QLYHUVLW\ RI )ORULGD LQ 3DUWLDO )XOILOOPHQW RI WKH 5HTXLUHPHQWV IRU WKH 'HJUHH RI 'RFWRU RI 3KLORVRSK\ '(&$<6 2) 7+( 76f ,172 $ 3+2721 $1' 7:2 &+$5*(' +$'5216 %\ /XLV %UHYD1HZHOO 'HFHPEHU &KDLU RKQ
PAGE 7
rf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f LQ GLIIHUHQW SURSRUWLRQV 'HPRFULWXV RQ WKH RWKHU KDQG EHOLHYHG WKDW WKH DSSDUHQWO\ FRQWLQXRXV REMHFWV LQ WKH QDWXUDO ZRUOG ZHUH QRW UHDOO\ FRQWLQXRXV EXW PDGH IURP YRLGV DQG LQGLYLVLEOH SDUWLFOHV FDOOHG DWRPV 2YHU WKH ODVW KDOI FHQWXU\ 3DUWLFOH SK\VLFV KDV DGYDQFHG WUHPHQGRXVO\ DQG ZH QRZ KDYH D EHDXWLIXO EXW LQFRPSOHWH WKHRU\ ILUPO\ JURXQGHG RQ H[SHULPHQW WKDW GHVFULEHV WKH IXQGDPHQWDO FRQVWLWXHQWV RI PDWWHU DQG KRZ WKH\ LQWHUDFW ZLWK HDFK RWKHU 7KLV WKHRU\ LV FDOOHG f7KH 6WDQGDUG 0RGHOf 7KH 6WDQGDUG 0RGHO $FFRUGLQJ WR WKH 6WDQGDUG 0RGHO WKH IXQGDPHQWDO EXLOGLQJ EORFNV RI PDWWHU DUH SRLQW OLNH SDUWLFOHV ZKLFK LQWHUDFW ZLWK HDFK RWKHU LQ DV PDQ\ DV WKUHH GLIIHUHQW ZD\V (DFK W\SH RI LQWHUDFWLRQ RU IRUFH LV LWVHOI FDUULHG E\ SRLQW OLNH SDUWLFOHV
PAGE 8
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r ,7 DQG :a SDUWLFOHV DQG ILQDOO\ WKH VWURQJ IRUFH LV FDUULHG E\ HLJKW NLQGV RI JOXRQV 7DEOH VXPPDUL]HV WKH VLWXDWLRQ 7DEOH *DXJH ERVRQV DQG WKH IRUFH WKH\ FDUU\ 6\PERO 1DPH )RUFH &DUULHG 3KRWRQ (OHFWURPDJQHWLF =r =r :HDN Z : :HDN Za :a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f UHPLQLVFHQW RI WKH ZD\ FKHPLVWV RUJDQL]HG WKH HOHPHQWV LQWR %RVRQV DUH GHILQHG DV SDUWLFOHV ZLWK LQWHJHU VSLQ LQ TXDQWXP PHFKDQLFV )HUPLRQV DUH GHILQHG DV SDUWLFOHV ZLWK KDOI RGG IUDFWLRQDO VSLQ
PAGE 9
WKH SHULRGLF WDEOH GXULQJ WKH VHFRQG KDOI RI WKH QLQHWHHQWK FHQWXU\ )HUPLRQV DUH GLYLGHG LQWR TXDUNV JHQHULFDOO\ UHSUHVHQWHG E\ WKH V\PERO Tf DQG OHSWRQV 7KH PDLQ GLIIHUHQFH EHWZHHQ WKHVH WZR JURXSV LV WKDW TXDUNV LQWHUDFW WKURXJK WKH VWURQJ IRUFH ZKLOH OHSWRQV GR QRW 7KHUH DUH VL[ W\SHV RI OHSWRQV DQG VL[ W\SHV RI TXDUNV DOVR FDOOHG WKH VL[ TXDUN IODYRUVf ZKLFK DUH JURXSHG LQWR WKUHH JHQHUDWLRQV (DFK JHQHUDWLRQ FRQVLVW RI WZR TXDUNV DQG WZR OHSWRQV $OO WKUHH JHQHUDWLRQV UHSOLFDWH WKH VDPH VHW RI IRUFH FKDUJHV WKH PDLQ GLIIHUHQFH EHWZHHQ JHQHUDWLRQV LV WKH PDVV RI WKH SDUWLFOHV IRU H[DPSOH WKH UDWLR RI WKH PDVVHV RI H U LV f )RU HDFK IHUPLRQ WKHUH LV DQ DQWLIHUPLRQ ZLWK HTXDO PDVV DQG VSLQ DQG RSSRVLWH FKDUJH 7DEOH )HUPLRQ V\PEROV FODVVLILHG LQWR TXDUNV OHSWRQV DQG WKH WKUHH JHQHUDn WLRQV DORQJ ZLWK WKH JHQHUDWLRQDO FRPPRQ FKDUJHV )LUVW *HQHUDWLRQ 6HFRQG *HQHUDWLRQ 7KLUG *HQHUDWLRQ (OHFWULF :HDN +DV &KDUJH &KDUJH &RORU 4XDUNV X G F V W E
PAGE 10
OHDYHV D TXDUN XQFKDQJHG 7KHUH DUH WKHUHIRUH fÂ§ JOXRQV 7KH LQGLYLGXDO JOXRQ FRORU VWDWHV FDQ EH ZULWWHQ DV IROORZV LWIr UE EIf 9 ! UU fÂ§ EEf f A\ ^UJ JUf MAUJJIf LEJ JEf 9 f te LUI IHf 9 DQG UHSUHVHQW WKH HLJKW GLIIHUHQW JOXRQV WKDW H[LVW LQ QDWXUH 7KH VLQJOH DQG XQLTXH FRORU VWDWH OHIW RXW LV FDOOHG D FRORU VLQJOHW aUU EE JJf f ZKLFK LV LQYDULDQW XQGHU D UHGHILQLWLRQ RI WKH FRORU D URWDWLRQ LQ FRORU VSDFHf ,Q JURXS WKHRU\ WKLV GHFRPSRVLWLRQ RI WKH FRORU VWDWHV LQWR DQ RFWHW DQG D VLQJOHW LV GHQRWHG E\ k ,W LV ZRUWK QRWLQJ KHUH WKDW D FRORUOHVV VDWH VXFK DV RU c LV QRW QHFHVVDULO\ D FRORU VLQJOHW VWDWH 7KLV VLWXDWLRQ LV DQDORJRXV WR WKH SHUKDSV PRUH IDPLOLDU H[DPSOH RI WZR VSLQ SDUWLFOHV (DFK SDUWLFOH FDQ KDYH WKHLU VSLQ XS 7f RU GRZQ If FRUUHVSRQGLQJ WR IRXU SRVVLEOH FRPELQDWLRQV HDFK JLYLQJ D WRWDO VSLQ 6 RU 7KH 6 VWDWHV
PAGE 11
IRUP D WULSOHW _ 7n7! aG WO! L!f f _! 8! DQG WKHUH LV D VLQJOHW VWDWH ZLWK 6 E_ 8! OW!f f 6LQFH JOXRQV WKHPVHOYHV FDUU\ FRORU WKH\ FDQ LQWHUDFW ZLWK HDFK RWKHU WKURXJK WKH VWURQJ IRUFH 7KLV LQWHUDFWLRQ DPRQJ WKH IRUFH FDUULHUV LV XQLTXH WR WKH VWURQJ IRUFH DQG ZKHQ LQFOXGHG LQ SHUWXUEDWLYH 4&' FDOFXODWLRQV OHDGV WR WZR LPSRUWDQW SURSHUWLHV REVHUYHG LQ QDWXUH FDOOHG fDV\PSWRWLF IUHHGRPf DQG fFRORU FRQILQHPHQWf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p ZKHUH ZH KDYH D VH[WHW DQG D WULSOHW EXW QR VLQJOHWf 2Q WKH RWKHU KDQG FRORU VLQJOHWV FDQ EH FRQVWUXFWHG IRU D TT RU TTT V\VWHP 7KH FRORU VLQJOHW IRU TT LV VLPSO\ WKH VWDWH VKRZQ RQ WKH ULJKW VLGH RI (TXDWLRQ 7KH FRORU VLQJOHW IRU TTT FDQ EH REWDLQHG IURP WKH GHFRPSRVLWLRQ !! kkk DQG LV ?TTT !FRLRU VLQJLHW aUJE JUE EUJ EJU JEU UEJf f 9 3DUWLFOHV WKDW DUH ERXQG VWDWHV RI TT DUH DEXQGDQW LQ QDWXUH DQG DUH FDOOHG PHVRQV WKRVH WKDW DUH ERXQG VWDWHV RI TTT DUH FDOOHG EDU\RQV DQG DOVR DERXQG LQ
PAGE 12
r QDWXUH %RWK JURXSV DUH FROOHFWLYHO\ FDOOHG KDGURQV 7KHUH KDYH EHHQ KXQGUHGV RI GLIIHUHQW KDGURQV REVHUYHG FRQILUPLQJ WKH YDOLGLW\ RI 4&' DQG WKH TXDUN IODYRUV 1HYHUWKHOHVV 4&' OHDYHV URRP IRU PRUH SRVVLELOLWLHV $ ERXQG VWDWH RI WZR JOXRQV JJ VRPHWLPHV FDOOHG D JOXHEDOOf FDQ EH LQ D FRORU VLQJOHW DQG LQ SULQFLSOH FRXOG EH REVHUYHG $OWKRXJK JOXHEDOOV DUH DOORZHG E\ 4&' WKHUH LV QR FRQYLQFLQJ H[SHULPHQWDO REVHUYDWLRQ RI RQH $QRWKHU SRVVLELOLW\ DUH ERXQG VWDWHV WKDW DUH D PL[WXUH RI WKH SUHYLRXV VWDWHV VXFK DV DTT cJJ ZLWK DUELWUDU\ D c 7KHVH VWDWHV DUH FDOOHG fK\EULG PHVRQVf RU VLPSO\ fK\EULGVf ,W LV EHOLHYHG >@ >@ >@ WKDW K\EULGV DUH QHFHVVDU\ WR H[SODLQ WKH VSHFWUXP RI OLJKW PHVRQV EHWZHHQ *H9 $V ZLWK DQ\ TXDQWXP PHFKDQLFDO WKHRU\ WKH VSHFWUXP RI ERXQG VWDWHV LV D IXQGDPHQWDO WHVW *OXHEDOOV DUH DOORZHG E\ 4&' \HW WKHUH LV QR FRQFOXn VLYH H[SHULPHQWDO HYLGHQFH RI WKHLU REVHUYDWLRQ GHVSLWH LQWHQVH H[SHULPHQWDO VHDUFKHV >@ >@ >@ FRPSOHPHQWHG E\ ODWWLFH 4&' FDOFXODWLRQV >@ >@ DQG RWKHU WKHRUHWLFDO FRQWULEXWLRQV OLNH EDJ PRGHOV >@ IOX[WXEH PRGHOV >@ 4&' VXP UXOHV >@ ZHDNO\ ERXQG ERXQGVWDWH PRGHOV >@ DQG 4&' IDFWRUL]DWLRQ IRUPDOLVP PRGHOV >@ 3K\VLFLVWV FDQQRW EH VXUH WKH\ XQGHUVWDQG 4&' XQWLO VXFK VWDWHV DUH REVHUYHG RU XQWLO WKH\ FDQ H[SODLQ ZK\ ZH FDQQRW REVHUYH WKHP ,QWURGXFWLRQ WR WKH 5DGLDWLYH 'HFD\V RI WKH <6f 7KH <6f LV D PHVRQ FRPSRVHG RI EE TXDUNV 0HVRQV RI WKLV NLQG FRPSRVHG RI D TXDUN DQG DQ DQWLTXDUN RI WKH VDPH IODYRU DUH LQ JHQHUDO FDOOHG TXDUNRQLD 7KHUH LV D FRQYHQWLRQ EHKLQG WKH QDPH RI WKH 76f 7KH 7 V\PERO LV UHVHUYHG IRU SDUWLFOHV FRPSRVHG RI EE ZKHUH WKH FRPELQHG VSLQ RI WKH TXDUN DQG DQWLTXDUN LV 7KH f,6f V\PERO LV ERUURZHG IURP DWRPLF VSHFWURVFRS\ ZLWK WKH ff PHDQLQJ WKDW WKH EE SDLU DUH LQ WKH ORZHVWHQHUJ\ ERXQGVWDWH DQG WKH f6f $ JRRG H[DPSOH RI WKLV LV WKH VXFFHVVIXO GHVFULSWLRQ RI WKH K\GURJHQ DWRPfV VSHFWUXP E\ TXDQWXP PHFKDQLFV
PAGE 13
PHDQLQJ WKDW WKH EE KDYH D UHODWLYH DQJXODU PRPHQWXP / $ GHVFULSWLRQ RI SDUWLFOH QDPLQJ FRQYHQWLRQV FDQ EH IRXQG LQ >@ 7KH 76f LV XQVWDEOH H[LVWLQJ IRU RQO\ DERXW f VHFRQGV DIWHU ZKLFK LW GHFD\V LQWR GDXJKWHU SDUWLFOHV ZKLFK LQ WXUQ GHFD\ WKHPVHOYHV LI WKH\ DUH QRW VWDEOHf 7KH WHUP fUDGLDWLYH GHFD\f LV UHVHUYHG WR DQ\ 76f GHFD\ ZKHUH RQH RI WKH VWDEOH GDXJKWHUV LV D SKRWRQ 7KH GLIIHUHQW ZD\V WKH <6f FDQ GHFD\ PXVW REH\ WKH V\PPHWULHV LQ QDWXUH ([DPSOHV RI VXFK V\PPHWULHV DUH fSDULW\f DQG fFKDUJH FRQMXJDWLRQf ERWK RI ZKLFK ZLOO EH GHVFULEHG VRRQ 6\PPHWULHV DUH YHU\ LPSRUWDQW LQ WKH VWDQGDUG PRGHO 7KH XVHIXOQHVV RI V\PPHWULHV FDQ EH VHHQ LQ 1RHWKHUVf WKHRUHP ZKLFK VWDWHV WKDW IRU HYHU\ V\PPHWU\ WKHUH LV DQ DVVRFLDWHG FRQVHUYHG TXDQWLW\ ([DPSOHV RI 1RHWKHUVf WKHRUHP DUH WKH FRQVHUYDWLRQ RI PRPHQWXP FRQQHFWHG WR WUDQVODWLRQDO LQYDULDQFH DQG WKH FRQVHUYDWLRQ RI DQJXODU PRPHQWXP DVVRFLDWHG ZLWK URWDWLRQDO LQYDULDQFH ,Q WKH QH[W VHFWLRQV ZH ZLOO XVH WKH SDULW\ DQG FKDUJH FRQMXJDWLRQ V\PPHWULHV WR ILQG 76f GHFD\V ZKLFK FRQVHUYH WKH DVVRFLDWHG V\PPHWU\ FRQVWDQWV DQG DUH DOORZHG E\ QDWXUH ,Q SDUWLFXODU ZH ZLOO VKRZ WKDW WKH UDGLDWLYH GHFD\ RI WKH <6f WKURXJK D SKRWRQ DQG WZR JOXRQV VHH )LJXUH f LV DOORZHG 7KH NH\ REVHUYDWLRQ LV WKDW WKH WZR JOXRQV PXVW EH LQ D FRORU VLQJOHW VLQFH ERWK WKH <6f DQG WKH UDGLDWHG SKRWRQ KDYH QR FRORU DQG FRORU PXVW EH FRQVHUYHG 7KLV PHDQV WKDW WKH WZR JOXRQV VDWLVI\ FRORU FRQILQHPHQW DQG FRXOG IRUP D JOXHEDOO DOWKRXJK PRUH FRQYHQWLRQDO PHVRQ VWDWHV RU KDGURQV LQ QR ERXQG VWDWH DW DOO DUH DOVR SRVVLEOH RXWFRPHV 5HJDUGOHVV RI ZKDW WKH JOXRQV GR WKHLU HQHUJ\ ZLOO HYHQWXDOO\ PDQLIHVWV LWVHOI DV KDGURQV 6RPHWLPHV WZR FKDUJHG KDGURQV RI RSSRVLWH FKDUJH ZnLOO HPHUJH 7KLV ZRUN H[DPLQHV WKRVH WZR KDGURQV IURP D UDGLDWLYH <6f GHFD\ WR H[SHULPHQWDOO\ SUREH WKH JJ VSHFWUXP
PAGE 14
3DULW\ 7KH SDULW\ RSHUDWRU 3 UHYHUVHV WKH VLJQ RI DQ REMHFWfV VSDWLDO FRRUGLQDWHV &RQVLGHU D SDUWLFOH _D ZLWK D ZDYH IXQFWLRQ ?ND[ Wf %\ WKH GHILQLWLRQ RI WKH SDULW\ RSHUDWRU 3r^[Wf 3D9D;Wf f ZKHUH 3D LV D FRQVWDQW SKDVH IDFWRU ,I ZH FRQVLGHU DQ HLJHQIXQFWLRQ RI PRPHQWXP 9IÂ£Wf Hprrf f WKHQ 3S^[Wf 3DbL[Wf 3DAS^[Wf f VR WKDW DQ\ SDUWLFOH DW UHVW ZLWK S UHPDLQV XQFKDQJHG XS WR D PXOWLSOLFDWLYH QXPEHU 3D XQGHU WKH SDULW\ RSHUDWRU 6WDWHV ZLWK WKLV SURSHUW\ DUH FDOOHG HLJHQVWDWHV ZLWK HLJHQYDOXH 3D 3D LV DOVR FDOOHG WKH LQWULQVLF SDULW\ RI SDUWLFOH D RU PRUH XVXDOO\ MXVW WKH SDULW\ RI SDUWLFOH D ZLWK WKH ZRUGV DW UHVW OHIW LPSOLFLW 6LQFH WZR VXFFHVVLYH SDULW\ WUDQVIRUPDWLRQV OHDYH WKH V\VWHP XQFKDQJHG 3b LPSO\LQJ WKDW WKH SRVVLEOH YDOXHV IRU WKH SDULW\ HLJHQYDOXH DUH 3D s ,Q DGGLWLRQ WR D SDUWLFOH DW UHVW D SDUWLFOH ZLWK GHILQLWH RUELWDO DQJXODU PRPHQWXP LV DOVR HQ HLJHQVWDWH RI SDULW\ 7KH ZDYH IXQFWLRQ IRU VXFK D SDUWLFOH LQ VSKHULFDO FRRUGLQDWHV LV AQLP^[Wf 5QO^Uf
PAGE 15
+HQFH 39QOPtWf 3DAQOP;Wf 3DOfOnQOP[Wf f SURYLQJ WKDW D SDUWLFOH ZLWK D GHILQLWH RUELWDO DQJXODU PRPHQWXP O LV LQGHHG DQ HLJHQVWDWH RI WKH SDULW\ RSHUDWRU ZLWK HLJHQYDOXH 3DfÂ§f 7KH SDULWLHV RI WKH IXQGDPHQWDO IHUPLRQV FDQQRW EH PHDVXUHG RU GHULYHG $OO WKDW QDWXUH UHTXLUHV LV WKDW WKH SDULW\ RI D IHUPLRQ EH RSSRVLWH WR WKDW RI DQ DQWLn IHUPLRQ $V D PDWWHU RI FRQYHQWLRQ IHUPLRQV DUH DVVLJQHG 3 DQG DQWLIHUPLRQV DUH DVVLJQHG 3 fÂ§ ,Q FRQWUDVW WKH SDULWLHV RI WKH SKRWRQ DQG JOXRQ FDQ EH GHULYHG E\ DSSO\LQJ 3 WR WKH ILHOG HTXDWLRQV UHVXOWLQJ LQ 3 fÂ§ DQG 3J fÂ§ 7KH 76f KDV 3 3E3E^ f/ VLQFH / 3DULW\ LV D JRRG TXDQWXP QXPEHU EHFDXVH LW LV D V\PPHWU\ RI WKH VWURQJ DQG HOHFWURPDJQHWLF IRUFH 7KLV PHDQV WKDW LQ DQ\ UHDFWLRQ LQYROYLQJ WKHVH IRUFHV SDULW\ PXVW EH FRQVHUYHG &KDUJH &RQMXJDWLRQ &KDUJH FRQMXJDWLRQ LV VLPSO\ WKH RSHUDWLRQ ZKLFK UHSODFHV DOO SDUWLFOHV E\ WKHLU DQWLSDUWLFOHV ,Q TXDQWXP PHFKDQLFV WKH FKDUJH FRQMXJDWLRQ RSHUDWRU LV UHSUHVHQWHG E\ & )RU DQ\ SDUWLFOH _D ZH FDQ ZULWH &MD FD_Â£ f ZKHUH FD LV D SKDVH IDFWRU ,I ZH OHW WKH & RSHUDWRU DFW WZLFH WR UHFRYHU WKH RULJLQDO VWDWH D _D &?D &FD?D !f FD&MD FDFV_D f ZKLFK VKRZV WKDW FDFD ,I DQG RQO\ LIf D LV LWV RZQ DQWLSDUWLFOH LW LV DQ HLJHQVWDWH RI & 7KH SRVVLEOH HLJHQYDOXHV DUH OLPLWHG WR & FD F s $OO V\VWHPV FRPSRVHG RI D WKH VDPH IHUPLRQ DQG DQ DQWLIHUPLRQ SDLU DUH $ 1 HLJHQVWDWHV RI & ZLWK HLJHQYDOXH & fÂ§f/6E 7KLV IDFWRU FDQ EH XQGHUVWRRG EHFDXVH RI WKH QHHG WR H[FKDQJH ERWK SDUWLFOHVf SRVLWLRQ DQG VSLQ WR UHFRYHU WKH
PAGE 16
RULJLQDO VWDWH DIWHU WKH FKDUJH FRQMXJDWLRQ RSHUDWRU LV DSSOLHG ([FKDQJLQJ WKH SDUWLFOHVf SRVLWLRQ JLYHV D IDFWRU RI fÂ§f/ DV ZDV VKRZQ LQ WKH SUHYLRXV VHFWLRQ H[FKDQJLQJ WKH SDUWLFOHV VSLQ JLYHV D IDFWRU RI fÂ§OfV DV FDQ EH YHULILHG E\ LQVSHFWLQJ (TXDWLRQV DQG DQG D IDFWRU RI f ZKLFK DULVHV LQ TXDQWXP ILHOG WKHRU\ ZKHQHYHU IHUPLRQV DQG DQWLIHUPLRQV DUH LQWHUFKDQJHG :LWK WKLV UHVXOW ZH FDQ FDOFXODWH WKH FKDUJH FRQMXJDWLRQ HLJHQYDOXH IRU WKH 7,6f DQG REWDLQ & fÂ§ VLQFH / 6 7KH SKRWRQ LV DQ HLJHQVWDWH RI & VLQFH LW LV LWV RZQ DQWLSDUWLFOH 7KH & HLJHQYDOXH IRU WKH SKRWRQ FDQ EH GHULYHG E\ LQVHUWLQJ & LQWR WKH ILHOG HTXDWLRQV DQG LV & )LQDOO\ ZH FRQVLGHU D V\VWHP FRPSRVHG RI WZR JOXRQV WKDW DUH LQ D FRORU VLQJOHW 7KH WZR JOXRQV DUH ERVRQV DQG WKH\ PXVW KDYH D V\PPHWULF ZDYH IXQFWLRQ 7J XQGHU D J[ S H[FKDQJH 8QGHU WKLV H[FKDQJH WKH RUELWDO DQJXODU PRPHQWXP SDUW RI WKH ZDYH IXQFWLRQ FRQWULEXWHV ZLWK D IDFWRU f/ WKH VSLQ SDUW RI WKH ZDYH IXQFWLRQ IRU WZR VSLQ SDUWLFOHV FRQWULEXWHV ZLWK D IDFWRU RI fÂ§OfV DQG WKH FRORU VLQJOHW SDUW RI WKH ZDYH IXQFWLRQ FRQWULEXWHV ZLWK D IDFWRU RI VLQFH LW LV V\PPHWULF 7R HQVXUH WKDW n6,T LV V\PPHWULF ZH QHHG / 6 WR EH HYHQ 7KLV LPSOLHV WKDW & fÂ§OfLV 7KH / DQG / SRVVLEOH JJ ERXQG VWDWHV DUH VKRZQ LQ 7DEOH 7KH 3& B LV SHFXOLDU EHFDXVH LW LV LPSRVVLEOH IRU D TT V\VWHP WR KDYH WKHVH TXDQWXP QXPEHUV ,I WKLV VWDWH LV HYHU REVHUYHG LW PXVW EH D JOXHEDOO ([SHULPHQWDO VHDUFKHV IRU VXFK D VWDWH KDYH EHHQ GRQH IRU H[DPSOH LQ >@ 7DEOH 3RVVLEOH JJ ERXQG VWDWHV ZLWK / fÂ§ RU / 7KH SRVVLEOH TXDQWXP QXPEHUV DUH OLPLWHG E\ WKH FRQGLWLRQ WKDW / 6 EH HYHQ ZKLFK LV QHHGHG WR HQVXUH D V\PPHWULF ZDYH IXQFWLRQ IRU WKH WZR JDXJH ERVRQV / 6 MUF Oa
PAGE 17
&KDUJH FRQMXJDWLRQ LV D V\PPHWU\ RI WKH VWURQJ DQG HOHFWURPDJQHWLF IRUFH )RU WKRVH SDUWLFOHV WKDW DUH HLJHQVWDWHV RI & & LV D JRRG TXDQWXP QXPEHU EHFDXVH LQ DQ\ UHDFWLRQ LQYROYLQJ WKHVH IRUFHV & PXVW EH FRQVHUYHG 7KH 3 DQG & YDOXHV RI YDULRXV SDUWLFOHV XVHG LQ WKLV DQDO\VLV DUH VKRZQ LQ 7DEOH DORQJ ZLWK WKHLU TXDUN FRPSRVLWLRQ RUELWDO DQJXODU PRPHQWXP DQG LQWHUQDO VSLQ 7DEOH 6\PERO QDPH TXDUN FRPSRVLWLRQ DQJXODU PRPHQWXP /f LQWHUQDO VSLQ 6f SDULW\ 3f DQG FKDUJH FRQMXJDWLRQ HLJHQYDOXHV &f IRU D IHZ RI WKH SDUWLFOHV XVHG LQ WKLV DQDO\VLV 6\PERO 1DPH 4XDUN &RPSRVLWLRQ / V 3 F 76f 8SVLORQO6f EE U 3LRQ XG ; Uf 3LRQ G ; .DRQ XV ; .a .DRQ 6 ; 3 3URWRQ XXG ; 3 $QWLSURWRQ G ; 3KRWRQ ; ; 3RVVLEOH 'HFD\V RI WKH 76f $W WKLV SRLQW ZH FDQ XQGHUVWDQG WKH GLIIHUHQW SRVVLEOH ZD\V WKH 76f FDQ GHFD\ 7KH 3RVVLEOH GHFD\V DUH OLPLWHG EHFDXVH WKH VWURQJ DQG HOHFWURPDJQHWLF IRUFH PXVW FRQVHUYH FRORU & DQG 3 7KH VLPSOHVW SRVVLELOLW\ LV IRU WKH EE SDLU WR LQWHUDFW HOHFWURPDJQHWLFDOO\ DQG DQQLKLODWH LQWR RQH YLUWXDO SKRWRQ 7KLV LV DOORZHG E\ SDULW\ FKDUJH FRQMXJDWLRQ V\PPHWU\ DQG FRORU FRQVHUYDWLRQ 7KH GHFD\ RI WKH
PAGE 18
DOVR IRUELGGHQ E\ FKDUJH FRQMXJDWLRQ <6f GHFD\V WR JOXRQV DUH DOORZHG JOXRQV FDQ IRUP D FRORU VLQJOHWf <6f GHFD\V WR SKRWRQV DUH DOVR DOORZHG EXW DUH ODUJHO\ VXSSUHVVHG E\ GHFD\V WR RQH SKRWRQ VLQFH VXFFHVVLYH HOHFWURPDJQHWLF LQWHUDFWLRQV DUH PXFK OHVV OLNHO\ WR RFFXU WKDQ D VLQJOH RQH )LQDOO\ <6f GHFD\V WZR RQH SKRWRQ DQG WZR JOXRQV DUH DOVR DOORZHG XQGHU WKH FRQGLWLRQ WKDW WKH WZR JOXRQV EH LQ D FRORU VLQJOHW VWDWH 7KH WKUHH GLIIHUHQW SRVVLEOH 76f GHFD\V ZLWK OHDVW DPRXQW RI LQWHUDFWLRQV DOVR FDOOHG ORZHVW RUGHU GHFD\Vf DUH VKRZQ LQ )LJXUH Df Ef &f )LJXUH /RZHVW RUGHU GHFD\V RI WKH 76f DOORZHG E\ FRORU FRQVHUYDWLRQ FKDUJH FRQMXJDWLRQ V\PPHWU\ DQG SDULW\ Df 6KRZV WKH GHFD\ LQWR WKUHH JOXRQV Ef VKRZV D UDGLDWLYH GHFD\ DQG Ff VKRZV WKH HOHFWURPDJQHWLF GHFD\ WKURXJK D YLUWXDO SKRn WRQ WKDW LQ WXUQ GHFD\V HOHFWURPDJQHWLFDOO\ LQWR D SDLU RI FKDUJHG IXQGDPHQWDO SDUWLFOHV VXFK DV TXDUNV RU FKDUJHG OHSWRQV WKH FKDUJHG OHSWRQV DUH UHSUHVHQWHG E\ WKH V\PERO Of 2EVHUYDEOH 5HVRQDQFHV ,Q WKLV ZRUN ZH VHDUFK IRU D UHVRQDQFH ; SURGXFHG LQ D UDGLDWLYH 76f GHFD\ 76f } r f
PAGE 19
$ 8VLQJ FKDUJH FRQMXJDWLRQ V\PPHWU\ RQ ERWK VLGHV fÂ§ fÂ§ &[ f 7KHUHIRUH &[ ,Q RUGHU IRU XV WR REVHUYH ; LW PXVW GHFD\ LQWR WZR FKDUJHG KDGURQV ; KKa f ZKHUH K WW S DUH WKH KDGURQV ZKRVH PRPHQWXP ZH DUH JRLQJ WR PHDVXUH $SSO\LQJ FKDUJH FRQMXJDWLRQ WR WKLV ODVW GHFD\ OfA f ZKHUH / DQG 6 DUH UHVSHFWLYHO\ WKH DQJXODU PRPHQWXP DQG VSLQ RI WKH KKa V\VWHP 7KLV ODVW HTXDWLRQ LPSOLHV WKDW / 6 PXVW EH HYHQ 7KLV KDV FRQVHTXHQFHV IRU WKH SRVVLEOH ; SDULWLHV ZH FDQ REVHUYH %\ SDULW\ FRQVHUYDWLRQ LQ 3[ f/ff OfL f )RU K 7 6 fÂ§ DQG / fÂ§ PXVW EH HYHQ ZKLFK LPSOLHV WKDW 3[ )RU K fÂ§ S 6 FDQ EH RU DQG SDUWLFOHV ZLWK ERWK SRVLWLYH DQG QHJDWLYH SDULWLHV FDQ EH GHWHFWHG 7DEOH 3RVVLEOH 6 / 3 DQG & YDOXHV IRU ; IURP WKH UDGLDWLYH GHFD\ 76f fÂ§! ; UHFRQVWUXFWHG LQ GLIIHUHQW GHFD\ PRGHV 'HFD\ 0RGH V / 3 & ; fÂ§! 7 7a HYHQ HYHQ ; ..a HYHQ HYHQ ;SS HYHQ HYHQ ;!SS RGG HYHQ DQG RGG
PAGE 20
5DGLDWLYH 'HFD\V RI 4XDUNRQLD 2YHUYLHZ 7KHRUHWLFDO PRGHOV H[LVW IRU JOXHEDOO SURGXFWLRQ LQ TXDUNRQLD GHFD\ >@ DQG IRU WKH JOXHEDOO VSHFWUXP )RU H[DPSOH D TXHQFKHG ODWWLFH FDOFXODWLRQ >@ SUHGLFWV D SF JOXHEDOO LQ WKH *H9F PDVV UHJLRQ VHH )LJXUH f $FFRUGLQJ WR 7DEOH LQ WKH FKDUJHG SLRQ DQG NDRQ PRGHV ZH DUH OLPLWHG WR GHWHFW JOXHEDOOV LQ WKH OHIWPRVW FROXPQ ZKHUH 3 & fÂ§ RI )LJXUH ZKLOH LQ WKH SURWRQ PRGH ZH DUH UHVWULFWHG WR WKH WZR OHIW PRVW FROXPQV &' & ( )LJXUH 4XHQFKHG ODWWLFH FDOFXODWLRQ UHVXOW IRU WKH JOXHEDOO VSHFWUXP IRU GLIIHUn HQW 3 DQG & YDOXHV 7KH PDVV VFDOH LV VKRZQ LQ WHUPV RI D VFDOLQJ SDUDPHWHU IURP WKH 4&' ODWWLFH FDOFXODWLRQ QDPHG U RQ WKH OHIW DQG LQ DEVROXWH WHUPV RQ WKH ULJKW E\ WDNLQJ US 0H9 +HUH VWDQGV IRU WKH LQWHUQDO DQJXODU PRPHQWXP VSLQf RI WKH JOXHEDOO /
PAGE 21
)RU VRPH WLPH ERXQG VWDWHV QRW FRQVLGHUHG WR EH SXUH JOXHEDOOV VXFK DV f f DQG UI KDYH EHHQ REVHUYHG LQ [S UDGLDWLYH GHFD\V DW WKH a SURGXFWLRQ OHYHO >@ ,Q WKH %(6 FROODERUDWLRQ FODLPHG WKH REVHUYDWLRQ RI D UHVRWLDQFH WKH\ FDOOHG WKH f SDUWLFOH LQ WKH UDGLDWLYH [S V\VWHP DW WKH OHYHO >@ VHH )LJXUH f $ ORW RI H[FLWHPHQW ZDV JHQHUDWHG DW WKH WLPH EHFDXVH LW LV SRVVLEOH WR LQWHUSUHW WKH f DV D JOXHEDOO +RZHYHU WKLV UHVXOW KDV QRW EHHQ FRQILUPHG )LJXUH 7KH PDVV VSHFWUXP REWDLQHG E\ WKH %(6 FROODERUDWLRQ LQ UDGLDWLYH [S GHFD\V LQWR GLIIHUHQW KDGURQLF PRGHV .H\ WR LGHQWLI\LQJ D SDUWLFOH DV D JOXHEDOO DUH Df VXSSUHVVHG SURGXFWLRQ LQ WZRSKRWRQ FROOLVLRQV XQOLNH TXDUNV JOXRQV GRQfW FDUU\ HOHFWULF FKDUJH DQG GR QRW 7KH [S SDUWLFOH LV D FF ERXQG VWDWH 6LQFH WKH VWURQJ IRUFH LV IODYRU EOLQG WKH VLWXDWLRQ LQ [S GHFD\V LV LQ SULQFLSOH VLPLODU WR WKDW LQ 76f GHFD\V
PAGE 22
FRXSOH WR SKRWRQVf DQD Ef IODYRU V\PPHWULF GHFD\V VLQFH D SXUH JOXHEDOO KDV QR YDOHQFH TXDUNV &/(2 KDV DOUHDG\ GRQH VHYHUDO VWXGLHV RI UDGLDWLYH GHFD\V RI WKH 7O6nf 1DLYHO\ RQH H[SHFWV WKHVH W\SHV RI GHFD\V WR EH VXSSUHVVHG E\ D IDFWRU >4ETFfPFPEf` } f ZLWK UHVSHFW WR WS UDGLDWLYH GHFD\V 3KLV FRPHV IURP QRWLFLQJ WKDW WKH TXDUN SKRWRQ FRXSOLQJ LV SURSRUWLRQDO WR WKH HOHFWULF FKDUJH DQG WKH TXDUN SURSDJDWRU LV DSSUR[LPDWHO\ P IRU ORZ PRPHQWXP TXDUNV ,Q &/(2 PDGH WKH ILUVW REVHUYDWLRQ RI D WZRERG\ 7,6nf UDGLDWLYH GHFD\ >@ 7KH VSLQ RI WKH REVHUYHG UHVRQDQFH FRXOG QRW EH PHDVXUHG EXW LWV PDVV DQG ZLGWK ZKHUH FRQVLVWHQW ZLWK WKH f SDUWLFOH 8QGHU WKLV DVVXPSWLRQ FRPSDULQJ WKH PHDVXUHG EUDQFKLQJ UDWLR RI 76f fÂ§! f WR WKH PHDVXUHG EUDQFKLQJ IUDFWLRQ RI WKH LS fÂ§ f D VXSSUHVVLRQ IDFWRU RI s ZDV REWDLQHG $IWHU WKH %(6 UHVXOW IRU WKH Mf LQ UDGLDWLYH [S GHFD\nV D FRUUHVSRQGLQJ VHDUFK ZDV SHUIRUPHG E\ &/(2 LQ WKH UDGLDWLYH 76f V\VWHP >@ 7KLV DQDO\VLV SXW OLPLWV RQ WKH Mf SURGXFWLRQ LQ UDGLDWLYH 76f GHFD\V ,Q WKLV ZRUN ZH DUH SULYLOHJHG WR KDYH DYDLODEOH WKH ODUJHVW FROOHFWLRQ RI UDGLDWLYH 76f GHFD\V LQ WKH ZRUOG :LWK LW ZH FDQ VWXG\ WKH VWUXFWXUH RI FRORU VLQJOHW JJ KDGURQL]DWLRQ DQG VKHG PRUH OLJKW RQ WKH IMf UHVXOW IURP %(6
PAGE 23
&+$37(5 (;3(5,0(17$/ $33$5$786 7R FDUU\ RXW RXU VWXG\ RI WKH GLKDGURQ VSHFWUXP ZH QHHG WR ILUVW SURGXFH WKH 76f UHVRQDQFH DQG VHFRQGO\ REVHUYH LWV GDXJKWHU SDUWLFOHV IO\LQJ DZD\ DW UHODWLYLVWLF VSHHGV 7KHVH WZR WDVNV DUH UHVSHFWLYHO\ DFFRPSOLVKHG E\ WKH &RUQHOO (OHFWURQ 6WRUDJH 5LQJ DQG WKH &/(2 ,,, GHWHFWRU 7KH &RUQHOO (OHFWURQ 6WRUDJH 5LQJ 7KH &RUQHOO (OHFWURQ 6WRUDJH 5LQJ &(65f ORFDWHG DW &RUQHOO 8QLYHUVLW\ LV D FLUFXODU SDUWLFOH DFFHOHUDWRU WKDW SURGXFHV HHa FROOLVLRQV ,Q RUGHU WR SURGXFH VXFK FROOLVLRQV HOHFWURQV DQG SRVLWURQV QHHG WR EH FUHDWHG DFFHOHUDWHG DQG VWRUHG &(65fV GLIIHUHQW FRPSRQHQWV VKRZQ LQ )LJXUH KDYH EHHQ FDUU\LQJ RXW WKLV WDVN VLQFH $ W\SLFDO &(65 UXQ EHJLQV DW WKH OLQHDU DFFHOHUDWRU /,1$&f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f PDVV *H9 2QFH WKH EHDPV DUH DW WKH GHVLUHG HQHUJ\ WKH\ DUH WUDQVIHUHG WR WKH VWRUDJH ULQJ ZKHUH WKH\ ZLOO UHPDLQ IRU DERXW DQ RXU $W RQH SRLQW LQ WKH VWRUDJH ULQJ WKH
PAGE 24
" ff$ rr fr} 6<1&+527521 :(67 75$16)(5 /,1( ($67 75$16)(5 /,1( /,1$& H EHDPV DUH IRUFHG WR FURVV SDWKV 7KLV LV WKH SRLQW ZKHUH HH FROOLVLRQV RFFXU DQG ZKHUH WKH FHQWHU RI WKH &/(2 ,,, GHWHFWRU LV ORFDWHG ,I WKH DFFHOHUDWRU LV f %81&+ 2) 326,75216 2 %81&+ 2) (/(&75216 )LJXUH 7KH :LOVRQ /DERUDWRU\ DFFHOHUDWRU ORFDWHG DERXW IHHW EHQHDWK &RUQHOO 8QLYHUVLW\fV $OXPQL )LHOGV SHUIRUPLQJ ZHOO D KLJK FROOLVLRQ UDWH UHVXOWV $ KLJK FROOLVLRQ UDWH LV FUXFLDO IRU WKH VXFFHVV RI DQ DFFHOHUDWRU DQG WKH H[SHULPHQWV LW VHUYHV 7KH LPSRUWDQW ILJXUH LV WKH QXPEHU RI SRVVLEOH FROOLVLRQV SHU VHFRQG SHU XQLW DUHD WKLV LV FDOOHG WKH OXPLQRVLW\ ,Q RUGHU WR PD[LPL]H WKH OXPLQRVLW\ WKH EHDPV DUH IRFXVHG DV VPDOO DV SRVVLEOH DW 7KLV SRLQW LV QRW IL[HG LQ VSDFH EXW YDULHV IURP HYHQW WR HYHQW LQVLGH D VPDOO YROXPH RI VSDFH FDOOHG WKH LQWHUDFWLRQ UHJLRQ ,5f
PAGE 25
WKH ,5 'XULQJ WKH &/(2 ,,, LQVWDOODWLRQ VHYHUDO PDJQHWLF TXDGUXSOHV ZHUH DGGHG WR &(65 WR LPSURYH WKH EHDP IRFXV &(65 KDV FRQVLVWHQWO\ RXWGRQH LWVHOI ZKLOH FROOHFWLQJ OXPLQRVLW\ RYHU WKH \HDUV VHH )LJXUH f A UfÂ§YfÂ§LfÂ§UfÂ§ W UfÂ§L XXX )LJXUH &(65 \HDUO\ OXPLQRVLW\ 7KH JDSV LQ DQG FRUUHVSRQG WR GRZQ WLPHV ZKHQ WKH &/(2 ,,9 DQG &/(2 ,,, GHWHFWRUV ZKHUH EHLQJ LQVWDOOHG 7KH &/(2 ,,, GHWHFWRU PHDVXUHV WKH WLPH LQWHJUDWHG OXPLQRVLW\ RYHU D SHULRG RI WLPH E\ FRXQWLQJ KRZ PDQ\ WLPHV D EHQFKPDUN SURFHVV RFFXUV GXULQJ D FHUWDLQ WLPH LQWHUYDO DW WKH ,5 )RU UHGXQGDQF\ WKHUH DUH WZR EHQFKPDUN SURFHVVHV WKDW DUH XVHG RQH ZKHUH WKH HHa SDUWLFOHV LQWHUDFW WR SURGXFH D QHZ HHfÂ§ SDLU DQG WKH RWKHU RQH ZKHUH WKH HH DQQLKLODWH DQG SURGXFH WZR SKRWRQV 8VLQJ WKH NQRZQ FURVVVHFWLRQ IRU HDFK SURFHVV WKH QXPEHU RI HYHQWV LV FRQYHUWHG WR D OXPLQRVLW\ FDOOHG WKH %KDEKD LQWHJUDWHG OXPLQRVLW\ IRU WKH ILUVW SURFHVV DQG WKH
PAGE 26
LQWHJUDWHG OXPLQRVLW\ IRU WKH VHFRQG RQH 7KH WHUP fLQWHJUDWHG LV VRPHWLPHV OHIW RXW DQG WKH WRWDO OXPLQRVLW\ LV UHIHUUHG WR DV VLPSO\ WKH %KDEKD RU OXPLQRVLW\ ZLWK WKH WLPH LQWHJUDWLRQ OHIW LPSOLFLW 7KH &/(2 ,,, 'HWHFWRU :KHQ WKH HHa FROOLVLRQ RFFXUV WKH WZR SDUWLFOHV DUH DQQLKLODWHG ZH HQWHU WKH ZRUOG RI SDUWLFOH SK\VLFV 1DWXUH GHFLGHV ZKDW WR GR ZLWK WKH HQHUJ\ IURP WKH DQQLKLODWLRQ :H KDYH QR FKDQFH RI GLUHFWO\ REVHUYLQJ ZKDW LV KDSSHQLQJ DW WKH DQQLKLODWLRQ SRLQW EXW HYHQWXDOO\ ORQJ OLYHG VHPLVWDEOH SDUWLFOHV DUH FUHDWHG WKDW IO\ RII DW UHODWLYLVWLF VSHHGV 7KHVH SDUWLFOHV FDUU\ LQIRUPDWLRQ DERXW ZKDW KDSSHQHG DIWHU WKH HHa FROOLVLRQ DQG FDQ WHOO XV ZKDW QDWXUH GLG 7KH &/(2 ,,, GHWHFWRU KDV WKH LPSRUWDQW PLVVLRQ RI GHWHFWLQJ DQG PHDVXULQJ VXFK SDU WLFOHV $V RQH FDQ VHH LQ )LJXUH WKH &/(2 ,,, GHWHFWRU LV D FRPSRVLWH RI PDQ\ GHWHFWRU HOHPHQWV 7KHVH VXEGHWHFWRUV DUH W\SLFDOO\ DUUDQJHG DV FRQFHQWULF F\OLQn GHUV 7KH HQWLUH GHWHFWRU LV DSSUR[LPDWHO\ FXEH VKDSHG ZLWK RQH VLGH PHDVXULQJ DERXW PHWHUV DQG ZHLJKV RYHU WKRXVDQG NLORJUDPV (QGFDS &DORULPHWHU ,URQ 3ROVSORFH 0DJQHW ,URQ %DUUHO 0XRQ &KDPEHUV &/(2 ,,, 6& 4XDGUXSROH 3\ORQ 6& 4XDGUXSROHV 5DUH (DUWK 4XDGUXSROH 6ROHQRLG &RLO %DUUHO &DORULPHWHU 5,&+ 'ULIW 6LOLFRQ EHDPSLSH )LJXUH 7KH &/(2 ,,, GHWHFWRU
PAGE 27
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b WR EHQG WKH SDWKV RI FKDUJHG SDUWLFOHV LQ WKH GHWHFWRU %\ PHDVXULQJ KRZ PXFK PXFK D FKDUJHG SDUWLFOH EHQGV H[SHULPHQWHUV FDQ PHDVXUH WKH PRPHQWXP RI WKH SDUWLFOH 7KH FRLO LQQHU UDGLXV LV P DQG LWV RXWHU UDGLXV LV P ZLWK D UDGLDO WKLFNQHVV RI P 7KH WRWDO OHQJWK RI WKH FRLO LQ ] LV P ,W LV ZRXQG IURP D PP [ PP VXSHUFRQGXFWLQJ FDEOH $ VXUURXQGLQJ &X1E7L VWUDQGVf ,W LV ZRXQG LQ OD\HUV ZLWK WXUQV SHU OD\HU RQ DQ DOXPLQXP VKHOO :KHQ LQ RSHUDWLRQ D FXUUHQW RI DPSV IORZV WKURXJK WKH FRLO 7UDFNLQJ 6\VWHP $IWHU SDUWLFOHV IURP WKH LQWHUDFWLRQ SRLQW SDVV WKURXJK WKH EHDP SLSH WKH\ EHJLQ WR HQFRXQWHU WKH DFWLYH GHWHFWRU HOHPHQWV RI WKH WUDFNLQJ V\VWHP 7KHUH DUH WZR VXEGHWHFWRUV UHVSRQVLEOH IRU WUDFNLQJ WKH FXUYLQJ SDWK RI FKDUJHG SDUWLFOHV 7KH ILUVW RQH HQFRXQWHUHG E\ SDUWLFOHV LV WKH VLOLFRQ YHUWH[ GHWHFWRU DQG WKH VHFRQG RQH LV WKH FHQWUDO GULIW FKDPEHU 7KH &/(2 ,,, WUDFNLQJ V\VWHP LV UHVSRQVLEOH IRU WUDFNLQJ D FKDUJHG SDUWLFOHfV SDWK DQG PHDVXULQJ LWV PRPHQWXP 7\SLFDO PRPHQWXP UHVROXWLRQ LV b IRU *H9 WUDFNV 7KH WUDFNLQJ V\VWHP DOVR PHDVXUHV LRQL]DWLRQ HQHUJ\ ORVVHV ZLWK DQ DFFXUDF\ RI DERXW b
PAGE 28
6LOLFRQ 9HUWH[ 'HWHFWRU 7KH VLOLFRQ YHUWH[ GHWHFWRU LQ &/(2 ,,, >@ DOVR FDOOHG 69' ,,, LV D IRXU OD\HU EDUUHORQO\ VWUXFWXUH ZLWK QR HQGFDSV WKDW VXUURXQGV WKH EHDP SLSH 7KLV GHWHFWRU VHH )LJXUH f SURYLGHV IRXU DQG IRXU ] PHDVXUHPHQWV FRYHULQJ RYHU RYHU b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fKLWf :KHQ FRPELQHG WRJHWKHU WKH KLW RQ WKH LQQHU VLGH RI D ZDIHU DQG WKH KLW RQ WKH RXWHU VLGH JLYH D PHDVXUHPHQW RI WKH ] f 7KH ZDIHU SRVLWLRQ LWVHOI GHWHUPLQHV U 7KH &HQWUDO 'ULIW &KDPEHU 7KH &/(2 ,,, FHQWUDO GULIW FKDPEHU '5 ,,,f LV IXOO RI D JDV PL[WXUH ZLWK b +HOLXP DQG b SURSDQH KHOG DW DERXW DQG DW D SUHVVXUH VOLJKWO\ DERYH RQH DWPRVSKHUH 7KH GULIW FKDPEHU LV VWUXQJ ZLWK DUUD\ RI DQRGH ZLUHV RI JROGSODWHG
PAGE 29
c! )LJXUH 9LHZ RI WKH 69' ,,, DORQJ WKH EHDPSLSH WXQJVWHQ RI ILP LQ GLDPHWHU DQG FDWKRGH ZLUHV RI JROGSODWHG DOXPLQXP WXEHV RI [P LQ GLDPHWHU 7KH DQRGHV DUH NHSW DW D SRVLWLYH YROWDJH DERXW 9f DQG WKH FDWKRGHV DUH NHSW JURXQGHG ZKLFK SURYLGHV DQ HOHFWULF ILHOG EHWZHHQ WKH DQRGH DQG WKH FDWKRGH ZLUHV $QRGH DQG FDWKRGH ZLUHV DUH RIWHQ FDOOHG fVHQVHf DQG fILHOGf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f :KHQ WKLV KDSSHQV WR D VHQVH ZLUH ZH VD\ WKDW WKHUH LV D fKLWf
PAGE 30
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f LV VWUXQJ LQ ZLWK D VOLJKW DQJOH DERXW PLOLUDGLDQVf ZLWK UHVSHFW WR WKH ] GLUHFWLRQ WR KHOS ZLWK WKH ] PHDVXUHPHQW :LUHV VWUXQJ LQ WKH ] GLUHFWLRQ DUH FDOOHG fD[LDOf ZLUHV ZKLOH WKRVH WKDW DUH VWUXQJ DW D VOLJKW DQJOH DUH FDOOHG fVWHUHRf ZLUHV 7KH '5 ,,, FRQVLVWV RI DQ LQQHU VWHSSHG VHFWLRQ ZLWK D[LDO OD\HUV DQG DQ RXWHU SDUW ZLWK FRQLFDO HQGSODWHV DQG VPDOO DQJOH VWHUHR OD\HUV 7KHUH DUH ILHOG ZLUHV SHU VHQVH ZLUH DQG WKH\ DSSUR[LPDWHO\ IRUP D FP VLGH VTXDUH 7KH GULIW UHVROXWLRQ LV DURXQG SP LQ U fÂ§ I! DQG DERXW PP LQ ] $OO ZLUHV DUH KHOG DW VXIILFLHQW WHQVLRQ WR KDYH RQO\ D SP JUDYLWDWLRQDO VDJ DW WKH FHQWHU ] f 7KHUH DUH D[LDO VHQVH ZLUHV DQG VWHUHR VHQVH ZLUHV D WRWDO &U\VWDO &DORULPHWHU 7KH &/(2 &U\VWDO &DORULPHWHU &&f LV FRPSRVHG RI WKDOOLXPGRSHG &VO FU\VWDOV (DFK FU\VWDO LV FP ORQJ UDGLDWLRQ OHQJWKVf ZLWK FP [ FP VTXDUH IURQW IDFH 7KH FU\VWDOV DEVRUE DQ\ LQFRPLQJ HOHFWURQ RU SKRWRQ ZKLFK FDVFDGHV LQWR D VHULHV RI HOHFWURPDJQHWLF VKRZHUV 7KH HOHFWURQLF V\VWHP FRPSRVHG RI SKRWRGLRGHV SUHVHQW DW WKH EDFN RI HDFK FU\VWDO DUH FDOLEUDWHG WR PHDVXUH WKH HQHUJ\ GHSRVLWHG E\ WKH LQFRPLQJ SDUWLFOH 2WKHU LQFRPLQJ SDUWLFOHV RWKHU WKDQ SKRWRQV DQG HOHFWURQV DUH SDUWLDOO\ DQG VRPHWLPHV IXOO\ DEVRUEHG E\ WKH FU\VWDO JLYLQJ DQ HQHUJ\ UHDGLQJ
PAGE 31
7KH && LV DUUDQJHG LQWR D EDUUHO VHFWLRQ DQG WZR HQGFDSV WRJHWKHU FRYHULQJ b RI WKH VROLG DQJOH 7KH && EDUUHO VHFWLRQ LV XQFKDQJHG VLQFH &/(2 ,, WKH HQGFDSV KDYH EHHQ UHEXLOW IRU &/(2 ,,, WR DFFRPPRGDWH WKH QHZ &(65 LQWHUDFWLRQ UHJLRQ TXDGUXSROHV 7KH EDUUHO GHWHFWRU FRQVLVWV RI DQ DUUD\ RI FU\VWDOV LQ If DQG LQ DUUDQJHG LQ DQ DOPRVWSURMHFWLYH EDUUHO JHRPHWU\ 7KDW LV WKH FU\VWDOV DUH WLOWHG LQ WR SRLQW WR D IHZ FP DZD\ IURP WKH LQWHUDFWLRQ SRLQW DQG WKHUH LV DOVR D VPDOO WLOW LQ 7KH && EDUUHO LQQHU UDGLXV LV P RXWHU UDGLXV LV P DQG WKH OHQJWK LQ DW WKH LQQHU UDGLXV LV P ,W FRYHUV WKH SRODU DQJOH UDQJH IURP WR GHJUHHV 7KH EDUUHO FU\VWDOV DUH WDSHUHG WRZDUGV WKH IURQW IDFH WKHUH DUH VOLJKWO\ GLIIHUHQW WDSHUHG VKDSHVf WKH HQGFDS FU\VWDOV DUH UHFWDQJXODU EXW VKDYHG QHDU WKH RXWHU UDGLXV WR ILW LQ WKH FRQWDLQHU 7KH && HQGFDSV FRQVLVW RI WZR LGHQWLFDO HQG SOXJV HDFK FRQWDLQLQJ FU\VWDOV RI VTXDUH FURVVVHFWLRQ DOLJQHG SDUDOOHO WR WKH EHDP OLQH QRW SURMHFWLYHf 7KHUH DUH FU\VWDOV LQ WKH ffIL[HGf SRUWLRQ RI WKH ffNH\VWRQHf SLHFH RI WKH HQGFDS LQ WKH SDUW WKDW VOLGHV 7KH NH\VWRQH LV PDGH XS RI WZR SDUWV RQH RQ WRS WKDW KDV FU\VWDOV WKDW IRU PHFKDQLFDO UHPRYDO UHDVRQV LV VHSDUDWH IURP D FRQWDLQHU KROGLQJ FU\VWDOV 7KH HQGFDS H[WHQGV IURP P WR P LQ U 7KH IURQW IDFHV DUH s P IURP WKH LQWHUDFWLRQ SRLQW ,W FRYHUV WKH SRODU DQJOH UHJLRQ IURP WR GHJUHHV LQ ] DQG WR LQ 7KH SKRWRQ HQHUJ\ UHVROXWLRQ LQ WKH EDUUHO HQGFDSf LV b bf IRU *H9SKRWRQV DQG JRHV GRZQ WR b bf IRU *H9SKRWRQV 5LQJ ,PDJLQJ &KHUHQNRY 'HWHFWRU 7KH 5LQJ ,PDJLQJ &KHUHQNRY 5,&+f GHWHFWRU >@ LV D QHZU GHWHFWRU VXEV\Vn WHP IRU &/(2 ,,, ,W UHSODFHV WKH &/(2 ,O9 WLPH RI IOLJKW V\VWHP GHVLJQHG WR PHDVXUH SDUWLFOHVf YHORFLWLHV
PAGE 32
&KHUHQNRY UDGLDWLRQ RFFXUV ZKHQ D SDUWLFOH WUDYHOV IDVWHU WKDQ WKH VSHHG RI OLJKW LQ D FHUWDLQ PHGLXP Y FQ f :KHUH Y LV WKH YHORFLW\ RI WKH SDUWLFOH F LV WKH VSHHG RI OLJKW LQ YDFXXP DQG Q LV WKH LQGH[ RI UHIUDFWLRQ RI WKH PHGLXP WKH SDUWLFOH LV WUDYHOLQJ LQ 7KH FKDUJHG SDUWLFOH SRODUL]HV WKH PROHFXOHV RI WKH PHGLXP ZKLFK WKHQ WXUQ EDFN UDSLGO\ WR WKHLU JURXQG VWDWH HPLWWLQJ UDGLDWLRQ 7KH HPLWWHG OLJKW IRUPV D FRKHUHQW ZDYHIURQW LI Y FQ DQG &KHUHQNRY OLJKW LV HPLWWHG XQGHU D FRQVWDQW &KHUHQNRY DQJOH Â• ZLWK WKH SDUWLFOH WUDMHFWRU\ IRUPLQJ D FRQH RI OLJKW 7KH FRQH KDOIDQJOH LV JLYHQ E\ WKH &KHUHQNRY DQJOH ZKLFK LV FRV fÂ§ YQ Q f ,I WKH UDGLDWLRQ DQJOH LV PHDVXUHG WKH VSHHG RI WKH LQFLGHQW SDUWLFOH LV NQRZQ 7KLV PHDVXUHPHQW FRPELQHG ZLWK WKH PRPHQWXP PHDVXUHPHQW IURP WKH WUDFNLQJ V\VWHP JLYHV D PHDVXUHPHQW RI WKH SDUWLFOHV PDVV DQG FDQ EH XVHG LQ SDUWLFOH LGHQWLILFDWLRQ 7KH WKUHVKROG YHORFLW\ DW ZKLFK &KHUHQNRY UDGLDWLRQ LV HPLWWHG LV YPLQ :KHQ D SDUWLFOH WUDYHOLQJ DW WKH WKUHVKROG YHORFLW\ WUDQVYHUVHV WKH PHGLXP D YHU\ VPDOO FRQH ZLWK m LV SURGXFHG 7KH PD[LPXP HPLVVLRQ DQJOH RFFXUV ZKHQ 9PD[ fÂ§ F DQG LV JLYHQ E\ &26 ÂPD[ f f Q 7KH 5,&+ VHH )LJXUH f FRQVLVWV RI PRGXOHV LQ SKL P ZLGH DQG P ORQJ 7KH GHWHFWRU VWDUWV DW D UDGLXV RI P DQG H[WHQGV WR P (DFK PRGXOH KDV SDQHV RI VROLG FU\VWDO /L) UDGLDWRU DW DSSUR[LPDWHO\ P UDGLXV P ZLGH P ORQJ FP WKLFN ,QQHU VHSDUDWLRQ EHWZHHQ UDGLDWRUV LV W\SLFDOO\ UP 7KH /L) LQGH[ RI UHIUDFWLRQ LV Q 7KH UDGLDWRUV FORVHVW WR ] LQ HDFK PRGXOH KDYH D GHJUHH VDZWRRWK RXWHU IDFH WR UHGXFH WRWDO LQWHUQDO
PAGE 33
UHIOHFWLRQ RI WKH &KHUHQNRY OLJKW IRU QRUPDO LQFLGHQW SDUWLFOHV VHH )LJXUH f 7KH UDGLDWRUV DUH IROORZHG E\ D FP UDGLDOf GULIW VSDFH ILOOHG ZLWK SXUH 1 7KH GULIW VSDFH LV IROORZHG E\ WKH SKRWRGHWHFWRU D WKLQJDS PXOWLZLUH SKRWRVHQVLWLYH SURSRUWLRQDO FKDPEHU )LJXUH 7KH 5,&+ GHWHFWRU VXEV\VWHP WUDFN ; PP ? 79" W /fÂ§ PP A WUDFN )LJXUH 7KH WZR NLQGV RI 5,&+ /L) UDGLDWRUV )RU QRUPDO LQFLGHQFH SDUWLFOHV ] a f D VDZWRRWK UDGLDWRU LV QHFHVVDU\ WR DYRLG LQWHUQDO UHIOHFWLRQ :LWK WKLV LQGH[ SDUWLFOHV LQ WKH /L) UDGLDWRU ZLWK EHWD SURGXFH &KHUHQNRY FRQHV RI KDOIDQJOH FRVBOQf UDGLDQV :LWK D FP GULIW
PAGE 34
VSDFH WKLV SURGXFHV D FLUFOH RI UDGLXV FP 7KH 5,&+ LV FDSDEOH RI PHDVXULQJ WKH &KHUHQNRY DQJOH ZLWK D UHVROXWLRQ RI D IHZ PLOLUDGLDQV VHH )LJXUH f 7KLV JUHDW UHVROXWLRQ DOORZV IRU JRRG VHSDUDWLRQ EHWZHHQ SLRQV DQG NDRQV XS WR DERXW *H9 DV )LJXUH VKRZV 7' FX ( fÂ§fÂ§nfÂ§ WfÂ§7fÂ§LfÂ§nfÂ§LfÂ§fÂ§LfÂ§nfÂ§U f 'DWD Â‘ R 0RQWH &DUOR L Â‘ Â‘ 7 Â‘ Â‘ R f OLOL Â‘ f W W 7 7 Â‘ L nLn f W Â‘ Â‘ f Â‘ Â‘ L / Â‘ 5DGLDWRU URZ )LJXUH &KHUHQNRY DQJOH UHVROXWLRQV SHU WUDFN DV D IXQFWLRQ RI UDGLDWRU URZ IRU %KDEKD HYHQWV 5RZ FRUUHVSRQGV WR WKH WZR URZV FORVHVW WR ] HWF 0XRQ 'HWHFWRUV 7KH PXRQ GHWHFWRUV 08f DUH WKH PRVW H[WHUQDO VXEV\VWHP RI WKH &/(2 ,,, GHWHFWRU 7KH\ UHPDLQ XQFKDQJHG IURP &/(2 ,, DQG DUH FRPSRVHG RI SODVWLF SURSRUWLRQDO WXEHV HPEHGGHG LQ WKH PDJQHW LURQ UHWXUQ \RNH 7KH\ FRYHU b RI WKH W VROLG DQJOH URXJKO\ GHJUHHV LQ SRODU DQJOHf ,I D VHULHV RI KLWV LV GHWHFWHG LQ WKH PXRQ FKDPEHU OD\HUV WKH\ PRVW OLNHO\ FRUUHVSRQG WR PXRQV EHFDXVH RWKHU SDUWLFOHV DUH EORFNHG E\ WKH LURQ %HVLGHV GHWHFWLQJ PXRQV WKH KHDY\ LURQ
PAGE 35
)LJXUH 3LQ IDNH UDWH DV D IXQFWLRQ RI SDUWLFOH PRPHQWXP IRU NDRQ HIILFLHQF\ RI b FLUFOHVf b VTXDUHVf DQG b WULDQJOHVf RI WKH UHWXUQ \RNH SURWHFWV WKH LQQHU VXEV\VWHPV RI WKH &/(2 ,,, GHWHFWRU IURP FRVPLF UD\ EDFNJURXQG H[FHSW IRU FRVPLF UD\ PXRQV RI FRXUVHf 7KHUH DUH WKUHH SODQHV RI FKDPEHUV LQ WKH EDUUHO VHFWLRQ DUUDQJHG LQ RFWDQWV LQ 7KH SODVWLF EDUUHO SODQHV OLH DW GHSWKV RI DQG FP RI LURQ DW QRUPDO LQFLGHQFHf FRUUHVSRQGLQJ WR URXJKO\ DQG KDGURQLF LQWHUDFWLRQ OHQJWKV FP LQ LURQf UHIHUUHG WR DV '37+08 7KHUH LV RQH SODQH RI FKDPEHUV LQ HDFK RI WKH WZR HQGFDS UHJLRQV DUUDQJHG LQ URXJK TXDGUDQWV LQ G! 7KH\ OLH DW ] fÂ§s P URXJKO\ FRYHULQJ WKH UHJLRQ _FRVf_ 7KH SODQDU WUDFNLQJ FKDPEHUV XVH SODVWLF SURSRUWLRQDO FRXQWHUV DW DERXW 9 ZLWK GULIW JDV RI b +H b SURSDQH LGHQWLFDO WR DQG VXSSOLHG E\ WKH VDPH V\VWHP DVf WKH GULIW FKDPEHU JDV ,QGLYLGXDO FRXQWHUV DUH P ORQJ DQG P ZLGH ZLWK D VSDFH
PAGE 36
UHVROXWLRQ DORQJ WKH ZLUH XVLQJ FKDUJH GLYLVLRQf RI FP 7KH WUDFNLQJ FKDPEHUV D[H PDGH RI H[WUXGHG SODVWLF FP ZLGH E\ FP WKLFN E\ P ORQJ FRQWDLQLQJ HLJKW WXEHV FRDWHG RQ VLGHV ZLWK JUDSKLWH WR IRUP D FDWKRGH ZLWK JP VLOYHUSODWHG &X%H DQRGH ZLUHV KHOG DW 9 7KH RUWKRJRQDO FRRUGLQDWH LV SURYLGHG E\ FP FRSSHU VWULSV UXQQLQJ SHUSHQGLFXODU WR WKH WXEHV RQ WKH VLGH QRW FRYHUHG E\ JUDSKLWH
PAGE 37
&+$37(5 $1$/<6,6 2) 7+( '$7$ 7KLV ZRUN EXLOGV RQ WKH WHFKQLTXHV GHYHORSHG E\ SUHYLRXV &/(2 UDGLDWLYH <6f DQDO\VHV >@ >@ PRGLILHG WR EH XVHG ZLWK &/(2 ,,, GDWD $ QHZ WHFKQLTXH EDVHG RQ NLQHPDWLF ILWWLQJ LV GHYHORSHG WR WRJHWKHU ZLWK WKH QHZ 5,&+ GHWHFWRU LPSURYH HIILFLHQF\ DQG SDUWLFOH LGHQWLILFDWLRQ :H VHDUFK IRU UDGLDWLYH 76f GHFD\V LQ WKH PRGHV 76f fÂ§Â‘! UUa..a DQG SS 7KH HHa FROOLVLRQ GDWD KDV ERWK UHVRQDQW HYHQWV ZKHUH WKH HHa DQQLKLODWH WR JLYH D 76f DQG FRQWLQXXP HYHQWV ZKHUH WKH HHa FROOLVLRQ GRHV QRW JLYH D <6f 7R EH VXUH ZH DUH REVHUYLQJ <6f DQG QRW D FRQWLQXXP SURFHVV WKH FRQWLQXXP PXVW EH VXEWUDFWHG E\ XVLQJ D SXUH VRXUFH RI FRUUHFWO\ VFDOHG FRQWLQXXP HYHQWV 3XUH FRQWLQXXP GDWD FDQ EH REWDLQHG E\ RSHUDWLQJ &(65 DW DQ HQHUJ\ GLIIHUHQW IRUP WKH <6f PDVV $IWHU VXEWUDFWLQJ WKH XQGHUO\LQJ FRQWLQXXP ZH H[DPLQH WKH GLKDGURQ LQYDULDQW PDVV VSHFWUXP LQ VHDUFK RI UHVRQDQFHV :H GHWHUPLQH WKH VSLQ DQG SURGXFWLRQ KHOLFLW\ WKH SURMHFWLRQ RI WKH VSLQ RQ WKH PRPHQWXP YHFWRU DW SURGXFWLRQ WLPHf RI DQ\ IRXQG UHVRQDQFHV E\ H[DPLQLQJ WKH SKRWRQ DQG KDGURQ DQJXODU GLVWULEXWLRQV 'DWD 6DPSOH 7KH DQDO\VLV SUHVHQWHG KHUH LV EDVHG RQ &/(2 ,,, GDWD 7KURXJKRXW WKLV GRFXPHQW DQG XQOHVV RWKHUZLVH VWDWHG ZH XVH OXPLQRVLW\ WKH OXPLQRVLW\ W\SHV XVHG E\ &/(2 DUH GHILQHG DW WKH HQG RI 6HFWLRQ f :H SUHIHU WR XVH WKH RYHU WKH %KDEKD OXPLQRVLW\ EHFDXVH WKH UHVRQDQW SURFHVV 76f fÂ§! HHa DUWLILFLDOO\ LQFUHDVHV WKH UHSRUWHG %KDEKD OXPLQRVLW\ E\ DERXW b LQ 76f GDWD 7KLV H[WUD FRQWULEXWLRQ ZRXOG QHHG WR EH DFFRXQWHG IRU ZKHQ GRLQJ D FRQWLQXXP VXEWUDFWLRQ &KRRVLQJ WKH OXPLQRVLW\ DYRLGV WKLV FRPSOLFDWLRQ
PAGE 38
7KH &/(2 ,,, GDWD LV GLYLGHG LQWR QXPEHUHG VHWV 6HWV DQG KDYH D OXPLQRVLW\ RI s IU LQ WKH EHDP HQHUJ\ UDQJH *H9 7KLV GDWD ZKLFK ZH FDOO WKH 76f GDWD RU VLPSO\ WKH ,6 GDWDf KDV ERWK UHVRQDQFH HHa fÂ§! <6f DQG FRQWLQXXP HYHQWV :H WDNH WKH QXPEHU RI UHVRQDQW HYHQWV IURP >@ 1ULVf s f [ 7KLV QXPEHU FRQWUDVWV ZLWK WKH SUHYLRXV JHQHUDWLRQ PHDVXUHPHQW RI &/(2 ,, ZKHUH 1ULVf a [ n ZHUH DYDLODEOH ,Q GDWDVHWV DQG DQG WKHUH DUH DOVR s f WDNHQ EHORZ WKH <6f EHDP HQHUJ\ *H9f 7KLV GDWD ZKLFK ZH FDOO WKH 7O6fRI) GDWD RU VLPSO\ WKH O6RII GDWDf KDV UHODWLYHO\ ORZ VWDWLVWLFV DQG FRUUHVSRQGV WR SXUHO\ FRQWLQXXP HYHQWV 7R LPSURYH RXU FRQWLQXXP VWDWLVWLFV ZH XVH s IEaO IURP GDWDVHWV DQG RI GDWD WDNHQ QHDU WKH FHQWHURIPDVV HQHUJ\ RI WKH <6f ZKLFK IRU RXU SXUSRVHV LV GHILQHG DV GDWD ZLWK EHDP HQHUJ\ LQ WKH *H9 UDQJH 7KLV VHW RI GDWD ZKLFK ZH FDOO WKH <6f GDWD RU VLPSO\ WKH 6 GDWDf LV D VRXUFH RI SLQH FRQWLQXXP EHFDXVH QR 76f fÂ§! %% UHVRQDQW HYHQW FDQ VXUYLYH RXU fFXWVf WKH FXWV DUH SUHVHQWHG LQ 6HFWLRQ f &RQWLQXXP 6XEWUDFWHG 'LVWULEXWLRQV :H XVH WKH FRQWLQXXP GDWD WDNHQ DW WKH 76f HQHUJ\ WR VXEWUDFW WKH XQGHUn O\LQJ FRQWLQXXP SUHVHQW LQ WKH 76f GDWD 7KLV LV LPSRUWDQW EHFDXVH FRQWLQXXP EDFNJURXQG SURFHVVHV OLNH HHf fÂ§r \S ZLWK S fÂ§r UUf HHf fÂ§! ZLWK fÂ§A ..a DQG GLUHFW HHf fÂ§! KKa ZH ZLOO XVH WKH FRQYHQWLRQ K U S IURP QRZ RQf ORRN OLNH WKH VLJQDO HYHQWV ZH DUH VHDUFKLQJ IRU 7R ILUVW RUGHU WKH FURVV VHFWLRQ RI WKHVH FRQWLQXXP SURFHVV VFDOHV OLNH V ZKHUH V LV WKH VTXDUH RI &XWV DUH VLPSO\ FRQGLWLRQV WKDW DQ HYHQW PXVW VDWLVI\ WR EH FRQVLGHUHG LQ WKH mDQDO\VLV &XWV DUH QHFHVVDU\ WR HOLPLQDWH EDFNJURXQG WKDW ZRXOG RWKHUZLVH PDNH D PHDVXUHPHQW GLIILFXOW RU HYHQ LPSRVVLEOH
PAGE 39
WKH FHQWHU RI PDVV HQHUJ\ RI WKH HH V\VWHP 7DNLQJ WKH OXPLQRVLW\ZHLJKWHG DYHUDJH EHDPHQHUJLHV RI HDFK LQWHUYDO DQG WKH OXPLQRVLWLHV VHH 7DEOH f ZH FDOFXODWH WKDW WKH 76f GDWD VFDOHG GRZQ E\ D IDFWRU RI UHSUHVHQWV WKH XQn GHUO\LQJ FRQWLQXXP LQ WKH 76f GDWD 7KLV LV WUXH XS WR GLIIHUHQFHV LQ PRPHQWXP GLVWULEXWLRQV DQG SKDVH VSDFH 7KH HUURU LQ WKH FRQWLQXXP VFDOH IDFWRU LV XQNQRZQ EHFDXVH WKH OXPLQRVLW\ UDWLR LV H[SHFWHG WR KDYH D VPDOO EXW XQGHWHUPLQHG V\Vn WHPDWLF HUURU :H PDNH WKH VRPHZKDW DUELWUDU\ GHFLVLRQ WR UHWDLQ WKUHH VLJQLILFDQW GLJLWV LQ WKH FRQWLQXXP VFDOH IDFWRU EHFDXVH LW LV VXIILFLHQW IRU RXU SXUSRVHV DQG WKHUH DUH b HIIHFWV IURP VHFRQG RUGHU WHUPV LQ WKH FURVV VHFWLRQ IRUPXODH 7R HOLPLQDWH WKH FRQWULEXWLRQ RI FRQWLQXXP HYHQWV IURP D 76f GDWD YDULDEOH GLVWULEXWLRQ HJ WKH LQYDULDQW PDVV RI WZR WUDFNV WKH SKRWRQ DQJXODU GLVWULEXWLRQf ZH SURFHHG DV IROORZV f 2EWDLQ WKH 76f GDWD GLVWULEXWLRQ IRU WKH VDPH YDULDEOH f (IILFLHQF\ FRUUHFW ERWK WKH <6f DQG 76f GLVWULEXWLRQV XVLQJ D *($17 >@ EDVHG 0RQWH &DUOR 0&f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a fÂ§! S FURVV VHF WLRQ DOVR KDV DQ LQVLJQLILFDQW HIIHFW RQ WKH VFDOH IDFWRU
PAGE 40
f 6XEWUDFW WKH 76f GLVWULEXWLRQ IURP WKH <6f GLVWULEXWLRQ XVLQJ D <6f VFDOH IDFWRU RI :H FDOO WKLV VHW RI VWHSV fFRQWLQXXP VXEWUDFWLRQf E\ GHILQLWLRQ )RU FRPn SDFWQHVV ZH FDOO VXFK D GLVWULEXWLRQ WKH fFRQWLQXXP VXEWUDFWHG YDULDEOH! GLVWULEXWLRQSORWf RU fYDULDEOH! FRQWLQXXP VXEWUDFWHG GLVWULEXWLRQSORWf ([FHSW IRU VWDWLVWLFDO IOXFWXDWLRQV DQG SKDVH VSDFH HIIHFWV WKH UHVXOWLQJ ,6 GLVWULEXWLRQ VKRXOG QRW KDYH DQ\ FRQWULEXWLRQ IURP FRQWLQXXP SURFHVVHV WKDW VFDOH DV V ,W LV LPSRUWDQW WR QRWLFH WKDW DQ\ FRQWLQXXP VXEWUDFWHG GLVWULEXWLRQ LV HIILFLHQF\ FRUUHFWHG 7KLV PHDQV WKDW D ILW WR D FRQWLQXXP VXEWUDFWHG GLVWULEXWLRQ IRU H[DPSOH WKH FRQWLQXXP VXEWUDFWHG LQYDULDQW PDVV GLVWULEXWLRQf JLYHV WKH HIILFLHQF\ FRUUHFWHG QXPEHU RI HYHQWV GLUHFWO\ 6WULFWO\ VSHDNLQJ WKLV QXPEHU RI HYHQWV LV RQO\ FRUUHFW LI DOO WKH RWKHU YDULDEOHV ZH FXW RQ KDYH WKH VDPH LQLWLDO GLVWULEXWLRQ LQ GDWD DQG 0& RU LI WKH HIILFLHQF\ GRHV QRW GHSHQG RQ WKHP ,Q WKLV QRWH ZH XVH fIODW 0&f GHILQHG DV 0& WKDW LV JHQHUDWHG ZLWK D IODW GLVWULEXWLRQ LQ WKH PDVV DQG WKH KHOLFLW\ DQJOHV Q WKHVH DQJOHV DUH GHILQHG LQ WKH DSSHQGL[f )RU H[DPSOH WKH QXPEHU RI HYHQWV REWDLQHG IURP WKH FRQWLQXXP VXEWUDFWHG LQYDULDQW PDVV ILW QHHGV WR EH FRUUHFWHG WR DFFRXQW IRU WKH IDFW WKDW WKH KHOLFLW\ DQJOH GLVWULEXWLRQV DUH QRW IODW LQ GDWD VHH IRU H[DPSOH )LJXUH DQG WKH HIILFLHQF\ LV KLJKO\ GHSHQGHQW RQ WKHVH YDULDEOHV VHH )LJXUH f 7KLV FRUUHFWLRQ LV GRQH LQ 6HFWLRQ $ VXPPDU\ RI WKH UHVXOWV IURP WKLV VHFWLRQ LV VKRZQ LQ 7DEOH (YHQW 6HOHFWLRQ (YHQW VHOHFWLRQ IRU <6f fÂ§! aIKKa LV VWUDLJKWIRUZDUG DQG FDQ EH WKRXJKW RI LQ WHUPV RI WKUHH PDMRU VWDJHV )LUVW ZH VNLP WKH GDWD NHHSLQJ RQO\ WKRVH HYHQWV WKDW FRQWDLQ H[DFWO\ RQH KLJKHQHUJ\ SKRWRQ DQG WZR WUDFNV 1H[W ZH UHTXLUH WKDW WKH WRWDO PRPHQWXP RI WKHVH WKUHH HOHPHQWV EH FRQVLVWHQW ZLWK WKH FROOLGLQJ HHa PRPHQWXP )LQDOO\
PAGE 41
7DEOH 6XPPDU\ RI WKH GDWD XVHG LQ WKLV DQDO\VLV 7KH FRQWLQXXP VFDOH IDFWRU LV REWDLQHG XVLQJ OXPLQRVLWLHV EHFDXVH WKH %KDEKD OXPLQRVLW\ LV DUWLILFLDOO\ KLJK GXULQJ 76f UXQQLQJ GXH WR WKH SURFHVV 76f fÂ§! HHa Ln nAW 76f 76f 7O6fRII 'DWDVHW $YHUDJH efHDP*H9f 5DQJH RI )EHDP*H9f ÂHH.LUf s s s Â‘&fIEff s s s 76f FRQWLQXXP VFDOH IDFWRU LQ WKH WKLUG VWDJH ZH SURMHFW WKH VXUYLYLQJ GDWD RQWR WKH WKUHH GLIIHUHQW KDGURQLF PRGHV YLD KDGURQ VHSDUDWLRQ DQG 4(' VXSSUHVVLRQ FXWV +RZHYHU FKHFNLQJ WKH PRPHQWXP LQYROYHV XVLQJ WKH WUDFNV PDVVHV 7KLV PHDQV WKDW WKH LQIRUPDWLRQ IURP VWDJH VKRXOG VRPHKRZ EH XVHIXO LQ VWDJH 7KLV LV LQGHHG WKH FDVH DQG WKH GHWDLOV RI KRZ ZH GR LW DUH UHYHDOHG LQ WKLV VHFWLRQ 6NLP &XWV :H VNLP RYHU WKH GDWD LQ WKH fKDUG*DPf VXEFROOHFWLRQ 7KH KDUG*DP VXEFROOHFWLRQ ZDV GHYHORSHG ZLWK WKLV W\SH RI DQDO\VLV LQ PLQG )RU DQ HYHQW WKDW SDVVHV WKH WULJJHUV WR EH FODVVLILHG DV KDUG*DP LW PXVW SDVV WKH IROORZLQJ FXWV f H*DPO f H6K f H2YHU3O f H9LV f D&RV7K ZKHUH fH*DPOf LV WKH KLJKHVW LVRODWHG VKRZHU HQHUJ\ UHODWLYH WR WKH EHDP HQHUJ\ fH6Kf LV WKH HQHUJ\ RI WKH VHFRQG KLJKHVW VKRZHU UHODWLYH WR WKH EHDP HQHUJ\ fH2YHU3Of LV WKH PDWFKHG VKRZHU HQHUJ\ UHODWLYH WR 7ULJJHUV DUH EDVLF FULWHULD WKDW DQ HYHQW PXVW VDWLVI\ WR UHFRUGHG GXULQJ WKH GDWD FROOHFWLRQ SURFHVVHV 7ULJJHUV DUH GHVLJQHG WR JHW ULG RI WUDVK DQG QRLVH DQG UHGXFH WKH VL]H RI WKH GDWD VDPSOH ZKLOH NHHSLQJ DOO RI WKH LPSRUWDQW LQIRUPDWLRQ
PAGE 42
LI WKH PRPHQWXP RI WKH WUDFN ZLWK KLJKHVW PRPHQWXP fH9LVf LV WKH WRWDO HQHUJ\ GHWHFWHG FKDUJHG WUDFNV DUH DVVXPHG WR EH SLRQVf UHODWLYH WR WKH FHQWHU RI PDVV HQHUJ\ DQG fD&RV7Kf LV WKH ] FRPSRQHQW RI WKH XQLWDU\ WRWDO PRPHQWXP YHFWRU 0RQWH&DUOR SUHGLFWV WKDW DERXW b RI WKH JHQHUDWHG 7 6f fÂ§ n\KKa VLJQDO SDVVHV WKH KDUGZDUH DQG VRIWZDUH WULJJHUV DQG JHWV FODVVLILHG DV KDUG*DP $V PHQWLRQHG DERYH ZH XVH WKLV GDWD WR PDOHH RXU VNLP 7R ZULWH DQ HYHQW IURP WKH KDUG*DP VXEFROOHFWLRQ LQWR RXU VNLP ZH UHTXLUH WKH IROORZLQJ WRSRORJLFDO FXWV f 7KHUH DUH H[DFWO\ WZR fJRRG WUDFNVf WKHUH FDQ EH DQ\ QXPEHU RI WUDFNV WKDW DUH QRW fJRRG WUDFNVf EXW WKHVH DUH QRW XVHG LQ WKH DQDO\VLV :H GHILQH D fJRRG WUDFNf DV D WUDFN WKDW VDWLVILHV WKH IROORZLQJ FXWV GULIW FKDPEHU WUDFN LRQL]DWLRQ HQHUJ\ ORVV G(G;f LQIRUPDWLRQ LV DYDLODEOH WKH UDWLR RI QXPEHU RI ZLUH KLWV WR WKRVH H[SHFWHG LV EHWZHHQ DQG WKH SLRQ ILW KDV [ cGRI KHUH GRI VWDQGV IRU GHJUHHV RI IUHHGRPf DQG WKH GLVWDQFH RI FORVHVW DSSURDFK WR WKH EHDP VSRW LQ WKH [\ SODQH FDOOHG '%&'f LV OHVV WKDQ fÂ§ 3 PPf LI 3 *H9F ZKHUH 3 LV WKH WUDFNV PRPHQWXP LQ *H9Ff DQG OHVV WKDQ PP IRU WUDFNV ZLWK 3 *H9F 7KLV '%&' FXW LV FRPPRQ LQ WKH PRUH VRSKLVWLFDWHG &/(2 ,,,, DQDO\VHV ,W SHUIRUPV EHWWHU WKDQ D VLPSOH '%&' PP FXW EHFDXVH LW WDNHV LQWR DFFRXQW WKH IDFW WKDW WUDNV ZLWK KLJKHU PRPHQWXP KDYH D EHWWHU PHDVXUHPHQW RI '%&' VLQFH WKH\ VFDWWHU OHVV f 7KHUH LV H[DFWO\ RQH fJRRG VKRZHUf WKHUH FDQ EH DQ\ QXPEHU RI VKRZHUV WKDW DUH QRW fJRRG VKRZUHUVf EXW WKHVH DUH QRW XVHG LQ WKH DQDO\VLV :H GHILQH D fJRRG VKRZHUf DV DQ XQPDWFKHG VKRZHU ZLWK HQHUJ\ *H9 7KHVH WRSRORJLFDO FXWV DUH DERXW b HIILFLHQW IRU JHQHUDWHG VLJQDO HYHQWV WKDW KDYH SDVVHG WKH WULJJHUV DQG KDYH EHHQ FODVVLILHG DV KDUG*DP 7KH RYHUDOO VNLP HIILFLHQF\ LV EHWZHHQ b GHSHQGLQJ RQ WKH PRGH VHH 7DEOH f
PAGE 43
$QDO\VLV &XWV $IWHU RXU VNLP ZH FDOO DQ\ FXWV ZH PDNH fDQDO\VLV FXWVf 7KHVH FXWV DUH GRQH DW DQDO\VLV WLPH DQG DUH PRGH GHSHQGHQW $V D FRQYHQWLRQ DQG XQOHVV RWKHUZLVH VWDWHG HIILFLHQFLHV IRU LQGLYLGXDO DQDO\VLV FXWV DUH UHSRUWHG UHODWLYH WR WKH HYHQWV LQ WKH VNLP QRW UHODWLYH WR WKH HYHQWV JHQHUDWHGf PRPHQWXP &XW $OO IXOO\ UHFRQVWUXFWHG HYHQWV VKRXOG KDYH WKH PRPHQWXP RI WKH H7Ha V\VWHP 7KLV FRQVWUDLQW LV XVXDOO\ LPSOHPHQWHG ZLWK D VLPSOH WZRGLPHQVLRQDO $(S ER[ FXW ZKHUH $( LV WKH GLIIHUHQFH EHWZHHQ WKH UHFRQVWUXFWHG HQHUJ\ IRU WKH HYHQW DQG WKH FROOLGLQJ HrHB HQHUJ\ (FPf DQG S LV WKH PDJQLWXGH RI WKH UHFRQVWUXFWHG WRWDO PRPHQWXP IRU WKH HYHQW 7\SLFDO YDOXHV IRU WKHVH FXWV DUH fÂ§ $((FP DQG S 0H9F WDNHQ IURP >@f 7KH WUDGLWLRQDO $(S ER[ FXW LV VRPHZKDW XVHIXO +RZHYHU LW GRHV QRW WDNH LQWR DFFRXQW WKH FRUUHODWLRQ EHWZHHQ WKH PHDVXUHG HQHUJ\ DQG PRPHQWXP ,QGHHG WKH VLJQDO OLHV LQ GLDJRQDO EDQGV LQ WKH $(S SODQH PDNLQJ D ER[VKDSHG FXW QRW RSWLPDO VHH )LJXUHV OD DQG OEf :H XVH DQ DOWHUQDWLYH DSSURDFK WR WKH PRPHQWXP FXW $IWHU D VLPSOH VXEVWLWXWLRQ S\ (FP fÂ§ (NK ZKHUH LV WKH SKRWRQfV HQHUJ\ S LV WKH PDJQLWXGH RI WKH SKRWRQV PRPHQWXP (FP LV WKH HQHUJ\ RI WKH HHa V\VWHP DQG (KK LV WKH HQHUJ\ RI WKH KDGURQ SDLUf ZH FDQ PLWH WKH (S FRQVHUYDWLRQ HTXDWLRQV DV 9KK 7 ^(FP (KK f3L fÂ§ 3FP f ZKHUH 3KK LV WKH GLKDGURQ PRPHQWXP S LV WKH SKRWRQfV PRPHQWXP XQLW YHFWRU DQG SFP LV WKH PRPHQWXP RI WKH H nUHa V\VWHP ZKLFK LV D IHZ 0H9 EHFDXVH RI WKH FURVVLQJ DQJOHf (TXDWLRQ LV D FRQVWUDLQW VXEVHW RI WKH PRPHQWXP FRQVWUDLQW DQG KDV WKH FRQYHQLHQW SURSHUW\ RI DYRLGLQJ WKH XVH RI WKH PHDVXUHG
PAGE 44
SKRWRQfV HQHUJ\ ZKLFK KDV QRQ*DXVVLDQ DV\PPHWULF HUURUV ,W LV LPSRUWDQW WR QRWLFH WKDW (TXDWLRQ FRQWDLQV WKH GLKDGURQ HQHUJ\ WKHUHIRUH LW FDQ KHOS GLVFULPLQDWH EHWZHHQ WKH YDULRXV SDUWLFOH K\SRWKHVHV :H SURFHHG DV IROORZV $IWHU YHUWH[LQJ WKH KDGURQ SDLU XVLQJ WKH EHDP VSRW ZLWK LWV HUURU PDWUL[ ZH FDOFXODWH WKH SKRWRQfV GLUHFWLRQ IURP WKH KDGURQ SDLUfV YHUWH[ DQG WKH VKRZHU SRVLWLRQ :H WKHQ ILW WKH HYHQW WR WKH FRQVWUDLQWV H[SUHVVHG LQ (TXDWLRQ f XVLQJ WKH WHFKQLTXHV RXWOLQHG LQ >@ DQG FXW RQ WKH [ RI WKH FRQVWUDLQW ILW ;HS: 7R FRPSOHWH WKH PRPHQWXP UHTXLUHPHQW ZH FDOFXODWH $(Kf (KK ( (&P ZKHUH (KK LV WKH XSGDWHG GLKDGURQ HQHUJ\ DIWHU WKH FRQVWUDLQW DQG (\ LV WKH PHDVXUHG SKRWRQfV HQHUJ\ DQG UHTXLUH r (FP $(Kf r (FP )XUWKHUPRUH ZH QRZ KDYH DYDLODEOH [_BSILf GLIIHUHQFHV EHWZHHQ GLIIHUHQW SDUWLFOH K\SRWKHVHV ZKLFK KHOS LQ SDUWLFOH LGHQWLILFDWLRQ ,'f 7KLV LV GLVFXVVHG LQ PRUH GHWDLO LQ WKH QH[W VHFWLRQ )LJXUH FRPSDUHV WKH SHUIRUPDQFH RI WKH ROG DQG QHZ DSSURDFKHV WR WKH (S FXW $W WKLV SRLQW LW LV D JRRG LGHD WR FKHFN WKDW WKH UDRPHQWXP FXW UHMHFWV EDFNJURXQG HYHQWV WKDW PDNH LW WKURXJK RXU VNLP FXWV VHH 6HFWLRQ f 7KHVH HYHQWV W\SLFDOO\ KDYH RQH KLJK HQHUJ\ VKRZHU WZR WUDFNV DQG DQf DGGLWLRQDO HOHPHQWVf 2QH VXFK EDFNJURXQG LV 76f fÂ§! UUfUr 2XW RI 7O6nf fÂ§! ZLWK L@ fÂ§! U7BU 0& HYHQWV RQO\ VXUYLYH RXU PRPHQWXP FXW :H FRQFOXGH WKDW RXU PRPHQWXP FXW LV JRRG DW UHMHFWLQJ EDFNJURXQG HYHQWV WKDW SDVV WKH VNLP FXWV EXW KDYH DGGLWLRQDO HOHPHQWV VXFK DV DQ H[WUD SKRWRQ Ur SDLU RI WUDFNV HWF +DGURQ 6HSDUDWLRQ :H GHILQH $;GAL fÂ§ (Âf IRU WKH SDUWLFOH K\SRWKHVHV K? DQG RI RXU FKDUJHG WUDFN SDLU HJ $[e!WW fÂ§ .ff DV IROORZV 7KDQNV WR 9LMD\ 3RWOLD IRU JHQHUDWLQJ WKHVH HYHQWV
PAGE 45
!9nY f r rnrÂ‘r } Y \ f .9n r ?Âf nRn ? R &/ & $( *H9f )LJXUH 'LVWULEXWLRQV IRU GLIIHUHQW PRPHQWXP FXWV 6LJQDO 0& HYHQWV 7/f fÂ§} 7Uff DUH UHSUHVHQWHG E\ WKH EODFN GRWV DQG fEDFNJURXQGf 0& HYHQWV HHf ..af DSSHDU DV UHGf WULDQJOHV 3ORW Df KDV QR FXWV 3ORW Ef KDV WKH ROG PRPHQWXP ER[FXW 3ORW Ff KDV WKH QHZ PRPHQWXP FXWV 3ORW Gf KDV WKH QHZ PRPHQWXP FXWV DQG DOVR D FXW GHILQHG DV [_S0 a ;HS.f 7KH SDUWLFOH ,' SRWHQWLDO RI WKH QHZO\ DYDLODEOH ;(aSKf LV HYLGHQW L a Af fÂ§ 9G(G;AWf a AG(G;AW f DG(G[K f a DG(G[A f RJeMXF+KWff ORJ &ULFKÂ+ff Af ORJeULFK.ff ORJe PFQLKnff ZKHUH WKH LGHD LV WR FRPELQH WKH G(G; DQG 5,&+ LQIRUPDWLRQ LQWR RQH QXPEHU 3DLUV RI WUDFNV ZLWK fÂ§ Kf D[H PRUH OLNHO\ WR EH RI W\SH K WKDQ RI W\SH K ,Q SUDFWLFH ZH RQO\ DGG D WUDFNV 5,&+ LQIRUPDWLRQ LI LWV PRPHQWXP LV DERYH WKH &KHUHQNRY UDGLDWLRQ WKUHVKROG IRU ERWK PDVV K\SRWKHVHV DQG WKHUH DUH DW OHDVW SKRWRQV ZLWKLQ M RI WKH &KHUHQNRY DQJOH IRU DW OHDVW RQH RI WKH PDVV K\SRWKHVLV
PAGE 46
:H DOVR UHTXLUH WKDW ERWK K\SRWKHVHV ZHUH DFWXDOO\ DQDO\]HG E\ 5,&+ GXULQJ SDVV ,Q DGGLWLRQ WR 5,&+ DQG G(G; DQG DV KLQWHG LQ WKH SUHYLRXV VHFWLRQ WKH GLIIHUHQFH LQ [bSKf IURP WKH FRQVWUDLQW H[SUHVVHG LQ (TXDWLRQ FDQ KHOS WKH SDUWLFOH ,' VHH )LJXUHV OF DQG OGf :H GHILQH $[/SIfL Kf ;(SIWLf ;HAf f (YHQWV ZLWK $\eBSLL f DUH PRUH OLNHO\ WR EH RI W\SH K? WKDQ R W\SH K ,Q WKLV DQDO\VLV WR VHOHFW K? DQG UHMHFW K WKH GHIDXOW FXW LV VLPSO\ fÂ§ Kf 7KLV VLPSOH FXW LV KLJKO\ HIILFLHQW KDV ORZ IDNH UDWHV DQG LV VRPHWLPHV VXIILFLHQW +RZHYHU RXW RI WKH VL[ SRVVLOGH FDVHV ZKHQ RQH KDGURQ IDNHV DQRWKHU WKHUH DUH WKUHH LPSRUWDQW FDVHV ZKHUH LW SD\V RII WR DOVR XVH $[_BSKL fÂ§ Kf WRJHWKHU ZLWK $['LL fÂ§ f LQ DQ RSWLPDO ZD\ 7 EDFNJURXQG WR 7KLV EDFNJURXQG FRPHV IURP WKH FRQWLQXXP SURFHVV HHB fÂ§ S S fÂ§ UUa 7 EDFNJURXQG WR S $JDLQ WKLV EDFNJURXQG FRPHV IURP WKH FRQWLQXXP SURFHVV HHf fÂ§! S S fÂ§} UUa EDFNJURXQG WR S 7KLV EDFNJURXQG FRPHV IURP WKH FRQWLQXXP SURFHVV HHB fÂ§ fÂ§! ..a ,Q RWKHU ZRUGV WKH LPSRUWDQW FDVHV RFFXU ZKHQ WKH OLJKWHU PDVV K\SRWKHVLV IDNHV WKH KHDYLHU PDVV K\SRWKHVLV 0DWKHPDWLFDOO\ RQH ZRXOG H[SHFW WKDW VLPSO\ DGGLQJ ERWK $;MG DQG $\_BS WRJHWKHU OLNH ZH MXVW GLG ZKHQ FRPELQLQJ 5,&+ DQG G(G;f DQG FXWWLQJ RQ WKH JUDQG $\ LV WKH ZD\ WR JR 8QIRUWXQDWHO\ WKLV VLPSOH DSSURDFK IDLOV EHFDXVH RI ODUJH QRQPDWKHPDWLFDO WDLOV LQ WKH LQGLYLGXDO ;HS GLVWULEXWLRQV ,QVWHDG IRU HDFK RI WKHVH WKUHH FDVHV ZH GHILQH WKH EHVW FXW YDOXHV f LQ $;LG FL D5G $;HS F DV WKRVH WKDW PD[LPL]H
PAGE 47
)&OFf fÂ§ 5f f A5H :r5 ZKHUH 5H 5If LV WKH HIILFLHQF\ IDNH UDWHf RI WKH SDUWLFOH ,' FXWV DQG : LV WKH URXJK UDWLR RI WKH EDFNJURXQG WR VLJQDO LQ WKH GDWD VDPSOH IRU HDFK FDVH (DFK : FDQfW EH NQRZQ D SULRUL EXW D URXJK LGHD RI LWV YDOXH FDQ EH REWDLQHG E\ GRLQJ D ILUVW LWHUDWLRQ RI WKH DQDO\VLV ZLWK IRU H[DPSOH FL F :H XVH : : fÂ§ DQG : fÂ§ IRU FDVHV UHVSHFWLYHO\ )LJXUH VKRZV )FL Ff IRU HDFK FDVH 7DEOH VKRZV WKH RSWLPL]HG FXW YDOXHV DQG WKHLU HIIHFW RQ SDUWLFOH ,' F )LJXUH &RQWRXU SORWV RI )FL Ff DV GHILQHG LQ (TXDWLRQ IRU GLIIHUHQW VLJn QDOV DQG EDFNJURXQGV IURP IODW 0& fRS KDV D VLJQDO DQG D U EDFNJURXQG &HQn WHU KDV D S VLJQDO DQG D YU EDFNJURXQG %RWWRP KDV D S VLJQDO DQG D EDFNJURXQG 4XDQWXP (OHFWURG\QDPLF 6XSSUHVVLRQ 4XDQWXP (OHFWURG\QDPLF 4('f EDFNJURXQG LQ WKLV DQDO\VLV FRPHV IURP WKH DEXQGDQW SURFHVVHV HHa fÂ§! HHff DQG HHf fÂ§ SgSa
PAGE 48
7DEOH &XW YDOXHV HIILFLHQFLHV DQG IDNH UDWHV IRU $[`' &? DQG $;HS F LQ IODW 0& 7KH GLIIHUHQW FL & YDOXHV DUH FKRVHQ VR ) IURP (TXDWLRQ LV PD[LPXQ (IILFLHQF\ DQG IDNH UDWHV RI HDFK FXW DUH UHSRUWHG UHODWLYH WR HYHQWV LQ WKH VNLP 6WDWLVWLFDO HUURUV LQ WKH HIILFLHQFLHV DUH b RU OHVV (UURUV LQ WKH IDNH UDWHV DUH VWDWLVWLFDO DQG DUH VKRZQ IRU FRPSOHWHQHVV RQO\ &XW 9DOXH &O F 6LJQDO HIILFLHQF\ bf D;G D;HY %RWK )DNH UDWH bf $[_ %RWK U IDNLQJ s s s U IDNLQJ S s s s IDNLQJ S s s s 7R UHMHFW HH fÂ§! AHHa ZH UHTXLUH ERWK WUDFNV WR KDYH D PDWFKHG VKRZHU ZLWK HQHUJ\ ( VXFK WKDW ?(SQf fÂ§ DQG WR KDYH $[M'K fÂ§ Hf 7R UHMHFW HHa fÂ§! MSSa ZH VLPSO\ UHTXLUH 37+ 08 37+ 08 ZDV GHILQHG LQ 6HFWLRQ f IRU ERWK WUDFNV LQ WKH DQG S PRGHV EHFDXVH SDUWLFOH ,' FXWV PDNH WKH SLRQ DQG WKHUHIRUH WKH PXRQf IDNH UDWH VPDOO VHH 7DEOHV DQG f )RU WKH U PRGH ZH FDQQRW XVH SDUWLFOH ,' LQ D SUDFWLFDO ZD\ EHFDXVH PXRQV DQG SLRQV KDYH VLPLODU PDVVHV ,QVWHDG WR VHSDUDWH SLRQV DQG PXRQV ZH XVH D PXFK VWURQJHU FXW UHTXLULQJ WKDW ERWK WUDFNV EH ZLWKLQ WKH EDUUHO SDUW RI WKH PXRQ FKDPEHU FRVf_ f ERWK KDYH 3 *H9F DQG ERWK KDYH 37+ 08 7R LPSURYH WKH RYHUDOO PXRQ VXSSUHVVLRQ FXW HIILFLHQF\ ZLWK YLUWXDOO\ QR LQFUHDVH LQ PXRQ IDNHV ZH IODJ DQ HYHQW DV fQRW PXRQLFf LI DQ\ RI WKH WUDFNV GHSRVLW PRUH WKDQ 0H9 LQ WKH && 7KLV LQFUHDVHV WKH FXW HIILFLHQF\ E\ DERXW b LQ WKH Q PRGH DQG PDNHV WKH GHWHFWRU PRUH KHUPHWLF 2WKHU DQDO\VHV IRU H[DPSOH >@f W\SLFDOO\ XVH 37+ 08 RXU &/(2 ,,, 0& KDV WRR PDQ\ SLRQ WUDFNV ZLWK '37+08 DQG WZR IHZ WUDFNV ZLWK '37+08
PAGE 49
&XW 6XPPDU\ (IILFLHQFLHV DQG )DNH 5DWHV 7DEOH VXPPDUL]HV WKH FXWV XVHG LQ WKLV DQDO\VLV )LJXUH VKRZV WKH RYHUDOO 0RQWH &DUOR HIILFLHQF\ DIWHU DOO FXWV )LJXUH VKRZV WKH IDNH UDWHV DFFRUGLQJ WR WKH 0& DQG WKH GDWD IRU GLIIHUHQW SDUWLFOH ,' FXWV 7KH GDWD IDNH UDWHV DQG OLPLW IDNH UDWHV ZKLFK DUH PHDVXUHG IRU SLRQV DQG NDRQV IDNLQJ RWKHU KDGURQV DUH FDOFXODWHG IURP WKH S DQG SHDNV LQ WKH FRQWLQXXP VHH )LJXUHV DQG f 7DEOHV WKURXJK VXPPDUL]H WKH UHVXOWV RI )LJXUHV DQG
PAGE 50
7DEOH &XWV XVHG LQ WKLV DQDO\VLV 0RWLYDWLRQ 'HILQLWLRQ IRU 76f fÂ§r KK ZKHUH K Q RU Sf 6NLP FXWV 'DWD DFTXLVLWLRQ (YHQW PXVW SDVV KDUGZDUH /HYHOOf DQG VRIWZDUH WULJJHUV SDVVf DQG EH RI W\SH KDUG*DP 7RSRORJLFDO 7KHUH DUH RQO\ WZR JRRG WUDFNV DQG RQO\ RQH LVRODWHG VKRZHU ZLWK ( *H9 $QDO\VLV FXWV 5HFRQVWUXFWHG HYHQW PXVW KDYH PRPHQWXP RI WKH FHQWHURIPDVV V\VWHP ;(3Kf DQG $(^Kf(&0 +DGURQ VHSDUDWLRQ 'HIDXOW LV $[M' 7KH WKUHH FDVHV ZLWK D ODUJH IDNH SRSn XODWLRQ EHFDXVH RI FRQWLQXXP SURFHVVHV XVH D VLPXOWDQHRXVO\ RSWLPL]HG FXW RQ $[M' DQO $;HS! DQG DUH VXPPDUL]HG LQ 7DEOH 4(' EDFNJURXQG HHa fÂ§r MHHa %RWK WUDFNV KDYH D PDWFKHG VKRZHU HQHUJ\ ( WKDW VDWLVILHV ?(SWWf fÂ§ DQG $;LGÂA a Hf 4(' EDFNJURXQG HH fÂ§! Sa )RU K S ERWK WUDFNV KDYH 37+ 08 )RU K Q $W OHDVW RQH WUDFN KDV D PDWFKHG VKRZHU HQHUJ\ 0H9f RU %RWK WUDFNV KDYH FRV2f DQG 3 *H9Ff DQG ERWK WUDFNV KDYH '37+08 ff
PAGE 51
7DEOH (IILFLHQFLHV LQ b IRU FXWV DV RXWOLQHG LQ 7DEOH f IRU IODW VLJQDO 0& (IILFLHQFLHV LQ WKH VHFRQG JURXS DUH UHSRUWHG UHODWLYH WR WKH QXPEHU RI FDQGLGDWHV WKDW PDNH LW WR WKH VNLP 7KH WKLUG SDUW RI WKH WDEOH VKRZV WKH RYHUDOO UHFRQVWUXFn WLRQ HIILFLHQF\ 6WDWLVWLFDO HUURUV DUH b RU OHVV D 6NLP FXWV 7 3 +DUGZDUH 7ULJJHU 6RIWZDUH 7ULJJHU KDUG*DP 7RSRORJLFDO QWUDFNV t ( $*H9f 2YHUDOO 6NLP HIILFLHQF\ $QDO\VLV FXWV 7 9 PRPHQWXP 4(' HHa fÂ§! \HH VXSSUHVLRQ 4(' QQa fÂ§! [La VXSSUHVLRQ +DGURQ VHSDUDWLRQ 2YHUDOO DQDO\VLV HIILFLHQF\ 2YHUDOO UHFRQVWUXFWLRQ HIILFLHQF\ 7DEOH (IILFLHQFLHV LQ b IRU FXWV DV RXWOLQHG LQ 7DEOH f IRU IODW 6 0& (IILn FLHQFLHV LQ WKH VHFRQG JURXS D[H UHSRUWHG UHODWLYH WR WKH QXPEHU RI FDQGLGDWHV WKDW PDNH LW WR WKH VNLP 7KH WKLUG SDUW RI WKH WDEOH VKRZV WKH RYHUDOO UHFRQVWUXFWLRQ HIILFLHQF\ 6WDWLVWLFDO HUURUV DUH b RU OHVV 6NLP FXWV U 3 +DUGZDUH 7ULJJHU 6RIWZDUH 7ULJJHU KDUG*DP 7RSRORJLFDO QWUDFNV t ( $*H9f 2YHUDOO 6NLP HIILFLHQF\ $QDO\VLV FXWV U 3 PRPHQWXP 4(' HHB fÂ§r HHa VXSSUHVLRQ 4(' LQa fÂ§! AnOnLB VXSSUHVLRQ +DGURQ VHSDUDWLRQ 2YHUDOO DQDO\VLV HIILFLHQF\ 2YHUDOO UHFRQVWUXFWLRQ HIILFLHQF\
PAGE 52
7DEOH )LQDO HIILFLHQFLHV DQG IDNH UDWHV DIWHU DOO FXWV LQ b 6WDWLVWLFDO HUURUV IRU HIILFLHQFLHV DUH b RU OHVV 6WDWLVWLFDO HUURUV IRU IDNH 0& UDWHV DUH VKRZQ IRU FRPSOHWHQHVV RQO\ 7KH ,6 '$7$ FRUUHVSRQGV WR WKH ,6 RII UHVRQDQFH GDWD VDPSOH 7KH 0& ZDV JHQHUDWHG IODW 7 FXWV FXWV S FXWV ,6 U 0& s 6 7 0& s ,6 S '$7$ 6 S '$7$ Â‘ s ,6 0& s 6 0& s ,6 M! '$7$ 6 S '$7$ s ,6 S 0& s s 6 S 0& s s ,QYDULDQW PDVV *H9Ff )LJXUH )LQDO HIILFLHQFLHV IRU HDFK PRGH DV D IXQFWLRQ RI LQYDULDQW PDVV IRU WKH ,6 VROLGf DQG 6 GDVKHGf 0RQWH &DUOR GDWD 7KH 0& ZDV JHQHUDWHG ZLWK D IODW DQJXODU GLVWULEXWLRQ
PAGE 53
)LJXUH )DNH UDWHV IRU ,6 KROORZ FLUFOHVf DQG 6 KROORZ VTXDUHVf DFFRUGLQJ WR IODW 0& 7KH UHGf GRZQZDUG SRLQWLQJ WULDQJOH LV REWDLQHG XVLQJ WKH S SHDN LQ 6 GDWD
PAGE 54
&+$37(5 (;&/86,9( 5$',$7,9( '(&$< 76f r \QLUa ,Q )LJXUH WKH QQa LQYDULDQW PDVV SORW LV VKRZQ IRU ERWK ,6 DQG 6 GDWD )LJXUH VKRZV WKH FRQWLQXXP VXEWUDFWHG 7ULUa LQYDULDQW PDVV GLVWULEXWLRQ DV GHILQHG LQ 6HFWLRQ f ZLWK WKH PRVW OLNHO\ VWDWLVWLFDO ILW RYHUOD\HG ZKLFK LV GHVFULEHG LQ WKH QH[W VHFWLRQf 7KH QXPEHU RI HYHQWV ZLWKLQ LU RI WKH S UHJLRQ fÂ§ *H9Ff OHIW DIWHU WKH FRQWLQXXP VXEWUDFWLRQ LV s DQG RI WKHVH EHORQJ WR WKH Rf ORZPDVV WDLO 5REXVWQHVV RI 7KH 0DVV 'LVWULEXWLRQ ,Q >@ WKH GHFD\ 76f UrUr LV DQDO\]HG DQG LW LV VKRZQ KRZ WKH DQDO\VLV VWUHDP ZDUSV WKH VKDSH RI WKH UHFRQVWUXFWHG UHVRQDQFH 7KLV HIIHFW ZKLFK DULVHV EHFDXVH RI WKH SDUWLFXODU Ur EHKDYLRU UDLVHV SUREOHPV ZKHQ ILWWLQJ WKH LQYDULDQW PDVV GLVWULEXWLRQ ,Q 6HFWLRQ ZH FODLPHG WKDW LI WKH GDWD DQG 0& KDG WKH VDPH DQG [ GLVWULEXWLRQV WKH ILW WR WKH FRQWLQXXP VXEWUDFWHG LQYDULDQW PDVV GLVWULEXWLRQ DXWRPDWLFDOO\ JLYHV WKH FRUUHFW HIILFLHQF\ FRUUHFWHG QXPEHU RI HYHQWV +HUH ZH WHVW WKLV FODLP 7R WKLV HQG ZH JHQHUDWH 76f fÂ§! f ZLWK Af fÂ§! UUa ZLWK IODW DQG Q GLVWULEXWLRQV :H WUHDW WKLV 0& DV GDWD DQG FDUU\ RXW WKH ILUVW WZR VWHSV RI WKH FRQWLQXXP VXEWUDFWLRQ SURFHVV 7KH UHVXOWLQJ SHDN KDV D PDVV RI s *H9F DQG D ZLGWK RI s *H9F FRQVLVWHQW ZLWK WKH JHQHUDWHG PDVV DQG ZLGWK RI *H9F DQG *H9F 0RUH LPSRUWDQWO\ WKH QXPEHU RI UHFRQVWUXFWHG HYHQWV IURP WKH ILW LV s ZKLFK LV FRQVLVWHQW ZLWK WKH QXPEHU RI JHQHUDWHG HYHQWV )LJXUH VKRZV WKH UHFRQVWUXFWHG DQG HIILFLHQF\ FRUUHFWHG HYHQWV D ILW WR WKHP DQG WKH JHQHUDWHG HYHQWV D
PAGE 55
:H FRQFOXGH WKDW WKHUH LV QR ZDUSLQJ RI WKH PDVV GLVWULEXWLRQ DQG WKDW WKH DQDO\VLV VWUHDP EHKDYHV OLNH ZH H[SHFW ZKHQ REWDLQLQJ WKH HIILFLHQF\ FRUUHFWHG QXPEHU RI HYHQWV IURP GDWD Â‘ 5HFRQVWUXFWHG DQG (II &RUUHFWHG UfÂ§0& *HQHUDWHG )LJXUH 5HFRQVWUXFWHG HYHQWV DQG HIILFLHQF\ FRUUHFWHG HYHQWV D ILW WR WKHP DQG WKH JHQHUDWHG HYHQWV IRU <6f fÂ§} f ZLWK f fÂ§! UUa ZLWK IODW DQG 2Q GLVWULEXWLRQV 6WDWLVWLFDO )LW RI WKH ,QYDULDQW 0DVV 'LVWULEXWLRQ 6LJQDO $UHDV DQG 7KHLU 6LJQLILFDQFH 7KH UHVXOWV RI WKLV VHFWLRQ DUH VXPPDUL]HG LQ 7DEOH )LJXUH VKRZV SRVVLEOH VLJQDOV IRU IRU 7 f fÂ§! f 7 f fÂ§! f DQG 7f fÂ§! f 7R GHWHUPLQH WKH QXPEHU RI HYHQWV LQ HDFK VLJQDO ZH ILW WKH LQYDULDQW QQa PDVV FRQWLQXXP VXEWUDFWHG GLVWULEXWLRQ ZLWK WKUHH VSLQGHSHQGHQW UHODWLYLVWLF %UHLW:LJQHU OLQH VKDSHV 7KH PDVVHV DQG ZLGWKV DUH DOORZHG WR IORDW H[FHSW IRU WKH ZLGWK RI WKH f ZKLFK KDV D YHU\ ODUJH HUURU LI DOORZHG WR IORDW DQG LV VHW WR LWV 3'* YDOXH 7KH 3'* YDOXHV>@ IRU WKH PDVV DQG ZLGWK RI WKH WKUHH UHVRQDQFHV DUH PMf s 0H9F 3Rf fÂ§ s 0H9F PK^f s 0H9F U0f A 0H9F UDf s 0H9F DQG 7f s 0H9F
PAGE 56
7R PHDVXUH WKH VWDWLVWLFDO VLJQLILFDQFH RI HDFK VLJQDO ZH GR PXOWLSOH [ ILWV IL[LQJ WKH VLJQDO DUHD WR GLIIHUHQW YDOXHV ZKLOH OHWWLQJ WKH PDVV DQG ZLGWK RI WKH VLJQDO ZKRVH VLJQLILFDQFH LV EHLQJ PHDVXUHG WR IORDW ZLWKLQ D RI WKH 3'* YDOXHV $W WKH VDPH WLPH WKH UHVW RI WKH ILW SDUDPHWHUV DUH IL[HG WR WKH YDOXHV WKDW RULJLQDOO\ PLQLPL]HG WKH [ :H DVVLJQ HDFK RI WKHVH PXOWLSOH ILWV D SUREDELOLW\ SURSRUWLRQDO WR Hr DQG WKHQ QRUPDOL]H :H FDOFXODWH WKH FKDQFH RI WKH VLJQDO EHLQJ GXH WR D UDQGRP IOXFWXDWLRQ E\ DGGLQJ WKH QRUPDOL]HG SUREDELOLWLHV IRU WKH ILWV ZLWK D QHJDWLYH RU VLJQDO 7KLV PHWKRG IDLOV IRU WKH KLJKO\ VLJQLILFDQW f VLJQDO EHFDXVH WKH Ha[ YDOXH RI ILWV ZLWK QHJDWLYH RU f VLJQDO LV RI WKH RUGHU RI DQG RXU FRPSXWLQJ VRIWZDUH FDQ RQO\ KDQGOH QXPEHUV DV VPDOO DV &7 )RU FRPSOHWHQHVV ZH VWDWH WKDW WKH VLJQLILFDQFH RI WKLV VLJQDO LV 7R PHDVXUH WKH XSSHU OLPLW IRU WKH Mf ZH DOVR GR PXOWLSOH [ ILWV IRU GLIIHUHQW IL[HG VLJQDO YDOXHV ZKLOH NHHSLQJ LWV PDVV DQG ZLGWK FRQVWDQW DW PIMf *H9F DQG Af 0H9F DV LQ >@ 7KH UHVXOWLQJ SUREDELOLW\ SORW LV VKRZQ LQ )LJXUH :H ILQG FOHDU HYLGHQFH IRU WKH f HYLGHQFH IRU WKH Rf DQG ZHDN HYLGHQFH IRU WKH f :H DOVR SXW D b FRQILGHQFH OHYHO XSSHU OLPLW RQ IM f SURGXFWLRQ )LW UHVXOWV DUH VKRZQ LQ 7DEOH 7DEOH 5HVXOWV IRU 7f fÂ§! UUa 7KH EUDQFKLQJ IUDFWLRQV RI f DQG f DUH WDNHQ IURP WKH 3'* >@ (UURUV VKRZQ DUH VDWLVWLFDO RQO\ 0RGH $UHD %),2f 6LJQLILFDQFH Rf Rf UUB Â6 ILXn f s s O2 Df f s s [ a Df fMf fÂ§! UU [ ,4 $QJXODU 'LVWULEXWLRQ RI 7KH 6LJQDO 7KH KHOLFLW\ DQJOH GLVWULEXWLRQV RI Q DQG DUH GHILQHG DQG GHVFULEHG LQ WKH 1 DSSHQGL[ ,Q WKLV VHFWLRQ ZH ILUVW REWDLQ WKH KHOLFLW\ DQJXODU GLVWULEXWLRQV RI WKH
PAGE 57
Rf f DQG f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f FXW DURXQG WKH PHDQ PDVV RI WKH UHVRQDQFH FDQ EH FKRVHQ DV D fVWDQGDUGf LQWHUYDO :H FRXOG SURFHHG WKLV ZD\ EXW LQ RXU FDVH EHFDXVH RI WKH ODUJH VXEWUDFWLRQV LQYROYHG ZKHQ REWDLQLQJ WKH DQJXODU GLVWULEXWLRQ D FRQVLGHUDEOH LQFUHDVH VWDWLVWLFDO VLJQLILFDQFH RI HDFK ELQ LQ WKH KHOLFLW\ DQJXODU GLVWULEXWLRQ FDQ EH DFKLHYHG E\ FKRRVLQJ WKH PDVV LQWHUYDO FDUHIXOO\ VHH WKH ODVW FROXPQ RI 7DEOH f /HWfV FRQVLGHU IRU H[DPSOH WKH Rf DQJXODU GLVWULEXWLRQ :H EHJLQ ZLWK WKH 7f DQG WKH 76f HIILFLHQF\ FRUUHFWHG GLVWULEXWLRQV %HIRUH WKH FRQWLQXXP VXEWUDFWLRQ HDFK ELQ LQ WKH Rf DQJXODU GLVWULEXWLRQ KDV FRQWULEXWLRQV IURP WKH KLJK HQG PDVV WDLO RI WKH S DQG WKH ORZ PDVV WDLO RI WKH f ,Q RUGHU WR
PAGE 58
JHW WKH ILQDO DQJXODU GLVWULEXWLRQ ERWK RI WKHVH FRQWULEXWLRQV DUH WDNHQ DZD\ E\ ILUVW GRLQJ D FRQWLQXXP VXEWUDFWLRQ XVLQJ WKH VFDOH IDFWRU D DQG WKHQ E\ VXEWUDFWLQJ WKH f GLVWULEXWLRQ RXWVLGH WKH PDVV LQWHUYDO VFDOHG E\ DQ DSSURSULDWH VFDOH IDFWRU HTXDO WR WKH UDWLR RI f LQVLGH WKH PDVV LQWHUYDO DQG RXWVLGH RI LW $IWHU DOO WKLV WKH FRQWULEXWLRQV WR WKH ELQV LQ WKH ILQDO DQJXODU GLVWULEXWLRQ DUH 1If fÂ§ "9Rf mS7O6ff 8If a :S76ff f f a QRf f ZKHUH QRfQLff LV WKH QXPEHU RI Rfff REWDLQHG E\ LQWHJUDWLQJ WKH ILWWHG VSLQGHSHQGHQW UHODWLYLVWLF %UHLW:LJQHU IXQFWLRQ LQVLGH WKH PDVV LQWHUYDO QS7LVffIWS76fff LV WKH QXPEHU RI SfV LQVLGH WKH PDVV LQWHUYDO IURP FRQWLQXXP HYHQWV DW WKH 7f76ff HQHUJ\ DQG QnIL? LV WKH QXPEHU RI f RXWVLGH WKH PDVV LQWHUYDO EHLQJ XVHG WR VXEWUDFW WKH FRQWULEXWLRQ RI f WR WKH Rf DQJXODU GLVWULEXWLRQ )ROORZLQJ WKLV ODVW GHILQLWLRQ Q0OfQnIff (DFK RI WKHVH WHUPV KDV DQ DVVRFLDWHG HUURU $VVXPLQJ WKDW $Q ?IQ WKDW WKH HIILFLHQF\ FRUUHFWLRQ KDV LQILQLWH VWDWLVWLFV DQG LJQRULQJ WKH HUURUV RQ WKH FRQWLQXXP VFDOH IDFWRU D WKH RYHUDOO HUURU LV $L9Rf ?7LIR f 4 fULS76ff fULILf :H DUULYH WR WKH FRQFOXVLRQ WKDW WKH PDVV LQWHUYDO P fÂ§ $P P $Pf ZKLFK SURGXFHV WKH KHOLFLW\ DQJOH GLVWULEXWLRQ ZLWK VPDOOHVW UHODWLYH ELQHUURUV LV WKH RQH 6WULFWO\ VSHDNLQJ WKLV LV FRUUHFW RQO\ WR ILUVW RUGHU 7KH f GLVWULEXWLRQ LWVHOI KDV D VPDOO FRQWDPLQDWLRQ IURP f DQG f 7KLV HIIHFW ZKLFK ZH FDOO FURVVFRQWDPLQDWLRQ LV LJQRUHG LQ (TXDWLRQ /DWHU LQ VHFWLRQ ZH ZLOO VKRZ KRZ WR HOLPLQDWH WKLV VPDOO FURVVFRQWDPLQDWLRQ XVLQJ DOO WKH UHVRQDQFHV 7KH GLIIHUHQFH LQ KRZ ZH DFWXDOO\ JHW WKH KHOLFLW\ GLVWULEXWLRQV DQG (TXDWLRQ KDV DQ LQVLJQLILFDQW HIIHFW ZKHQ FDOFXODWLQJ WKH RSWLPDO PDVV LQWHUYDO
PAGE 59
\ WKDW PD[LPL]HV )IfP $Pf QRf ?$Rf r mfmS76ff AfQLf f 7KH SORW RI )2fP $Pf LV VKRZQ LQ ILJXUH 7KLV VDPH WHFKQLTXH FDQ EH DSSOLHG WR WKH f DQG f UHVRQDQFHV 5HVXOWV DUH VKRZQ LQ 7DEOH 7DEOH 0HDQ PDVVHV ZLGWKV LQ *H9F DQG LQYHUVH RI WKH DYHUDJH ELQ UHODWLYH HUURU )f IURP EDFNJURXQG VXEWUDFWLRQV IRU WKH DQJXODU GLVWULEXWLRQ RI GLIIHUHQW UHVRQDQFHV 6WDQGDUG PHDQ PDVVHV DQG ZLGWKV FRUUHVSRQGLQJ WR ,7 DUH WDNHQ IURP WKH ILW LQ )LJXUH DQG DUH ODEHOHG ZLWK WKH VXEVFULSW fVf ZKLOH WKRVH WKDW PD[LPL]H ) DUH ODEHOHG ZLWK WKH VXEVFULSW fPf 7KH ODVW FROXPQ VKRZV WKH IDFWRU E\ ZKLFK WKH HIIHFWLYH VWDWLVWLFV LQFUHDVH 5HVRQDQFH PV $PVf )PV $Pf 7QP $PPf )PP $UQPf )PV$PDf Rf f f f f f f f f %DFNJURXQG 6XEWUDFWLRQ 7KH FRQWLQXXP VXEWUDFWLRQ RI WKH KHOLFLW\ DQJOH GLVWULEXWLRQ LV GHILQHG LQ 6HFWLRQ 7KH VXEWUDFWLRQ RI WKH WDLOV IURP RWKHU UHVRQDQFHV UHTXLUHV D FORVHU ORRN /HW XV FDOO WKH FRQWLQXXP VXEWUDFWHG KHOLFLW\ DQJOH GLVWULEXWLRQ RI WKH HYHQWV LQ WKH Rf f DQG f PDVV LQWHUYDOV &Rf &LRf DQG Ff UHVSHFWLYHO\ /HW XV FDOO WKH KHOLFLW\ DQJOH GLVWULEXWLRQ RI WKH HYHQWV WKDW FRPH H[FOXVLYHO\ IURP WKH UHVRQDQFH ZH DUH WU\LQJ WR VHOHFW LQ WKH VDPH PDVV LQWHUYDOV WKDW LV WKH WUXH KHOLFLW\ DQJOH GLVWULEXWLRQ RI WKH UHVRQDQFHf Â2f eLf! DQG ÂKLf UHVSHFWLYHO\ 7KH FRQWLQXXP VXEWUDFWHG KHOLFLW\ DQJOH GLVWULEXWLRQ RI D
PAGE 60
UHVRQDQFH LV EHLQJ FRQWDPLQDWHG E\ WKH WDLOV RI WKH RWKHU UHVRQDQFHV .HHSLQJ WKLV LQ PLQG ZH ZULWH FRf LRf SWI f n &Lf fÂ§ Rf Af WI$f &f Â2f f :KHUH WKH VPDOO QXPEHUV DQG H DUH WKH UDWLRV RI WKH QXPEHU RI HYHQWV IURP D UHVRQDQFH LQ WKH PDVV LQWHUYDO ZKHUH WKH FRQWDPLQDWLRQ LV WDNLQJ SODFH WR WKH QXPEHU RI HYHQWV IURP WKH VDPH UHVRQDQFH LQ WKH PDVV LQWHUYDO XVHG WR VHOHFW LW 8VLQJ WKH PP $UDPf YDOXHV LQ 7DEOH DQG WKH ILW LQ )LJXUH ZH REWDLQ 3 [ a [ O2 [ .7 DQG H 7R REWDLQ WKH EDFNJURXQG VXEWUDFWHG KHOLFLW\ DQJXODU GLVWULEXWLRQV ZH VLPSO\ LQYHUW WKH V\VWHP RI HTXDWLRQV H[SUHVVHG LQ (TXDWLRQ 7KH VROXWLRQ FDQ EH FRQYHQLHQWO\ H[SUHVVHG DV Af a7A&Rf ARf a Â&ff eRf &f a f f Af &f fÂ§ HD f $V D FKHFN LI ZH LJQRUH WKH VHFRQG RUGHU WHUPV WKH SUHYLRXV VROXWLRQ EHFRPHV +HUH WKH FURVV FRQWDPLQDWLRQ EHWZHHQ WKH Rf DQG WKH f LV LJQRUHG 7KHUH LV QR PDWKHPDWLFDO SUREOHP LQ LQFOXGLQJ WKHVH FRQWDPLQDWLRQ WHUPV EXW WKH V\VWHP RI HTXDWLRQV ZRXOG EH PRUH FRPSOLFDWHG WKDQ LW QHHGV WR EH VLQFH WKLV W\SH RI FURVV FRQWDPLQDWLRQ LV QHJOLJLEOH
PAGE 61
UÂ GI?r r \ f r \ f f &f a &Rf f Â&f Rf FRf A& f Af &f a HFOf ZKLFK LV LQGHHG LV WKH VROXWLRQ ZKHQ FURVVFRQWDPLQDWLRQ LV LJQRUHG 6WDWLVWLFDO )LW RI WKH +HOLFLW\ $QJXODU 'LVWULEXWLRQ $VVLJQn PHQWV DQG WKHLU 3UREDELOLW\ 'LVWULEXWLRQ )RU HDFK UHVRQDQFH ZH ILW WKH GDWD WR WKH VLPXOWDQHRXV FRV DQG FRV Q KHOLFLW\ DQJOH GLVWULEXWLRQ REWDLQHG IURP GDWD XVLQJ (TXDWLRQ WR WKH KHOLFLW\ IRUPDOLVP SUHGLFWLRQ (TXDWLRQV SURMHFWHG RQ HDFK DQJOH DQG IROGHG LQ RSSRVLWH GLUHFWLRQV DURXQG WKHLU V\PPHWU\ D[LV LQ RUGHU WR VKRZ ERWK GLVWULEXWLRQV RQ WKH VDPH SORW 7KH FRUUHVSRQGLQJ YDOXH RI WKH EHVW ILW IRU HDFK UHVRQDQFH VKRZQ LQ )LJXUHV WKURXJK LV GHILQHG DV WKH DVVLJQPHQW Xf :H REWDLQ D IRU WKH Rf ZKLFK LV LQFRQVLVWHQW ZLWK WKH NQRZQ VSLQ RI WKH Rf ZKLFK LV fÂ§ )RU WKH f DQG WKH f ZH REWDLQ D fÂ§ DQG D UHVSHFWLYHO\ ZKLFK LV FRQVLVWHQW ZLWK WKHLU NQRZQ VSLQV 7R KDYH DQ LGHD RI KRZ ZHOO WKH DQJXODU GLVWULEXWLRQ GHWHUPLQHV DPRQJ WKH K\SRWKHVHV ZH GR D VWDWLVWLFDO ILW IRU HDFK K\SRWKHVLV DQG DVVLJQ HDFK RQH D SUREDELOLW\ SURSRUWLRQDO WR HALnDAA ZKHUH GRI DUH WKH GHJUHHV RI IUHHGRP LQ WKH ILW 7KH UHVXOWLQJ QRUPDOL]HG SUREDELOLW\ GLVWULEXWLRQV JLYH DQ LGHD RI WKH DVVLJQHG VLJQLILFDQFH DQG DUH VKRZQ LQ )LJXUH ,Q SDUWLFXODU WKLV ILJXUH VKRZV WKDW WKH D IRU WKH Rf LQFRQVLVWHQF\ FDQ EH GXH WR D VWDWLVWLFDO IOXFWXDWLRQ
PAGE 62
)LOH fERKUXVHQOXLVGHRQWSGDWDBQWS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 5,& + ( 0DVV *H9Ff )LJXUH ,QYDULDQW PDVV RI UU IRU ,6 WRSf DQG 6 ERWWRPf GDWD )RU WKH 6 GDWD ZH VKRZ D EORZXS RI WKH PDVV UHJLRQ fÂ§ *H9F ZKHUH WKH Sr FDQ EH VHHQ
PAGE 63
9L \rr r n 0,18,7 [ )LW WR 3ORW t 8O6*DPPD K K PDVVSLSL D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW 8/ )LW 6WDWXV ('0 ( ; IRU GRI &/ b (UURUV )XQFWLRQ I 3DUDEROLF 0LQRV 02' s 0($1 s ( ( ( :,'7+ ( s ( ( n ( r 63,1 222222(22 s 22222(22 22222(22 22222(22 r 7+5(6+2/' s 22222(22 22222(22 )XQFWLRQ 02' 0($1 s ( ( :,'7+ ( r 63,1 s 2222(A22 22222(22 22222(22 r 7+5(6+2/' s 22222(22 ( 22222(22 )XQFWLRQ W 02' s 0($1 s ( ( ( r :,'7+ W ( W 2222(22 22222(22 r 63,1 s ( 22222(22 22222(22 r 7+5(6+2/' s 2222(22 22222(22 22222(22 )LJXUH &RQWLQXXP VXEWUDFWHG DV GHILQHG LQ 6HFWLRQ f LQYDULDQW PDVV RI UUf IURP 7nf fÂ§A UUa
PAGE 64
(YHQWV 0H9F f )LOH *HQHUDWHG LQWHUQDOO\ ,' ,' 6\PE 'DWH7LQH $UHD 0HDQ 506 A 0DVV *H9Ff )LJXUH %ORZ XS RI WKH LQYDULDQW PDVV RI UU LQ WKH f PDVV UHJLRQ LQFOXGLQJ WKH XSSHU OLPLW RQ WKH IMf VKRZQ DV D GDVKHG OLQH
PAGE 65
3UREDELOLW\ GLVWULEXWLRQ )LOH $PSBPQWERKUXVHUOXLVFOHRPQLWXVJDPPDKKWHPSWPS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 $UHD )LJXUH 1RUPDOL]HG SUREDELOLW\ GLVWULEXWLRQ IRU GLIIHUHQW IMf QDO DUHDV 7KH VKDGHG DUHD VSDQV b RI WKH SUREDELOLW\ U7n VLJ
PAGE 66
)LJXUH &RQWRXU SORW RI WKH LQYHUVH RI WKH DYHUDJH UHODWLYH ELQHUURU IURP EDFNn JURXQG VXEWUDFWLRQV LQ WKH f WRSf f PLGGOHf DQG f ERWWRPf DQJXODU GLVWULEXWLRQ !
PAGE 67
0,18,7 [ )LW WR 3ORW t 86}*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW [ IRU GRI (UURUV 3DUDEROLF )XQFWLRQ 2 $5($ $8* )LW 6WDWXV (20 ( &/ b 0LQRV ,FRV%A, _FRVQ_ )LJXUH $QJXODU GLVWULEXWLRQ IRU WKH H[FHVV HYHQWV LQ WKH Rf PDVV UHJLRQ 7KH ILW FRUUHVSRQGV WR G1 G_FRVUH_f
PAGE 68
0,18,7 [ )LW WR 3ORW t 86!*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW ; IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $5($ s k ( s $8* )LW 6WDWXV ('0 ( &/ b 0LQRV 22222(22 _FRV7O _FRVQ_ )LJXUH $QJXODU GLVWULEXWLRQ IRU WKH H[FHVV HYHQWV LQ WKH Rf PDVV UHJLRQ 7KH ILW FRUUHVSRQGV WR G1 G_FRVf
PAGE 69
0,18,7 [ )LW WR 3ORW t 86!*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW $8* )LW 6WDWXV ('0 ( ; IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $5($ W! s ( s ( s &/ b 0LQRV 22222(22 ( _FRV\_ ,FRV2)LJXUH $QJXODU GLVWULEXWLRQ IRU WKH H[FHVV HYHQWV LQ WKH f PDVV UHJLRQ 7KH ILW FRUUHVSRQGV WR G1 GOFRV*f
PAGE 70
0,18L7 b )LW WR 3ORW t 86}*DPPD K K $%6&263+27+(U$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW ;] IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $5($ c! s s ( s $8* )LW 6WDWXV ('0 ( &/ b 0LQRV ( ( 22222(22 22222(22 ,FRV*A, ,FRV*)LJXUH $QJXODU GLVWULEXWLRQ IRU WKH H[FHVV HYHQWV LQ WKH f PDVV UHJLRQ 7KH ILW FRUUHVSRQGV WR G1 G_FRVf
PAGE 71
)LJXUH D SUREDELOLW\ GLVWULEXWLRQ IRU UHVRQDQFHV LQ WKH Rf WRSf f PLGGOHf DQG f ERWWRPf LQYDULDQW PDVV UHJLRQ ZKHQ RQO\ WKH K\SRWKHVHV D DUH FRQVLGHUHG
PAGE 72
&+$37(5 (;&/86,9( 5$',$7,9( '(&$< 7f ..a ,Q ILJXUH WKH .. a LQYDULDQW PDVV SORW LV VKRZQ IRU ERWK UHVRQDQFH DQG FRQWLQXXP UXQQLQJ 7KLV ILJXUH DOVR KDV DQ LQVHW VKRZLQJ WKH G(G[.f IRU ERWK WUDFNV IRU HYHQWV LQ WKH *H9F UHJLRQ 7KLV LQVHW LQGLFDWHV WKDW PRVW RI WKH HYHQWV LQ WKLV PDVV UHJLRQ KDYH LQGHHG WZR NDRQV DQG WKDW WKH DPRXQW RI S UHIOHFWLRQ LV VPDOO )XUWKHUPRUH ZKHQ WKH UUB LQYDULDQW PDVV LV SORWWHG IRU WKHVH HYHQWV RQO\ RXW RI FDQ EH ILW XQGHU D S SHDN WKLV ILW LV ZKDW ZH XVH WR FDOFXODWH WKH SLRQ IDNLQJ NDRQ UDWH LQ 7DEOH f )LJXUH VKRZV RXU ILW WR WKH .a LQYDULDQW PDVV FRQWLQXXP VXEWUDFWHG SORW DV GHILQHG LV 6HFWLRQ ZLWK WKH PRVW OLNHO\ VWDWLVWLFDO ILW RYHUOD\HG ZKLFK LV GHVFULEHG LQ WKH QH[W VHFWLRQf 7KH QXPEHU RI HYHQWV QHDU WKH UHJLRQ fÂ§ *H9Ff OHIW DIWHU WKH FRQWLQXXP VXEWUDFWLRQ LV s 6WDWLVWLFDO )LW RI WKH ,QYDULDQW 0DVV 'LVWULEXWLRQ 6LJQDO $UHDV DQG 7KHLU 6LJQLILFDQFH 7KH UHVXOWV RI WKLV VHFWLRQ DUH VXPPDUL]HG LQ 7DEOH )URP WKH PHDVXUHPHQW RI WKH SUHYLRXV VHFWLRQ ZH H[SHFW D VPDOO FRQWULEXWLRQ m HYHQWV DVVXPLQJ QR LQWHUIHUHQFHf IURP f fÂ§! ..a :H GR ILQG VRPH HYLGHQFH IRU f fÂ§! ..a HYHQWV s f LQ WKH ILW :H DOVR ILQG VWURQJ HYLGHQFH IRU WKH UHVRQDQFH f DQG ZHDN HYLGHQFH IRU WKH Rf UHVRQDQFH 7KH f DQG Rf DUH ILWWHG ZLWK WKHLU ZLGWKV IL[HG WR WKHLU 3'* YDOXHV>@ EHFDXVH WKH\ KDYH ODUJH HUURUV LI DOORZHG WR IORDW 7KH UHVW RI WKH UHVRQDQFHV SDUDPHWHUV DUH FRQVLVWHQW ZLWK WKHLU 3'* YDOXHV>@ ZKLFK DUH frf s 0H9F UnLVf s 0H9F URPf s
PAGE 73
0H9F DQG 7APRf s 0H9F :H DOVR REVHUYH DQ H[FHVV RI HYHQWV LQ WKH fÂ§ *H9F UHJLRQ ZKLFK ZH FDQfW DWWULEXWH WR DQ\ NQRZQ UHVRQDQFHV 6LJQLILFDQFHV RI WKH VLJQDOV RI WKH LGHQWLILHG UHVRQDQFHV LQ WKH ILW DUH FDOFXODWHG DV GHVFULEHG LQ WKH SUHYLRXV VHFWLRQ 7KH <6f Mf IMf fÂ§! ..a XSSHU OLPLW LV DOVR FDOFXODWHG XVLQJ PXOWLSOH ILWV H[FHSW WKDW WKLV WLPH WKH HYHQWV XQGHU WKH IMf DUH RI XQNQRZQ RULJLQ VR ZH XVH D ILUVW RUGHU SRO\QRPLDO DOORZHG WR IORDW 7KH VLJQLILFDQFH RI WKH H[FHVV RI HYHQWV LQ WKH fÂ§ *H9F LQYDULDQW PDVV UHJLRQ LV FDOFXODWHG DVVXPLQJ D QRUPDO GLVWULEXWLRQ ZH VLPSO\ DGG XS WKH QXPEHU RI HYHQWV LQ HDFK ELQ DORQJ ZLWK LWV HUURU 5HVXOWV DUH VKRZQ LQ 7DEOH 7DEOH 5HVXOWV IRU 7,6nf fÂ§ A..a 0RGH $UHD %UDQFKLQJ )UDFWLRQ f 6LJQLILFDQFH f f s s [ Mf r r FUf Rf Rf fÂ§ ..a s s [ aFUf A..a fÂ§ *H9Ff s s X MfMf fÂ§! ..a [ a $QJXODU 'LVWULEXWLRQ RI 7KH 6LJQDO ,Q WKLV VHFWLRQ ZH DGDSW WKH LGHDV SUHVHQWHG LQ 6HFWLRQ WR WKH 7,6nf fÂ§} ..a VLWXDWLRQ 7KH GHULYHG VWDWLVWLFDO HUURUV IURP WKH VLJQDO DQG EDFNJURXQG VXEWUDFWLRQV FDQ EH XVHG WR FDOFXODWH WKH PDVV LQWHUYDO ZKLFK EHVW UHSUHVHQWV WKH KHOLFLW\ DQJXODU GLVWULEXWLRQ 7KH LQYHUVH RI WKH H[SHFWHG DYHUDJH UHODWLYH ELQ HUURU DV D IXQFWLRQ RI WKH PDVV LQWHUYDO LV VKRZQ LQ )LJXUH DQG WKH PDVV LQWHUYDO WKDW PD[LPL]HV LW DUH WDEXODWHG LQ 7DEOH 7KH WDLOV IURP WKH UHVRQDQFHV FRQWULEXWH WR WKH FRQWLQXXP VXEWUDFWHG KHOLFLW\ GLVWULEXWLRQV
PAGE 74
7DEOH 0HDQ PDVVHV ZLGWKV LQPDVV DQG LQYHUVH RI WKH DYHUDJH ELQ UHODWLYH HUURU )f IURP EDFNJURXQG VXEWUDFWLRQV IRU WKH DQJXODU GLVWULEXWLRQ RI GLIIHUHQW UHVRQDQFHV 6WDQGDUG PHDQ PDVVHV DQG ZLGWKV FRUUHVSRQGLQJ WR ,) DUH WDNHQ IURP WKH ILW LQ )LJXUH DQG DUH ODEHOHG ZLWK WKH VXEVFULSW fVf ZKLOH WKRVH WKDW PD[LPL]H ) D[H ODEHOHG ZLWK WKH VXEVFULSW fPf 7KH ODVW FROXPQ VKRZV WKH IDFWRU E\ ZKLFK WKH HIIHFWLYH VWDWLVWLFV LQFUHDVH 5HVRQDQFH PV $ PVf )PV $PVf PP $ PPf )^PP $PPf )PV$UULVf f f f Lf f f Rf f f &Df f SWIn f &n@f enf Af b!f &f Âf Hen f :KHUH DJDLQ WKH VPDOO QPQEHUV DQG H DUH WKH UDWLRV RI WKH QXPEHU RI HYHQWV IURP D UHVRQDQFH LQ WKH PDVV LQWHUYDO ZKHUH WKH FRQWDPLQDWLRQ LV WDNLQJ SODFH WR WKH QXPEHU RI HYHQWV IURP WKH VDPH UHVRQDQFH LQ WKH PDVV LQWHUYDO XVHG WR VHOHFW LW 8VLQJ WKH PP $PPf YDOXHV LQ 7DEOH DQG WKH ILW LQ )LJXUH ZH REWDLQ [ DQG H 7KH EDFNJURXQG VXEWUDFWHG KHOLFLW\ DQJXODU GLVWULEXWLRQV DUH nf LOS?H &f a &f f Â&R ff L Âf Ff a f Af ARf FRf a An f 7KH WKH EHVW ILW IRU HDFK UHVRQDQFH DQG WKH H[FHVV RI HYHQWV LQ WKH fÂ§ *H9F DUH VKRZQ LQ )LJXUHV WKURXJK 7KH EHVW VSLQ DVVLJQPHQW IRU WKH f LV D fÂ§ IRU WKH f LW LV D IRU WKH LW LV Rf DQG LW LV D fÂ§ IRU WKH H[FHVV RI HYHQWV LQ WKH fÂ§ *H9F PDVV UHJLRQ 7KH D YDOXH IRU WKH Rf
PAGE 75
LV LQFRQVLVWHQW ZLWK LWV NQRZQ VSLQ $OVR H[DPLQDWLRQ RI WKH QRUPDOL]HG KHOLFLW\ DPSOLWXGHV IRU WKH f UHYHDOV WKDW WKH\ DUH LQFRQVLVWHQW ZLWK WKRVH REWDLQHG IRU WKH f LQ WKH UUa PRGH 7KH DVVLJQHG SUREDELOLW\ GLVWULEXWLRQV DUH VKRZQ LQ )LJXUH 7KH\ UHYHDO A WKDW WKH LQFRQVLVWHQFLHV LQ WKH f DQG Rf DUH QRW VLJQLILFDQW DQG FDQ EH DWWULEXWHG WR WKH VWDWLVWLFDO XQFHUWDLQW\ )LOH ERKUXVHU ,XLVGHRQWSGDWDBQWS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 )LJXUH ,QYDULDQW PDVV RI ..a IRU ,6 WRSf DQG 6 ERWWRPf GDWD )RU WKH 6 GDWD WKH LQVHW VKRZV WKH FUL(G[.f IRU ERWK WUDFNV IRU HYHQWV LQ WKH fÂ§ *H9F PDVV UHJLRQ 7KLV LQVHW LQGLFDWHV WKDW PRVW RI WKH HYHQWV DUH FRQn VWDQW ZLWK KDYLQJ WZR NDRQV
PAGE 76
(YHQWV 0H9F f 0,18,7 [ )LW WR 3ORW 86!*DPPD K K PDVVNN D[LV t )LOH *HQHUDWHG LQWHUQDOO\ $8* 3ORW $UHD 7RWDO)LW )LW 6WDWXV )XQF $UHD 7RWDO)LW ('0 ( \ IRU GRI } &/ b (UURUV 3DUDEROLF 0LQRV )XQFWLRQ IBf $5($ s 0($1 s ( ( ( r :,'7+ s 222(2 22222(22 22222(22 r 63,1 s ( 22222(22 22222(22 r 7+5(6+2/' s ( 22222(22 22222(22 )XQFWLRQ IBnf 02' s 0($1 s ( ( ( :,'7+ ( s ( ( ( r 63,1 s 22222(22 22222(22 22222(22 r 7+5(6+2/' 2222(22 22222(22 22222(22 )XQFWLRQ IBf 02' s 0($1 s ( ( ( r :,'7+ s 22222(22 22222(22 22222(22 m63,1 222222(22 s 22222(22 22222(22 22222(22 }7+5(6+2/' s 22222(22 22222(22 22222(22 0DVV *H9Ff )LJXUH &RQWLQXXP VXEWUDFWHG LQYDULDQW PDVV RI ..
PAGE 77
(YHQWV 0H9F f )LOH *HQHUDWHG LQWHUQDOO\ ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 0DVV *H9Ff )LJXUH fÂ§ %ORZ XS RI WKH Mf UHJLRQ ZLWK WKH b &/ XSSHU OLPLW RYHUn ODLG 7KH PDVV DQG ZLGWK DUH WDNHQ WR EH Af *H9F DQG 7M f 0H9F DV LQ >@
PAGE 78
3UREDELOLW\ GLVWULEXWLRQ )LOH $PSBPQWERKUXVHUOXLVFOHR&9PQLWXOVJDPPDKKnWHPSWPS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ IL06 O / )LJXUH 1RUPDOL]HG SUREDELOLW\ GLVWULEXWLRQ IRU GLIIHUHQW IMf VLJQDO DUHDV 7KH VKDGHG DUHD VSDQV b RI WKH SUREDELOLW\ ..
PAGE 79
0HDQ 0DVV 0H9F f )LJXUH &RQWRXU SORW RI WKH LQYHUVH RI WKH DYHUDJH UHODWLYH ELQHUURU IURP EDFNn JURXQG VXEWUDFWLRQV LQ WKH f WRSf f PLGGOHf DQG Rf ERWn WRPf DQJXODU GLVWULEXWLRQV
PAGE 80
A 0O 18 ,7 b] )LW WR 3ORW t 86}*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW ; IRU GRIf (UURUV )XQFWLRQ 3DUDEROLF $5($ s ( s ! s $8* )LW 6WDWXV ('0 ( &/ b 0LQRV 22222(22 42222(W22 ,FRV*A_FRVQ_ )LJXUH %DFNJURXQG VXEWUDFWHG .. DQJXODU GLVWULEXWLRQ LQ WKH f PDVV UHJLRQDV GHILQHG LQ WKH WH[W 7KH ILW FRUUHVSRQGV WR G1 GOFRV%f
PAGE 81
0,18,7 [ )LW WR 3ORW t 8 6!*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW [] IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $5($ ( s s $8* )LW 6WDWXV ('0 ( &/ b 0LQRV 22222(22 ,FRV2,FRV2)LJXUH %DFNJURXQG VXEWUDFWHG .r .a DQJXODU GLVWULEXWLRQ LQ WKH f PDVV UHJLRQ DV GHILQHG LQ WKH WH[W 7KH ILW FRUUHVSRQGV WR G1 G_FRVUW_f
PAGE 82
0,18,7 [ )LW WR 3ORW t 86!*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW $8* )LW 6WDWXV ('0 ( bn IRU GRI (UURUV )XQFWLRQ 3DUDEROLF $5($ ( s `! s &/ b 0LQRV I 22222(22 22222(22 ,FRV4A,FRV2)LJXUH %DFNJURXQG VXEWUDFWHG . DQJXODU GLVWULEXWLRQ LQ WKH Rf PDVV UHJLRQ DV GHILQHG LQ WKH WH[W 7KH ILW FRUUHVSRQGV WR G1 W GMFRV2f
PAGE 83
0,18,7 [ )LW WR 3ORW t 86!*DPPDK K $%6&&63+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW b IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $5($ W $8* )LW 6WDWXV (20 ( &/ b 0LQRV f 22222(W22 ,FRV4A,FRV%)LJXUH ..a DQJXODU GLVWULEXWLRQ IRU HYHQWV ZLWKLQ WKH fÂ§ *H9F 7TVRR,S f13
PAGE 84
0,18,7 [ )LW WR 3ORW t 86a!*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW \ IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $8* )LW 6WDWXV ('0 ( &/ b 0LQRV ,FRVHAO _FRVUW_ )LJXUH %DFNJURXQG VXEWUDFWHG ..a DQJXODU GLVWULEXWLRQ LQ WKH Rf PDVV UHJLRQ DV GHILQHG LQ WKH WH[W 7KH ILW FRUUHVSRQGV WR G1 G_FRV7&_f
PAGE 85
;f 2 Fc 2 D )LJXUH D SUREDELOLW\ GLVWULEXWLRQ IRU UHVRQDQFHV LQ WKH fWRSf f PLGGOHWRSf Rf PLGGOHERWWRPf LQYDULDQW PDVV UHJLRQ DQG WKH H[FHVV RI HYHQWV LQ WKH fÂ§ *H9F ERWWRPf UHJLRQ ZKHQ RQO\ WKH K\SRWKHVHV D fÂ§ DUH FRQVLGHUHG
PAGE 86
&+$37(5 (;&/86,9( 5$',$7,9( '(&$< 76f SS ,Q ILJXUH WKH SS LQYDULDQW PDVV SORW LV VKRZQ IRU ERWK ,6 DQG 6 GDWD ZLWK DQ LQVHW VKRZLQJ WKDW PRVW RI WKH HYHQWV LQGHHG KDYH D SURWRQ DQG DQ DQWLSURWRQ DQG WKDW WKH S DQG S UHIOHFWLRQV DUH VPDOO 7KH HQKDQFHPHQW DW *H9F LQ WKH 6 SORW FRUUHVSRQGV WR WKH SURFHVV HHB fÂ§ LS ZLWK LS fÂ§! SS 7KLV HQKDQFHPHQW LV QRW DV SURQRXQFHG LQ WKH ,6 SORW EHFDXVH WKH ,6 GDWD RQO\ KDV b RI WKH OXPLQRVLW\ RI WKH 6 GDWD 7KH QXPEHU RI HYHQWV LQ WKH LS LQYDULDQW PDVV UHJLRQ DIWHU FRQWLQXXP VXEWUDFWLRQ LV s DQG LV FRQVLVWHQW ZLWK )LJXUH VKRZV WKH FRQWLQXXP VXEWUDFWHG LQYDULDQW SS PDVV GLVWULEXWLRQ DV GHILQHG LQ 6HFWLRQ f ZLWK D b FRQILGHQFH OHYHO XSSHU OLPLW IRU IM^f RYHUODLG $ GLUHFW ILW WR WKH IMf \LHOGV s HYHQWV 7KHUH LV DQ H[FHVV RI HYHQWV LQ WKH FRQWLQXXP VXEWUDFWHG LQYDULDQW PDVV SORW :H PHDVXUH WKLV H[FHVV DQG WKH XSSHU OLPLW RI HHB fÂ§! Mf Mf fÂ§ SS WKH VDPH ZD\ ZH PHDVXUHG WKH H[FHVV RI HYHQWV LQVLGH *H9F P..af *H9F UHJLRQ DQG WKH IMf XSSHU OLPLW LQ 6HFWLRQ 5HVXOWV DUH VKRZQ LQ 7DEOH 7KH SS DQJXODU GLVWULEXWLRQ IRU WKH PDVV UDQJH *H9F PSSf *H9F LV VKRZQ LQ )LJXUH ZLWK WKH PRVW OLNHO\ DVVLJQPHQW D fÂ§ ILW RYHUOD\HG 7KH SUREDELOLW\ GLVWULEXWLRQ IRU D LV VKRZQ LQ )LJXUH 7DEOH 5HVXOWV IRU <,6nf fÂ§ SS 0RGH $UHD %) RU b 8/ af 6LJQLILFDQFH SS fÂ§ *H9Ff s s D f Mf fÂ§A SS [
PAGE 87
(YHQWV 0H9F f )LOH ERKUXVHU ,XLVFOHRQWSGDWDBQWS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 0DVV *H9Ff )LJXUH ,QYDULDQW PDVV RI SS IRU ,6 WRSf DQG 6 ERWWRPf UXQQLQJ )RU WKH 6 UXQQLQJ WKH LQVHW LV FRQVLVWHQW ZLWK WKH HYHQWV KDYLQJ D SURWRQ DQG DQ DQWLSURWRQ
PAGE 88
)LOH *HQHUDWHG LQWHUQDOO\ ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 0DVV *H9Ff )LJXUH ,QYDULDQW PDVV RI SS 7KH SORW LV FRQWLQXXP VXEWUDFWHG DQG HIILFLHQF\ FRUUHFWHG $Q RYHUOD\ ZLWK WKH b FRQILGHQFH OHYHO XSSHU OLPLW IRU Mf LV VKRZQ 7KH PDVV DQG ZLGWK DUH WDNHQ WR EH Af fÂ§ *H9F DQG AIMf 0H9F DV LQ >@
PAGE 89
5H $PSBPQWERKUXVHUOXLVGHRPQ LWX VJDPPDKKWHPSWPS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 6 2 U + FR [ & ;f $UHD )LJXUH 1RUPDOL]HG SUREDELOLW\ GLVWULEXWLRQ IRU GLIIHUHQW IMf DUHDV 7KH VKDGHG DUHD VSDQV b RI WKH SUREDELOLW\ SS VLJQDO
PAGE 90
0,18,7 b )LW WR 3ORW t 86!*DPPD K K $%6&263+27+(7$ff D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW b IRU GRI (UURUV 3DUDEROLF )XQFWLRQ $5($ s ( s $8* )LW 6WDWXV ('0 ( &/ b 0LQRV 22222(22 ,FRV*A,FRV*)LJXUH SS DQJXODU GLVWULEXWLRQ IRU WKH H[FHVV RI HYHQWV LQ WKH PDVV UDQJH *H9F PSSf *H9F G1 G_FRV*f
PAGE 91
)LJXUH D SUREDELOLW\ GLVWULEXWLRQ IRU WKH H[FHVV RI HYHQWV LQ WKH PDVV UDQJH *H9F UQSSf *H9F ZKHQ RQO\ WKH K\SRWKHVHV D DUH FRQVLGHUHG
PAGE 92
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n TXLUHPHQW D[H QHDUO\ b HIILFLHQW 7KHUHIRUH VXFK FXWV VKRXOG QRW EH D VRXUFH RI V\VWHPDWLF XQFHUWDLQW\ 7KH RQO\ VNLP FXW ZRUWK WDNLQJ D FORVHU ORRN DW LV WKH KDUG*DP FXW ZKLFK LV DERXW b HIILFLHQW LQ 0& 0HDVXULQJ WKLV HIILFLHQF\ LQ RXU GDWD LV QRW SRVVLEOH EHFDXVH HYHQWV QRW FODVVLILHG DV KDUG*DP DUH QRW LQ WKH GDWD WR EHJLQ ZLWK &ORVHU H[DPLQDWLRQ RI RXU 0& UHYHDOV WKDW WKDW PRVW RI WKH b LQHIILFLHQF\ LQ KDUG*DP FRPHV IURP WKH H2YHU3O FXW bf VRPH IURP WKH 6K FXW bf DQG WKH UHVW bf IURP WKH RWKHU FXWV SUHVHQW LQ KDUG*DP :H FDQ PHDVXUH WKH HIILFLHQF\ RI H2YHU3O LQ GDWD E\ ORRNLQJ DW WKH H2YHU3 GLVWULEXWLRQ RI GDWD WUDFNV IURP S DQG Mf GHFD\ :H DUH VDWLVILHG E\ WKLV FKHFN RQ WKH 0& PRGHOLQJ VHH 7DEOHV DQG f DQG ZH ZRQfW PHDVXUH KRZ ZHOO 0& PRGHOV WKH UHVW RI WKH KDUG*DP FXWV ZKLFK DUH b HIILFLHQW LQ 0&
PAGE 93
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b &XWV 0& HII 'DWD HII 6\VWHPDWLF (UURU H2YHU3 4(' HHa fÂ§ n\HHa VXSSUHVLRQ 4(' S3S7 fÂ§r SSa VXSSUHVLRQ +DGURQ VHSDUDWLRQ 2YHUDOO DQDO\VLV FXW V\VWHPDWLF HUURU s 0& HIILFLHQFLHV IURP 7DEOH (IILFLHQFLHV IRU 7f fÂ§! UUa IRU IODW VLJQD GDWD Sf DQG WKH GHULYHG V\VWHPDWLF HUURU LQ b (IILFLHQFLHV DUH UHSRUWHG DV WKH QXPEHU RI VLJQDO HYHQWV DIWHU DOO FXWV GLYLGHG E\ WKH QXPEHU RI VLJQDO HYHQWV ZLWK DOO FXWV H[FHSW WKH RQH XQGHU FRQVLGHUDWLRQ 6WDWLVWLFDO HUURUV D[H b RU OHVV )RU WKH SURWRQ FDVH ZH GRQfW KDYH D FOHDQ VDPSOH ZLWK KLJK VWDWLVWLFV RI HHB fÂ§! SS HYHQWV LQ GDWD %\ H[WHQVLRQ ZH WDNH WKH V\VWHPDWLF HUURU LQ WKLV PRGH WR EH b XVWLILFDWLRQ RI WKH '37+08 &XW 7KH UHDVRQ ZH SUHIHU WR XVH '37+08 LQVWHDG RI WKH PRUH WUDGLWLRQDO VHH >@f '37+08 XVHG LQ &/(2 ,, LV WKDW IRU VRPH XQNQRZQ UHDVRQ RXU &/(2 ,,, 0& KDV WRR PDQ\ U WUDFNV ZLWK '37+08 $ VWXG\ XVLQJ .J RU $ VLJQDOV IURP KDGURQLF HQYLURQPHQWV ZRXOG KDYH ODUJHU VWDWLVWLFV EXW LV SUREOHPDWLF EHFDXVH WKH ODUJH QXPEHU RI WUDFNV DQG VKRZHUV DUWLILn FLDOO\ GHFUHDVH WKH FXW HIILFLHQF\ 6HH >@ $SSHQGL[ $ IRU DQ H[DPSOH
PAGE 94
&XWV 0& (II 'DWD (II 6\VWHPDWLF (UURU H2YHU3 4(' HHf THH VXSSUHVLRQ 4(' SSa fÂ§! SSa VXSSUHVLRQ +DGURQ VHSDUDWLRQ 2YHUDOO DQDO\VLV FXW V\VWHPDWLF HUURU s 7DEOH (IILFLHQFLHV IRU 7O6nf fÂ§ ..a IRU IODW VLJQDO 0& HIILFLHQFLHV IURP GDWD ff DQG WKH GHULYHG V\VWHPDWLF HUURU LQ b (IILFLHQFLHV DUH UHSRUWHG DV WKH QXPEHU RI VLJQDO HYHQWV DIWHU DOO FXWV GLYLGHG E\ WKH QXPEHU RI VLJQDO HYHQWV ZLWK DOO FXWV H[FHSW WKH RQH XQGHU FRQVLGHUDWLRQ 6WDWLVWLFDO HUURUV DUH b RU OHVV 7R REVHUYH WKLV IDFW ZH ILUVW VHOHFW D UHODWLYHO\ FOHDQ VDPSOH RI SLRQ WUDFNV E\ UHTXLULQJ WKH HYHQW WR KDYH D QQa LQYDULDQW PDVV FRQVLVWHQW ZLWK WKH S PDVV DQG WR SDVV DOO RXU DQDO\VLV FXWV H[FHSW IRU WKH FXW RQ 37+ 08 RQ RQH WUDFN )RU VXFK HYHQWV ZH SORW WKH UDWH DV D IXQFWLRQ RI PRPHQWXP DW ZKLFK WKH U WUDFN ZKRVH '37+08 FXW ZH UHOHDVHG KDV 37+ 08 DQG '37 + 08 7R LQFUHDVH RXU VWDWLVWLFV ZH GR WKLV SURFHGXUH WZLFH RQFH IRU HDFK WUDFN DQG DYHUDJH WKH IDNH UDWH )LJXUH VKRZV WKH UHVXOWV &OHDUO\ WKH &/(2 ,,, 0& ZH DUH XVLQJ KDV VRPH SUREOHP PRGHOLQJ WKH '37+08 FXW 7R NHHS WKH V\VWHPDWLF HUURU ORZ ZH FKRRVH D FXW DW '37+08 '37+08 JLYHV D V\VWHPDWLF HUURU RI DERXW bf 7KLV GRHV QRW FKDQJH WKH HIILFLHQF\ LQ GDWD YHU\ PXFK EXW LW LQFUHDVHV WKH HIILFLHQF\ UL4(7WHG E\ 0& EULQJn LQJ LW FORVHU WR UHDOLW\ 7KH LQFUHDVH LQ S IDNHV DIWHU ORRVHQLQJ WKH FXW LV HVWLPDWHG WR EH ORZ XVLQJ 4(' 0& VHH )LJXUH f $QJXODU 'LVWULEXWLRQ RI 6LJQDO 7KH SKRWRQ DQG WUDFNV IURP WKH SURFHVV 7,6nf fÂ§! ; ZLWK ; fÂ§ KKa KDYH D GLIIHUHQW DQJXODU GLVWULEXWLRQ WKDQ WKDW RI IODW 0& ([DPSOHV RI SRVVLEOH DQJXODU GLVWULEXWLRQV DUH VKRZQ LQ WKH DSSHQGL[
PAGE 95
)LOH fERKUXVHUOXcVFOHRQWSGDWDBQWS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 ( ( 2 7UDFN 0RPHQWD *H9f )LJXUH 3LRQ IDNLQJ PXRQ IDNH UDWHV IRU D FXW RQ 37+ 08 WRSf DQG 37+ 08 ERWWRPf 0& LV VKRZQ DV VROLG FLUFOHV ZKLOH S IURP GDWD LV VKRZQ DV KROORZ FLUFOHV )DNH UDWHV DUH UHSRUWHG UHODWLYH WR DOO HYHQWV WKDW SDVV DOO FXWV H[FHSW IRU WKH 37+ 08 FXW IRU RQH RI WKH WUDFNV )LJXUH VKRZV WKH HIILFLHQF\ LQ IODW 0& DV D IXQFWLRQ RI FRV DQG FRV GKn 1RWH WKDW WKH DQG S PRGHV DUH QHDUO\ LQVHQVLWLYH WR WKH WUDFN DQn JXODU GLVWULEXWLRQ ZKLOH WKH U PRGH LV PRUH VHQVLWLYH WR K 7KLV KDSSHQV EHFDXVH RI WKH VWURQJHU PXRQ UHMHFWLRQ FXW LQ WKH U PRGH :H PHDVXUH WKH V\VWHPDWLF HIIHFWV RI IODW 0& HIILFLHQF\ E\ FRQYROXWLQJ HDFK SORW LQ )LJXUH ZLWK GLIIHUHQW SRVVLEOH DQJXODU GLVWULEXWLRQV FDOFXODWHG LQ WKH / DQG K DUH WKH KHOLFLW\ DQJOHV RI WKH VHTXHQWLDO GHFD\ GHILQHG LQ WKH DSn SHQGL[ VHH )LJXUH f
PAGE 96
5OHfERKUXVHUOXLVGHRQWSPFSDVVBPXPXBJDPPDBQHZQWS ,' ,'% 6\PE 'DWH7LPH $UHD 0HDQ 506 ,' fÂ§r F FV S+ 7UDFN 0RPHQWD *H9f )LJXUH 0XRQ IDNLQJ SLRQ IDNH UDWH LQ bf IRU 37+ 08 VROLG VTXDUHVf DQG 37+ 08 VROLG FLUFOHVf 7KHUH LV D FRPSOHWH RYHUODS LQ WKH ODVW ELQ DSSHQGL[ 7DEOHV VKRZ WKH QHFHVVDU\ FRUUHFWLRQ IDFWRUV UHODWLYH WR IODW 0& HIILFLHQF\ IRU GHFD\V ZLWK GHILQLWH DQG ; KHOLFLWLHV $ DQG $A GXH WR WKH QRQIODW SKRWRQ DQG KDGURQ GLVWULEXWLRQV :H FDOO WKHVH IDFWRUV f DQG H\f$[ Kf UHVSHFWLYHO\ 7KH HIILFLHQF\ RI D GHFD\ ZLWK GHILQLWH $ $A FDQ EH REWDLQHG XVLQJ WKH IODW 0& HIILFLHQF\ FRUUHFWHG E\ D IDFWRU Â?A?[ fÂ§ f?;?; f [ H?r?[Kf ,Q JHQHUDO WKH ILQDO VWDWH LV D PL[WXUH RI DOO SRVVLEOH $ ?[ SDLUV DQG WKH HIILFLHQF\ FRUUHFWLRQ IDFWRU LV H[ FRV FRV 7HR VLQ TI FRV VLQ f
PAGE 97
R G R R 8L : !! R G R r ( WX FRV\ &26)LJXUH )ODW 0& HIILFLHQF\ DV D IXQFWLRQ RI FRV2 OHIW FROXPQf DQG DV D IXQFn WLRQ RI FRVK ULJKW FROXPQf 7RS URZ FRUUHVSRQGV WR WKH SLRQ PRGH PLGGOH URZ WR WKH NDRQ PRGH DQG ERWWRP URZ WR WKH SURWRQ PRGH 7KH ILWV LQ )LJXUHV DQG DQG fÂ§fÂ§ PHDVXUH WKH SDLU I!f 7KHVH YDOXHV DUH VXPPDUL]HG LQ 7DEOH ZKHUH WKH\ DUH XVHG WR REWDLQ H[ IRU HDFK PRGH 7KH SDLU !f FDUU\ DQ HUURU ZKLFK LV D VRXUFH RI V\VWHPDWLF XQFHUWDLQW\ :H FDOFXODWH WKLV V\VWHPDWLF XQFHUWDLQW\ E\ LQVSHFWLQJ WKH GLIIHUHQFHV LQ HIILFLHQF\ ZKHQ !f PRYH DZD\ IURP WKH YDOXH ZKLFK JLYHV WKH PLQLPXP FKLVTXDUHG [/Q!
PAGE 98
0RGH DLR RU DL ,, c" 6_ L U 3 7DEOH W?;?[ f HFLHQFL\ FRUUHFWLRQ ZLWK UHVSHFW WR IODW 0& IDFWRUV GXH WR WKH QRQIODW DQJXODU GLVWULEXWLRQ RI WKH SKRWRQ LQ
PAGE 99
5HVXOWV IRU WKH FRUUHFWLRQ IDFWRU DQG LWV V\VWHPDWLF HUURU D[H VKRZQ LQ 7DEOH 8SSHU OLPLWV RQ <6f fÂ§} f Af fÂ§r KKa DUH FKDQJHG WR LQFOXGH WKH DQJXODU GLVWULEXWLRQfV HIIHFW RQ HIILFLHQF\ 6LQFH ZH FDQfW PHDVXUH WKH KHOLFLW\ DPSOLWXGHV LQ WKLV FDVH ZH FKRRVH WKH ZRUVW SRVVLEOH FDVH ZKHUH WKH FRUUHFWHG HIILFLHQF\ LV ORZHVW 7KLV DOZD\V FRUUHVSRQGV WR k 5HVXOWV DUH LQ 7DEOH UDGLDQVf )LJXUH 6XUIDFHV LQ WKH k Kf SODQH XVHG WR GHWHUPLQH WKH V\VWHPDWLF XQFHUn WDLQW\ LQ WKH HIILFLHQF\ FRUUHFWLRQ IDFWRU IRU WKHPRGHV ZLWK f f DQG f 7KH f VXUIDFH PD\ VHHP ODUJH EXW ZKHQ GUDZQ LQ VSKHULFDO FRRUGLQDWHV LW LV D VPDOO fQRUWK SROH FDSf
PAGE 100
W! UDGLDQVf )LJXUH 0HDVXUHU f VXUIDFHV IRU WKH f KLJK DQG ORZ PDVV UHJLRQV
PAGE 101
2 UDGLDQVf )LJXUH 0HDVXUHU I!f VXUIDFHV IRU WKH If KLJK DQG ORZ PDVV UHJLRQV
PAGE 102
7DEOH 0HDVXUHG !f FDOFXODWHG H[ HIILFLHQF\ FRUUHFWLRQ IDFWRU LQWHUYDO ZKHQ G!f PRYH DZD\ IURP WKHLU PHDVXUHG YDOXH XQGHU WKH FRQGLWLRQ WKDW ? VWD\ ZLWKLQ RQH XQLW RI LWV PLQLPXP YDOXH DQG WKH V\VWHPDWLF XQFHUWDLQW\ RQ WKH FRUUHFn WLRQ IDFWRU 0RGH [ k I! H[ ,QWHUYDO 6\V (UU bf Rf f f f Af Rf ..a *H9F P..f *H9F SS *H9F PSSf *H9F sJ 22TRR c RR XXXB L ORRR sLJ s sL_ s s s s 7DEOH :RUVW FDVH HIILFLHQF\ FRUUHFWLRQ IDFWRUV DQG WKHLU HIIHFW RQ XSSHU OLPLWV RI 7Of fÂ§} Mf f fÂ§r KK GHFD\V 0RGH :RUVW FDVH FRUUHFWLRQ 8SSHU OLPLW LQFUHDVH bf IM^f 77 s IMf r .. s Mf SS s
PAGE 103
'LIIHUHQW +DGURQLF )DNH 5DWHV %HWZHHQ ,6 DQG 6 'LIIHUHQFHV LQ WKH IDNH UDWHV RI KDGURQLF HYHQWV EHWZHHQ WKH DQG GDWD DUH DQRWKHU SRWHQWLDO VRXUFH RI V\VWHPDWLF HUURUV ZKHQ GRLQJ D FRQWLQXXP VXEWUDFWLRQ ,Q SULQFLSOH WKLV LV D VHFRQG RUGHU HIIHFW VLQFH WKH FRQWDPLQDWLRQ LV SURSRUWLRQDO WR WKH GLIIHUHQFH RI WZR VPDOO QXPEHUV +RZHYHU VLQFH ZH KDYH VXFK D ODUJH QXPEHU RI S DQG Mf HYHQWV LQ WKH XQGHUO\LQJ ,6 FRQWLQXXP ZH VKRXOG TXDQWLI\ DQ\ SRVVLEOH VLJQDO FRQWDPLQDWLRQ 6HFWLRQ RXWOLQHV WKH WKUHH FDVHV ZH QHHG WR ZRUU\ DERXW S HYHQWV FRQWDPLQDWLQJ WKH ..a LQYDULDQW PDVV SORW S FRQWDPLQDWLQJ WKH SS LQYDULDQW PDVV SORW DQG HYHQWV FRQWDPLQDWLQJ WKH SS LQYDULDQW PDVV SORW ,W LV UHDVRQDEOH WR DVVXPH WKDW WKH ,6 DQG 6 IDNH UDWHV VWD\ ZLWKLQ b RI HDFK RWKHU VHH 6HFWLRQ IRU VRPH H[DPSOHV IURP 0& PHDVXUHPHQWVf :LWK WKLV DVVXPSWLRQ WDNLQJ QS P DQG Q a LQ WKH ,6 GDWD DQG WKH IDNH UDWHV PHDVXUHG LQ GDWD VHH 7DEOH f ZH FDQ FDOFXODWH WKH V\VWHPDWLF XQFHUWDLQWLHV :H H[SHFW s HYHQWV IURP S WR FRQWDPLQDWH WKH ..a LQYDULDQW PDVV SORW +RZHYHU WKHVH HYHQWV FDQ EH LJQRUHG DV VRXUFH RI V\VWHPWLF XQFHUWQW\ EHFDXVH b RI WKHP IDOO LQ WKH fÂ§ *H9F UHJLRQ EHORZ WKH Af SHDN :H DOVR H[SHFW s IURP S DQG s HYHQWV IURP FRQWDPLQDWLQJ WKH SS LQYDULDQW PDVV SORW 7KLV UHSUHVHQWV D sb V\VWHPDWLF HUURU IRU WKH 7f fÂ§! n\SS PRGH 2WKHU 6\VWHPDWLF 6RXUFHV %HVLGHV WKH V\VWHPDWLF XQFHUWDLQWLHV GHVFULEHG DERYH ZH DOVR DGG D b V\VWHPDWLF HIIHFW IURP WUDFN ILQGLQJ b SHU WUDFNf DQG b IURP WKH QXPEHU RI 7f :H ILQG QR HYLGHQFH IRU QRQUHVRQDQW KDGURQ SDLUV WKDW FRXOG LQWHUIHUH ZLWK WKH UHVRQDQFHV 7KHUH LV D SRVVLEOH LQWHUIHUHQFH EHWZHHQ f DQG f EHFDXVH WKH\ ERWK DUH LQ WKH VWDWH )URP WKH f PHDVXUHPHQW LQ WKH UUB PRGH ZH H[SHFW HYHQWV IURP WKLV VRXUFH LQ WKH f PDVV UHJLRQ fÂ§
PAGE 104
*H9F ,Q WKH PRVW H[WUHPH FDVHV ZH KDYH 1 RF __O\nf__ ,$ILf A __Af Af ZKHUH $\nLM DQG $IAPRf DUH WKH DPSOLWXGHV DVVRFLDWHG ZLWK Âf DQG R f UHVSHFWLYHO\ )URP WKH GDWD ZH KDYH 1 m DQG XVLQJ WKH IDFW WKDW __ALf__ FRQWULEXWHV ZLWK HYHQWV WKH FRQWULEXWLRQ IURP ??$I f + FDQ UDQJH ERP WR HYHQWV 7KLV UHSUHVHQWV D b YDULDWLRQ IURP WKH FHQWUDO YDOXH 6LQFH WKLV LV WKH H[WUHPH LQWHUIHUHQFH FDVH ZH WDNH KDOI RI WKLV YDOXH b DV D UHDVRQDEOH V\VWHPDWLF XQFHUWDLQW\ IURP SRVVLEOH LQWHUIHUHQFH LQ WKH Âf 7KH SRVVLEOH GHFD\ 7f fÂ§! SLU FRXOG FRQWDPLQDWH RXU U Ua LQYDULDQW PDVV SORW VLQFH D Ur ZLWK D PRPHQWXP JUHDWHU WKDQ *H9F ORRNV YHU\ PXFK OLNH D SKRWRQ 8VLQJ SKDVH VSDFH 0& DQG LVRVSLQ V\PPHWU\ ZH HVWLPDWH WKDW DERXW b RI SRVVLEOH <,6nf fÂ§! IP KDYH VXFK D 7r 7KH ODWHVW PHDVXUHG XSSHU OLPLW >@ IRU WKLV GHFD\ PRGH LV [ 7KLV JLYHV DQ XSSHU OLPLW RI HYHQWV LQ WKH HIILFLHQF\ FRUUHFWHG U Wa LQYDULDQW PDVV SORW RI ZKLFK HYHQWV IDOO LQ WKH S PDVV UHJLRQ EHKLQG WKH Rf PDVV UHJLRQ 7KH UHPDLQLQJ SRWHQWLDO HYHQWV FRXOG FRQWDPLQDWH WKH UHVW RI WKH LQYDULDQW PDVV UHJLRQ 1RQH RI WKH VLJQDO UHJLRQV LV DIIHFWHG E\ PRUH WKDQ b 7R WHVW WKH UREXVWQHVV RI WKH VWDWLVWLFDO ILWV ZH UHGR WKHP ZLWK DOO PDVVHV DQG ZLGWKV RI SRVVLEOH UHVRQDQFHV UHTXLUHG WR VWD\ ZLWKLQ D RI WKHLU QRPLQDO 3'* YDOXHV >@ 7KH UHVXOWV IURP ILWWLQJ WKH GDWD LQ WKLV ZD\ DUH ZLWKLQ WKH VWDWLVWLFDO HUURUV ZKHQ FRPSDUHG WR WKH ILW UHVXOWV VXPPDUL]HG LQ 7DEOH DQG :H GR QRW REVHUYH D V\VWHPDWLF GLIIHUHQFH EHWZHHQ WKHVH WZR ILWWLQJ WHFKQLTXHV 7KH EUDQFKLQJ UDWLRV IRU f fÂ§! UUa f fÂ§ UUB DQG Âf fÂ§! ..a DUH WDNHQ IURP WKH 3'* >@ DQG FRQWULEXWH WR WKH V\VWHPDWLF XQFHUWDLQW\ ZLWK LJ s DQG s SHUFHQW UHVSHFWLYHO\
PAGE 105
2YHUDOO 6\VWHPDWLF 8QFHUWDLQWLHV :H FRPELQH WKH V\VWHPDWLF HUURUV RI HDFK VHFWLRQ LQ TXDGUDWXUH 7DEOHV VXPPDUL]H WKH V\VWHPDWLF HUURUV 7DEOH 6\VWHPDWLF HUURUV H[SUHVVHG DV b IURP WKH GLIIHUHQW VRXUFHV GHVFULEHG LQ WKH WH[W 1RW VKRZQ LQ WKH WDEOH EXW LQFOXGHG LQ WKH WRWDO DUH D V\VWHPDWLF XQFHUWDLQW\ RI b IURP WKH QXPEHU RI 7f D b V\VWHPDWLF XQFHUWDQW\ IURP 0& WUDFNLQJ LQ DOO WKH PRGHV D b V\VWHPDWLF XQFHUWDLQW\ LQ WKH f IURP SRVVLEOH LQWHUIHUHQFH ZLWK f DQ b V\VWHPDWLF XQFHUWDLQW\ LQ WKH SS PRGH IURP SRVVLEOH KDGURQLF FRQWDPLQDWLRQ DQG WKH V\VWHPDWLF XQFHUWDQWLHV LQ WKH f f DQG Mf KDGURQLF EUDQFKLQJ IUDFWLRQV 0RGH $QDO\VLV &XWV $QJXODU 'LVWULEXWLRQ 7RWDO Rf s s f s s s f s s f s s s Rf s s LI$n *H9Ff s s SS fÂ§ *H9Ff s s 7DEOH ,QFUHDVH LQ WKH XSSHU OLPLWV RI
PAGE 106
&+$37(5 5(68/76 $1' &21&/86,21 :H UHSRUW RQ D QHZ VHDUFK IRU WZRERG\ UDGLDWLYH 76f GHFD\V :H SODFH VWULQJHQW OLPLWV RQ WKH SURGXFWLRQ RI WKH Mf SDUWLFOH LQ WKH SLRQ DQG NDRQ PRGHV DQG D OHVV VWULQJHQW OLPLW LQ WKH SURWRQ PRGH ZKHUH VRPH H[FHVV RI HYHQWV LV REVHUYHG LQ WKH UHJLRQ RI LQWHUHVW ,Q WKH GHFD\ FKDQQHO 7,6nf fÂ§ UUB ZH ILQG FOHDU HYLGHQFH IRU WKH UHVRQDQFH f DQG PHDVXUH D EUDQFKLQJ IUDFWLRQ RI s f [ a ZKLFK LV FRQVLVWHQW ZLWK WKH HDUOLHU &/(2 PHDVXUHPHQW RI OAf [ 8 >@ 7KH DQJXODU GLVWULEXWLRQV RI WKH SKRWRQ DQG WUDFNV VWURQJO\ LQGLFDWH WKDW WKH KDGURQ SDLUV ZH DVVLJQ WR WKH f DUH LQGHHG LQ D VWDWH DQG WKDW QDWXUH SUHIHUV WR SURGXFH WKH f ZLWK KHOLFLW\ ,Q FRQWUDVW IRU LE fÂ§! f LW ZDV IRXQG >@ WKDW WKH f LV SURGXFHG DW HTXDO UDWHV ZLWK ERWK KHOLFLW\ DQG KHOLFLW\ DQG DW D UDWH FRQVLVWHQW ZLWK IRU KHOLFLW\ EXW PRUH UHFHQWO\ >@ PHDVXUHG KHOLFLW\ GRPLQDQFH IRU WKLV VDPH PRGH 7KHUH LV D EDUHO\ VLJQLILFDQW Df H[FHVV RI HYHQWV LQ WKH Rf fÂ§! UUa LQYDULDQW PDVV UHJLRQ 7KH DQJXODU GLVWULEXWLRQV RI WKH SKRWRQ DQG WUDFNV LQGLFDWH WKDW WKLV H[FHVV RI HYHQWV LV LQ D VWDWH UDWKHU WKDQ ZKLFK ZRXOG EH WKH FDVH LI WKH H[FHVV ZHUH GXH WR Rf GHFD\V ,W PD\ EH WKDW WKLV H[FHVV LV GXH WR QRQUHVRQDQW f\FQa SURGXFWLRQ :H FRQFOXGH WKDW PRUH GDWD LV QHHGHG WR UHVROYH WKLV VLWXDWLRQ :H ILQG ZHDN HYLGHQFH IRU WKH SURGXFWLRQ RI WKH UHVRQDQFH Lf LQ WKH 7f 7U GHFD\ FKDQQHO ,Q WKH GHFD\ FKDQQHO 7Of fÂ§! ..a ZH ILQG VWURQJ HYLGHQFH IRU WKH SURGXFWLRQ RI WKH UHVRQDQFH >f DQG ZHDN HYLGHQFH IRU WKH SURGXFWLRQ RI
PAGE 107
Rf 7KH SKRWRQ DQG WUDFN DQJXODU GLVWULEXWLRQV VKRZ WKDW WKH WZR WUDFNV DWWULEXWHG WR WKH f DUH LQGHHG LQ D VWDWH DQG WKDW WKH f LV SURGXFHG PRVWO\ ZLWK KHOLFLW\ 7KHUH LV DOVR HYLGHQFH RI DQ H[FHVV RI HYHQWV LQ WKH fÂ§ *H9F ..a LQYDULDQW PDVV UHJLRQ ZKLFK ZH FDQQRW DWWULEXWH WR DQ\ NQRZQ UHVRQDQFHV )LQDOO\ ZH DOVR ILQG HYLGHQFH RI DQ H[FHVV RI HYHQWV LQ WKH FKDQQHO 7,6nf fÂ§! SS ZKLFK ZH FDQQRW DWWULEXWH WR DQ\ NQRZQ UHVRQDQFHV :H ILQG QR HYLGHQFH RI DQ HQKDQFHPHQW QHDU SS WKUHVKROG 6XFK DQ HQKDQFHPHQW ZDV UHFHQWO\ UHn SRUWHG E\ >@ LQ WKH [Mf V\VWHP DQG LV FXUUHQWO\ EHLQJ LQWHUSUHWHG 6HH HJ >@ >@ >@f :KHQ FRPSDULQJ WKH UDGLDWLYH 76f GHFD\V VWXGLHG LQ WKLV DQDO\VLV WR UDn GLDWLYH [MM GHFD\V ZH REVHUYH D VXSSUHVVLRQ UDWLR RI WKDW LV LQ UHDVRQDEOH DJUHHPHQW ZLWK ZLWK WKH QDLYH H[SHFWDWLRQ RI REWDLQHG IURP VFDOLQJ DUJXn PHQWV 7DEOH VXPPDUL]HV RXU UHVXOWV :KLOH ZH ILQG KDGURQLF UHVRQDQFHV DW D OHYHO FRQVLVWHQW ZLWK WKH HVWLPDWH RI 6HFWLRQ ZH GR QRW REVHUYH DQ\ JOXHEDOO VWDWHV 7KLV LV QRW WKH UHVXOW ZH ZRXOG H[SHFW IURP D QDLYH LQWHUSUHWDWLRQ RI 4&' DV WKH WZR JOXRQV IURP D 76f UDGLDWLYH GHFD\ FDQ IRUP D ERXQG VWDWH GLUHFWO\ DQG QHHG WR JR WKURXJK DW OHDVW WZR VWURQJ LQWHUDFWLRQ YHUWH[HV WR IRUP D KDGURQ ,QWHUQDO FRQVLVWHQF\ LQ 4&' SUHGLFWV WKH H[LVWHQFH RI JOXHEDOOV (YHQ WKRXJK WKHRUHWLFDO FDOFXODWLRQV RI WKH JOXHEDOO PDVV VSHFWUXP H[LVW D FOHDU JOXHEDOO REVHUYDWLRQ KDV QRW EHHQ PDGH \HW :H FRQFOXGH WKDW PRUH JOXHEDOO VHDUFKHV VXFK DV WKRVH SURSRVHG LQ WKH *OXH; SURMHFW DUH QHFHVVDU\ WR FODULI\ WKLV VLWXDWLRQ
PAGE 108
7DEOH )LQDO PHDVXUHG EUDQFKLQJ UDWLRV PHDVXUHG EUDQFKLQJ UDWLRV UHSRUWHG UHODWLYH WR LS EUDQFKLQJ UDWLRV DQG VWDn WLVWLFDO VLJQLILFDQFH IRU HDFK GHFD\ FKDQQHO 7KH PHDVXUHG EUDQFKLQJ UDWLRV KDYH EHHQ FRUUHFWHG E\ WKH IDFWRUV FDOFXODWHG LQ $SSHQGL[ $ )RU WKH EUDQFKLQJ UDWLRV WKH ILUVW XQFHUWDLQW\ LV VWDWLVWLFDO DQG WKH VHFRQG LV V\VWHPDWLF &KDQQHO Â‘ %UDQFKLQJ )UDFWLRQ [ f < 6f n K3 6LJQLILFDQFH Rf f 77 LLR"sRL f s s s D f s s s FU f Â J s A Rf Rf r ..a s s rrf aI.. s s 33 s s s Mf IMf A U7 Mf IMf .. Af Mf rÂ‘33
PAGE 109
$33(1',; +(/,&,7< )250$/,60 )25 7:2 %2'< '(&$<6 +HOLFLW\ IRUPDOLVP FDQ EH XVHG WR REWDLQ WKH DQJXODU GLVWULEXWLRQ RI D GHFD\ SURFHVV ,Q WKLV DSSHQGL[ ZH UHVWULFW RXUVHOYHV WR WKH FDVH RI WZR ERG\ GHFD\V 7KH VLWXDWLRQ LV DV IROORZV SDUWLFOH $ ZLWK VSLQ DQG ]D[LV VSLQ SURMHFWLRQ GHFD\V WR WZR SDUWLFOHV $ fÂ§A % & ZLWK GHILQLWH KHOLFLW\ $ DQG $ /HW WKH GLUHFWLRQ RI WKH RI WKH GHFD\ EH Q I!f QV !Ef SV fÂ§SF VHH )LJXUH f 7KH DPSOLWXGH IRU WKLV GHFD\ LV $=b A%?D?+?$0$ f :H KDYH OLWWOH FKDQFH RI GLUHFWO\ FDOFXODWLQJ %XW ZH FDQ H[SORLW WKH IDFW WKDW + FRQVHUYHV DQJXODU PRPHQWXP E\ FRQVLGHULQJ WKH EDVLV ^ MP$V$F_` ZKHUH M LV WKH WRWDO DQJXODU PRPHQWXP RI % & DQG LQ LWV SURMHFWLRQ DORQJ WKH ] D[LV ,QVHUWLQJ WKLV EDVLV LQWR (TXDWLRQ fÂ§ A MP?%?F?+?D0D MP fÂ§ ; AAE$F,WR$E$F 6MM$6P0$$?%[F MP f?E?F?D0D?E;F $?E[F 7KH URWDWLRQDO LQYDULDQFH RI KHOLFLWLHV LV D QHFHVVDU\ FRQGLWLRQ WR FRQVWUXFW ^ MP;%?F?` DV D EDVLV IRU WKH WZRSDUWLFOH VWDWH % & 5RWDWLRQDO LQYDULDQFH LV RQH RI WZR NH\ LGHDV LQ KHOLFLW\ IRUPDOLVP 7KH RWKHU NH\ LGHD ZKLFK LV ERRVW LQYDULDQFH LV!XVHIXO LQ VHTXHQWLDO GHFD\V
PAGE 110
7KH DERYH HTXDWLRQ FRQWDLQV WZR IDFWRUL]HG WHUPV 7KH ILUVW WHUP LV LQGHSHQn GHQW RI WKH LQWHUDFWLRQ + DQG FRQWDLQV WKH DQJXODU GLVWULEXWLRQ LQIRUPDWLRQ ZH DUH LQWHUHVWHG LQ 7KH VHFRQG WHUP $?%?F LV XVXDOO\ FDOOHG WKH fKHOLFLW\ FRXSOLQJ DPn SOLWXGHf ,W LV LQGHSHQGHQW RI WKH DQJXODU GLVWULEXWLRQ DQG FRQWDLQV WKH SK\VLFV RI WKH GHFD\ LQ SDUWLFXODU LW GHSHQGV RQ $J DQG $F EXW QRW RQ 0Df ,Q WKH KHOLFLW\ IRUPDOLVP WKH KHOLFLW\ FRXSOLQJ DPSOLWXGHV DUH XQNQRZQ SDUDPHWHUV )RU VWURQJ DQG HOHFWURPDJQHWLF GHFD\V WKHUH LV DQRWKHU LQYDULDQFH RI + ZH FDQ H[SORLW SDULW\ )RU WKHVH NLQG RI LQWHUDFWLRQV LW FDQ EH VKRZQ E\ LQVHUWLQJ WKH SDULW\ RSHUDWRU WKDW $?E?F a afDaEaFUM$U@%ULF$?%?F f ZKHUH WLDAE9F LV WKH SURGXFW RI WKH SDULW\ RI SDUWLFOHV $ % DQG & 7KLV XVHIXO SURSHUW\ RI $?%?F UHGXFHV WKH QXPEHU RI XQNQRZQ SDUDPHWHUV LQ WKH DQJXODU GLVWULEXWLRQ RI VWURQJ DQG HOHFWURPDJQHWLF GHFD\V 7KH QH[W VWHS LV WR GHULYH H[SOLFLW IRUPXODH IRU cf;E;F?D $W D $ E $F IRU WKH % & V\VWHP 6XFK GHULYDWLRQ ZKLFK LV RXW RI WKH VFRSH RI WKLV DSSHQGL[ ZLOO
PAGE 111
EH RPLWWHG KHUH 7KH UHVXOW FDQ EH H[SUHVVHG LQ WHUPV RI WKH :LJQHU IXQFWLRQV ?E?F?D0D?E?F f ZKHUH ; ;E fÂ§ ;F DQG !f HP0$;fGL?0 : 7KH IXQFWLRQV GÂc$ [f KDYH UHDO YDOXHV DQG VRPH RI WKHP DUH JLYHQ LQ WKH SDUWLFOH GDWD ERRN :H KDYH ILQDOO\ DUULYHG WR WKH IROORZLQJ LPSRUWDQW UHODWLRQ ?AOf $OO KHOLFLW\ GLVWULEXWLRQ FDOFXODWLRQV DUH UHGXFHG WR KDQGOLQJ WKHVH EDVLF DPSOLn WXGHV +HUH DUH VRPH RI WKHLU YHU\ LPSRUWDQW SURSHUWLHV f;N LI _$V $FO L f $nY b f SQ R G$0$ $r0D fÂ§ WW6n0Da0A$0D $r0D fA$E$Fr ;n;n a ÂU?;nf A$E$FADn 9 %A&f f } D$$R LL $E$FADn $N $fÂ§$nf f 7KH LQWHUHVWHG UHDGHU FDQ ILQG WKLV GHULYDWLRQ LQ >@
PAGE 112
; 3URSHUW\ LV D UHIOHFWLRQ RI WKH SK\VLFDO IDFW WKDW WKH SURMHFWLRQ RI WKH WRWDO DQJXODU PRPHQWXP DORQJ D SDUWLFXODU D[LV FDQQRW EH JUHDWHU WKDQ WKH WRWDO DQJXODU PRPHQWXP LWVHOI __ $A;U __ LV WKH DQJXODU GLVWULEXWLRQ RI D ILQDO VWDWH ZKHUH 0$ DQG WKH KHOLFLWLHV RI % DQG & DUH NQRZQ VHH (TXDWLRQ f +RZHYHU LW LV XVXDOO\ WKH FDVH WKDW $W\$A$H DUH QRW NQRZQ ,Q VXFK FDVHV ZH FDQ RQO\ PHDVXUH WKH RYHUDOO DQJXODU GLVWULEXWLRQ ZKLFK LV VLPSO\ WKH DYHUDJH RYHU SRVVLEOH LQLWLDO VWDWHV RI WKH LQFRKHUHQW VXP RI ILQDO VWDWHV G3B Ga% 0D?E?F .LF G&OQ ( L f ZKHUH 3PD LV WKH ZHLJKW RI WKH LQLWLDO VWDWH $W\ DQG GW\ G^FRV^VffGS% 1H[W FRQVLGHU D VHTXHQWLDO GHFD\ $ fÂ§} % & ZLWK & r ( VHH )LJXUH f 7KH URWDWLRQDO DQG ERRVW LQYDULDQFH RI KHOLFLWLHV PDNHV LW HDV\ WR H[WHQG WKH KHOLFLW\ IRUPDOLVP WR WKLV FDVH 7R GR VR ZH FRQVLGHU D QHZ UHIHUHQFH IUDPH 2f DW UHVW ZLWK & VXFK WKDW ]n fÂ§ Q^% !Ef Â‘ $LO SDUWLFOH KHOLFLWLHV UHPDLQ WKH VDPH LQ 2f DQG LQ SDUWLFXODU 0=&f fÂ§ fÂ§$F WKH QHJDWLYH VLJQ DULVHV IURP WKH IDFW WKDW % !Ef LV WKH GLUHFWLRQ LQ ZKLFK % LV WUDYHOLQJf ,W LV QRZ VWUDLJKWIRUZDUG WR FDOFXODWH WKH LQWHUPHGLDWH DPSOLWXGH DW 2f 2I FRXUVH LV VWLOO FDOFXODWHG DW fV RULJLQDO UHVW IUDPH 6LQFH $F LV DQ LQWHUPHGLDWH VWDWH WKH KHOLFLW\ DPSOLWXGH IRU WKH ILQDO VWDWH ;E?GAH LV $&0D ;EAGAH (D0D QfÂ§?& fA;E;FA;R;H ;F f 7KLV FRQGLWLRQ GRHV QRW XQOLNHO\ GHWHUPLQH 2f :H FDQ EH PRUH SUHFLVH E\ GHILQLQJ 2n fÂ§ /SFf5W!% E f ZKHUH / LV WKH /RUHQ] %RRVW RSHUDWRU 5 LV WKH URWDWLRQ RSHUDWRU LQnWHUPV RI WKH (XOHU DQJOHV DQG 2 LV 7fV RULJLQDO UHVW IUDPH
PAGE 113
)LJXUH $ fÂ§! % & DQG & r ( 7KH GUDZLQJ LV SHFXOLDU LQ WKDW &V GDXJKn WHUV DUH GUDZQ DV VHHQ LQ &fV UHVW IUDPH EDFNWREDFNf ZKHUH ZH KDYH FRKHUHQWO\ VXPPHG RYHU WKH LQWHUPHGLDWH VWDWH & ,I 0DAEAGAH DUH QRW NQRZQ G3 ( 0DAEAGAE S 0L GWL%GLO' ( ,, &t2$} 0 D;EAGAH f 7KH VHTXHQWLDO GHFD\ DPSOLWXGHV KDYH WKH IROORZLQJ LPSRUWDQW SURSHUWLHV $&a0D f f7 rLU m GIH0F;YVOO MU(0LKFOOOOFL8,,n Qf R R ;F 7KLV NLQG RI ORJLF FDQ EH H[WHQGHG WR D JHQHUDO WZRERG\ GHFD\ FKDLQ ([DPSOH 7 6f Wff WWWWff )URP RXU SUHYLRXV GLVFXVVLRQ ZH FDQ LPPHGLDWHO\ ZULWH VHH (TXDWLRQ f A$Â7O6f B 9n A$W7O6f e$! $f
PAGE 114
,Q WKLV FDVH $r $A $ s ,Q &/(2 0[LVf s WKHUH LV QR 0\ LVf fÂ§ VLQFH DW WKH 7/6f HQHUJ\ HOHFWURQV FRXSOH RQO\ WR SRVLWURQV RI WKH RSSRVLWH KHOLFLW\ WKLV ZD\ 3? fÂ§ 3BL fÂ§ a DQG 3T fÂ§ LV DQ H[FHOOHQW DSSUR[LPDWLRQ 7KHUH DUH WKHUHIRUH IRXU DPSOLWXGHV ZH PXVW FRQVLGHU $&OP $&A $& O DQG Â‘AORR 7KH RYHUDOO DQJXODU GLVWULEXWLRQ LV VHH (TXDWLRQ f G3 G4aG4Z $& ORR ,3 ,, rAALRR Q O ,, rALRR ,, ,, ALRR ,, f LO $&@ c, $& ZKHUH ZH KDYH XVHG SURSHUW\ 7KH WZR UHOHYDQW DPSOLWXGHV DUH fÂ§ $>T&AT AOOA22 n$A rA a fAA AOUOA22 7 AA ZKHUH ZH KDYH XVHG WKH IDFW WKDW _$ fÂ§ $Lf L UL6f WR VXP RYHU DOO SRVVLEOH $Lf $OO WKDW UHPDLQV LV WR VXEVWLWXWH AORR OOaf ,, ARARR 7 LO AQARR ,, ,, AKARG cf ^AfLRARRAOOERR AAAA $LOARRAA FF` ,, $& ,, ,, +8 ,, Â‘ $@ & LQnn nn LO A ,, ,, fÂ§A L ^AAAOOA22 fAO2A22rA A AOAAOA &&` f LQWR (TXDWLRQ 'RLQJ VR ZLOO JLYH XV WKH IXOO DQJXODU GLVWULEXWLRQ RI WKH GHFD\ FKDLQ +RZHYHU XVLQJ SURSHUW\ DOO WKH FURVV WHUPV LQVLGH ^` YDQLVK LI ZH LQWHJUDWH RYHU SQ RU SaW )RU WKH WLPH EHLQJ ZH LQWHJUDWH RYHU ERWK "U DQG :H ZLOO UHWXUQ WR WKH
PAGE 115
7>__ Â6R ,, $/: If ,, &MR ,, ,0OL ,, A,,f ,, FMR ,, Â, A ,, ,, r ,3f A ,+ ILOO $O ____ &P ,, GLmf GLBmffG6Rmf ,, $Q O__ &RR __ GcfJ Jf!: ,, $ ____ &RR ,, cOLHf ÂcLmffGÂR} f@ a ,, &RR ,, >,, $LR ,, MO FRV}fn FRV f Of ,, $Q __ VLQIL"VLQUFRV[f ,, $L MO FRV f_ VLQ }f@ $W WKLV SRLQW LW LV FRQYHQLHQW WR GHILQH WKH QRUPDOL]HG KHOLFLW\ DPSOLWXGHV LR f, 9S7RO))$Q ,, ,, $Q ,, BB A fX 73LRO)076O ,, D ,3 $Q f f Y7$7R ,, ,, $Q ,, ,, $Q __ WHUPV LQVLGH ^` ODWHU UQ UQ G3 ; ; a ZKLFK VDWLVI\ c, mLR __ __ DQ __ c, DL __ f 7KH QRUPDOL]HG SUREDELOLW\ GLVWULEXWLRQ Â^AVGAfGc&26 FDQ EH REWDLQHG E\ GLYLGLQJ WKH SUREDELOLW\ GLVWULEXWLRQ E\ WKH EUDQFKLQJ IUDFWLRQ RI WKH VHTXHQWLDO GHFD\
PAGE 116
3 __ &TR __ $LR __ ,, $Q __ ,, $ ,,f ,, &RR ____ $LR __ __ &RR ____ $Q __ __ &RR ____ $LB __ ,, &RR L___ $BLR __ __ &TR O___ $BLBL __ __ &TR Oc_ $BLB __ 7KLV ZD\ G8 GFRV fGFRV Qf DLR __ FRV f_FRV W Of 2 2 DQ _> VLQAVLQ7 FRVAZf DL __ cUO HRV faVLQQ ,W LV XVHIXO WR SURMHFW WKH DQJXODU GLVWULEXWLRQ RQ DQG : G8fÂ§ __ DLR __ iFRVUOf __ DQ __ AVLQAI FRV!Uf __ DQ __ AVLQA G FRV f ,, fOR ,3 ,, rL __f_ FRVf __ DQ __ aVLQ f /HW XV QRZ UHWXUQ WR WKH WHUPV LQVLGH ^` LQ (TXDWLRQ ,W WXUQV RXW WKDW IRU WKHVH WHUPV D VXEVWDQWLDO VLPSOLILFDWLRQ RFFXUV LI ZH LQWHJUDWH RYHU FRV $IWHU VRPH DOJHEUD ZH DUULYH DW WKH IROORZLQJ H[SUHVVLRQ GID FW!Qf A f _^DLRDr DrrDA FRVA a 0' f ([DPSOH 7f ff WWW7f )RU WKH FDVH f WKH FDOFXODWLRQ LV YHU\ VLPLODU WR WKH f FDVH GLVFXVVHG DERYH 7KH DOORZHG KHOLFLWLHV IRU f DUH WKH VDPH DV WKH DOORZHG KHOLFLWLHV IRU f 7KLV PDNHV WKH DQJXODU GLVWULEXWLRQ FDOFXODWLRQ YHU\ VLPLODU
PAGE 117
,OO WR WKH f FDVH WKH RQO\ GLIIHUHQFH EHLQJ WKDW WKH IXQFWLRQV AD^fR DUH UHSODFHG E\ 22f n 7KH SKRWRQfV DQJXODU GLVWULEXWLRQ FRPHV RXW WR EH WKH VDPH 7KH UHVXOW IRU WKH SLRQfV QRUPDOL]HG DQJXODU GLVWULEXWLRQ LV G8 f G FRV m FRV "f fÂ§ FRV Q f DQ f UUVLQA FRVAf FRV Q fÂ§ f m __ fÂ§ VLQ A FRV Q Of f )LW RI WKH +HOLFLW\ $QJXODU 'LVWULEXWLRQ 2EWDLQHG IURP 'DWD ,Q SUDFWLFH D SDUWLFXODU ELQ LQ WKH KHOLFLW\ DQJXODU GLVWULEXWLRQ REWDLQHG IURP GDWD KDV ELQV ZLWK G1 fÂ§ 1G8 HYHQWV ZKHUH 1 LV WKH WRWDO QXPEHU RI HYHQWV ZKLFK LV DOORZHG WR IORDW GXULQJ WKH ILWf DQG G8 LV WKH SRUWLRQ RI WKH QRUPDOL]HG SUREDELOLW\ DVVLJQHG WR WKH ELQ E\ WKH KHOLFLW\ IRUPDOLVP SUHGLFWLRQ 7R LQFRUSRUDWH (TXDWLRQ GLUHFWO\ LQWR WKH ILW ZH PDNH WKH IROORZLQJ FKDQJH RI YDULDEOHV LO mLR __fÂ§ FRVFRV __ DL M_fÂ§ FRVkVLQI! __ DQ __ VLQk f %HFDXVH RI WKLV GHILQLWLRQ ERWK DQG L! VKRXOG EH UHTXLUHG WR EH ZLWKLQ WKH LQWHUYDO _f GXULQJ WKH ILW 7KH QHZ YDULDEOHV KDYH WKH FRQYHQLHQW SURSHUW\ WKDW IRU fÂ§ DQG IRU fÂ§ 7fÂ§ ,Q JHQHUDO WKHUH DUH DQG GHJUHHV RI IUHHGRP IRU WKH ILWV ZLWK fÂ§ fÂ§ DQG UHVSHFWLYHO\ 2QH GHJUHH RI IUHHGRP DOZD\V FRUUHVSRQGV WR WKH QXPEHU RI HYHQWV 1 DQG WKH RWKHU WZR
PAGE 118
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fÂ§ FRV Q DQG \ FRV Ef G1M R G[G\ 1M JL \fA f L9LFRV i \fA[ VLQ Lf_O ]ff f 2EWDLQLQJ WKH HUURU RI WKH QRUPDOL]HG KHOLFLW\ DPSOLWXGHV UHTXLUHV VRPH PDWUL[ DLR1 PXOWLSOLFDWLRQ )RU FRPSDFWQHVV OHW XV GHILQH D PDWUL[ IRU WKH QRUPDOL]HG KHOLFLW\ DPSOLWXGHV LV R? DQG OR a 9DLL n1 7KH HUURU 6D6D XL7 A 2nMRX ZKHUH 6MXM7 LV WKH HUURU PDWUL[ REWDLQHG IURP WKH ILW
PAGE 119
1M FRV FRV a \f_[ f G[G\ VLQ \ffÂ§LO [f f FRVHVLQr DfL_O [ff n OR 1M FRV k FRV a \fA[ [f VLQ A \ffA^ [f[ f f FRV VLQ \fA7O [ff OR $M FRV k FRV L! UfAA [ f D[D\ /] VLQ k A B \fA[O a [f[ f f W 2 = FRV kVLQ L! A I \fJAO fÂ§ [f[ fÂ§ Off ,Q WKLV DQDO\VLV EHFDXVH RI RXU ORZ VWDWLVWLFV DQG E\ ORZ VWDWLVWLFV ZH PHDQ WKDW WKHUH D[H QXPHURXV ELQV ZLWK HYHQWV LQ WKH SORWVf ZH SURMHFW (TXDn WLRQV RQ [ FRV A DQG \ FRV :H DOVR IROG HDFK GLVWULEXWLRQ DURXQG LWV V\PPHWU\ D[LV [ DQG \ WR LQFUHDVH HDFK ELQnV VWDWLVWLFV )LQDOO\ ZH GR D VLPXOWDQHRXV ILW WR ERWK GLVWULEXWLRQV DQG REWDLQ 1 k DQG $GPLWWHGO\ WKLV SURFHGXUH KDV WKH SUREOHPV GHVFULEHG SUHYLRXVO\ ZKHQ ILWV ZKHUH DGYRFDWHG ZKLFK FDQ JLYH ULVH WR V\VWHPDWLF HIIHFWV LQ WKH PHDVXUHG KHOLFLW\ DPSOLWXGHV ,Q HVVHQFH ZH QHHG WR XVH D XQLWDU\ ZHLJKW IXQFWLRQ SURSRUWLRQDO WR WKH HIILFLHQF\ GLVWULEXWLRQ VHH )LJXUH f LQ WKH LQWHJUDOV XVHG WR PDNH WKH SURMHFWLRQV 7KLV DIIHFWV WKH UHODWLYH VWUHQJWK RI HDFK KHOLFLW\ DPSOLWXGH DQG WKH RYHUDOO HIILFLHQF\ EXW GRHV QRW FKDQJH HDFK SURMHFWHG VKDSH +RZHYHU WKH KHOLFLW\ DPSOLWXGHV WKHPVHOYHV DUH RQO\ XVHG WR FDOFXODWH V\VWHPDWLF HIIHFWV RQ WKH HIILFLHQF\ G1MA G[G\
PAGE 120
RI UHVRQDQFHV VR E\ VLPXOWDQHRXVO\ ILWWLQJ WKH XQZHLJKWHG ,' SURMHFWLRQV ZH DUH MXVW LJQRULQJ WKH V\VWHPDWLFV RQ WKH V\VWHPDWLFV )LQDOO\ OHW XV FRQVLGHU WKH fÂ§ Sm GLVWULEXWLRQ IRU (TXDWLRQ EHFRPHV MUMfÂ§fÂ§U fÂ§AO A^FRVVLQ,!FRV$`FRV>e fÂ§ I!Qf@f f G?S\ Â7 9 ZKHUH WKH QHZ GHJUHH RI IUHHGRP $ LV WKH UHODWLYH SKDVH EHWZHHQ DLR DQG 7KH GHYLDWLRQ IURP WKH IODW GLVWULEXWLRQ LV FKDUDFWHUL]HG E\ 5 7HMHRV VLQ 7 FRV $` 7 7KH ELJJHU WKLV IDFWRU LV WKH HDVLHU LW LV WR REVHUYH LQ GDWD )LJXUH VKRZV WKH FRQWLQXXP VXEWUDFWHG fÂ§ !f GLVWULEXWLRQ IRU HYHQWV QHDU WKH f UHVRQDQFH 7KH IDFWRU 5 REWDLQHG IURP ILWWLQJ WKH GLVWULEXWLRQ LV QRQVLJQLILFDQW ,W LV D ELW WHGLRXV EXW VWUDLJKWIRUZDUG WR FDOFXODWH WKH fÂ§ MfQ GLVWULEXWLRQ IRU RWKHU YDOXHV :H ZLOO QRW SUHVHQW WKRVH UHVXOWV KHUH 7KH\ DUH QRW YHU\ LQWHUHVWLQJ EHFDXVH ZH QHHG ODUJH VWDWLVWLFV WR REVHUYH GHYLDWLRQV IURP WKH IODW GLVWULEXWLRQ DQG WKHUH LV QR HIIHFW RQ WKH HIILFLHQF\ IURP QRQILDW fÂ§ IfQ GLVWULEXWLRQV
PAGE 121
0,18,7 [r )LW WR 3ORW t 86!*DPPD K K SLSLSKL D[LV )LOH *HQHUDWHG LQWHUQDOO\ 3ORW $UHD 7RWDO)LW )XQF $UHD 7RWDO)LW \ IRU GRI (UURUV 3DUDEROLF )XQFWLRQ &20,6 )XQFWLRQ ;01&01 s 5 ( $8* )LW 6WDWXV ( '0 ( &/ b 0LQRV 22222(22 22222(22 2( 2222(22 0r )LJXUH &RQWLQXXP VXEWUDFWHG A fÂ§ cfQ GLVWULEXWLRQ IRU HYHQWV QHDU WKH f UHVRQDQFH ZLWK D ILW WR (TXDWLRQ RYHU ID\HG
PAGE 122
([DPSOH &RQWLQXXP (YHQWV 2XU SUHYLRXV H[DPSOHV WDFNOHG UDGLDWLYH GHFD\V RI WKH 76f 1RZ OHW XV FRQVLGHU WKH FRQWLQXXP fUDGLDWLYH EHDPf SURFHVV HH fÂ§} UU VHH )LJXUH f $W ILUVW JODQFH LW PLJKW VHHP WKDW WKH KHOLFLW\ IRUPDOLVP FDQQRW EH DSSOLHG LQ WKLV FDVH %XW D FORVHU ORRN DW WKH GLDJUDP RI WKLV SURFHVV VKRZQ LQ )LJXUH LQGLFDWHV RWKHUZLVH 7KH NH\ SRLQW LV WKDW UHJDUGOHVV RI WKH SDUWLFXODU FRXSOLQJ KHOLFLW\ IRUPDOLVP FDQ VWLOO EH DSSOLHG DW WKH &0 RI WKH UUf YHUWH[ ZKHUH WKH LQLWLDO VWDWH KDV S fÂ§ a 7 )LJXUH HH fÂ§! UUf LQ FRQWLQXXP UW SDLU SURGXFWLRQ XVXDOO\ RFFXUV WKURXJK S fÂ§ FRXSOLQJ :H RQO\ QHHG WR NQRZ WKDW ZH DUH GHDOLQJ ZLWK fÂ§} DW WKH WU YHUWH[ 7KHUH LV RQO\ RQH KHOLFLW\ DPSOLWXGH ZH QHHG WR FRQVLGHU f 96fnf f ZKLFK DIWHU D FRXSOH VLPSOH VWHSV OHDGV WR G1 W D fÂ§ ,9 VLQ Q D FRV Q 7KLV UHVXOW LV DOVR YDOLG IRU WKH NDRQ FDVH f 2Q WKH RWKHU KDQG WKH SKRWRQfV DQJXODU GLVWULEXWLRQ FDQQRW EH FDOFXODWHG XVLQJ WZRERG\ KHOLFLW\ IRUPDOLVP $ JRRG DSSUR[LPDWLRQ LV VHH >@ $SSHQGL[ %f
PAGE 123
G3 FRV M 7 D 7f Ef G FRV S fÂ§ FRV IF&\ ZKLFK LV QRW YDOLG DW ORZ DQJOHV fÂ§! fÂ§ Uf EHFDXVH RI WKH HIIHFWV RQ WKH DSSUR[LPDWLRQ RI WKH QHJOHFWHG HOHFWURQ PDVV 9 )LJXUH VKRZV WKHVH DQJXODU GLVWULEXWLRQV LQ 6 fUDGLDWLYH EHDPf HYHQWV 7KH DJUHHPHQW EHWZHHQ WKHRU\ DQG H[SHULPHQW LV H[FHOOHQW 7KLV LV UHDVVXULQJ EHFDXVH LQ WKH PDLQ DQDO\VLV ZH UHO\ RQ RXU DQJXODU GLVWULEXWLRQ PHDVXUHPHQWV WR JHW WKH FRUUHFW HIILFLHQF\ LQ GDWD DQG WR LGHQWLI\ WKH YDOXH RI HDFK UHVRQDQFH
PAGE 124
)LJXUH +HOLFLW\ DQJOH GLVWULEXWLRQ IRU fUDGLDWLYH EHDPf FRQWLQXXP HYHQWV HHf fÂ§ S DQG HHf fÂ§! M! LQ WKH 6 GDWD 7RS OHIW ULJKWf SORW VKRZV LQ JUD\ WKH HYHQWV VHOHFWHG DV S fÂ§! UU fÂ§! ..af 0LGGOH OHIW ULJKWf SORW VKRZV WKH 7UB ..af KHOLFLW\ DQJOH GLVWULEXWLRQ IRU WKH VHOHFWHG HYHQWV DQG KDV WKH KHOLFLW\ IRUPDOLVP SUHGLFWLRQ (TXDWLRQ RYHUOÂ£LG %RWWRP OHIW ULJKWf SORW VKRZV WKH DQJXODU GLVWULEXWLRQ IRU WKH SKRWRQ DFFRPSDQ\LQJ WKH S f DQG KDV WKH ILUVW RUGHU 4(' SUHGLFWLRQ (TXDWLRQ RYHUOD\HG
PAGE 125
5()(5(1&(6 >@ $ )LOLSSL f*OXHEDOOV DQG +\EULGV LQ WKH *H9 WR *H9 5HJLRQ $Q ([SHULPHQWDO 2YHUYLHZf H&RQI & )5:3 f >@ + 6WRFN &/(2 LQWHUQDO QRWH &%; f XQSXEOLVKHGf >@ )( &ORVH DQG $ .LUN (XU 3K\V & f >@ &$ 0H\HU f3URFHHGLQJV RI WKH :RUNVKRS RQ *OXRQLF ([FLWDWLRQVf 1HZSRUW 1HZV 9LUJLQLD $,3 &RQI 3URF f >@ .. 6HWK 1XF 3K\V % 3URF 6XSSOf f >@ %LQJ6RQJ =RX 1XF 3K\V $ f F >@ &0RUQLQJVWDU DQG 03HDUGRQ 3K\V 5HY f >@ $ 9DFDULQR DQG :HLQJDUWHQ 3K\V 5HY f >@ &( &DUOVRQ 7+ +DQVRQ DQG &3HWHUVRQ 3K\V 5HY f >@ 1 ,VJXU DQG 3DWQ 3K\V 5HY f >@ 6 1DULVRQ 1XF 3K\V % f >@ 0 0HOLV ) 0XUJLD DQG 3DULVL 3K\V 5HY f >@ 6 %URGVN\ $ 6 *ROGKDEHU DQG /HH 3K\V 5HY /HWW f >@ '+ 3HUNLQV f,QWURGXFWLRQ WR +LJK (QHUJ\ 3K\VLFVf WK (GLWLRQ &DPn EULGJH 8QLYHUVLW\ 3UHVV ,6%1 >@ 9 $PHOLQ HW DO 9(6 &ROODERUDWLRQf KHSH[ f >@ 6 (LGHOPDQ HW DO 3DUWLFOH 'DWD *URXS &ROODERUDWLRQf 3K\V /HWW % f >@ = %DL HW DO %(6 &ROODERUDWLRQf 3K\V 5HY /HWW f >@ $ $QDVWDVVRY HW DO &/(2 &ROODERUDWLRQf 3K\V 5HY /HWW f >@ 0DVHN HW DO &/(2 &ROODERUDWLRQf 3K\V 5HY f >@ 7 &RDQ 1XFO ,QVWUXP 0HWK $ f
PAGE 126
>@ < .XERWD HW DO &/(2 &ROODERUDWLRQf 1XF ,QVWUXP 0HWK $ f >@ 0 @ 0 $UWXVR HW DO &/(2 &ROODERUDWLRQf 1XF ,QVWUXUQ 0HWK $ f >@ %. +HOWVOH\ + 0DKONH.UXHJHU &/(2 LQWHUQDO QRWH &%; f XQSXEOLVKHGf >@ *($17 'HWHFWRU 'HVFULSWLRQ DQG 6LPXODWLRQ 7RRO &(51 3URJUDP /LEUDU\ /RQJ :ULWHXS : >@ 3 $YHU\ &/(2 LQWHUQDO QRWH &%; f XQSXEOLVKHGf >@ + 6WRFN 0 @ 5 *DOLN 1 3RZHOO 9LHKKDXVHU &/(2 LQWHUQDO QRWH &%; f XQSXEOLVKHGf >@ 5 0 %DOWUXVDLWLV HW DO 0DUN ,,, &ROODERUDWLRQf 3K\V 5HY f >@ ; < 6KHQ HW DO %(6 &ROODERUDWLRQf fLS 5DGLDWLYH 'HFD\Vf H&RQI & 7+$7 f >@ = %DL HW DO %(6 &ROODERUDWLRQf 3K\V 5HY /HWW f >@ & + &KDQJ DQG + 5 3DQJ KHSSK f >@ ; $ /LX ; 4 =HQJ < % 'LQJ ; 4 /L + 6KHQ DQG 3 1 6KHQ KHSSK f >@ 9 %XJJ 3K\V /HWW % f >@ 5LFKPDQ f$Q ([SHULPHQWHUfV *XLGH WR WKH +HOLFLW\ )RUPDOLVPf &$/7 f XQSXEOLVKHGf >@ 9 )HGH\HY .RURONRY %HVVRQ 5 *DOLN + 3DDU &/(2 LQWHUQDO QRWH &%; f XQSXEOLVKHGf
PAGE 127
%,2*5$3+,&$/ 6.(7&+ /XLV %UHYD1HZHOO ZDV ERUQ LQ %RXOGHU &RORUDGR RQ 'HFHPEHU :KHQ KH ZDV WZR \HDUV ROG KLV IDPLO\ PRYHG WR WKH %DVTXH &RXQWU\ UHJLRQ RI 6SDLQ +H JUDGXDWHG +LJK 6FKRRO IURP *(7;2 ,,, DQG PRYHG WR 0DGULG ZKHUH KH REWDLQHG D \HDU GHJUHH LQ 3K\VLFV IURP WKH 8QLYHUVLGDG &RPSOXWHQVH GH 0DGULG +H ZDV DFFHSWHG WR DWWHQG JUDGXDWH 6FKRRO DW WKH 8QLYHUVLW\ RI )ORULGD LQ
PAGE 128
, FHUWLI\ WKDW KDYH UHDG WKLV VWXG\ DQG WKDW LQ P\ RSLQLRQ LW FRQIRUPV WR DFFHSWDEOH VWDQGDUGV RI VFKRODUO\ SUHVHQWDWLRQ DQG LV IXOO\ DGHTXDWH LQ VFRSH DQG TXDOLW\ DV D GLVVHUWDWLRQ IRU WKH GHJUHH RI 'RFWRU RI 3KLORVRSK\ RKQ
PAGE 129
, FHUWLI\ WKDW KDYH UHDG WKLV VWXG\ DQG WKDW LQ P\ RSLQLRQ LW FRQIRUPV WR DFFHSWDEOH VWDQGDUGV RI VFKRODUO\ SUHVHQWDWLRQ DQG LV IXOO\ DGHTXDWH LQ VFRSH DQG 7KLV GLVVHUWDWLRQ ZDV VXEPLWWHG WR WKH *UDGXDWH )DFXOW\ RI WKH 'HSDUWPHQW RI 3K\VLFV LQ WKH &ROOHJH RI /LEHUDO $UWV DQG 6FLHQFHV DQG WR WKH *UDGXDWH 6FKRRO DQG ZDV DFFHSWHG DV SDUWLDO IXOILOOPHQW RI WKH UHTXLUHPHQWV IRU WKH GHJUHH RI 'RFWRU RI 3KLORVRSK\ 'HFHPEHU 'HDQ *UDGXDWH 6FKRRO
PAGE 130
/' 2
DECAYS OF THE T(1S) INTO A PHOTON AND TWO CHARGED HADRONS
By
LUIS BREVANEWELL
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2004
To my Parents.
ACKNOWLEDGMENTS
I owe my gratitude to many people without whom this work would not have
been possible. First of all, 1 would like to thank my supervisor, Dr. Yelton. His
office door was always open for me, and he spent long hours of his time listening
to my ideas even though only a few of them ever worked. He gave me freedom to
explore and branch off in different directions while steadily guiding me forward at
the same time. After these years of working together I consider him more than a
mentor, and a think of him as a close friend.
During my stay at the University of Florida I have met many people inside
and outside the physics world. I would like to thank my first year graduate teachers
Dr. Sikivie and Dr. Woodard who were a source of inspiration for me. My thanks
also go to my fellow graduate students Vijay Potlia, Rukshana Patel, Necula
Valentin, Jennifer Sippel, Suzette Atienza, G. Suhas, and many more for all our fun
physics discussions and late homework sessions. I would also like to thank the good
friends 1 made outside the physics building, James Power, Yaseen Afzal (Paki),
Ramji Kamakoti (Ramjizzle), Dan DeKee (Double Down), and Fernando Zamit
(Fernizzle) who always reminded me that there is more to life than Physics.
The most important person I have met during my graduate career is my wife,
Jennifer. I thank her for enduring all those endless days when I would answer her
questions with only grunts and nods while my attention remained fixed on the
computer screen. Her love and support are invaluable to me.
Finally, I would like to thank my parents Manuel and Charlene, my sister
Teresa, and my brother Gaizka. They have always been there for me.
iii
TABLE OF CONTENTS
page
ACKNOWLEDGMENTS Ã¼i
ABSTRACT vi
CHAPTER
1 THEORY 1
1.1 Particle Physics 1
1.1.1 The Standard Model 1
1.1.2 Quantum Chromodynamics 3
1.1.3 Introduction to the Radiative Decays of the T(1S) .... 6
1.2 Radiative Decays of Quarkonia Overview 14
2 EXPERIMENTAL APPARATUS 17
2.1 The Cornell Electron Storage Ring 17
2.2 The CLEO III Detector 20
2.2.1 Superconducting Coil 21
2.2.2 Tracking System 21
2.2.3 Crystal Calorimeter 24
2.2.4 Ring Imaging Cherenkov Detector 25
2.2.5 Muon Detectors 28
3 ANALYSIS OF THE DATA 31
3.1 Data Sample 31
3.1.1 Continuum Subtracted Distributions 32
3.2 Event Selection 34
3.2.1 Skim Cuts 35
3.2.2 Analysis Cuts 37
3.2.3 Cut Summary, Efficiencies, and Fake Rates 43
4 EXCLUSIVE RADIATIVE DECAY T(1S) > ttt+tT 48
4.1 Robustness of The Mass Distribution 48
4.2 Statistical Fit of the Invariant Mass Distribution: Signal Areas
and Their Significance 49
4.3 Angular Distribution of The Signal .' . . . 50
4.3.1 Optimum Mass Interval 51
4.3.2 Background Subtraction 53
IV
4.3.3Statistical Fit of the Helicity Angular Distribution: J AsÂ¬
signments and their Probability Distribution 55
5 EXCLUSIVE RADIATIVE DECAY T(1S) > 7K+K~ 66
5.1 Statistical Fit of the Invariant Mass Distribution: Signal Areas
and Their Significance 66
5.2 Angular Distribution of The Signal 67
6 EXCLUSIVE RADIATIVE DECAY T(1S) 7PP 80
7 SYSTEMATIC UNCERTAINTIES 86
7.1 Cuts 86
7.1.1 Justification of the DPTHMU Cut 87
7.2 Angular Distribution of Signal 88
7.3 Different Hadronic Fake Rates Between IS and 4S 97
7.4 Other Systematic Sources 97
7.5 Overall Systematic Uncertainties 99
8 RESULTS AND CONCLUSION 100
APPENDIX HELICITY FORMALISM FOR TWO BODY DECAYS .... 103
REFERENCES 119
BIOGRAPHICAL SKETCH 121
v
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DECAYS OF THE T(1S) INTO A PHOTON AND TWO CHARGED HADRONS
By
Luis BrevaNewell
December 2004
Chair: John Yelton
Major Department: Physics
Using the CLEO III detector we report on a new study of exclusive radiative
T(15) decays into the final states 77r+/T~, 7K+K~ and 7pp. We present branching
ratios for the decay modes T(1S) â€”> 7/2(1270), T(lS') â€”> 7/3(1525), T(IS') â€”â–º
7/o(1710) with /o(1710) K+K~ and T(15) > 7/4(2050).
vi
CHAPTER 1
THEORY
1.1 Particle Physics
Particle physics is the branch of physics dedicated to the study of matter
and energy at the most fundamental level. This means that the job of a particle
physicist is to identify the smallest constituents of matter and describe how they
interact with each other.
Humankind has been interested in this subject since ancient times. Two of the
first particle physicists in recorded history are the Greek thinkers Empedocles and
Democritus from the fifth century BC. Empedocles stated that our complex world
was made from combining four fundamental elements (earth, air, fire and water) in
different proportions. Democritus on the other hand, believed that the apparently
continuous objects in the natural world were not really continuous, but made from
voids and indivisible particles called atoms.
Over the last half century Particle physics has advanced tremendously and we
now have a beautiful, but incomplete, theory firmly grounded on experiment that
describes the fundamental constituents of matter and how they interact with each
other. This theory is called â€œThe Standard Model.â€
1.1.1 The Standard Model
According to the Standard Model, the fundamental building blocks of matter
are point like particles which interact with each other in as many as three different
ways. Each type of interaction, or force, is itself carried by point like particles
1
2
called force carriers. The particles which are force carriers are bosons 1 and are
collectively called gauge bosons because they are needed for the theory to be gauge
invariant. The nonforce carrying particles are fermions. 2
The three interactions described by the Standard Model are called the
electromagnetic force, the weak force, and the strong force. The force of gravity is
not included in the Standard Model, and this is one of the reasons the Standard
Model is not yet complete. The electromagnetic force carriers are photons, the
weak force carriers are the ZÂ°, IT+, and W~ particles, and finally, the strong force
is carried by eight kinds of gluons. Table 11 summarizes the situation.
Table 11: Gauge bosons and the force they carry.
Symbol
Name
Force Carried
7
Photon
Electromagnetic
ZÂ°
ZÂ°
Weak
w+
W+
Weak
w~
W~
Weak
9
gluon
Strong
Particles which interact through a particular force are said to couple to it
and to carry an associated charge. The nomenclature is as follows, particles that
interact through the electromagnetic force have an electromagnetic charge called
electric charge, those which interact through the weak force have a weak charge
called weak isospin, and those which interact strongly have a strong charge called
color.
The rest of the particles in the Standard Model which are not force carriers,
the fermions, are subdivided into different groups depending on their properties
(See table 12) , reminiscent of the way chemists organized the elements into
1 Bosons are defined as particles with integer spin in quantum mechanics.
2 Fermions are defined as particles with half odd fractional spin.
3
the periodic table during the second half of the nineteenth century. Fermions are
divided into quarks (generically represented by the symbol q) and leptons. The
main difference between these two groups is that quarks interact through the strong
force while leptons do not. There are six types of leptons and six types of quarks
(also called the six quark flavors) which are grouped into three generations. Each
generation consist of two quarks and two leptons. All three generations replicate
the same set of force charges, the main difference between generations is the mass
of the particles (for example, the ratio of the masses of e : /x : t is 1 : 200 : 3500).
For each fermion there is an antifermion with equal mass and spin and opposite
charge.
Table 12: Fermion symbols classified into quarks, leptons, and the three generaÂ¬
tions along with the generational common charges.
First
Generation
Second
Generation
Third
Generation
Electric Weak Has
Charge Charge Color
Quarks
u
d
c
s
t
b
(2/3 +1/2 Yes
1/3 1/2 Yes
Leptons
Ve
e
vn
V
vr
T
0 +1/2 No
1 1/2 No
1.1.2 Quantum Chromodynamics
Quantum chromodynamics (QCD) is the part of the Standard Model that
describes the strong force. QCD is based on local gauge invariance and color
symmetry. There are three possible color charges for quarks called r (red), b (blue),
and g (green). Antiquarks have opposite colors called f, b, g. The strong force
between quarks only depends on their colors and is independent of their flavor.
A very important characteristic of the strong force is that gluons themselves
carry a color charge and an anticolor charge so they can interact with other quarks
through the strong force and change their color. Since there axe three colors and
three anticolors one might think that there are nine different gluons. However,
there is one linear combination of color anticolor states that has no net color and
4
leaves a quark unchanged. There are therefore 9 â€” 1 = 8 gluons. The 8 individual
gluon color states can be written as follows,
Â° *â€¢
1 > =
2 > =
3 > =
4 > =
72
(rb + br)
(â– rb â€” br)
(rf â€” bb)
72
J_
75
j=(rg + gf)
(1.1)
5 > = f={rggr)
6 > = \{bg + gb)
V2
7 > â€¢= 7/=(&Â£ 
8 > = =(rf + gg  266)
V6
and represent the eight different gluons that exist in nature. The single and unique
color state left out is called a color singlet,
9 >=~(rr+ bb +gg) (1.2)
which is invariant under a redefinition of the color (a rotation in color space). In
group theory this decomposition of the color states into an octet and a singlet is
denoted by 3 0 3 = 8 Â® 1. It is worth noting here that a colorless sate, such as 3 >
or Â¡8 >, is not necessarily a color singlet state.
This situation is analogous to the perhaps more familiar example of two spin
1/2 particles. Each particle can have their spin up (T) or down (f) corresponding to
four possible combinations, each giving a total spin S = 0 or 1. The S = 1 states
5
form a triplet,
1 > =  T'T>
2 > = J~( tl> +1 U>) (1.3)
3> =  U>
and there is a singlet state with S = 0,
4 >= b( Tl> I lt>). (14)
Since gluons themselves carry color they can interact with each other through
the strong force. This interaction among the force carriers is unique to the strong
force and, when included in perturbative QCD calculations, leads to two important
properties observed in nature called â€œasymptotic freedomâ€ and â€œcolor confinement.â€
Asymptotic freedom means that the interaction gets weaker at short distances.
Color confinement is the requirement that observed states have neutral color, or in
other words, they must be in a color singlet state.
Confinement explains why free quarks or free gluons, which have a net color
charge, have never been observed. It also explains why no fractional charged
particles made from a qq bound state have never been observed since it is not
possible to construct a color singlet for such a state (in group theory terms
3 3 = 6 Â® 3 where we have a sextet and a triplet, but no singlet). On the other
hand, color singlets can be constructed for a qq or qqq system. The color singlet for
qq is simply the state shown on the right side of Equation 1.2. The color singlet for
qqq can be obtained from the decomposition 3<8>3<8>3 = 10Â©8Â®8Â©1 and is,
\qqq >coior singlet= A={rgb  grb + brg bgr + gbr  rbg). (1.5)
V6
Particles that are bound states of qq are abundant in nature and are called
mesons, those that are bound states of qqq are called baryons and also abound in
6
ft
nature. Both groups are collectively called hadrons. There have been hundreds
of different hadrons observed, confirming the validity of QCD and the quark
flavors. Nevertheless, QCD leaves room for more possibilities. A bound state of two
gluons gg (sometimes called a glueball) can be in a color singlet and in principle
could be observed. Although glueballs are allowed by QCD there is no convincing
experimental observation of one. Another possibility are bound states that are
a mixture of the previous states, such as aqq + /3gg with arbitrary a, (3. These
states are called â€œhybrid mesonsâ€ or simply â€œhybridsâ€. It is believed [1], [2], [3] that
hybrids are necessary to explain the spectrum of light mesons between 12 GeV.
As with any quantum mechanical theory, the spectrum of bound states is
a fundamental test. 3 Glueballs are allowed by QCD, yet there is no concluÂ¬
sive experimental evidence of their observation, despite intense experimental
searches [4], [5], [6] complemented by lattice QCD calculations [7], [8] and other
theoretical contributions like bag models [9], fluxtube models [10], QCD sum
rules [11], weakly bound boundstate models [12], and QCD factorization formalism
models [13]. Physicists cannot be sure they understand QCD until such states are
observed, or until they can explain why we cannot observe them.
1.1.3 Introduction to the Radiative Decays of the Y(1S)
The Y(1S) is a meson composed of bb quarks. Mesons of this kind, composed
of a quark and an antiquark of the same flavor, are in general called quarkonia.
There is a convention behind the name of the T(1S) . The T symbol is
reserved for particles composed of bb where the combined spin of the quark and
antiquark is 1. The â€œISâ€ symbol is borrowed from atomic spectroscopy with the
â€œ1â€ meaning that the bb pair are in the lowestenergy boundstate, and the â€œSâ€
3 A good example of this is the successful description of the hydrogen atomâ€™s
spectrum by quantum mechanics.
7
meaning that the bb have a relative angular momentum L â€” 0. A description of
particle naming conventions can be found in [14].
The T(1S) is unstable, existing for only about 10â€œ24 seconds after which it
decays into daughter particles (which in turn decay themselves if they are not
stable). The term â€œradiative decayâ€ is reserved to any T(1S) decay where one of
the stable daughters is a photon.
The different ways the Y(1S) can decay must obey the symmetries in nature.
Examples of such symmetries are â€œparityâ€ and â€œcharge conjugationâ€, both of which
will be described soon. Symmetries are very important in the standard model. The
usefulness of symmetries can be seen in Noethersâ€™ theorem, which states that for
every symmetry there is an associated conserved quantity. Examples of Noethersâ€™
theorem are the conservation of momentum connected to translational invariance
and the conservation of angular momentum associated with rotational invariance.
In the next sections we will use the parity and charge conjugation symmetries
to find T(1S) decays which conserve the associated symmetry constants and are
allowed by nature. In particular, we will show that the radiative decay of the
Y(1S) through a photon and two gluons (see Figure 11) is allowed. The key
observation is that the two gluons must be in a color singlet since both the Y(1S)
and the radiated photon have no color and color must be conserved. This means
that the two gluons satisfy color confinement and could form a glueball, although
more conventional meson states, or hadrons in no bound state at all, are also
possible outcomes. Regardless of what the gluons do, their energy will eventually
manifests itself as hadrons. Sometimes, two charged hadrons of opposite charge will
emerge. This work examines those two hadrons from a radiative Y(1S) decay to
experimentally probe the gg spectrum.
8
Parity
The parity operator, P, reverses the sign of an objectâ€™s spatial coordinates.
Consider a particle a > with a wave function \ka(x, t). By the definition of the
parity operator,
P*{x,t)=PaVa(X,t) (16)
where Pa is a constant phase factor. If we consider an eigenfunction of momentum
'S>f{x,t) = ei&2Et) (1.7)
then
P$p{x,t) = Pa%i~x,t) = Pa^p{x,t), (1.8)
so that any particle at rest, with p = 0, remains unchanged up to a multiplicative
number, Pa, under the parity operator. States with this property are called
eigenstates with eigenvalue Pa. Pa is also called the intrinsic parity of particle a,
or more usually just the parity of particle a, with the words at rest left implicit.
Since two successive parity transformations leave the system unchanged, P% = 1,
implying that the possible values for the parity eigenvalue are Pa = Â±1.
In addition to a particle at rest, a particle with definite orbital angular
momentum is also en eigenstate of parity. The wave function for such a particle in
spherical coordinates is,
W2.0 = Mr)17"(M. (19)
where (r, 0, ) are spherical polar coordinates, Rni(r) is a function of the radial
variable r only, and the Yjm(0, (Â¡>) is a spherical harmonic.
The spherical harmonics are well known functions which have the following
property,
>7(0, Â«Hi/vn*Â«,* + <Â»â€¢
(1.10)
9
Hence
PVnlm&t) == Pa^mm(X,t) = Pa(l)l^nlm(x,t) (1.11)
proving that a particle with a definite orbital angular momentum l is indeed an
eigenstate of the parity operator with eigenvalue Pa(â€”1).
The parities of the fundamental fermions cannot be measured or derived. All
that nature requires is that the parity of a fermion be opposite to that of an antiÂ¬
fermion. As a matter of convention fermions are assigned P = +1 and antifermions
are assigned P = â€”1. In contrast, the parities of the photon and gluon can be
derived by applying P to the field equations resulting in P7 = â€” 1 and Pg = â€”1.
The T(1S) has P = PbPb{ 1)L = 1 since L = 0.
Parity is a good quantum number because it is a symmetry of the strong and
electromagnetic force. This means, that in any reaction involving these forces,
parity must be conserved.
Charge Conjugation
Charge conjugation is simply the operation which replaces all particles by
their antiparticles. In quantum mechanics the charge conjugation operator is
represented by C. For any particle a > we can write,
C\a>=ca\a> (1.12)
where ca is a phase factor. If we let the C operator act twice to recover the original
state  a >,
a >= C2\a >= C(ca\a >) = caCja >= cacsa > (1.13)
which shows that caca = 1. If (and only if) a is its own antiparticle, it is an
eigenstate of C. The possible eigenvalues are limited to C = ca = c5 = Â±1.
All systems composed of a the same fermion and an antifermion pair are
A >
eigenstates of C with eigenvalue C = (â€”l)(i+sb This factor can be understood
because of the need to exchange both particlesâ€™ position and spin to recover the
10
original state after the charge conjugation operator is applied. Exchanging the
particlesâ€™ position gives a factor of (â€”1)L as was shown in the previous section,
exchanging the particles spin gives a factor of (â€”l)s+1 as can be verified by
inspecting Equations 1.3 and 1.4, and a factor of (1) which arises in quantum field
theory whenever fermions and antifermions are interchanged. With this result we
can calculate the charge conjugation eigenvalue for the T(IS) and obtain C â€” â€”l
since L + S = 1.
The photon is an eigenstate of C since it is its own antiparticle. The C
eigenvalue for the photon can be derived by inserting C into the field equations and
is C1 = â€”1.
Finally, we consider a system composed of two gluons that are in a color
singlet. The two gluons are bosons and they must have a symmetric wave function,
Tg under a gx exchange. Under this exchange, the orbital angular momentum
part of the wave function contributes with a factor ( 1)L, the spin part of the
wave function for two spin 1 particles contributes with a factor of (â€”l)s, and the
color singlet part of the wave function contributes with a factor of +1 since it is
symmetric. To ensure that 'I'g is symmetric we need L + S to be even. This implies
that C = (â€”l)i+s = +1. The L â€” 0 and Lâ€” 1 possible gg bound states are shown
in Table 13. The JPC = 1â€œ+ is peculiar because it is impossible for a qq system to
have these quantum numbers. If this state is ever observed, it must be a glueball.
Experimental searches for such a state have been done, for example, in [15].
Table 13: Possible gg bound states with L â€” 0 or L = 1. The possible quantum
numbers are limited by the condition that J = L + S be even, which is needed to
ensure a symmetric wave function for the two gauge bosons.
L
S
JFC
0
0
0++
0
2
2++
1
1
0+,l~+,2~+
11
Charge conjugation is a symmetry of the strong and electromagnetic force. For
those particles that are eigenstates of O, C is a good quantum number because in
any reaction involving these forces C must be conserved.
The P and C values of various particles used in this analysis are shown in
Table 14 along with their quark composition., orbital angular momentum, and
internal spin.
Table 14: Symbol, name, quark composition, angular momentum (L), internal spin
(S), parity (P), and charge conjugation eigenvalues (C) for a few of the particles
used in this analysis.
Symbol
Name
Quark Composition
L
S
P
c
T(1S)
Upsilon(lS)
bb
0
1
1
1
7r+
Pion
ud
0
0
1
X
7râ€œ
Pion
dÃ¼
0
0
1
X
K+
Kaon
us
0
0
1
X
K~
Kaon
su
0
0
1
X
P
Proton
uud
0
1/2
+1
X
P
Antiproton
Ã¼Ã±d
0
1/2
1
X
7
Photon
X
X
1
1
1
Possible Decays of the T(1S)
At this point we can understand the different possible ways the T(1S) can
decay. The Possible decays are limited because the strong and electromagnetic
force must conserve color, C and P.
The simplest possibility is for the bb pair to interact electromagnetically and
annihilate into one virtual photon 4 . This is allowed by parity, charge conjugation
symmetry and color conservation. The decay of the Y(lS) to one gluon is not
allowed by color conservation and is therefore forbidden. Y(1S) decays to two
photons are forbidden by charge conjugation. T(1S) decays to two gluons are
4 Such a photon is called virtual because it cannot conserve the 4momentum of
bb and is unstable, only existing for a brief period of time, as allowed by the uncerÂ¬
tainty principle, after which it decays.
12
also forbidden by charge conjugation. Y(1S) decays to 3 gluons are allowed (3
gluons can form a color singlet). Y(1S) decays to 3 photons are also allowed, but
are largely suppressed by decays to one photon since 3 successive electromagnetic
interactions are much less likely to occur than a single one. Finally, Y(1S) decays
two one photon and two gluons are also allowed under the condition that the two
gluons be in a color singlet state.
The three different possible T(1S) decays with least amount of interactions
(also called lowest order decays) are shown in Figure 11.
(a)
(b)
(C)
Figure 11: Lowest order decays of the T(1S) allowed by color conservation, charge
conjugation symmetry, and parity, (a) Shows the decay into three gluons, (b) shows
a radiative decay, and (c) shows the electromagnetic decay through a virtual phoÂ¬
ton that in turn decays electromagnetically into a pair of charged fundamental
particles, such as quarks or charged leptons (the charged leptons are represented by
the symbol l).
Observable Resonances
In this work we search for a resonance X produced in a radiative T(1S) decay,
T(1S) Â» 7X.
(.1.14)
Using chaige conjugation symmetry on both sides,
â€” 1 = â€”1 Cx
(1.15)
Therefore, Cx = +1.
In order for us to observe X it must decay into two charged hadrons,
X 4 h+h~, (1.16)
where h = tt, K, p are the hadrons whose momentum we are going to measure.
Applying charge conjugation to this last decay,
+1 = (1)^+*, (1.17)
where L and S are respectively the angular momentum and spin of the h+h~
system. This last equation implies that L + S must be even. This has consequences
for the possible X parities we can observe. By parity conservation in 1.16,
Px = (1)L(1)(1) = (l)i. (1.18)
For h = 7T, K, S â€” 0 and L â€” J must be even, which implies that Px = +1. For
h â€” p, S can be 0 or 1 and particles with both positive and negative parities can be
detected.
Table 15: Possible 5, L, J, P and C values for X from the radiative decay
T(1S) â€”> 7X reconstructed in different, decay modes.
Decay Mode
s
L
J
P
C
X 4 7T + 7T~
0
even
even
+1
+1
X 4 K+K~
0
even
even
+1
+1
X >pp
0
even
even
+1
+1
x > PP
1
odd
even and odd
1
+1
14
1.2 Radiative Decays of Quarkonia Overview
Theoretical models exist for glueball production in quarkonia decay [12]
and for the glueball spectrum. For example, a quenched lattice calculation [7]
predicts a Jpc = 2++ glueball 5 in the 2.2 GeV/c2 mass region (see Figure 12).
According to Table 15, in the charged pion and kaon modes, we are limited to
detect glueballs in the leftmost column where P = C = +1 of Figure 12, while in
the proton mode we are restricted to the two left most columns.
4
3
>
CD
o
C!
2 E
1
0
Figure 12: Quenched lattice calculation result for the glueball spectrum for differÂ¬
ent P and C values. The mass scale is shown in terms of a scaling parameter from
the QCD lattice calculation named r0 on the left and in absolute terms on the right
by taking rp1 = 410 MeV.
0 Here J stands for the internal angular momentum (spin) of the glueball,
J = L + 5.
%
15
For some time bound states not considered to be pure glueballs, such as
/2(1270), /4(2050), 77, and rf have been observed in J/ip radiative decays at
the 10~3 production level [16] 6 . In 1996, the BES collaboration claimed the
observation of a resotiance they called the /./(2220) particle in the radiative J/ip
system at the 10~5 level [17] (see Figure 13). A lot of excitement was generated at
the time because it is possible to interpret the /j(2220) as a glueball. However, this
result has not been confirmed.
27 70301 *004
Figure 13: The mass spectrum obtained by the BES collaboration in radiative
JÂ¡ip decays into different hadronic modes.
Key to identifying a particle as a glueball are (a) suppressed production in
twophoton collisions (unlike quarks, gluons donâ€™t carry electric charge and do not
6 The J/ip particle is a cc bound state. Since the strong force is flavor blind the
situation in J/ip decays is in principle similar to that in T(1S) decays.
couple to photons), and (b) flavor symmetric decays, since a pure glueball has no
valence quarks.
V;/*
CLEO has already done several studies of radiative decays of the T(lS').
Naively, one expects these types of decays to be suppressed by a factor
[(Qb/qc)(mc/mb)}2 Â» 0.025 (1.19)
with respect to J/xp radiative decays. This comes from noticing that the quark
photon coupling is proportional to the electric charge and the quark propagator
is approximately 1/m for low momentum quarks. In 1999, CLEO made the first
observation of a twobody T(15) radiative decay [18]. The spin of the observed
resonance could not be measured, but its mass and width where consistent with the
/2(1270) particle. Under this assumption, comparing the measured branching ratio
of T(1S) â€”> 7/2(2220) to the measured branching fraction of the J/xp â€”> 7/2(1270),
a suppression factor of 0.06 Â± 0.03 was obtained. After the BES result for the
fj (2220) in radiative J/xp decay's, a corresponding search was performed by CLEO
in the radiative T(1S) system [19]. This analysis put limits on the fj(2220)
production in radiative Y(1S) decays.
In this work we are privileged to have available the largest collection of
radiative T(1S) decays in the world. With it, we can study the structure of color
singlet gg hadronization, and shed more light on the fj(2220) result from BES.
CHAPTER. 2
EXPERIMENTAL APPARATUS
To carry out our study of the dihadron spectrum we need to first produce
the T(1S) resonance and secondly observe its daughter particles flying away at
relativistic speeds. These two tasks are respectively accomplished by the Cornell
Electron Storage Ring and the CLEO III detector.
2.1 The Cornell Electron Storage Ring
The Cornell Electron Storage Ring (CESR), located at Cornell University, is a
circular particle accelerator that produces e+e~ collisions.
In order to produce such collisions electrons and positrons need to be created,
accelerated and stored. CESRâ€™s different components, shown in Figure 21, have
been carrying out this task since 1979.
A typical CESR run begins at the linear accelerator (LINAC) where electrons
and positrons are produced. To create positrons, electrons are evaporated off a
filament and linearly accelerated by electromagnetic fields towards a tungsten
target. The collision creates a spray of electrons, positrons and photons. The
electrons are cleared away with magnetic fields and the positrons are introduced
into the synchrotron. The filling procedure is identical, except that the tungsten
target is removed.
Once the electron and positron beams are introduced into the synchrotron,
they are accelerated to the operating energy. In hour case they are accelerated to
the point where their combined energy is the Y(1S) mass, 9.46 GeV.
Once the beams are at the desired energy they are transfered to the storage
ring, where they will remain for about an our. At one point in the storage ring the
17
18
SYNCHROTRON
WEST
TRANSFER
LINE
EAST
TRANSFER
LINE
LINAC e
beams are forced to cross paths. This is the point where e+e collisions occur 1
and where the center of the CLEO III detector is located. If the accelerator is
â€¢ BUNCH OF POSITRONS
O BUNCH OF ELECTRONS
Figure 21: The Wilson Laboratory accelerator located about 40 feet beneath
Cornell Universityâ€™s Alumni Fields.
performing well a high collision rate results. A high collision rate is crucial for the
success of an accelerator and the experiments it serves. The important figure is the
number of possible collisions per second per unit area; this is called the luminosity.
In order to maximize the luminosity, the beams are focused as small as possible at
1 This point is not fixed in space, but varies from event to event inside a small
volume of space called the interaction region (IR).
19
the IR. During the CLEO III installation several magnetic quadrupÃ³les were added
to CESR to improve the beam focus. CESR has consistently outdone itself while
collecting luminosity over the years (see Figure 22).
9000
8000
7000
5000
5000
13
a
~ 4000
CA
o
c
E
3
300G
2000
1000
Figure 2 2: CESR yearly luminosity. The gaps in 1995 and 1999 correspond to
down times when the CLEO II.V and CLEO III detectors where being installed.
The CLEO III detector measures the time integrated luminosity over a period
of time by counting how many times a benchmark process occurs during a certain
time interval at the IR. For redundancy, there are two benchmark processes that
are used, one where the e+e~ particles interact to produce a new e+eâ€” pair,
and the other one where the e+e~ annihilate and produce two photons. Using
the known crosssection for each process, the number of events is converted to a
luminosity called the Bhabha integrated luminosity for the first process, and the 77
integrated luminosity for the second one. The term â€œintegrated" is sometimes left
out and the total luminosity is referred to as simply the Bhabha or 77 luminosity,
with the time integration left implicit.
2.2 The CLEO III Detector
When the e+e~ collision occurs, the two particles are annihilated we enter
the world of particle physics. Nature decides what to do with the energy from
the annihilation. We have no chance of directly observing what is happening at
the annihilation point, but eventually long lived semistable particles are created
that fly off at relativistic speeds. These particles carry information about what
happened after the e+e~ collision, and can tell us what nature did. The CLEO III
detector has the important mission of detecting and measuring such par ticles.
As one can see in Figure 23, the CLEO III detector is a composite of many
detector elements. These subdetectors are typically arranged as concentric cylinÂ¬
ders. The entire detector is approximately cube shaped, with one side measuring
about 6 meters, and weighs over 500 thousand kilograms.
CLEO III
SC Quadrupole
Pylon
SC
Quadrupoles
Rare Earth
Quadrupole
Solenoid Coil
Barrel
Calorimeter
RICH
Drift
Silicon /
oeampipe
Magnet
Iron
Endcap
Calorimeter
Iron
Polepioce
Barrel Muon
Chambers
Figure 23: The CLEO III detector.
21
As implied by the name. CLEO ill is not the only CLEO detector. CLEO
III was preceded by CLEO II.V, CLEQ^JI, CLEO I.V and the original CLEO
>
detectors. The CLEO III detector was a major upgrade compared with the
previous version of CLEO [20], [21], and. has an improved particle identification
system together with a new drift, chamber and a new silicon vertex detector.
2.2.1 Superconducting Coil
All the detector subsystems except for the muon chambers are located inside
a superconduction coil. The coil remains unchanged since CLEO II. It is kept in
a superconducting state by liquid helium. The purpose of the coil is to provide a
1.5 Tesla magnetic field, which is uniform to 0.2%, to bend the paths of charged
particles in the detector. By measuring how much much a charged particle bends,
experimenters can measure the momentum of the particle.
The coil inner radius is 1.45m and its outer radius is 1.55 m, with a radial
thickness of 0.10 m. The total length of the coil in z is 3.50 m. It is wound from
a 5mm x 16 mm superconducting cable (A1 surrounding CuNbTi strands). It
is wound in 2 layers, with 650 turns per layer, on an aluminum shell. When in
operation a current of 3300 amps flows through the coil.
2.2.2 Tracking System
After particles from the interaction point pass through the beam pipe, they
begin to encounter the active detector elements of the tracking system. There are
twro subdetectors responsible for tracking the curving path of charged particles.
The first one encountered by particles is the silicon vertex detector, and the second
one is the central drift chamber. The CLEO III tracking system is responsible
for tracking a charged particleâ€™s path and measuring its momentum. Typical
momentum resolution is 0.3% for 1 GeV tracks. The tracking system also measures
ionization energy losses with an accuracy of about 6%.
22
Silicon Vertex Detector
The silicon vertex detector in CLEO III [22], also called SVD III, is a four
layer barrelonly structure with no endcaps that surrounds the beam pipe. This
detector (see Figure 24) provides four 0 and four z measurements covering over
over 93% of the solid angle. The average radius of inner surface of the four layers is
25 mm, 37.5 mm, 72 mm, 102 mm. The detector is constructed from 447 identical
doublesided silicon wafers, each 27.0 mm in 0, 52.6 mm in z and 0.3 nun thick.
The wafers are instrumented and read out on both sides. The instrumentation on
each side consists of an array of aluminum strips on the wafer surface. These strips
are connected to preamplifiers at the end of the detector. The inner side has 512
strips in the z direction and the outer side has 512 in the 0 direction. Therefore
each wafer contains 512+512 sensors. The 447 wafers are arranged in the 4 layers,
as follows: 7 sections in 0, each with 3 wafers in z, total = 21 wafers in the first
layer; 10 sections in 0, each with 4 wafers in z, total = 40 wafers in the second
layer; 18 sections in 0, each with 7 wafers in z, total = 126 wafers in the third
layer; 26 sections in phi, each with 10 in Z, total = 260 wafers in the fourth layer.
Charged particles traversing the wafer lose energy and create electron hole
pairs. Approximately 3.6 eVis required to create a single electronhole pair. The
electrons and holes then travel in opposite directions in the electric field applied
to the surfaces of the wafers until they end up on the aluminum strips, and the
detector registers a â€œhitâ€. When combined together, the hit on the inner side of a
wafer and the hit on the outer side give a measurement of the (z, 0). The wafer
position itself determines r.
The Central Drift Chamber
The CLEO III central drift chamber (DR III) is full of a gas mixture with 60%
Helium and 40% propane held at about 270 K and at a pressure slightly above one
atmosphere. The drift chamber is strung with array of anode wires of goldplated
ff
23
Figure 24: View of the SVD III along the beampipe.
tungsten of 20 Â¿Â¿m in diameter and cathode wires of goldplated aluminum tubes
of 130 /iin in diameter. The anodes are kept at a positive voltage (about 2000 V),
and the cathodes are kept grounded, which provides an electric field between the
anode and the cathode wires. Anode and cathode wires are often called â€œsenseâ€
and â€œfieldâ€ wires respectively.
As a charged particle passes through the DR III, it interacts electromagneti
cally with the gas molecules giving energy to the outer electrons which become free
in a process called ionization. The free electrons from the ionized gas molecules
drift in the electric field toward the nearest anode wire. As the electrons get close
to the anode, the electric field becomes very strong which causes an avalanche
as further ionization is induced. The result of the avalanche is a large number of
electrons collapsing upon the sense wire in a very short amount of time (less than
one nanosecond). When this happens to a sense wire, we say that there is a â€œhitâ€.
24
The current on the anode wire from the avalanche is amplified and collected at
the end of the anode wire. Both the amount of charge and the time it takes it to
move to the end of the detector are measured. A calibration of the drift chamber
is used to convert the amount of charge to a specific ionization measurement of the
incident particle. A calibrated drift chamber can also convert the time to roughly
measure the position along the sense wire where the charge was deposited.
The wires are strung along the 2: direction. About 2/3 of the outer part of
the drift chamber (the farthest part from the interaction point) is strung in with a
slight angle (about 25 miliradians) with respect to the 2 direction to help with the
z measurement. Wires strung in the z direction are called â€œaxialâ€ wires, while those
that are strung at a slight angle are called â€œstereoâ€ wires.
The DR III consists of an inner stepped section with 16 axial layers, and an
outer part with conical endplates and 31 small angle stereo layers. There are 3
field wires per sense wire and they approximately form a 1.4 cm side square. The
drift resolution is around 150 pm in r â€” 4> and about 6 mm in 2. All wires are held
at sufficient tension to have only a 50 pm gravitational sag at the center (2 = 0).
There are 1696 axial sense wires and 8100 stereo sense wires, a 9796 total.
2.2.3 Crystal Calorimeter
The CLEO Crystal Calorimeter (CC) is composed of 7784 thalliumdoped
Csl crystals. Each crystal is 30 cm long (16.2 radiation lengths) with 5cm x 5cm
square front face. The crystals absorb any incoming electron or photon which
cascades into a series of electromagnetic showers. The electronic system composed
of 4 photodiodes present at the back of each crystal are calibrated to measure the
energy deposited by the incoming particle. Other incoming particles other than
photons and electrons are partially, and sometimes fully, absorbed by the crystal
giving an energy reading.
25
'
i
The CC is arranged into a barrel section and two endcaps, together covering
95% of the solid angle. The CC barrel section is unchanged since CLEO II; the
endcaps have been rebuilt for CLEO III to accommodate the new CESR interaction
region quadrupoles. The barrel detector consists of an array of 6144 crystals, 128
in cf> and 48 in 2, arranged in an almostprojective barrel geometry. That is, the
crystals are tilted in 2 to point to a few cm away from the interaction point, and
there is also a small tilt in . The CC barrel inner radius is 1.02 m, outer radius
is 1.32 m, and the length in 2 at the inner radius is 3.26 m. It covers the polar
angle range from 32 to 148 degrees. The barrel crystals are tapered towards the
front face (there are 24 slightly different tapered shapes), the endcap crystals are
rectangular, but shaved near the outer radius to fit in the container. The CC
endcaps consist of two identical end plugs, each containing 820 crystals of square
crosssection, aligned parallel to the beam line (not projective). There are 60
crystals in the â€™â€™fixedâ€ portion of the â€™â€™keystoneâ€ piece of the endcap. 760 in the
part that slides. The keystone is made up of two parts, one on top that has 12
crystals that for mechanical removal reasons is separate from a container holding 48
crystals. The endcap extends from 0.434 m to 0.958 m in r. The front faces are 2 =
Â±1.308 m from the interaction point. It covers the polar angle region from 18 to 34
degrees in +z, and 146 to 162 in 2.
The photon energy resolution in the barrel (endcap) is 1.5% (2.5%) for 5
GeVphotons, and goes down to 3.8% (5.0%) for 0.1 GeVphotons.
2.2.4 Ring Imaging Cherenkov Detector
The Ring Imaging Cherenkov (RICH) detector [23] is a newr detector subsysÂ¬
tem for CLEO III. It replaces the CLEO II.V time of flight system designed to
measure particlesâ€™ velocities.
26
Cherenkov radiation occurs when a particle travels faster than the speed of
light in a certain medium,
v > c/n. (2.1)
Where v is the velocity of the particle, c is the speed of light in vacuum, and n
is the index of refraction of the medium the particle is traveling in. The charged
particle polarizes the molecules of the medium, which then turn back rapidly
to their ground state, emitting radiation. The emitted light forms a coherent
wavefront if v > c/n and Cherenkov light is emitted under a constant Cherenkov
angle, 5, with the particle trajectory forming a cone of light. The cone halfangle is
given by the Cherenkov angle which is,
cos 5 = â€”
vn
1
/3n
(2.2)
If the radiation angle, 5, is measured, the speed of the incident particle is known.
This measurement, combined with the momentum measurement from the tracking
system, gives a measurement of the particles mass, and can be used in particle
identification.
The threshold velocity at which Cherenkov radiation is emitted is vmin =
When a. particle traveling at the threshold velocity transverses the medium a very
small cone with Ã³ Â« 0 is produced. The maximum emission angle occurs when
Vmax = c and is given by
COS Â¿max = â€¢ (2.3)
n
The RICH (see Figure 25) consists of 30 modules in phi, 0.192 m wide and
2.5 m long. The detector starts at a radius of 0.80 m and extends to 0.90 m. Each
module has 14 panes of solid crystal LiF radiator at approximately 0.82 m radius,
0.192 m wide, 0.17 m long, 1 cm thick. Inner separation between radiators is
typically 50/im. The LiF index of refraction is n = 1.5. The radiators closest to z
= 0 in each module have a 45 degree sawtooth outer face, to reduce total internal
27
reflection of the Cherenkov light for normal incident particles (see Figure 26). The
radiators are followed by a 15.7 cm (radial) drift space filled with pure N2. The
drift space is followed by the photodetector, a thingap multiwire photosensitive
proportional chamber.
Figure 25: The RICH detector subsystem.
track
X
/
10 mm \
TV? 1
t Lâ€”
170
mm ^
track
Figure 26: The two kinds of RICH LiF radiators. For normal incidence particles
(z ~ 0) a sawtooth radiator is necessary to avoid internal reflection.
With this index, particles in the LiF radiator with beta = 1 produce
Cherenkov cones of halfangle cos_1(l/n) = 0.84 radians. With a 16 cm drift
28
space, this produces a circle of radius 13 cm. The RICH is capable of measuring
the Cherenkov angle with a resolution of a few miliradians (see Figure 27). This
great resolution allows for good separation between pions and kaons up to about 3
GeV as Figure 28 shows.
TD
cu
E
â€”1â€”'â€”
tâ€”Tâ€”iâ€”'â€”iâ€”Tâ€”iâ€”'â€”r
1



â€¢ Data
â–
o Monte Carlo
1 
â–
T â–
â–

o
"
â€œ

lili
â–
â€¢
t I T T
â–
" 1
* 'ii

â€¢ t
â–
0

â–
â–


â€¢
â–
â–
â–
I
J . 1 . 1 . 1 i L
â– 1
1 2 3 4 5 6 7
Radiator row
Figure 27: Cherenkov angle resolutions per track as a function of radiator row for
Bhabha events. Row 1 corresponds to the two rows closest to z = 0, etc.
2.2.5 Muon Detectors
The muon detectors (MU) are the most external subsystem of the CLEO III
detector. They remain unchanged from CLEO II, and are composed of plastic
proportional tubes embedded in the magnet iron return yoke. They cover 85%
of the 47t solid angle (roughly 30150 degrees in polar angle). If a series of hits is
detected in the muon chamber layers they most likely correspond to muons because
other particles are blocked by the iron. Besides detecting muons, the heavy iron
29
Figure 28: PiÃ³n fake rate as a function of particle momentum for kaon efficiency of
80% (circles), 85% (squares) and 90% (triangles).
of the return yoke protects the inner subsystems of the CLEO III detector from
cosmic ray background (except for cosmic ray muons of course).
There are three planes of chambers in the barrel section, arranged in 8 octants
in (f>. The plastic barrel planes lie at depths of 36, 72, and 108 cm of iron (at
normal incidence), corresponding to roughly 3, 5, and 7 hadronic interaction
lengths (16.8 cm in iron) referred to as DPTHMU. There is one plane of chambers
in each of the two endcap regions, arranged in 4 rough quadrants in d>. They lie
at z â€”Â± 2.7 m, roughly covering the region 0.80 < cos(0) < 0.85. The planar
tracking chambers use plastic proportional counters at about 2500 V with drift gas
of 60% He, 40% propane, identical to (and supplied by the same system as) the
drift chamber gas. Individual counters are 5 m long and 8.3 m wide, with a space
30
resolution (along the wire, using charge division) of 2.4 cm. The tracking chambers
arc made of extruded plastic, 8cm wide by 1 cm thick by 5 m long, containing eight
tubes, coated on 3 sides with graphite to form a cathode, with 50 gm silverplated
CuBe anode wires held at 2500 V. The orthogonal coordinate is provided by
8cm copper strips running perpendicular to the tubes on the side not covered by
graphite.
CHAPTER 3
ANALYSIS OF THE DATA
This work builds on the techniques developed by previous CLEO radiative
Y(1S) analyses [18], [19], modified to be used with CLEO III data. A new
technique based on kinematic fitting is developed to, together with the new
RICH detector, improve efficiency and particle identification.
We search for radiative T(1S) decays in the modes T(1S) â€”â– > 77r+7r~,7K+K~,
and 7pp. The e+e~ collision data has both resonant events, where the e+e~
annihilate to give a T(1S) , and continuum events, where the e+e~ collision does
not give a Y(1S) . To be sure we are observing Y(1S) and not a continuum process,
the continuum must be subtracted by using a pure source of correctly scaled
continuum events. Pure continuum data can be obtained by operating CESR at an
energy different form the Y(1S) mass. After subtracting the underlying continuum,
we examine the dihadron invariant mass spectrum in search of resonances. We
determine the spin and production helicity (the projection of the spin on the
momentum vector at production time) of any found resonances by examining the
photon and hadron angular distributions.
3.1 Data Sample
The analysis presented here is based on CLEO III data. Throughout this
document, and unless otherwise stated we use 77 luminosity (the luminosity types
used by CLEO are defined at the end of Section 2.1). We prefer to use the 77 over
the Bhabha luminosity because the resonant process T(1S) â€”> e+e~ artificially
increases the reported Bhabha luminosity by about 3% in T(1S) data. This extra
contribution would need to be accounted for when doing a continuum subtraction.
Choosing the 77 luminosity avoids this complication.
31
32
The CLEO III data is divided into numbered sets. Sets 18 and 19 have a
luminosity of 1.13 Â± 0.02/fr1 in the beam energy range 4.7274.734 GeV. This
data, which we call the T(1S) data (or simply the IS data), has both resonance
e+e~ â€”> Y(1S) and continuum events. We take the number of resonant events
from [24], Nr(is) = (21 Â± 0.1) x 107. This number contrasts with the previous
generation measurement of CLEO II, where Nr(is) ~ 0.15 x 10' were available.
In datasets and 18 and 19 there are also 0.192 Â± 0.004/6â€œ1 taken below the
Y(1S) beam energy (4.7144.724 GeV). This data, which we call the T(lS)ofF data
(or simply the lSoff data), has relatively low statistics and corresponds to purely
continuum events.
To improve our continuum statistics we use 3.49 Â± 0.07fb~l from datasets
9, 10, 12, 13, and 14 of data taken near the centerofmass energy of the Y(4S),
which for our purposes is defined as data with beam energy in the 5.2705.300 GeV
range. This set of data, which we call the Y(4S) data (or simply the 4S data), is a
source of pine continuum because no T(4S) â€”> BB resonant event can survive our
â€œcutsâ€ 1 (the cuts are presented in Section 3.3).
3.1.1 Continuum Subtracted Distributions
We use the continuum data taken at the T(4S) energy to subtract the underÂ¬
lying continuum present in the T(1S) data. This is important because continuum
background processes like e+eâ€œ â€”* yp with p â€”* 7r+7râ€œ, e+eâ€œ â€”> 74> with
4> â€”> K+K~, and direct e+e~ â€”> 7h+h~ (we will use the convention h = 7r, K, p
from now on), look like the signal events we are searching for. To first order, the
cross section of these continuum process scales like 1/s, where s is the square of
1 Cuts are simply conditions that an event must satisfy to be considered in the
Â«analysis. Cuts are necessary to eliminate background that would otherwise make a
measurement difficult or even impossible.
33
the center of mass energy of the e+e~ system. Taking the luminosityweighted
average beamenergies of each interval,, and the 77 luminosities (see Table 31), we
calculate that the T(4S) data scaled down by a factor of 0.404 2 represents the unÂ¬
derlying continuum in the T(1S) data. This is true up to differences in momentum
distributions and phase space. The error in the continuum scale factor is unknown
because the luminosity ratio is expected to have a small but undetermined sysÂ¬
tematic error. We make the somewhat arbitrary decision to retain three significant
digits in the continuum scale factor because it is sufficient for our purposes and
there are 0.5% effects from second order terms in the cross section formulae.
To eliminate the contribution of continuum events from a T(1S) data variable
distribution (e.g., the invariant mass of two tracks, the photon angular distribution)
we proceed as follows,
â€¢ Obtain the T(4S) data distribution for the same variable.
â€¢ Efficiency correct both the Y(1S) and T(4S) distributions, using a
GEANT [25] based Monte Carlo (MC) simulation of the detector. Examples of
MC efficiency distributions are shown in Figures 33 and 73.
2 If we are to be mathematically strict we should calculate the scale factor as
with obvious notation. This is equivalent to redefining the average energy as
However, our energy intervals are sufficiently narrow and this calculation does
not change the last significant digit of the original scale factor. Similarly, taking
into account second order terms in the energy dependence of the cross section, like
the one that appears as m2/s in the explicit formula for the e+e~ â€”> 7p cross sec
tion, also has an insignificant effect on the scale factor.
34
â€¢ Subtract the T(4S) distribution from the Y(1S) distribution using a Y(4S)
scale factor of 0.404.
We call this set of steps â€œcontinuum subtractionâ€ by definition. For comÂ¬
pactness, we call such a distribution the â€œcontinuum subtracted
distribution/plotâ€ or â€œ continuum subtracted distribution/plotâ€. Except
for statistical fluctuations and phase space effects, the resulting IS distribution
should not have any contribution from continuum processes that scale as 1/s.
It is important to notice that any continuum subtracted distribution is
efficiency corrected. This means that a fit to a continuum subtracted distribution
(for example, the continuum subtracted invariant mass distribution) gives the
efficiency corrected number of events directly. Strictly speaking, this number of
events is only correct if all the other variables we cut on have the same initial
distribution in data and MC, or if the efficiency does not depend on them. In this
note we use â€œflat MCâ€, defined as MC that is generated with a flat distribution in
the mass and the helicity angles #7, 8n (these angles are defined in the appendix).
For example, the number of events obtained from the continuum subtracted
invariant mass fit needs to be corrected to account for the fact that the helicity
angle distributions are not flat in data (see for example Figure 49, and the
efficiency is highly dependent on these variables (see Figure 73). This correction is
done in Section 7.2.
A summary of the results from this section is shown in Table 31.
3.2 Event Selection
Event selection for Y(1S) â€”> ~fh+h~ is straightforward and can be thought of
in terms of three major stages.
First we skim the data, keeping only those events that contain exactly one
highenergy photon and two tracks. Next, we require that the total 4momentum
of these three elements be consistent with the colliding e+e~ 4momentum. Finally,
35
Table 31: Summary of the data used in this analysis. The continuum scale factor
is obtained using 77 luminosities because the Bhabha luminosity is artificially high
during T(1S) running due to the process T(1S) â€”> e+e~.
T(1S)
T(4S)
T(lS)off
Dataset
18, 19
9, 10, 12, 13, 14
18, 19
Average Â¿Â£wi(GeV)
4.730
5.286
4.717
Range of Ffceam(GeV)
4.7274.734
5.270  5.300
4.7144.724
Â¿(e+eK/ir1)
1.20 Â± 0.02
3.56 Â± 0.07
0.201 Â± 0.004
â– C(77)(fbâ€˜)
1.13 Â±0.02
3.49 Â± 0.07
0.192 Â±0.004
T(1S) continuum scale factor
1
0.404
5.84
in the third stage, we project the surviving data onto the three different hadronic
modes via hadron separation and QED suppression cuts.
However, checking the 4momentum involves using the tracks masses. This
means that the information from stage 2 should somehow be useful in stage 3. This
is indeed the case, and the details of how we do it are revealed in this section.
3.2.1 Skim Cuts
We skim over the data in the â€œhardGamâ€ subcollection. The hardGam
subcollection was developed with this type of analysis in mind. For an event that
passes the triggers 3 to be classified as hardGam it must pass the following cuts,
â€¢ eGaml > 0.5
â€¢ eSh2 < 0.7
â€¢ eOverPl < 0.85
â€¢ eVis >0.4
â€¢ aCosTh < 0.95 where â€œeGamlâ€ is the highest isolated shower energy
relative to the beam energy, â€œeSh2â€ is the energy of the second highest shower
relative to the beam energy, â€œeOverPlâ€ is the matched shower energy relative to
3 Triggers are basic criteria that an event must satisfy to recorded during the
data collection processes. Triggers are designed to get rid of trash and noise and
reduce the size of the data sample while keeping all of the important information.
36
if
the momentum of the track with highest momentum, â€œeVisâ€ is the total energy
detected (charged tracks are assumed to be pions) relative to the center of mass
energy, and â€œaCosThâ€ is the z component of the unitary total momentum vector.
MonteCarlo predicts that about 75% of the generated T(1 S) â€”â–º 'yh+h~ signal
passes the hardware and software triggers and gets classified as hardGam.
As mentioned above, we use this data to malee our skim. To write an event
from the hardGam subcollection into our skim we require the following topological
cuts,
â€¢ There are exactly two â€œgood tracksâ€; there can be any number of tracks that
are not â€œgood tracksâ€ but these are not used in the analysis. We define a â€œgood
trackâ€ as a track that satisfies the following cuts; drift chamber track ionization
energy loss (dE/dX) information is available, the ratio of number of wire hits to
those expected is between 0.5 and 1.5, the pion fit has x2/do.f. < 20 (here d.o.f.
stands for degrees of freedom), and the distance of closest approach to the beam
spot in the xy plane (called DBCD) is less than 5 â€” 3.8P (mm) if P < 1 GeV/c
(where P is the tracks momentum in GeV/c) and less than 1.2 mm for tracks with
P > 1 GeV/c. This DBCD cut is common in the more sophisticated CLEO II/II.5
analyses. It performs better than a simple DBCD < 5 mm cut, because it takes
into account the fact that traks with higher momentum have a better measurement
of DBCD since they scatter less.
â€¢ There is exactly one â€œgood showerâ€, there can be any number of showers
that are not â€œgood showrersâ€ but these are not used in the analysis. We define a
â€œgood showerâ€ as an unmatched shower with energy > 4 GeV.
These topological cuts are about 85% efficient for generated signal events that
have passed the triggers and have been classified as hardGam.
The overall skim efficiency is between 6065%, depending on the mode (see
Table 34).
37
3.2.2 Analysis Cuts
After our skim we call any cuts we make â€œanalysis cutsâ€. These cuts are done
at analysis time and are mode dependent. As a convention, and unless otherwise
stated, efficiencies for individual analysis cuts are reported relative to the events in
the skim (not relative to the events generated).
4momentum Cut
All fully reconstructed events should have the 4momentum of the eTe~
system. This constraint is usually implemented with a simple twodimensional
AEp box cut, where AE is the difference between the reconstructed energy for
the event and the colliding e4e~ energy (Ecm), &nd P is the magnitude of the
reconstructed total momentum for the event. Typical values for these cuts are
â€”0.03 < AE/Ecm < 0.02 and p < 150 MeV/c (taken from [19]).
The traditional AEp box cut is somewhat useful. However, it does not take
into account the correlation between the measured energy and momentum. Indeed,
the signal lies in diagonal bands in the AEp plane, making a boxshaped cut not
optimal (see Figures 3la and 3lb).
We use an alternative approach to the 4momentum cut. After a. simple
substitution, = p^ = Ecm ~ Ek+h (where is the photonâ€™s energy, p7 is
the magnitude of the photons momentum, Ecm is the energy of the e+e~ system,
and Eh+h is the energy of the hadron pair), we can mite the Ep conservation
equations as:
Vh+h T {Ecm  Eh+h )Pi â€” Pcm (3.1)
where, Ph+h is the dihadron momentum, p7 is the photonâ€™s momentum unit
vector, and pcm is the momentum of the e 're~ system (which is a few MeV because
of the crossing angle). Equation 3.1 is a 3constra.int subset of the 4momentum
constraint and has the convenient property of avoiding the use of the measured
38
photonâ€™s energy, which has nonGaussian asymmetric errors. It is important to
notice that Equation 3.1 contains the dihadron energy, therefore, it can help
discriminate between the various particle hypotheses.
We proceed as follows. After vertexing the hadron pair using the beam spot
with its error matrix, we calculate the photonâ€™s direction from the hadron pairâ€™s
vertex and the shower position. We then fit the event to the 3 constraints expressed
in Equation (1) using the techniques outlined in [26] and cut on the x2 of the
3constraint fit, XepW <100. To complete the 4momentum requirement, we
calculate AE(h) = Eh+h + E1  ECm, where Eh+h is the updated dihadron
energy after the constraint, and Ey is the measured photonâ€™s energy, and require
0.050 * Ecm < AE(h) < 0.025 * Ecm Furthermore, we now have available
X2Ep[h) differences between different particle hypotheses, which help in particle
identification (ID). This is discussed in more detail in the next section. Figure 31
compares the performance of the old and new approaches to the Ep cut.
At this point it is a good idea to check that the 4raomentum cut rejects
background events that make it through our skim cuts (see Section 3.1). These
events typically have one high energy shower, two tracks, and (an) additional
element(s). One such background is T(1S) â€”> 77r+7r~7rÂ°. Out of 25000 T(lS') â€”Â» 77,
with i] â€”> 7r+7r_7rÂ° MC events 4 only 4 survive our 4momentum cut. We conclude
that our 4momentum cut is good at rejecting background events that pass the
skim cuts but have additional elements such as an extra photon, 7rÂ°, pair of tracks,
etc.
Hadron Separation
We define Ax2Â£>(/q â€” h2) for the particle hypotheses h\ and h2 of our charged
track pair (e.g. AxjD(^ â€” K)) as follows,
4 Thanks to Vijay Potlia for generating these events
39
0.50
0.25
0.00
0.25
TT
\
0 0.00
o
CL
0.25
C.00
0.25
0.00
0.50 0.40 0.30 0.20 0.10 0.00 0.10 0.20
AE (GeV)
Figure 31: Distributions for different 4momentum cuts. Signal MC events
(T(L9) â€”Â» 77T+7râ€œ) are represented by the black dots, and â€œbackgroundâ€ MC events
(e+e~ 70, 0 â€”> K+ K ~) appear as (red) triangles. Plot a) has no cuts. Plot b)
has the old 4momentum boxcut. Plot c) has the new 4momentum cuts. Plot d)
has the new 4momentum cuts, and also a cut defined as XepM ~ Xep(K) < 0.
The particle ID potential of the newly available X2E~p(h) is evident.
1 ~ ^2) â€” adE/dx(hi ) ~ adE/dx(^2 ) + adE/dx(h 1 ) ~ adE/dx(^2 )
2 \og(C rich (ht)) +2 log (Crich^z)) fi.2)
2 log(C,rich(K)) P 2log(Â£ Ricnih'i)).
where the idea is to combine the dE/dX and RICH information into one number.
Pairs of tracks with AxjD(hi â€” h<Â¿) <0 axe more likely to be of type hi than of type
h2.
In practice, we only add a tracks RICH information if its momentum is above
the Cherenkov radiation threshold for both mass hypotheses and there are at least
3 photons within 3
We also require that both hypotheses were actually analyzed by RICH during
pass2.
In addition to RICH and dE/dX, and as hinted in the previous section, the
difference in XepW from the constraint expressed in Equation 3.1 can help the
particle ID (see Figures 3lc and 3ld). We define,
Ax_p(/ii â€œ M = xlP(^i)  (33)
Events with Ax\^p{hi  h^) < 0 are more likely to be of type h\ than oÃ type hi.
In this analysis, to select hi and reject hi the default cut is simply AxjD{hi â€”
hi) < 0. This simple cut is highly efficient, has low fake rates, and is sometimes
sufficient. However, out of the six possible cases when one hadron fakes another,
there are three important cases where it pays off to also use Ay_p(hi â€” hi)
together with ~ ^2) in an optimal way;
1. 7T background to K. This background comes from the continuum process
e+e_ â€”â–º 7p, p â€”â–º 7r+7r~.
2. 7r background to p. Again, this background comes from the continuum
process e+eâ€œ â€”> yp, p â€”> 7r+7r~.
3. K background to p. This background comes from the continuum process
e+e_ â€”> 70, 0 â€”> K+K~.
In other words, the important cases occur when the lighter mass hypothesis
fakes the heavier mass hypothesis.
Mathematically, one would expect that simply adding both Ax]D and Axep
together (like we just did when combining RICH and dE/dX), and cutting on the
grand Ay2 is the way to go. Unfortunately, this simple approach fails because of
large nonmathematical tails in the individual Xep distributions.
Instead, for each of these three cases we define the best cut values (C1.C2) in
Ax2d < ci and AXep < c2 as those that maximize
41
F(cl5c2) *=â€¢
1L
(3.4)
/Re + W * Rfâ€™
where Re (Rf) is the efficiency (fake rate) of the particle ID cuts and IT is the
rough ratio of the background to signal in the data sample for each case. Each W
canâ€™t be known a priori, but a rough idea of its value can be obtained by doing
a first iteration of the analysis with, for example, c\ â€” c2 = 0. We use W = 20,
W â€” 60, and W â€” 30 for cases 13 respectively. Figure 32 shows F(ci,c2) for each
case. Table 32 shows the optimized cut values and their effect on particle ID.
10 5 0 5 10
c,
Figure 32: Contour plots of F(ci, c2) as defined in Equation 3.4 for different sigÂ¬
nals and backgrounds from flat MC. flop has a K signal and a tt background. CenÂ¬
ter has a p signal and a vr background. Bottom has a p signal and a K background.
Quantum Electrodynamic Suppression
Quantum Electrodynamic (QED) background in this analysis comes from the
abundant processes e+e~ â€”> 7e+e'â€œ and e+eâ€œ â€”â–º 7p^pT.
42
Table 32: Cut values, efficiencies, and fake rates for Ax}D < C\ and AXep < c2 in
flat MC. The different ci, C2 values are chosen so F from Equation 3.4 is maximun.
Efficiency and fake rates of each cut are reported relative to events in the skim.
Statistical errors in the efficiencies are 0.1% or less. Errors in the fake rates are
statistical! and are shown for completeness only.
Cut Value
Cl c2
Signal efficiency (%)
aX/d A Xev Both
Fake rate (%)
Axiâ€ž
Both
7r faking K
3
2
94.2
95.2
89.9
1.73 Â±0.09
14.5 Â± 0.3
0.31 Â± 0.04
7r faking p
1
2
99.0
99.8
98.9
0.89 Â± 0.06
1.03 Â± 0.07
0.03 Â±0.01
K faking p
1
3
98.8
99.3
98.1
1.89 Â±0.09
1.09 Â± 0.07
0.10 Â±0.02
To reject e+e â€”* ^e+e~ we require both tracks to have a matched shower with
energy E such that \E/p(n) â€” 0.95 > 0.1, and to have AXw(h â€” e) < 0.
To reject e+e~ â€”> ryp+p~ we simply require D PTH MU < 5 (D PTH MU
was defined in Section 2.2.5) for both tracks in the K and p modes because particle
ID cuts make the pion (and therefore the muon) fake rate small (see Tables 32
and 36). For the 7r mode we cannot use particle ID in a practical way because
muons and pions have similar masses. Instead, to separate pions and muons we
use a much stronger cut requiring that both tracks be within the. barrel part of
the muon chamber ( cos(0) < 0.7), both have P > 1 GeV/c and both have
D PTH MU < 55 .
To improve the overall muon suppression cut efficiency with virtually no
increase in muon fakes, we flag an event as â€œnot muonicâ€ if any of the tracks
deposit more than 600 MeV in the CC, This increases the cut efficiency by about
90% in the n mode and makes the detector more hermetic.
5 Other analyses (for example [18]) typically use D PTH MU < 3, our CLEO III
MC has too many pion tracks with 3 < I) PTH MU < 5 and two few tracks with
DPTHMU = 0.
43
3.2.3 Cut Summary, Efficiencies, and Fake Rates
Table 33 summarizes the cuts used in this analysis. Figure 33 shows the
overall Monte Carlo efficiency after all cuts. Figure 34 shows the fake rates
according to the MC and the data for different particle ID cuts. The data fake
rates and limit fake rates, which are measured for pions and kaons faking other
hadrons, are calculated from the p and Ã© peaks in the continuum (see Figures 42
and 51). Tables 34 through 36 summarize the results of Figures 33 and 34.
Table 33: Cuts used in this analysis.
Motivation
Definition for T(1S) â€”* 7h+h (where h = tt, K, or p)
Skim cuts
Data acquisition
Event must pass hardware (Levell) and software triggers
(pass2) and be of type hardGam
Topological
There are only two good tracks and only one isolated shower
with E > 4 GeV
Analysis cuts
Reconstructed event must
have 4momentum of the
centerofmass system
XEp(h) < 100.0 and 0.050 < AE{h)/ECM < 0.025
Hadron separation
Default is, AxjD < 0 The three cases with a large fake popÂ¬
ulation because of continuum processes use a simultaneously
optimized cut on AxjD an(l AXepi aiid are summarized in
Table 32
QED background
e+e~ â€”* je+e~
Both tracks have a matched shower energy E that satisfies
\E/p(tt) â€” 0.95 > 0.1, and AXidÃ^1 ~ e) < 0
QED background
e+e â€”> 7p.~
For h = K, p both tracks have D PTH MU < 5. For
h = n (At least one track has a matched shower energy >
600 MeV) or (( Both tracks have cos(0) < 0.7 and P >
1 GeV/c) and (both tracks have DPTHMU < 5))
45
r
Table 3 4: Efficiencies in % for cuts (as outlined in Table 33) for flat signal MC.
Efficiencies in the second group are reported relative to the number of candidates
that make it to the skim. The third part of the table shows the overall reconstrucÂ¬
tion efficiency. Statistical! errors are 0.1% or less.
. .a... ..... .
Skim cuts
7T
K
P
Hardware Trigger
89.7
89.0
90.9
Software Trigger
96.5
96.2
96.6
hardGam
78.4
79.9
76.5
Topological (ntracks = 2 &. E1 > 4GeV)
73.3
67.6
72.3
Overall Skim efficiency
64.3
60.3
62.4
Analysis cuts
7T
K
V
4momentum
98.6
98.5
99.0
QED e+e~ â€”> ye+e" suppresion
93.9
87.4
93.1
QED /D"/x~ â€”> 7/x+/i~ suppresion
74.7
93.0
98.3
Hadron separation
97.1
89.0
97.8
Overall analysis efficiency
66.9
79.1
89.1
Overall reconstruction efficiency
43.0
47.6
55.6
Table 35: Efficiencies in % for cuts (as outlined in Table 33) for flat 4S MC. EffiÂ¬
ciencies in the second group are reported relative to the number of candidates that
make it to the skim. The third part of the table shows the overall reconstruction
efficiency. Statistical! errors are 0.1% or less.
Skim cuts
7T
K
V
Hardware Trigger
88.4
88.0
89.0
Software Trigger
94.6
94.6
94.8
hardGam
79.2
80.5
77.9
Topological (ntracks â€” 2 & E1 > 4GeV)
75.2
69.5
75.8
Overall Skim efficiency
66.9
62.7
66.6
Analysis cuts
7T
K
V
4momentum
97.1
97.0
97.4
QED e+e_ â€”* 7e+e~ suppresion
93.4
89.1
.93.2
QED /â€”> 7^'l'/i_ suppresion
75.8
92.5
98.1
Hadron separation
98.0
91.4
96.6
Overall analysis efficiency
67.4
80.2
87.3
Overall reconstruction efficiency
45.0
50.3
58.2
46
Table 36: Final efficiencies and fake rates after all cuts in %. Statistical errors
for efficiencies are 0.1% or less. Statistical errors for fake MC rates are shown for
completeness only. The IS DATA corresponds to the IS off resonance data sample.
The MC was generated flat.
7T cuts.
K cuts
p cuts
IS 7r MC
43.0
0.14 Â±0.01
< 0.007
4S 7T MC
45.0
0.29 Â±0.02
<0.01
IS p DATA

< 1.2
< 0.2
4S p DATA
 â–
0.20 Â± 0.06
< 0.06
IS K MC
1.27 Â± 0.03
47.6
< 0.02
4S K MC
0.92 Â± 0.04
50.3
< 0.06
IS <Â¡> DATA
< 3.8

< 2.0
4S
4.14 Â±0.69

< 0.4
IS p MC
0.08 Â± 0.02
0.34 Â± 0.04
55.6
4S p MC
0.05 Â±0.01
0.37 Â±0.01
58.2
0 12 3
Invariant mass (GeV/c2)
Figure 33: Final efficiencies for each mode as a function of invariant mass for the
IS (solid) and 4S (dashed) Monte Carlo data. The MC was generated with a flat
angular distribution.
47
Figure 34: Fake rates for IS (hollow circles) and 4S (hollow squares) according to
flat MC. The (red) downward pointing triangle is obtained using the p peak in 4S
data.
CHAPTER 4
EXCLUSIVE RADIATIVE DECAY T(1S) * ^+'n~
In Figure 42 the 7r+7râ€œ invariant mass plot is shown for both IS and 4S data.
Figure 43 shows the continuum subtracted 7r+7r~ invariant mass distribution (as
defined in Section 3.2.1) with the most likely statistical fit overlayed (which is
described in the next section). The number of events within ir of the p region
(0.62 â€” 0.92 GeV/c2) left after the continuum subtraction is 200 Â± 300, and 50 of
these belong to the /o(980) lowmass tail.
4.1 Robustness of The Mass Distribution
In [27] the decay T(1S) 77rÂ°7rÂ° is analyzed, and it is shown how the analysis
stream warps the shape of the reconstructed resonance. This effect, which arises
because of the particular 7rÂ° behavior, raises problems when fitting the invariant
mass distribution.
In Section 3.2.1 we claimed that if the data and MC had the same 07 and
6k distributions, the fit to the continuum subtracted invariant mass distribution
automatically gives the correct efficiency corrected number of events.
Here we test this claim. To this end, we generate 10000 T(1S) â€”Â» 7/2(1270),
with ^(1270) â€”* 7r+7r~ with flat 07 and 0* distributions. We treat this MC
as data and carry out the first two steps of the continuum subtraction process.
The resulting peak has a mass of 1.278 Â± 0.002 GeV/c2 and a width of 0.193 Â±
0.006 GeV/c2, consistent with the generated mass and width of 1.275 GeV/c2 and
0.185 GeV/c2. More importantly, the number of reconstructed events from the fit is
10040 Â± 180, which is consistent with the number of generated events.
Figure 41 shows the reconstructed and efficiency corrected events, a fit to
them, and the generated events.
.1 ;
48
49
We conclude that there is no warping of the mass distribution, and that the
analysis stream behaves like we expect when obtaining the efficiency corrected
number of events from data.
â– Reconstructed and Eff. Corrected
râ€”i MC Generated
Figure 41: Reconstructed events and efficiency corrected events, a fit to them, and
the generated events for Y(1S) â€”Â» 7/2(1270), with /2(1270) â€”> 7r+7r~ with flat 67
and On distributions.
4.2 Statistical Fit of the Invariant Mass Distribution: Signal Areas and
Their Significance
The results of this section are summarized in Table 41.
Figure 43 shows possible signals for for T(15) â€”> 7/o(980), T(IS') â€”â–º
7/2(1270) and T(15) â€”> 7/4(2050). To determine the number of events in each
signal we fit the invariant n+n~ mass continuum subtracted distribution with three
spindependent, relativistic BreitWigner line shapes. The masses and widths are
allowed to float, except for the width of the /4(2050) which has a very large error
if allowed to float and is set to its PDG value. The PDG values[16] for the mass
and width of the three resonances are nif0^8o) â€” 980 Â± 10 MeV/c2, F/0(98o) â€”
70 Â± 15 MeV/c2, mh{1270) = 1275.4 Â± 1.2 MeV/c2, r/2(127o) = 185.ll; MeV/c2,
ra/4(2050) = 2034 Â± 11 MeV/c2, and T/42050) = 222 Â± 19 MeV/c2.
50
To measure the statistical significance of each signal we do multiple x2 fits
fixing the signal area to different values while letting the mass and width of
the signal whose significance is being measured to float within 2a of the PDG
values. At the same time the rest of the fit parameters are fixed to the values that
originally minimized the x2 We assign each of these multiple fits a probability
proportional to e*2/2 and then normalize. We calculate the chance of the signal
being due to a random fluctuation by adding the normalized probabilities for the
fits with a negative or 0 signal. This method fails for the highly significant /2(1270)
signal because the e~x2/2 value of fits with negative or 0 /2(1270) signal is of the
order of 1054 and our computing software can only handle numbers as small as
1CT45. For completeness we state that the significance of this signal is < 1045.
To measure the upper limit for the /j(2220) we also do multiple x2 fits
for different fixed signal values, while keeping its mass and width constant at
mfj(2220) = 2.234 GeV/c2 and ^(2220) = 17 MeV/c2 as in [17]. The resulting
probability plot is shown in Figure 45.
We find clear evidence for the /2(1270), evidence for the /o(980) and weak
evidence for the /4(2050). We also put a 90% confidence level upper limit on
fj (2220) production. Fit results are shown in Table 41.
Table 41: Results for T(15) â€”> 77r+7r~. The branching fractions of /2(1270) and
/4(2050) are taken from the PDG [16]. Errors shown are satistical only.
Mode
Area
B.F.(IO5)
Significance
7/o(980), /o(980) â€”> 7r+7r
340ÃS
1 fi+u'2
7/2(1270)
1230 Â± 100
10.4 Â±0.8
< lO"45 (> 14a)
7/4(2050)
85 Â±30
3.6 Â±1.3
5.2 x 10~3 (2.6a)
7//(2220),/j(2220) â€”> 7r+7r
< 13
< 6.2 x IQ2

4.3 Angular Distribution of The Signal
The helicity angle distributions of 6n and #7 are defined and described in the
N
appendix. In this section we first obtain the helicity angular distributions of the
51
/o(980), /2(980), and /4(2050) and then fit them to the predictions of the helicity
formalism.
In practical terms, obtaining the helicity angle distribution of a particular
resonance from data consists of two steps. First, we choose an invariant mass
interval around the resonance peak to select events from the resonance and obtain
a helicity angular distribution which has both signal and background events.
Second, we subtract the contribution to the helicity angular distribution of the
background events in the chosen mass interval to obtain what we want; the helicity
angle distribution of the resonance.
Choosing the mass interval is not a trivial thing. If its too wide there will
be too much background, and if its to narrow there will not be enough signal.
To choose the optimum mass interval we need to know how much signal and
background we are selecting. Therefore, the two steps described in the previous
paragraph are related. How we deal with this is revealed in the next two sections.
4.3.1 Optimum Mass Interval
As described above, the first step in obtaining the helicity angle distribution
for a resonance is choosing an invariant mass interval to select events from such a
resonance. A standard ir (which corresponds to 1.6tr for a spin 0 BreitWigner)
cut around the mean mass of the resonance can be chosen as a â€œstandardâ€ interval.
We could proceed this way, but in our case because of the large subtractions
involved when obtaining the angular distribution, a considerable increase statistical
significance of each bin in the helicity angular distribution can be achieved by
choosing the mass interval carefully (see the last column of Table 42).
Letâ€™s consider for example the /o(980) angular distribution. We begin with
the T(15) and the T(4S) efficiency corrected distributions. Before the continuum
subtraction each bin in the /o(980) angular distribution has contributions from
the high end mass tail of the p and the low mass tail of the /2(1270). In order to
52
get the final angular distribution, both of these contributions are taken away by
first doing a continuum subtraction using the scale factor a = 0.404, and then
by subtracting the /2(1270) distribution outside the mass interval scaled by an
appropriate scale factor 8 equal to the ratio of /2(1270) inside the mass interval
and outside of it 1 . After all this, the contributions to the bins in the final angular
distribution are,
Nf0(980) â€” ?Vo(980) + Â«p(T(lS)) + Uf2(i270) ~
where TC/o(980)(n/2(i270)) is the number of /o(980)(/2(1270)) obtained by integrating
the fitted spindependent, relativistic BreitWigner function inside the mass
interval, np^[(is))(^p(r(4S))) is the number of pâ€™s inside the mass interval from
continuum events at the T(15)(T(4S)) energy, and n^(1270j is the number of
/2(1270) outside the mass interval being used to subtract the contribution of
/2(1270) to the /o(980) angular distribution. Following this last definition, (3 =
nMl270)/n'f2(1270)
Each of these terms has an associated error. Assuming that An = \fn, that
the efficiency correction has infinite statistics, and ignoring the errors on the
continuum scale factor a, the overall error is,
AN/o(980) = \J Tifo (980) + (Q + 22)ftp(T(4S)) + (2 + (3 + (32)rif2( i270)
We arrive to the conclusion that the mass interval (m â€” Am, m + Am) which
produces the helicity angle distribution with smallest relative binerrors is the one
1 Strictly speaking this is correct only to first order. The /2(1270) distribution
itself has a small contamination from /0(980) and /4(2050). This effect, which we
call crosscontamination, is ignored in Equation 4.1. Later, in section 4.2.2, we will
show how to eliminate this small crosscontamination using all the resonances. The
difference in how we actually get the helicity distributions and Equation 4.1 has an
insignificant effect when calculating the optimal mass interval
53
y
that maximizes,
Ff0(9so){m, Am)
n/o(980)
\A/o(980) + (<* + Â«2)Â«p(T(4S)) + (2 + (5 + /?2)n/2(i270)
(4.2)
The plot of F/O(980)(m, Am) is shown in figure 46.
This same technique can be applied to the /2(1270) and /4(2050) resonances.
Results are shown in Table 42.
Table 42: Mean masses, widths in GeV/c2 and inverse of the average bin relative
error (F) from background subtractions for the angular distribution of different
resonances. Standard mean masses and widths, corresponding to IT, are taken
from the fit in Figure 43, and are labeled with the subscript â€œsâ€, while those that
maximize F are labeled with the subscript â€œmâ€. The last column shows the factor
by which the effective statistics increase.
Resonance
(ms, Ams)
F(ms, Ams)
(Tnm, Amm)
F(mm, Amm)
F'Â¿(mm,Amm)
F2(ms,Arris)
/o(980)
(0.970,0.070)
5.7
(0.985,0.060)
6.0
1.11
/2(1270)
(1.270,0.120)
18.3 .
(1.590,0.420)
20.6
1.27
/4(2050)
(2.120,0.220)
3.2
(2.240,0.250)
3.4
1.13
4.3.2 Background Subtraction
The continuum subtraction of the helicity angle distribution is defined in
Section 2.1. The subtraction of the tails from other resonances requires a closer
look. Let us call the continuum subtracted helicity angle distribution of the events
in the /o(980), /2(1270), and /4(2050) mass intervals c/o(980), C/2(mo), and c/4(2050)
respectively. Let us call the helicity angle distribution of the events that come
exclusively from the resonance we are trying to select in the same mass intervals
(that is, the true helicity angle distribution of the resonance) Â¿/O(980), Â¿/2(i270)> and
2050) respectively. The continuum subtracted helicity angle distribution of a
54
resonance is being contaminated by the tails of the other resonances. Keeping this
in mind we write 2
c/o(980) = i/o(980) + fit f2(1270)
< C/2( 1270) â€” Ã/o(1270) + 7Ã/o(980) + 5tfA(2050)
C/4(2050) = Ã/O(2050) + ^/2( 1270)
Where the small numbers /?, 7, <5, and e are the ratios of the number of events
from a resonance in the mass interval where the contamination is taking place to
the number of events from the same resonance in the mass interval used to select
it. Using the (mm, Amm) values in Table 42 and the fit in Figure 43 we obtain
/3 == 8.5 x 1CT3, 7 = 6.0 x 1(T2, 5 = 9.7 x 1(T2, and e = 0.13.
To obtain the background subtracted helicity angular distributions we simply
invert the system of equations expressed in Equation 4.3. The solution can be
conveniently expressed as,
i/2(1270) = (C/o( 1270)  7^/o(980) ~ Â¿C/4(2050))
Â£/o(980) = C/0(980) ~ /%a(1270)
(4.4)
^/4(2050) = C/0(2050) â€” ei/2( 1270)
As a check, if we ignore the second order terms the previous solution becomes,
2 Here the cross contamination between the /o(980) and the /4(2050) is ignored.
There is no mathematical problem in including these contamination terms, but the
system of equations would be more complicated than it needs to be since this type
of cross contamination is negligible.
4
i y ' .
* t â–
55
Ã/2(1270) = C/â€ž(1270) ~ 7C/o(980) â€œ Â¿C/4(2050)
<
Ã/4(2050) â€” C/0(2050) ~ ec/2(l270)
which is indeed is the solution when crosscontamination is ignored.
4.3.3 Statistical Fit of the Helicity Angular Distribution: J AssignÂ¬
ments and their Probability Distribution
For each resonance, we fit the data to the simultaneous cos 81 and cos 9n
helicity angle distribution obtained from data using Equation 4.4 to the helicity
formalism prediction, Equations 2327. projected on each angle and folded in
opposite directions around their symmetry axis in order to show both distributions
on the same plot. The corresponding J value of the best fit for each resonance,
shown in Figures 48 through 410, is defined as the J assignment (Ju). We obtain
Ja = 1 for the /o(980) which is inconsistent with the known spin of the /o(980)
which is J â€” 0. For the /2(1270) and the /4(2050) we obtain Ja â€” 2 and Ja = 4
respectively, which is consistent with their known spins.
To have an idea of how well the angular distribution determines J among the
hypotheses J = 0, 1, 2, 3, 4, we do a statistical fit for each hypothesis and assign
each one a probability proportional to e_^x2+do ^^2 where d.o.f. are the degrees
of freedom in the fit. The resulting normalized probability distributions give an
idea of the assigned J significance and are shown in Figure 411. In particular,
this figure shows that the Ja â€” 1 for the /o(980) inconsistency can be due to a
statistical fluctuation.
56
RÃe: â€¢/bohr/usen/luis/deo3/ntp/data_50.ntp
ID
IDB
Symb
Date/Time
Area
Mean
R.M.S
1
1
1
040807/0140
9112
0.8080
0.2895
1
2
1
040807/0140
2.1195E+04
0.7647
0.2072
98765
981
1
040807/0203
569.1
1.227
0.7650
Mass (GeV/c2)
Figure 42: Invariant mass of 7r+7r for IS (top) and 4S (bottom) data. For the 4S
data we show a blowup of the mass region 1.5 â€” 3 GeV/c2 where the p* can be
seen.
y â€¢
i
i
/'* *
MINUIT x2 Fit to Plot 9&3 '
UlS>Gamma h h. masspipi axis
File: Generated internally
Plot Area Total/Fit 1544.1 /1542.1
Func Area Total/Fit 1359.7 /1359.7
22JUL2004 11:51
Fit Status 3
E.D.M. 8.843E06
x2= 44.7 for 54 8d.o.fâ€ž
Errors
Function 1: f_0
Parabolic
Minos
C.L.= 52.7%
MOD
435.70
t
165.4
 167.0
+
177.3
MEAN
0.96672
t
1.5714E02
1.2374E02
1.8160E02
WIDTH
7.05219E02
Â±
3.2912E 02
4.3455E02
+ â–
4.4879E02
* SPIN
O.OOOOOE+OO
Â±
O.OOOOErOO
 O.OOOOE+OO
+
O.OOOOE+OO
* THRESHOLD
Function 2:1 2
028000
Â±
0 00006+00
 O.OOOOE+OO
+
O.OOOOE+OO
MOD
998.59
Â±
77.96
 7812
+
77.92
MEAN
1.2678
Â±
7 59056 03
 7.1319E03
+
8.0978E03
WIDTH
0.12353
Â±
1.4273E02
 1.3459602
1 5382E02
* SPIN
2.0000
Â±
0 OOOOEâ€™OO
 O.OOOOE+OO
+
O.OOOOE+OO
* THRESHOLD
Function 3: f 4
028000
Â±
O.OOOOE+OO
 0 0000E+00
+
O.OOOOE+OO
MOD
104.37
+
38.84
 38.76
+
38.91
MEAN
2.1191
Â±
5.4515E02
5.8893E02
+
4 9096E02
* WIDTH
0.22200
Â±
0 0000E t00
 0 OOOOE+OO
+
O.OOOOE+OO
* SPIN
4 0000
Â±
O.OOOOE+OO
 O.OOOOE+OO
+
O.OOOOE+OO
* THRESHOLD
0.28000
Â±
0 OOOOE+OO
O.OOOOE+OO
+
O.OOOOE+OO
Figure 43: Continuum subtracted (as defined in Section 3.2.1) invariant mass of
7r+7râ€œ from T(15l) â€”â™¦ 77r+7r~.
Events / ( 50.0 MeV/c )
f
58
1.
File: Generated internally
ID
ID8 Syrr.b
Qete/Ti'rie
Area
Mean
R.M.S.
9
3 32
040806/1839
1544.
1.408
0.4996
98765
981 1
040806/1845
1360.
1.416
0.4800
98765
982 2
040806/18*5
1371.
1.422
0.4836
1.5 2.0 2.5 3.0
Mass (GeV/c2)
Figure 44: Blow up of the invariant mass of 7r+7r in the /i(2050) mass region,
including the upper limit on the fj(2220) shown as a dashed line.
Probability distribution
59
File: Amp_mnt/boMuser1.1uis/deo3/mnJit/u1sgammahh/temp.tmp
ID
IDB Symb
Date/Time
Area
Mean
R.M.S.
1
29 1
040806/1845
1.000
6.294
5008
1
25 1
040806/1845
0 9001
5.132
3.650
0.080
0.060
0.040
0.020
0.000
0
25
50
Area
Figure 45: Normalized probability distribution for different fj(2220)
nal areas. The shaded area spans 90% of the probability.
7T+7T' sig
60
>u tz_j iimimiiii .Mb i l i i , i j l=
1990 2090 2190 2290 2390
Mean Mass (MeV/c )
Figure 46: Contour plot of the inverse of the average relative binerror from backÂ¬
ground subtractions in the /â€ž(980) (top), /2(1270) (middle), and /4(2050) (bottom)
angular distribution.
J>
61
MINUIT x2 Fit to Plot 9&82
U1S>Gamma h h. ABS(COS(PHOTHFTA'i) axis
File: Generated internally
Plot Area Total/Fit 606.18 / 606.18
Func Area Total/Fit 436.31 / 436.31
x2= 25.6 for 20  1 d.o.f.,
Errors Parabolic
Function 1: JO
AREA 218.15 + 42.27 .
5AUG2004 23:06
Fit Status 3
E.D.M. 6.301 E13
C.L.= 14.1%
Minos
42.27 + 42.27
IcosB^I cos6n
Figure 47: Angular distribution for the excess events in the /o(980) mass region.
The fit corresponds to J = 0.
dN/( 0.1 dcos0re)
62
MINUIT x2 Fit to Plot 9&82
U1S>Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 606.13/606.18
Func Area Total/Fit 478.89 / 478.89
x2= 15.2for 20 2d.o.f.,
Errors Parabolic
Function 1: J1
AREA 239.45 Â± 39.31
Â© 1.32483E03 Â± 0 3852
5AUG2004 23:06
Fit Status 3
E.D.M. 1.115E05
C.L.= 65.1%
Minos
39.33 + 39.29
O.OOOOE+OO + 0.3537
cos0Tl cos0n
Figure 48: Angular distribution for the excess events in the /o(980) mass region.
The fit corresponds to J = 1.
dN/( 0.1 dcos0J)
63
MINUIT x2 Fit to Plot 9&72
U1S>Gamma h h. A3S(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 1673.4 /1673.4
Func Area Total/Fit 1733.8 /1733.8
X2= 27.3 for 20  3 d.o.f.,
Errors Parabolic
Function 1: J2
AREA
0
866.97 Â± 43.10
4.25267E04 Â± 7.7843E02
0.29838 Â± 0.1098
5AUG2004 23:06
Fit Status 3
E.D.M 9.701 E06
C.L.= 5.4%
Minos
45.85 + 46.07
O.OOOOE+OO + 0.1238
0.1478 + 9.9862E02
1.0
0.5 0.0 0.5 1.0
cos6y IcosOJ
Figure 49: Angular distribution for the excess events in the /2(1270) mass region.
The fit corresponds to J = 2.
dN/( 0.1 dcos0J)
64
MINUiT %2 Fit to Plot 9&92
U1SÂ»Gamma h h. ABS(COS(PHOTHErA)) axis
File: Generated internally
Plot Area Total/Fit 159 93 /159.93
Func Area Total/Fit 127.08 /127.08
Xz= 19.6 for 20 3d.o.f.,
Errors Parabolic
Function 1: J4
AREA
0
<Â¡>
63.528 Â± 1.408
1.5697 Â± 5.8174E02
0.16919 Â± 0.6887
5AUG2004 23:06
Fit Status 3
E.D.M. 5.300E07
C.L.= 29.3%
Minos
14.91 + 0.0000E+00
0.3398 + 0.0000E+00
O.OOOOE+OO + O.OOOOE+OO
1.0 0.5 0.0 0.5 1.0
IcosG^I IcosGJ
Figure 410: Angular distribution for the excess events in the /4(2250) mass region.
The fit corresponds to J = 4.
dN/( 0.1 dcos6J)
65
<*
Figure 411: Ja probability distribution for resonances in the /o(980) (top),
/2(1270) (middle), and /4(2050) (bottom) invariant mass region when only the
hypotheses Ja = 0, 1, 2, 3, 4 are considered.
CHAPTER 5
EXCLUSIVE RADIATIVE DECAY T(15) â–º 7K+K~
In figure 51 the K+K ~ invariant mass plot is shown for both resonance and
continuum running. This figure also has an inset showing the
tracks for events in the 1.1  3 GeV/c2 region. This inset indicates that most
of the events in this mass region have indeed two kaons and that the amount of
p reflection is small. Furthermore, when the 7r+7r_ invariant mass is plotted for
these events only 40 out of 700 can be fit under a p peak (this fit is what we use to
calculate the pion faking kaon rate in Table 36).
Figure 52 shows our fit to the K+ K~ invariant mass continuum subtracted
plot as defined is Section 3.2.1 with the most likely statistical fit overlayed (which
is described in the next section). The number of events near the region (1.01 â€”
1.03 GeV/c2) left after the continuum subtraction is 50 Â± 70.
5.1 Statistical Fit of the Invariant Mass Distribution: Signal Areas and
Their Significance
The results of this section are summarized in Table 51.
From the measurement of the previous section we expect a small contribution
(Â« 50 events assuming no interference) from /2(1270) â€”> K+K~. We do find
some evidence for /2(1270) â€”> K+K~ events (110 Â± 40) in the fit. We also find
strong evidence for the resonance /2(1525) and weak evidence for the /o(1710)
resonance. The /2(1270), and /o(1710) are fitted with their widths fixed to their
PDG values[16] because they have large errors if allowed to float. The rest of
the resonances parameters are consistent with their PDG values[16], which are
"V'(1525) = 1525 Â± 5 MeV/c2, r/'(1525) = 76 Â± 10 MeV/c2, m/o(mo) = 1715 Â±
66
67
6 MeV/c2, and T^mo) = 125 Â± 10 MeV/c2. We also observe an excess of events in
the 2 â€” 3 GeV/c2 region which we canâ€™t attribute to any known resonances.
Significances of the signals of the identified resonances in the fit are calculated
as described in the previous section. The Y(IS') â€”â–º 7/j(2220), /j(2220) â€”> K+K~
upper limit is also calculated using multiple fits, except that this time the events
under the fj{2220) are of unknown origin, so we use a first order polynomial
allowed to float. The significance of the excess of events in the 2 â€” 3 GeV/c2
invariant mass region is calculated assuming a normal distribution; we simply add
up the number of events in each bin along with its error. Results are shown in
Table 51.
Table 51: Results for T(IS') â€”> ^K+K~
Mode
Area
Branching Fraction (10 5)
Significance
7/2(1270)
7/2(1525)
110 Â±40
23 Â±8
5.4 x 10 4(3.3(j)
360/7Â°
39 ij?
< 1045(> 14cr)
7/o(1710), /o(1710) â€”> K+K~
75 Â±30
0.35 Â±0.14
7.5 x 10~4(3.2cr)
^K+K~(2 â€” 3 GeV/c2)
220 Â± 20
1.03 Â±0.12
8.8u
7/7(2220), fj(2220)*K+K
< 10
< 5 x 10~2

5.2 Angular Distribution of The Signal
In this section we adapt the ideas presented in Section 3.4.3 to the T(IS') â€”*
7K+K~ situation.
The derived statistical errors from the signal and background subtractions can
be used to calculate the mass interval which best represents the helicity angular
distribution. The inverse of the expected average relative bin error as a function of
the mass interval is shown in Figure 55, and the mass interval that maximizes it
are tabulated in Table 52.
The tails from the resonances contribute to the continuum subtracted helicity
distributions,
68
Table 52: Mean masses, widths inmass and inverse of the average bin relative
error (F) from background subtractions for the angular distribution of different
resonances. Standard mean masses and widths, corresponding to IF, are taken
from the fit in Figure 52, and are labeled with the subscript â€œsâ€, while those that
maximize F axe labeled with the subscript â€œmâ€. The last column shows the factor
by which the effective statistics increase.
Resonance
(ms, A ms)
F(ms, Ams)
(mm, A mm)
F{mm,Amm)
F2(ms,Arris)
/2(1270)
(1.276,0.185)
3.5
(1.300,0.100)
4.1
1.37
/i(1525)
(1.540,0.085)
9.6
(1.565,0.100)
9.8
1.04
/o(1710)
(1.760,0.125)
3.1
(1.780,0.095)
3.3
1.13
C/a (1270) = 1270) + /3t/'( 1525)
< C/'(]525) = Â£/'(1525) + 7^/2(1270) + <%>(1710)
C/0(1710) = Â¿/0(1710) + eÂ£/'( 1525)
Where again the small numbers 8, 7, 5, and e are the ratios of the number
of events from a resonance in the mass interval where the contamination is taking
place to the number of events from the same resonance in the mass interval used
to select it. Using the (mm, Amm) values in Table 52 and the fit in Figure 52 we
obtain /3 = 1.3 x 102, 7 = 0.14, 8 = 0.11, and e = 0.19.
The background subtracted helicity angular distributions are,
Ã/'(1525) = (C/2(1525) ~ 7C/2(1270) â€œ Â¿C/o(m0))
i Â¿/2(1270) = c/2(1270) ~ /%Â£( 1525) (^2)
^/o(1710) = c/o(1710) ~ 1525)
The the best fit for each resonance and the excess of events in the 2â€”3 GeV/c2
are shown in Figures 56 through 59. The best spin assignment for the /2(1270) is
Ja â€” 2, for the /2(1525) it is Ja = 2, for the it is /o(1710), and it is Ja = 1 for the
excess of events in the 2 â€” 3 GeV/c2 mass region. The Ja = 2 value for the /o(1710)
69
is inconsistent with its known spin. Also, examination of the normalized helicity
amplitudes for the /2(1270) reveals that they are inconsistent with those obtained
for the /2(1270) in the 7r+7T_ mode.
The assigned J probability distributions are shown in Figure 511. They reveal
that the inconsistencies in the /2(1270) and /o(1710) are not significant and can be
attributed to the statistical uncertainty.
File: 7bohr/user1 /Iuis/deo3/ntp/data_50.ntp
ID IDB Symb Date/Time
1 1 1 040807/0219
1 2 1 040807/0219
1 10 1 040807/0235
Area
Mean
R.MS.
1274.
1.320
0.4283
2031.
1.156
0.2585
705.0
0.2246
1.208
0.1480
1.170
Figure 51: Invariant mass of K+K~ for IS (top) and 4S (bottom) data. For
the 4S data the inset shows the (JdE/dxi.^) fÂ°r both tracks for events in the
1.1 â€” 3 GeV/c2 mass region. This inset indicates that most of the events are conÂ¬
stant with having two kaons.
Events / ( 25.0 MeV/c )
70
MINUIT x Fit to Plot
U1S>Gamma h h. masskk axis
9&3
File: Generated internally
7AUG2004 14:23
Plot Area Total/Fit
794.78/575.02
Fit Status 3
Func Area Total/Fit
494.50/429.38
E.D.M. 1.627E06
y2= 37.5 for 40 
7 d.o.f.
Â»
C.L.= 26.9%
Errors
Parabolic
Minos
Function 1: f_2(1270)
AREA
197.37
Â±
64.92
 64.93
+
64.91
MEAN
1.2757
Â±
3.3189E02
 8.9556E02
+
3.0045E02
* WIDTH
0.18510
Â±
0 OOO0E+0O
 O.OOOOE+OO
+
O.OOOOE+OO
* SPIN
2.0000
Â±
0.0000E+00
 O.OOOOE+OO
+
O.OOOOE+OO
* THRESHOLD
0.98000
Â±
0 0000E+00
 O.OOOOE+OO
+
O.OOOOE+OO
Function 2:f_2'(1510)
MOD
318.76
Â±
64.61
 59.53
+
72.90
MEAN
1.5384
Â±
7 8867E03
 7.5447E03
+
8.8799E03
WIDTH
8.54208E02
Â±
2.7982E02
 2.2981 E02
+
3.6400E02
* SPIN
2.0000
Â±
O.OOOOE+OO
 O.OOOOE+OO
+
O.OOOOE+OO
* THRESHOLD
0.98800
+
0 OOOOE+OO
 O.OOOOE+OO
+
O.OOOOE+OO
Function 3: f_0(1710)
MOD
96.117
Â±
38.27
 38.70
+
37.77
MEAN
1.7622
Â±
2.2476E02
 2.0985E02
+
2.6333E02
* WIDTH
0.12500
Â±
O.OOOOE+OO
 O.OOOOE+OO
+
O.OOOOE+OO
* SPIN
O.OOOOOE+OO
Â±
O.OOOOE+OO
 O.OOOOE+OO
+
O.OOOOE+OO
Â»THRESHOLD
0.98800
Â±
O.OOOOE+OO
 O.OOOOE+OO
+
O.OOOOE+OO
0.9 1.4 1.9 2.4 2.9
Mass (GeV/c2)
Figure 52: Continuum subtracted invariant mass of K+K .
Events / ( 25.0 MeV/c'1)
71
File: Generated internally
ID
IDB
Symb
Date/Time
Area
Mean
R.M.S.
9
3
32
040807/1422
794.8
1.739
0.4861
98765
981
1
040807/1426
308.2
2.086
0.5922
98765
982
2
040807/1426
318.3
2.091
0.5835
1.80 2.00 2.20 2.40 2.60
Mass (GeV/c2)
Figure 5â€”3: Blow up of the /j(2220) region, with the 90% CL upper limit overÂ¬
laid. The mass and width are taken to be m/j(2220) = 2.234 GeV/c2 and
F/j(2220) = 17 MeV/c2 as in [17].
Probability distribution
72
File: Amp_mnt/bohr/user1/luis/deo3/mnJit/u1sgammahh/temp.tmp
ID IDB Symb Date/Time Area Mean
1 29 1 040807/1426 1.000 4.746
1 25 1 040807/1426 0.9054 3.971
fi.M.S.
3.582
2.723
0.20
0.15
0.10
0.05
0.00
l I L.
50
Figure 54: Normalized probability distribution for different fj(2220)
signal areas. The shaded area spans 90% of the probability.
K+K
150
100
50
1700
1750 1800 1850 1900
2
Mean Mass (MeV/c )
Figure 55: Contour plot of the inverse of the average relative binerror from backÂ¬
ground subtractions in the /2(1270) (top), /2(1525) (middle), and /o(1710) (botÂ¬
tom) angular distributions.
tf
74
â™¦
MI NU IT %2 Fit to Plot 9&B2
U1S>Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 517.49/517.49
Func Area Total/Fit 129 01 /129.01
X2= 17.8 for 20 3d.o.fâ€ž
Errors Parabolic
Function 1:J2
AREA 64.505 Â± 17.78
0 7.95385E03 Â± 1.347
5 AUG2004 23:16
Fit Status 3
E.D.M. 5.475E05
C.L. 40.4%
Minos
17.78 + 17.78
O.OOOOE+OO + 0.5745
0.2689 + Q.OOOOEtOO
1.0
0.5 0.0 0.5 1.0
IcosG^J cos0n
Figure 5 6: Background subtracted K+K angular distribution in the /2(1270)
mass regionas defined in the text. The fit corresponds to J = 2.
dN/( 0.1 dlcosBJ)
75
MINUIT x2 Fit to Plot 9&72
U1 S>Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 284.44 / 284.44
Func Area Total/Fit 367.87 / 367.87
x2= 20.9 for 20  3 d.o.f.,
Errors Parabolic
Function 1:J2
AREA 183.94 + 18.70
0 6.76443E04 Â± 0.1885
5AUG2004 23:16
Fit Status 3
E.D.M. 3.771 E06
C.L.= 23.3%
Minos
19.29 + 19.25
O.OOOOE+OO + 0.2757
0.1593 + 0.1248
1.0 0.5 0.0 0.5 1.0
ICOS0.J ICOS0J
Figure 57: Background subtracted K* K~ angular distribution in the /2(1525)
mass region as defined in the text. The fit corresponds to J = 2.
dN/( 0.1 dcos0rt)
76
MINUIT x2 Fit to Plot 9&92
U1S>Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 129.47 /129.47
Func Area Total/Fit 79.031 / 79.031
5AUG2004 23:16
Fit Status 3
E.D.M 1 721E05
%'= 28.0 for 20  3 d.o.f.,
Errors
Function 1: J2
Parabolic
AREA 39.516 Â± 10.80
9 2.99860E04 Â± 0.3847
<}> 1.5696 Â± 0.2633
C.L.â€” 4.5%
Minos
10.81 ^ 10.80
O.OOOOE+OO + 0.4645
0.3060 + O.OOOOE+OO
1.0
0.5 0.0 0.5 1.0
IcosQ^J IcosOJ
Figure 58: Background subtracted K+K angular distribution in the /o(1710)
mass region as defined in the text. The fit corresponds to J â€” 2.
dN/( 0. f djcosBJ )
MINUIT x2 Fit to Plot 9&102
U1S>Gammah h. ABS(CCS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 411.73 / 411.73
Fane Area Total/Fit 370.86 / 370.86
%2= 18.9 for 20 2d.o.f.,
Errors Parabolic
Function 1: J1
AREA 185.43 Jt 17.19
0 015542 + 0.3003
5AUG2004 23:16
Fit Status 3
E.O.M. 4.735E09
C.L.= 40.0%
Minos
 17.19 + 17.18
â€¢ O.OOOOEtOO + 0.1921
1.0 0.5 0.0 0.5 1.0
IcosQ^J IcosBJ
Figure 59: K+K~ angular distribution for events within the 2 â€” 3 GeV/c
( TqsooIp 10 )/NP
78
MINUIT x2 Fit to Plot 9&92
U1S~>Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 129.47 /129.47
Func Area Total/Fit 62.194 / 62.194
y2= 33.3 for 201 d.o.f.,
Errors Parabolic
Function 1: JO
5AUG2004 23:16
Fit Status 3
E.D.M. 3.484E15
C.L.= 2.2%
Minos
1.0 0.5 0.0 0.5 1.0
Icose^l cos0rt
Figure 510: Background subtracted K+K~ angular distribution in the /o(1710)
mass region as defined in the text. The fit corresponds to J = 0.
dN/( 0.1 dcos0TC)
79
83
â– C
O
Cm
Cm
I
0 2
4
J
a
Figure 511: Ja probability distribution for resonances in the /%{1270)(top),
/2(1525) (middletop), /o(1710) (middlebottom) invariant mass region and the
excess of events in the 2 â€” 3 GeV/c2 (bottom) region when only the hypotheses Ja
â€” 0, 1, 2, 3, 4 are considered.
CHAPTER 6
EXCLUSIVE RADIATIVE DECAY T(1S) > 7pp
In figure 61 the pp invariant mass plot is shown for both IS and 4S data, with
an inset showing that most of the events indeed have a proton and an antiproton
and that the p and
the 4S plot corresponds to the process e+e_ â€”â–º 7 J/ip with J /ip â€”> pp. This
enhancement is not as pronounced in the IS plot because the IS data only has
32.4% of the luminosity of the 4S data. The number of events in the J/ip invariant
mass region after continuum subtraction is 6 Â± 3 and is consistent with 0.
Figure 62 shows the continuum subtracted invariant pp mass distribution
(as defined in Section 3.2.1) with a 90% confidence level upper limit for fj{2220)
overlaid. A direct fit to the fj(2220) yields 12 Â± 5 events.
There is an excess of events in the continuum subtracted invariant mass plot.
We measure this excess and the upper limit of e+e_ â€”> 7/j(2220), fj(2220) â€”â–º pp
the same way we measured the excess of events inside 2 GeV/c2 < m(K+K~) <
3 GeV/c2 region and the fj(2220) upper limit in Section 5.1. Results are shown in
Table 61.
The pp angular distribution for the mass range 2 GeV/c2 < m(pp) < 3 GeV/c2
is shown in Figure 64 with the most likely J assignment, Ja â€” 1, fit overlayed.
The probability distribution for Ja is shown in Figure 65.
Table 61: Results for Y(IS') â€”+ 7pp.
Mode
Area
B.F. or 90 % U.L. (10~5)
Significance
7pp (2 â€” 3 GeV/c2)
85 Â± 18
0.41 Â± 0.08
4.85cr
7/7(2220), /j(2220) â€”â™¦ pp
<20
< 9 x 102

80
Events / ( 25.0 MeV/c )
81
File: 7bohr/user1 /Iuis/cleo3/ntp/data_50.ntp
ID
IDB
Symb
1
Date/Time
Area
Mean
R.M.S.
1
1
040807/0303
104.0
2.334
0.3628
1
2
1
040807/0303
141.0
2.187
0.3419
1
10
1
040807/0304
139.0
0.2583
0.1224
1.259
1.119
1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2
Mass (GeV/c2)
Figure 61: Invariant mass of pp for IS (top) and 4S (bottom) running. For the 4S
running the inset is consistent with the events having a proton and an antiproton.
82
File: Generated internally
ID
IDB
Symb
Date/Time
Area
Mean
R.M.S.
9
3
32
040807/1446
85.43
2.521
0.3093
98765
981
2
040807/1446
72.17
2.685
0.3291
98765
982
1
040807/1446
55.91
2.813
0.2553
Mass (GeV/c2)
Figure 62: Invariant mass of pp. The plot is continuum subtracted and efficiency
corrected. An overlay with the 90% confidence level upper limit for /j(2220)
is shown. The mass and width are taken to be 771^(2220) â€” 2.234 GeV/c2 and
T/j(2220) = 17 MeV/c2 as in [17].
83
RÃe: Amp_mnt/bohr/user1/luis/deo3/mn Jit/u1 sgammahh/temp.tmp
ID
IDB
Symb
Date/Time
Area
Mean
R.M.S.
1
29
1
040807/1446
1.000
12.78
5.430
1
25
1
040807/1446
0.9110
11.80
4.590
S3
O
3
r2
H
43
XA
'S
C3
2
0,
0.060
0.040
0.020
0.000
Area
Figure 63: Normalized probability distribution for different /j(2220)
areas. The shaded area spans 90% of the probability.
pp signal
84
MINUIT x2 Fit to Plot 9&22
U1S>Gamma h h. ABS(COS(PHOTHETA)) axis
File: Generated internally
Plot Area Total/Fit 256.69/256.69
Func Area Total/Fit 134.57 /134.57
X2= 19.5 for 20  2 d.o.f.,
Errors Parabolic
Function 1:J1
AREA 67.287 Â± 10.91
0 1.20611E03 Â± 0.4148
2AUG2004 18:12
Fit Status 3
E.D.M. 1.845E05
C.L.= 36.4%
Minos
10.93 + 10.88
O.OOOOE+OO + 0.3544
1.0 0.5 0.0 0.5 1.0
IcosG^j cos6n
Figure 64: pp angular distribution for the excess of events in the mass range
2 GeV/c2< m(pp) < 3 GeV/c2.
(lVÂ°oP ro )/NP
85
Figure 65: Ja probability distribution for the excess of events in the mass range
2 GeV/c2< rn(pp) < 3 GeV/c2 when only the hypotheses Ja â€” 0, 1, 2, 3, 4 are
considered.
CHAPTER 7
SYSTEMATIC UNCERTAINTIES
Systematic uncertainties are any sources of experimental uncertainty other
than the statistical ones. Limits on the accuracy of our detector simulation and any
physical processes that interferes with the experimental measurement are typical
examples of systematic uncertainties. These uncertainties need to be identified,
quantified, and if possible, corrected.
7.1 Cuts
From now on we report individual cut efficiencies relative to the events that
survive all the other cuts.
According to our MC, most of the skim cuts, except for the hardGam reÂ¬
quirement, axe nearly 100% efficient. Therefore, such cuts should not be a source
of systematic uncertainty. The only skim cut worth taking a closer look at is the
hardGam cut, which is about 93% efficient in MC. Measuring this efficiency in our
data is not possible because events not classified as hardGam are not in the data
to begin with. Closer examination of our MC reveals that that most of the 7%
inefficiency in hardGam comes from the eOverPl cut (4%), some from the Sh2 cut
(2%), and the rest (1%) from the other cuts present in hardGam. We can measure
the efficiency of eOverPl in data by looking at the eOverP2 distribution of data
tracks from p and
Tables 71 and 72), and we wonâ€™t measure how well MC models the rest of the
hardGam cuts, which are 97% efficient in MC.
86
87
To quantify the quality in the MC modeling of our analysis cuts in the 7r
and K modes we use the p and 0 signals present in our data1 . In this case the
process is straightforward. We measure the p and
over a floating background function with all cuts in place, and with all in place
cuts except the one under consideration. Prom these numbers we calculate the
effective efficiency of the cut. The differences between data and MC are taken as
the systematic errors which are added in quadrature. Results axe shown in Tables.
71 72. The 4rnomentum cut does not appear because its efficiency is close to
100%.
Cuts
MC eff.
Data eff.
Systematic Error
eOverP2
96.2
96.6
0.4
QED e+e~ â€”â–º ye+eâ€ suppresion
93.1
94.2
1.2
QED pPpT â€”> 7p+p~ suppresion
75.3
77.7
3.2
Hadron separation
96.9
96.8
0.1
Overall analysis cut systematic error
Â±3
MC, efficiencies from
Table 71: Efficiencies for T(15) â€”> 'yn+n~ for flat signa
data (p), and the derived systematic error in %. Efficiencies are reported as the
number of signal events after all cuts divided by the number of signal events with
all cuts except the one under consideration. Statistical! errors axe 0.1% or less.
For the proton case we donâ€™t have a clean sample with high statistics of
e+e_ â€”> 7pp events in data. By extension we take the systematic error in this mode
to be 10%.
7.1.1 Justification of the DPTHMU Cut
The reason we prefer to use DPTHMU < 5, instead of the more traditional
(see [18]) DPTHMU < 3 used in CLEO II, is that, for some unknown reason, our
CLEO III MC has too many 7r tracks with 3 < DPTHMU < 5.
1 A study using Kg or A signals from hadronic environments would have larger
statistics, but is problematic because the large number of tracks and showers artifiÂ¬
cially decrease the cut efficiency. See [28] Appendix A.3 for an example.
88
Cuts
MC Eff.
Data Eff.
Systematic Error
eOverP2
98.5
100
1.5
QED e+e~ â€”â–º qe+e suppresion
98.1
99.3
1.3
QED p+p~ â€”> 7P+/r suppresion
93.5
96.4
3.1
Hadron separation
88.4
82.2
7.0
Overall analysis cut systematic error
Â±8
Table 72: Efficiencies for T(lS') â€”â– > 7K+K~ for flat signal MC, efficiencies from
data (
number of signal events after all cuts divided by the number of signal events with
all cuts except the one under consideration. Statistical! errors are 0.1% or less.
To observe this fact we first select a relatively clean sample of pion tracks
by requiring the event to have a n+n~ invariant mass consistent with the p mass,
and to pass all our analysis cuts except for the cut on D PTH MU on one track.
For such events we plot the rate as a function of momentum, at which the 7r track
whose DPTHMU cut we released has D PTH MU > 3 and DPT H MU > 5. To
increase our statistics we do this procedure twice, once for each track, and average
the fake rate. Figure 71 shows the results. Clearly the CLEO III MC we are using
has some problem modeling the DPTHMU < 3 cut.
To keep the systematic error low we choose a cut at DPTHMU < 5
(.DPTHMU < 3 gives a systematic error of about 20%). This does not change the
efficiency in data very much, but it increases the efficiency rQEfted by MC, bringÂ¬
ing it closer to reality. The increase in p fakes after loosening the cut is estimated
to be low using QED MC (see Figure 72).
7.2 Angular Distribution of Signal
The photon and tracks from the process T(IS') â€”â–º 7X with X â€”â–º h+h~ have
a different angular distribution than that of flat MC. Examples of possible angular
distributions are shown in the appendix.
89
File: â€¢/bohr/user1/luis/deo3/ntp/data_50.ntp
ID
IDB
Symb
Date/Time
Area
Mean
R.M.S.
1
4
31
040525/1408
0.2350
2.758
0.8547
1
3
31
040525/1408
0.4317
2.684
0.8088
1
6
31
040525/1408
9.6769E02
2.996
0.7400
1
5
31
040525/1408
6.0516E02
2.792
0.6995
0 1.6 3.2
Track Momenta (GeV)
Figure 71: Pion faking muon fake rates for a cut on D PTH MU < 3 (top) and
D PTH MU < 5 (bottom). MC is shown as solid circles while p from data is shown
as hollow circles. Fake rates are reported relative to all events that pass all cuts,
except for the D PTH MU cut for one of the tracks.
Figure 73 shows the efficiency in flat MC as a function of cos 61 and
cos dh+'1 . Note that the K and p modes are nearly insensitive to the track anÂ¬
gular distribution, while the tt mode is more sensitive to 6h+. This happens because
of the stronger muon rejection cut in the 7r mode.
We measure the systematic effects of flat MC efficiency by convoluting each
plot in Figure 73 with different possible angular distributions calculated in the
2 6L and 6h+ are the helicity angles of the sequential decay, defined in the apÂ¬
pendix (see Figure 2).
90
Rle:'/bohr/user1/luis/deo3/ntp/mcpass2_mumu_gamma_new.ntp
ID IDB Symb Date/Time Area Mean R.M.S.
1 3 31 040518/1402 3.676 2.361 0.9088
1 30 32 040518/1402 3.022 2.549 0.8590
ID
4â€”*
c3
cs
pH
Track Momenta (GeV)
Figure 72: Muon faking pion fake rate (in %) for D PTH MU < 3 (solid squares)
and D PTH MU < 5 (solid circles). There is a complete overlap in the last bin.
appendix. Tables 7374 show the necessary correction factors relative to flat MC
efficiency for decays with definite 7 and X helicities A7 and A^, due to the nonflat
photon and hadron distributions. We call these factors eAxAx (7) and ey)Ax (h)
respectively.
The efficiency of a decay with definite A7, A^ can be obtained using the flat
MC efficiency corrected by a factor eA*Ax â€” â‚¬x^xx (7) x eAxAx(h). In general, the
final state is a mixture of all possible A7; \x pairs, and the efficiency correction
factor is,
eJx = cos2 0 cos2 Te/o + sin2 0qf + cos2 0 sin2 .
(7.1)
91
o
tÃ
o
o
Ui
ttÃ
>>
o
tÃ
O
*5
E
tu
cos0y
COS0JJ
Figure 73: Flat MC efficiency as a function of cosO7 (left column) and as a funcÂ¬
tion of cos6h+ (right column). Top row corresponds to the pion mode, middle row
to the kaon mode and bottom row to the proton mode.
The fits in Figures 48 and 410 and 5â€”65â€”8 measure the pair (0, ). These
values are summarized in Table 75, where they are used to obtain eJx for each
mode.
The pair (0.4>) carry an error which is a source of systematic uncertainty.
We calculate this systematic uncertainty by inspecting the differences in efficiency
when (0.4>) move away from the value which gives the minimum chisquared, xLn>
92
Mode
aio or ai2
0 = 0
II Â§
S5=i
7T
0.910
1.18
K
0.925
1.15
P
0.922
1.16
Table 73: Â£\x\x (7), eÃÃ±cienciy correction with respect to flat MC factors due to
the nonflat angular distribution of the photon in Y(15) â€”â–º 7X for different X spin
values (Jx) and 7, X helicities (A7, Ax)
Mode
Jx
Qio
0 = 0, $ = 0
an
0 = i
w 2
G12
0 = 0, $ = f
0
1.00


1
0.785
1.11

X â€”> 7T+7r_
2
0.829
0.934
1.15
3
0.898
0.853
1.03
4
0.899
0.863
0.922
0
1.00


1
0.944
1.03

X > K+K
2
0.950
0.988
1.04
3
0.957
0.971
1.01
4
0.965
0.962
0.992
Table 74: ex*Xx(h), eÃÃ±cienciy correction factors with respect to nonflat MC due
to the track angular distribution in T(lS') â€”> 7X, with X â€”> h+h~ for different X
spin values (Jx) and 7, X helicities (A7, A*).
under the condition x2 < Xmin + 1 3 â€¢ For Jx > 1 both (Â©, $) are free to move,
defining a surface in the (0,
the /2(1270), /4(2050), and /Ã¡(1525).
At this point we can check whether (0, ) depend on the mass of the deÂ¬
cay. We split the /2(1270) and the /2(1525) into a high mass and a low mass
region. The plots of the error surfaces of the measured (0, ) show no significant
separation for the different mass regions (see Figures 75 and 76).
3 In a two dimensional linear problem such a set of points defines the surface of
the standard error ellipse.
93
Results for the correction factor and its systematic error axe shown in Table 7
5.
Upper limits on Y(1S) â€”* 7/2(2220), ^(2220) â€”> h+h~ are changed to
include the angular distributionâ€™s effect on efficiency. Since we canâ€™t measure
the helicity amplitudes in this case, we choose the worst possible case where the
corrected efficiency is lowest. This always corresponds to 0 = $ = 0. Results are in
Table 76.
0 (radians)
Figure 74: Surfaces in the (Â©,
tainty in the efficiency correction factor for the.modes with /2(1270), /i(2050),
and /2(1525). The /i(2050) surface may seem large, but when drawn in spherical
coordinates it is a small â€œnorth pole capâ€.
(radians)
94
Figure 75: Measurer. (0, $) surfaces for the /2(1270) high and low mass regions.
(radians)
95
Figure 76: Measurer (0, ) surfaces for the f2(1525) high and low mass regions.
Table 75: Measured (0, 4>), calculated eJx, efficiency correction factor interval when (0, 4>) move away from their measured
value under the condition that \2 stay within one unit of its minimum value, and the systematic uncertainty on the correcÂ¬
tion factor.
Mode
Jx
0
â‚¬Jx
Interval
Sys. Err. (%)
7/o(980)
7/0 (980)
7/2(1270)
7/4(2050)
7/^(1525)
7/o(1710)
ryK+K~, 2 GeV/c2 < m(K+K") < 3 GeV/c2
7pp, 2 GeV/c2 < m(pp) < 3 GeV/c2
0
1
2
4
2
0
1
1
0
o.oÂ±8;g
O.OO/oioo
1 C7+0.00
10'0.34
0.00/Â°;2'
0
016Â±8;ii
900/o!oo
0
0
0.30/â€; 10
0.2 Â±0.7
OSOÃaÃe
0
0
0
0.91
0.71
0.78
1.02
0.90
0.93
0.88
0.92
0.91  0.91
0.71  0.80
0.76  0.80
1.00  1.02
0.89  0.91
0.930.93
0.87  0.91
0.92  0.95
0
+ 13
Â±3
2
Â±1
0
+3
1
+3
Table 76: Worst .case efficiency correction factors and their effect on upper limits of Y(l5) â€”Â» 7/j(2220), fj(2220) â€”Â» h+h
decays.
Mode
Worst case correction
Upper limit increase (%)
fj{2220) > 7T+7T
0.75
+33
fj(2220)  K+K
0.88
+14
/H2220)  pp
0.92
+9
97
7.3 Different Hadronic Fake Rates Between IS and 4S
Differences in the fake rates of hadronic events between the 15 and 45
data are another potential source of systematic errors when doing a continuum
subtraction.
In principle this is a second order effect since the contamination is proportional
to the difference of two small numbers. However since we have such a large number
of p and 4> events in the underlying IS continuum, we should quantify any possible
signal contamination. Section 3.2.2 outlines the three cases we need to worry
about; p events contaminating the K+K~ invariant mass plot, p contaminating the
pp invariant mass plot, and 0 events contaminating the pp invariant mass plot.
It is reasonable to assume that the IS and 4S fake rates stay within 50%
of each other (see Section 3.3.2 for some examples from MC measurements).
With this assumption, taking np Â« 20000 and n$ ~ 2000 in the IS data, and
the fake rates measured in data (see Table 36) we can calculate the systematic
uncertainties. We expect Â±27 events from p to contaminate the K+K~ invariant
mass plot. However, these events can be ignored as source of systemtic uncertnty
because 90% of them fall in the 1 â€” 1.5 GeV/c2 region, below the 7^(1525) peak.
We also expect Â±7 from p and Â±3 events from 0 contaminating the pp invariant
mass plot. This represents a Â±8% systematic error for the T(15) â€”> 'ypp mode.
7.4 Other Systematic Sources
Besides the systematic uncertainties described above, we also add a 2%
systematic effect from track finding (1% per track) and 5% from the number of
T(15).
We find no evidence for nonresonant hadron pairs that could interfere with
the resonances. There is a possible interference between /2(1275) and /2(1525)
because they both are in the J = 2 state. From the ^(1275) measurement in the
7r+7r_ mode we expect 3 events from this source in the /2(1525) mass region 1.45 â€”
98
1.6 GeV/c2. In the most extreme cases we have, N oc 7ly'(1525)2 + I IAf2(i270) 112 ^
2^4(1525) 1111^4/2(1270) 11> where Ay'i,1525j and Af^mo) are the amplitudes associated
with /2(1525) and /2(1270) respectively. From the data we have N Â« 300, and
using the fact that 1/2(i270) 112 contributes with 3 events, the contribution from
P/; '(1525) 112 can range from 400 to 200 events. This represents a 30% variation from
the central value. Since this is the extreme interference case, we take half of this
value, 15%, as a reasonable systematic uncertainty from possible interference in the
Â¿(1525).
The possible decay T(IS') â€”> pir could contaminate our 7r+ 7r~ invariant mass
plot, since a 7rÂ° with a momentum greater than 4 GeV/c2 looks very much like a
photon. Using phase space MG and isospin symmetry, we estimate that about 40%
of possible Y(IS') â€”> pit have such a 7TÂ°. The latest measured upper limit [24]
for this decay mode is 1.9 x 10 "6. This gives an upper limit of 16 events in the
efficiency corrected ir+ n~ invariant mass plot, of which 10 events fall in the p mass
region, behind the /o(980) mass region. The remaining potential 6 events could
contaminate the rest of the invariant mass region. None of the signal regions is
affected by more than 1%.
To test the robustness of the statistical fits, we redo them with all masses and
widths of possible resonances required to stay within 1
values [16]. The results from fitting the data in this way are within the statistical
errors when compared to the fit results summarized in Table 41 and 51. We do
not observe a systematic difference between these two fitting techniques.
The branching ratios for /2(1270) â€”> 7r+7r~, /4(2050) â€”â–º 7T+7T, and /Ã¡(1525) â€”>
K+K~ are taken from the PDG [16] and contribute to the systematic uncertainty
with jj, Â±9, and Â±3 percent respectively.
99
7.5 Overall Systematic Uncertainties
We combine the systematic errors of each section in quadrature. Tables 77
78 summarize the systematic errors.
Table 77: Systematic errors expressed as % from the different sources described
in the text. Not shown in the table, but included in the total, are a systematic
uncertainty of 5% from the number of Y(1S), a 2% systematic uncertanty from
MC tracking in all the modes, a 15% systematic uncertainty in the 7/2(1525) from
possible interference with 7/2 (1270), an 8% systematic uncertainty in the 7pp
mode from possible hadronic contamination, and the systematic uncertanties in the
/2(1270), /4(2050), and ^(1525) hadronic branching fractions.
Mode
Analysis
Cuts
Angular
Distribution
Total
7/o(980)
Â±3
0
Â±6
7/2(1270)
Â±3
Â±3
Â±7
7/4(2050)
Â±3
+2
Â±7
7/5(1525)
Â±8
Â±1
Â±20
7/o(1710)
Â±8
0
Â±10
7if+A'(23 GeV/c2)
Â±8
+1
3
Â±10
7pp(2 â€” 3 GeV/c2)
Â±10
3
Â±14
Table 78: Increase in the upper limits of T(lS') â€”â–º 7/j(2220) for different fj(2220)
decay modes due to the possible systematic effects described in the text added in
quadrature. Not shown in the table, but included in the total, are a 5% contribuÂ¬
tion from the number of Y(IS') and of 2% contribution from MC tracking.
Mode
Analysis
Cuts
Angular
Distribution
Total
fj(2220) + 7T+7T
Â±3%
Â±33%
Â±34%
/j(2220) â€”> K+K~
Â±8%
Â±14%
Â±17%
fj(2220)  pp
Â±10%
Â±9%
Â±14%
CHAPTER 8
RESULTS AND CONCLUSION
We report on a new search for twobody radiative T(1S) decays. We place
stringent limits on the production of the /j(2220) particle in the pion and kaon
modes, and a less stringent limit in the proton mode where some excess of events is
observed in the region of interest.
In the decay channel T(IS') â€”â–º 77r+7r_ we find clear evidence for the resonance
/2(1270) and measure a branching fraction of (13.3 Â± 1.4) x 10~5, which is
consistent with the earlier CLEO measurement of (7.4/^) x 1(U5 [18]. The angular
distributions of the photon and tracks strongly indicate that the hadron pairs we
assign to the /2(1270) are indeed in a J = 2 state, and that nature prefers to
produce the /2(1270) with 0 helicity. In contrast, for J/ij) â€”> 7/2(1270) it was
found [29] that the /2(1270) is produced at equal rates with both helicity 0 and
helicity 1, and at a rate consistent with 0 for helicity 2, but more recently [30]
measured helicity 0 dominance for this same mode.
There is a barely significant (4.3a) excess of events in the /o(980) â€”> 7r+7r~
invariant mass region. The angular distributions of the photon and tracks indicate
that this excess of events is in a J = 1 state, rather than J = 0 which would be the
case if the excess were due to /o(980) decays. It may be that this excess is due to
nonresonant â€¢y7c+n~ production. We conclude that more data is needed to resolve
this situation.
We find weak evidence for the production of the resonance /4(2050) in the
T(1S) 4 77T+7r~ decay channel.
In the decay channel T(15l) â€”> 7K+K~ we find strong evidence for the
production of the resonance /[(1525) and weak evidence for the production of
100
101
/o(1710). The photon and track angular distributions show that the two tracks
attributed to the (1525) are indeed in a J = 2 state and that the /2(1525) is
produced mostly with helicity 0.
There is also evidence of an excess of events in the 2 â€” 3 GeV/c2 K+K~
invariant mass region, which we cannot attribute to any known resonances.
Finally, we also find evidence of an excess of events in the channel Y(IS') â€”>
7pp, which we cannot attribute to any known resonances. We find no evidence
of an enhancement near pp threshold. Such an enhancement was recently reÂ¬
ported by [31] in the J/t/j system, and is currently being interpreted (See
e.g. [32], [33], [34]).
When comparing the radiative T(1S) decays studied in this analysis to raÂ¬
diative J/i/> decays we observe a suppression ratio of 0.09 that is in reasonable
agreement with with the naive expectation of 0.04 obtained from scaling arguÂ¬
ments.
Table 81 summarizes our results.
While we find hadronic resonances at a level consistent with the estimate
of Section 1.2, we do not observe any glueball states. This is not the result we
would expect from a naive interpretation of QCD. as the two gluons from a T(1S)
radiative decay can form a bound state directly and need to go through at least
two strong interaction vertexes to form a hadron.
Internal consistency in QCD predicts the existence of glueballs. Even though
theoretical calculations of the glueball mass spectrum exist, a clear glueball
observation has not been made yet. We conclude that more glueball searches, such
as those proposed in the GlueX project, are necessary to clarify this situation.
Table 81: Final measured branching ratios, measured branching ratios reported relative to J/tp branching ratios, and staÂ¬
tistical significance for each decay channel. The measured branching ratios have been corrected by the factors calculated in
Appendix A2. For the branching ratios the first uncertainty is statistical and the second is systematic.
Channel
Branching Fraction x(10 5)
Y(1 S)
J/+
Significance
7/o(980), /0(980) >
7T+7T
i8io:?Â±o.i

4.3(7
7/2(1270)
13.3 Â±1.0 Â±0.9
0.09 Â± 0.01
> 14 a
7/4(2050)
3.5 Â± 1.3 Â±0.2
0.019 Â± 0.006
2.6cr
7/2(1525)
4.3/Â¿ g Â± 0.9
0.09ÃÂ£$
> 14(7
7/o(1710), /o(1710)
K+K~
0.38 Â±0.16 Â±0.04
0.007/^005
3.2a
1k+k~
1.14 Â± 0.14 Â±0.11

9.1(7
7PP
0.41 Â±0.08 Â±0.10
0.011 Â±0.005
4.8(7
7//(2220), fj(2220) 
Â» 7r+7T
< 0.08


7/j(2220), fj(2220) >
K+K
< 0.06


7//(2220), fj(2220)
PP
< 0.11


APPENDIX
HELICITY FORMALISM FOR TWO BODY DECAYS
Helicity formalism can be used to obtain the angular distribution of a decay
process. In this appendix we restrict ourselves to the case of two body decays.
The situation is as follows: particle A with spin J4, and zaxis spin projection
decays to two particles, A â€”â™¦ B + C with definite helicity A# and Ac. Let the
direction of the of the decay be n(6, ) = n(0s, 4>b) = Ps â€” â€”Pc (see Figure 1).
The amplitude for this decay is,
AZ% = < 0^b\c\H\JaMa > . (1)
We have little chance of directly calculating 1. But we can exploit the fact that
H conserves angular momentum by considering the basis {< jmAsAcI}, where j
is the total angular momentum of B + C and in its projection along the z axis1 .
Inserting this basis into Equation 1,
c ~ < ^â€˜Mb'Mj^iAbAc >< jm\B\c\H\ JaMa >
jm
â€” X/ < ^^bAcIj'^AbAc > 5jjASmMAA\Bxc
jm
= < 9(})\b\c\JaM\\b\c >
1 The rotational invariance of helicities is a necessary condition to construct
{< jmXB\c\} as a basis for the twoparticle state B + C. Rotational invariance
is one of two key ideas in helicity formalism. The other key idea, which is boost
invariance, is>useful in sequential decays.
103
104
The above equation contains two factorized terms. The first term is indepenÂ¬
dent of the interaction H and contains the angular distribution information we are
interested in. The second term. A\g\c. is usually called the â€œhelicity coupling amÂ¬
plitudeâ€ . It is independent of the angular distribution, and contains the physics of
the decay (in particular it depends on Ag and Ac, but not on Ma). In the helicity
formalism, the helicity coupling amplitudes are unknown parameters.
For strong and electromagnetic decays there is another invariance of H we
can exploit; parity. For these kind of interactions it can be shown by inserting the
parity operator that,
^AsAc = (~ 1) â€œ Jc nAr]B r]CAxB \c (2)
where tia^bVc is the product of the parity of particles A, B and C. This useful
property of A\B\c reduces the number of unknown parameters in the angular
distribution of strong and electromagnetic decays.
The next step is to derive explicit formulae for < 9(Â¡)XbXc\JaMaAbAc > for
the B + C system. Such derivation, which is out of the scope of this appendix, will
a. J 
105
be omitted here2 . The result can be expressed in terms of the Wigner functions,
< 64,\b\c\JaMa\b\c >= ÃMt) (3)
where X = Xb â€” Xc and,
Â». 4>) = W
The functions dJ^A x(0) have real values, and some of them are given in the particle
data book.
We have finally arrived to the following important relation,
All helicity distribution calculations are reduced to handling these basic ampliÂ¬
tudes. Here are some of their very important properties,
4^=0 if Ab  Ac > Ja,
(6)
A'"v 11=11 <%
(7)
/â– 27T
Jo
dd>AMA A*Ma â€” 2ttS('Ma~M^AMa A*Ma
a(P'A\BXc'Axâ€™x'  Z7r0(XX') ^XbXc^x'x'
B^Câ€™
(8)
/
dnAâ€œ\â€žXf?, = ^ H!
XbXc^x' x'c (AA')
(9)
2 The interested reader can find this derivation in [35]
Property 6 is a reflection of the physical fact that the projection of the total
angular momentum along a particular axis cannot be greater than the total angular
momentum itself.
 A^Xr 2 is the angular distribution of a final state where MA and the
helicities of B and C are known (see Equation 1). However, it is usually the case
that Ma^b^c are not known. In such cases we can only measure the overall
angular distribution, which is simply the average over possible initial states of the
incoherent sum of final states,
dP_
dÃœB
Ma\b^c
d
dCln
E Ãvj ii2
Ma^b^c
(10)
where Pma is the weight of the initial state Ma and dBts â€” d(cos(0B))d
Next, consider a sequential decay, A â€”> 13 4 C with C * D + E (see Figure
2). The rotational and boost invariance of helicities makes it easy to extend the
helicity formalism to this case. To do so we consider a new reference frame, Oâ€™,
at rest with C, such that z' â€” n{6B, 4>b)Z â– Ail particle helicities remain the same
in Oâ€™, and in particular MZ{C)  Ac (the negative sign arises from the fact
that n(#B, (j)B) is the direction in which B is traveling). It is now straightforward
to calculate the intermediate amplitude at Oâ€™. Of course is still
calculated at A's original rest frame. Since Ac is an intermediate state, the helicity
amplitude for the final state Xb\d^e is,
ACMa
EaMa nâ€”\C
Ãa
(11)
3 This condition does not mxlikely determine Oâ€™. We can be more precise by
defining O' â€” L(pc)R(B, #b> 0)0, where L is the Lorenz Boost operator, R is the
rotation operator in'terms of the Euler angles, and O is Aâ€™s original rest frame.
107
Figure 2: A â€”> B +C and C â–º D + E. The drawing is peculiar in that Cs daughÂ¬
ters are drawn as seen in Câ€™s rest frame (backtoback).
where we have coherently summed over the intermediate state C. If Ma^b^d^e
are not known.
dP
E
Ma^b^d^b
P
Mi iiladik,
= E II
MaXbXd^e
(12)
The sequential decay amplitudes have the following important properties,
ac~Ma
XbXdXe
1=11
(13)
p2n r2n
/ #Â«/ #D^VJ2=4*2Ell
Jo Jo >c
This kind of logic can be extended to a general twobody decay chain.
Example: T(1 S) > 7/2(1270)(/2(1270) > tt+ttâ€™)
From our previous discussion we can immediately write (see Equation 11),
^Afr(is) _ V"' ^Atr(is) r~xM1270)
^r\r + \â€” ZJ \+\
A/a(1270)
108
In this case, A* = A^+ = 0, A7 = Â±1. In CLEO Mx(is) = Â±1, there is no
Mr (is) â€” 0 since at the T(l,9) energy electrons couple only to positrons of the
opposite helicity, this way â€” P_i = ~ and Pq â€” 0 is an excellent approximation..
There are therefore four amplitudes we must consider, AClm. AC^, AC l 100 and
4Â£ioo
The overall angular distribution is (see Equation 12),
dP = \ II AC
dÂ£l1dÂ£llt+
loo IP +7
2 II
n i
I2 +2 II 4Â£ioo II2 +2 II ACtfjQ Â¡I"
(15)
=11 ^Cloo Â¡Â¡2 +  AC1
100
where we have used property 13. The two relevant amplitudes are,
a:1
100
â€” ^io^oo T A\iCfu( Al^Cm
n''oo
l12'~'00
*^ioo ~ ^io^oo + â€œ^Aii^oo + v41_1_2CÂ§0
where we have used the fact that A7 â€” A/2(i270) I < Jr(iS) = 1 to sum over all
possible A/2(i270)
All that remains is to substitute,
^loo ll2~â€˜ II ^Ão^oo I!2 T ll ^n^oo1 II" + II A\2C0q Â¡1â€œ +
{^â€¢Ão^oo^lÃboo 1 + ^10^00^12^00 2 + Ail^oo1 ^12^00 + cc}
II AC1 II2
II HUioo II "
A] C
0
10''00
+ II AlA
00
+ l AL^Cqo Â¡i? +
{410^00â€¢'^ll^'OO + ^lO^OO*^1! 2^'00 + ^â– 11^'00â– ^12^'00 + ccâ€¢}
(16)
into Equation 15.
Doing so will give us the full angular distribution of the decay chain. However,
using property 8, all the cross terms inside {...} vanish if we integrate over (pn+ or
7. We will return to the
109
47T2[( A\0 !2 + II ALW f) II cÂ°0 II2 +
(IMli II2+ Â¡1 4111II2) II Cll2 +
(ÃI 412 l2 + II Â«4Ã.12 l2) I! ^00 lH
= fill Al0 2 Cm II2 (dln(ej2 + Â¿Ã_1(Â«1)2)dSo(Â«,+)2 +
II An 2 Coo 2 (Â¿Â¡â€ž((
II a12 2 Coo II2 (Â¿U(Â«,)2 + <Â¡}_1(ft,)2)4)(Mil]
= ~ II Coo II2 [II Aio II2 i(l + cos2 #.,)(? cos2 0Â»+  l)2 +
II An II2 sin2 sin 0â€ž+ cos0x+)2 +
II Ai2 !2 i(l + cos2 0,)( sin2Â«Â»+)].
At this point it is convenient to define the normalized helicity amplitudes,
Aio
â€œI0 = v4l ATolF+ifln II2 + II Alo II2
__ 4n
â€œâ€œ " '7prFTri7p+ii a12 ip
4x2
â€œ'2 â€œ /liATo II2 + II An II2 + II A ,2 II2
terms inside {...} later.
[2* Â¡in dp
which satisfy,
Â¡I aio 2 + II an 2 + Â¡I ai2 2= 1. (17)
The normalized probability distribution, Â¿^g^cose +). can be obtained by
dividing the probability distribution by the branching fraction of the sequential
decay,
110
P = 2  Cqo 2 (II A 10 2 +  An 2 +  Ayi 2)
= II Coo 2 A10 2 + II Coo 2 An 2 +  Coo 2 A\i Â¡2 +
II Coo 2 <4io II2 +  Coo 2 Ann 2 +  Cqo 2l A_i_2 II"
This way,
dU
d(cos07)d(cos07r+)
aio 2 (1 + cos207)(3cos29t+  l)2 +
O O
an [2 7 sin2 61^(sm0n+ cos^t)2 +
aw 2 ^(1 + eos2 07)~sin40,r+
It is useful to project the angular distribution on 07 and 0V+,
dU~ = aw II2 ^(3cos207r+â€”1)2+  an 2 ^(sin^ cos07r+)2+  an 2 ^sin40ff+
d cos 9.
8
16
(18)
JcosFâ€ž = ^ a"â€˜ ^ + H â€œ1! (+'1 + cos2 II 011 II2 i S1IIâ€˜ F (19)
Let us now return to the terms inside {...} in Equationl6. It turns out that for
these terms a substantial simplification occurs if we integrate over cos 9K+. After
some algebra, we arrive at the following expression,
dfa  (t>n+) = 2^(1 â€œ 7{aioa*2 + aiÂ°Gl2} coÂ®[2(^7 ~ (2Â°)
Example: T(L9) Â» 7/4(2050)(/4(2050) + tt+tT)
For the case /4(2050) the calculation is very similar to the /Â¿(1270) case
discussed above. The allowed helicities for /4(2050) are the same as the allowed
helicities for /2(1270). This makes the angular distribution calculation very similar
I
111
to the /2(1270) case, the only difference being that the functions ^a/220270)o are
replaced by di/4(a050,o '
The photonâ€™s angular distribution comes out to be the same. The result for
the pionâ€™s normalized angular distribution is.
=  aio 2 (35 cos4 0*+  30 cos2 6n+ + 3)2 +
45
II an IP 'rx(sin07T+ cos07r+)2(7cos20n+ â€” 3)2 f
45
II 0,12 II2 rj sin4 0w+(7 cos2 0*+  l)2.
04
(21)
Fit of the Helicity Angular Distribution Obtained from Data
In practice, a particular bin in the helicity angular distribution obtained from
data has bins with dN â€” NdU events, where N is the total number of events
(which is allowed to float during the fit), and dU is the portion of the normalized
probability assigned to the bin by the helicity formalism prediction.
To incorporate Equation 17 directly into the fit we make the following change
of variables,
it flio â€” cos0cos$;  ai2 jâ€” cosÂ©sinÃÂ»;  an = sinÂ©. (22)
Because of this definition, both 0 and i> should be required to be within the
interval (0, ) during the fit. The new variables have the convenient property that
for J 0, i> = 0 â€” 0, and for J â€” 1, Tâ€”0. In general there are 1, 2 and 3 degrees
of freedom for the fits with J â€” 0, J â€” 1, and J > 1 respectively. One degree
of freedom always corresponds to the number of events, N. and the other two
112
correspond to 0 and T, which determine the magnitude of the normalized helicity
amplitudes through Equation 22 4 .
When fitting, it is preferable to choose the full 2D helicity angular distribution
which is a function of 0f and 6W+ instead of its one dimensional projections. The
2D choice is better not only because the 2D distribution carries the maximum
amount of information, but because in general the efficiency depends on both the
pion and photon helicity angles, which can lead to complications when efficiency
correcting the onedimensional projections. Building on the two previous examples
the 2D helicity angular distributions for J = 0, 1, 2, 3, 4 are (for compactness we
define x = cos 6n+ and y â€” cos 07),
dNJ=o
dxdy
= Wj=o(l + y2)
(23)
dNJ=1
dxdy
= NJ=i(cos2 0 5(1 + y2)^x2 +
sin2 0 5(1y2)5(lx2))
(24)
Obtaining the error of the normalized helicity amplitudes requires some matrix
Ã flio
multiplication. For compactness let us define a â€” O12 ) and u
\ant
matrix for the normalized helicity amplitudes is,
'0N
. The error
< 5'dSa1 >â€” < 8Ã¼SÃ¼j >
Ouj
:..T ^
da/
dCo
where < 8Â¿oÃ³ujr > is the error matrix obtained from the fit.
113
*
= NJ=2( cos2 0cos2 $ ~(1 1 y2)(3x2  1)2+
dxdy 8 8
sin20 "(1  y2)â€”i2(l  x2)+ (25)
cos2(Wsf(l+y2))f(l x2)2)
8 ' lo
Nj=3(cos2 0cos2 $ ?(1 + y2)^(5x3  3x)2+
sin2 0 5(i _ y2)â€˜^{ 1  x2)(5x2  1)2+ (26)
cos2 0sin2 $ (1 + y2)^.T2(l  x2)2)
8 lo
= NJt=4( cos2 Â© cos2 5(1 + r)7^(35^  30x2 + 3)2+
axa?/ 8 Lz8
sin2 Â© 5(i _ j/2)~x2( 1  x2)(7x2  3)2+ (27)
t: O Z
cos2 Â©sin2 5(1 + y2)^5(i _ x2)2(7x2 â€” l)2).
In this analysis, because of our low statistics (and by low statistics we mean
that there axe numerous bins with < 1 events in the 2D plots), we project EquaÂ¬
tions 2327 on x = cos 6^, and y = cos 07. We also fold each distribution around its
symmetry axis x = 0 and y = 0 to increase each bin's statistics. Finally, we do a
simultaneous fit to both distributions and obtain N, 0, and $.
Admittedly, this procedure has the problems described previously when 2D fits
where advocated, which can give rise to systematic effects in the measured helicity
amplitudes. In essence, we need to use a unitary weight function proportional
to the efficiency distribution (see Figure 73) in the integrals used to make the
projections. This affects the relative strength of each helicity amplitude and the
overall efficiency, but does not change each projected shape. However, the helicity
amplitudes themselves are only used to calculate systematic effects on the efficiency
dNj^
dxdy
114
of resonances, so by simultaneously fitting the unweighted ID projections we are
just ignoring the systematics on the systematics.
Finally, let us consider the 7 â€” 4>n+ distribution for J = 2. Equation 20
becomes,
jrjâ€”â€”r = ~~~(1 Â¡={cos2 0 sin 24Â» cos A} cos[2(d>7 â€” (f>n+)]) (28)
d\(p7 *p7T^') V6
where the new degree of freedom A is the relative phase between aio and 0,20 The
deviation from the flat distribution is characterized by R â€” T?{cos2 0 sin 2T cos A} <
The bigger this factor is, the easier it is to observe in data. Figure 3 shows the
continuum subtracted â€” â€žâ– + distribution for events near the /,2(1270) resonance.
The factor R obtained from fitting the distribution is nonsignificant. It is a bit
tedious but straightforward to calculate the 7 â€” (j)n+ distribution for other J values.
We will not present those results here. They are not very interesting because we
need large statistics to observe deviations from the flat distribution, and there is no
effect on the efficiency from nonfiat â€” n+ distributions.
115
MINUIT x* Fit to Plot 9&3
U1Sâ€”>Gamma h h. pipiphi axis
File: Generated internally
Plot Area Total/Fit 779.39 / 779.39
Func Area Total/Fit 769.52 / 769.52
yz= 5.2 for 10  2 d.o.f.,
Errors Parabolic
Function 1: COMIS Function XMNCMJ
N 769.49 Â± 56.78
R 6.87430E02 t 0.1062
7AUG2004 00.36
Fit Status 3
E D.M. 1 509E06
C.L.= 74.0%
Minos
O.OOOOE+OO + O.OOOOE+OO
O.OOOOE+OO + 0 OOOOE+OO
M*
Figure 3: Continuum subtracted â€” T+ distribution for events near the /2(1270)
resonance with a fit to Equation 28 over fayed.
116
Example: Continuum Events.
Our previous examples tackled radiative decays of the T(1S). Now let us
consider the continuum â€œradiative beamâ€ process e+e" â€”Â» 77r+7r (see Figure 4).
At first glance it might seem that the helicity formalism cannot be applied in this
case. But a closer look at the diagram of this process, shown in Figure 4, indicates
otherwise. The key point is that, regardless of the particular coupling, helicity
formalism can still be applied at the CM of the 7r+7râ€œ vertex, where the initial state
has Jp â€” 1~.
7T+
Figure 4: e+e â€”> 77r+7râ€˜ in continuum. 7r+7r pair production usually occurs
through p â€” 7 coupling.
We only need to know that we are dealing with 1 â€”Â» 0 0 at the 7r+7r
vertex. There is only one helicity amplitude we need to consider,
K>= VS23'Â»'4â„¢ (29)
which after a couple simple steps leads to,
dN . t 3 . a.
â€” = N sm2i4+.
a cos 0n+ 4
This result is also valid for the kaon case.
.. (30)
On the other hand, the photonâ€™s angular distribution cannot be calculated
using twobody helicity formalism. A good approximation is (see [36] Appendix B),
117
dP l + cos2#7 /01,
j T ^ 1 27Tâ€™ b31)
d cos p7 1 â€” cos21/7
which is not valid at low 6 angles (6 â€”> 0, 9 â€”â–º 7r) because of the effects on the
approximation of the neglected electron mass,
V
Figure 5 shows these angular distributions in 4S â€œradiative beamâ€™ events. The
agreement between theory and experiment is excellent. This is reassuring, because
in the main analysis we rely on our angular distribution measurements to get the
correct efficiency in data and to identify the J value of each resonance.
118
Figure 5: Helicity angle distribution for â€œradiative beamâ€ continuum events
e+eâ€œ â€”â–º 7p and e+eâ€œ â€”> 7 in the 4S data. Top left (right) plot shows in gray
the events selected as p â€”> 7r+7r (0 â€”> K+K~). Middle left (right) plot shows
the 7T+7r_ (K+K~) helicity angle distribution for the selected events, and has the
helicity formalism prediction, Equation 30, overlÃ¡id. Bottom left (right) plot shows
the 7 angular distribution for the photon accompanying the p (0), and has the first
order QED prediction, Equation 31, overlayed.
REFERENCES
[1] A. Filippi, â€œGlueballs and Hybrids in the 1GeV to 2GeV Region: An
Experimental Overview,â€ eConf C0309101, FRWP012 (2003).
[2] H. Stock, CLEO internal note CBX 0235 (2002) (unpublished).
[3] F.E. Close and A. Kirk, Eur. Phys. J. C 21, 531543 (2001).
[4] C.A. Meyer, â€œProceedings of the Workshop on Gluonic Excitationsâ€, Newport
News, Virginia 2003, AIP Conf. Proc. 698, 554 (2004).
[5] K.K. Seth, NucÃ. Phys. B (Proc. Suppl.) 96 205 (2001).
[6] BingSong Zou, NucÃ. Phys. A644 (1999) 41c.
[7] C.J. Morningstar and M.J. Peardon, Phys. Rev. D 60 034509 (1999).
[8] A. Vacarino and D. Weingarten, Phys. Rev. D 60 114501 (1999).
[9] C.E Carlson, T.H. Hanson, and C.Peterson, Phys. Rev. D 27 1556 (1983).
[10] N. Isgur and J. PatÃ³n, Phys. Rev. D 31 2910 (1985).
[11] S. Narison, NucÃ. Phys. B 96 244 (2001).
[12] M. Melis, F. Murgia and J. Parisi, Phys. Rev. D 70, 034021 (2004).
[13] S. J. Brodsky, A. S. Goldhaber and J. Lee, Phys. Rev. Lett. 91, 112001
(2003).
[14] D.H. Perkins, â€œIntroduction to High Energy Physicsâ€, 4th Edition, CamÂ¬
bridge University Press,2000, ISBN 0521621968.
[15] D. V. Amelin et al. (VES Collaboration), hepex/9810013 (1998).
[16] S. Eidelman et al. (Particle Data Group Collaboration), Phys. Lett. B 592, 1
(2004).
[17] J.Z. Bai et al., (BES Collaboration), Phys. Rev. Lett. 76 3502 (1996).
[18] A. Anastassov et al., (CLEO Collaboration), Phys. Rev. Lett. 82 286 (1999).
[19] G. Masek et al., (CLEO Collaboration), Phys. Rev. D 65 072002 (2002).
[20] T. Coan, Nucl. Instrum. Meth. A 453, 49 (2000).
119
120
[21] Y. Kubota et al. (CLEO Collaboration), NucÃ. Instrum. Meth. A 320, 66
(1992).
[22] M. Yurko et al. (CLEO Collaboration), NucÃ. Phys. Proc. Suppl. 61B, 212
(1998).
[23] M. Artuso et al. (CLEO Collaboration), NucÃ. Instrurn. Meth. A 502, 91
(2003).
[24] B.K. Heltsley, H. MahlkeKrueger, CLEO internal note CBX 0329 (2003)
(unpublished).
[25] GEANT  Detector Description and Simulation Tool, CERN Program Library
Long Writeup W5013.
[26] P. Avery, CLEO internal note CBX 9837 (1998) (unpublished).
[27] H. Stock, J. M Yelton, CLEO internal note CBX 0317 (2003) (unpublished).
[28] R. Galik, N. Powell, G. Viehhauser, CLEO internal note CBX 0150 (2001)
(unpublished).
[29] R. M. Baltrusaitis et al. (Mark III Collaboration), Phys. Rev. D 35 2077
(1987).
[30] X. Y. Shen et al. (BES Collaboration), â€œJ/iÂ¡) Radiative Decaysâ€ eConf
C020620, THAT07 (2002).
[31] J. Z. Bai et al. (BES Collaboration), Phys. Rev. Lett. 91, 022001 (2003).
[32] C. H. Chang and H. R. Pang, hepph/0407188 (2004).
[33] X. A. Liu, X. Q. Zeng, Y. B. Ding, X. Q. Li, H. Shen and P. N. Shen,
hepph/0406118 (2004).
[34] D. V. Bugg, Phys. Lett. B 598, 8 (2004).
[35] J. D. Richman, â€œAn Experimenterâ€™s Guide to the Ilelicity Formalismâ€,
CALT681148 (1995) (unpublished).
[36] V. Fedeyev, I. Korolkov, D. Besson, R. Galik, H. Paar, CLEO internal note
CBX 9816 (1998) (unpublished).
BIOGRAPHICAL SKETCH
Luis BrevaNewell was born in Boulder, Colorado on December 11, 1975.
When he was two years old his family moved to the Basque Country region of
Spain. He graduated High School from GETXO III and moved to Madrid where he
obtained a 5year degree in Physics from the Universidad Complutense de Madrid.
He was accepted to attend graduate School at the University of Florida in 1999.
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Doctor of Philosophy.
John Yelton
Professor of Physics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Doctor of Philosophy.
David banner
Distinguished Professor of Physics
1 certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree
Professor of Physics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
quality, as a dissertation for the degree of Doctor of Philosophy.
Pierre Sikivie
Professor of Phvsics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adeqK^te, in scope and
quality, as a dissertation for the degree of Doctor of rhv
Charle:
PpfysÃ¡br of Civil and
Coastal Engineering
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and
This dissertation was submitted to the Graduate Faculty of the
Department of Physics in the College of Liberal Arts and Sciences and
to the Graduate School and was accepted as partial fulfillment of the
requirements for the degree of Doctor of Philosophy.
December 2004
Dean, Graduate School
LD
20<2it

