<%BANNER%>

Ab Initio Density Functional Theory for Open-Shell Systems, Excited States and Response Properties

University of Florida Institutional Repository
Permanent Link: http://ufdc.ufl.edu/UFE0021221/00001

Material Information

Title: Ab Initio Density Functional Theory for Open-Shell Systems, Excited States and Response Properties
Physical Description: 1 online resource (135 p.)
Language: english
Creator: Bokhan, Denis
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: density, effective, excited, functional, hyperpolarizabilities, open, optimized, potential, properties, response, shell, states, system, theory
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Ab initio density functional theory (DFT) based on the optimized effective potential (OEP) method is a new approach to study the electronic structure of atomic, molecular and^M solid state systems. It contains elements of both wave function and density functional theories and is free from limitations of conventional DFT because of using orbital-dependent functionals derived from systematic approximations of the wave function theory. Ab initio DFT methods with exchange-correlation functionals based on many-body perturbation theory (MBPT) have been derived and implemented recently. The exchange-correlation potentials derived from MBPT have a complicated structure and their derivation in higher-order of MBPT by the use of the chain-rule for functional differentiation requires significant effort. To facilitate such derivations, I developed a special diagrammatic formalism for taking functional derivatives. An alternative way to construct OEP MBPT exchange-correlation potentials is to use the density condition. It makes it possible to obtain potentials for different partitionings of the full molecular Hamiltonian with MBPT. Using the diagrammatic formalism developed for taking functional derivatives, we show an order-by-order equivalence between the functional derivative and the density condition approaches to OEP MBPT for the case of the Kohn-Sham partitioning of the molecular Hamiltonian. For any other partitionings, different exchange-correlation potentials are produced by the functional derivative and density condition approaches. The time-dependent extension of OEP in the adiabatic approximation with exchange-only potentials and kernels was recently implemented and applied to some molecular systems. The corresponding excitation energies and polarizabilities are in good agreement with time-dependent Hartree-Fock results. However, such results cannot be used for comparison with experimental values due to lack of a description of electron correlation effects. In order to account for correlation effects, a correlation kernel corresponding to the MBPT(2) potentials has been derived. Its structure and properties are described in detail. The first numerical results for excitation energies with pure ab initio potentials and kernels are presented. The recently implemented OEP MBPT(2) method has been extended to the case of spin-polarized open-shell systems. The total energies obtained for several open-shell systems are very close to the corresponding values obtained with the highly-correlated coupled cluster singles and doubles with perturbative triples (CCSD(T)) method. Comparison with results obtained with the OEP MBPT(2) exchange-correlation potentials and some density functionals has shown a qualitatively incorrect shape for some widely used exchange-correlation potentials. Higher-order response properties, such as hyperpolarizabilities, are described very poorly with conventional functionals. The typical relative errors described in the literature are about 100 percent. The reason for the poor description is an incorrect long-range asymptotic behavior and an incomplete cancellation of the coulombic self-interaction. The OEP method is free from those drawbacks and calculated values of static hyperpolarizabilities with the exchange-only potential are close to those derived from Hartree-Fock theory. The second kernel required for calculations of hyperpolarizabilities within the DFT framework has been derived for OEP potentials by using the developed diagrammatic technique. The structure and properties of the second OEP kernel are discussed.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Denis Bokhan.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Bartlett, Rodney J.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021221:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021221/00001

Material Information

Title: Ab Initio Density Functional Theory for Open-Shell Systems, Excited States and Response Properties
Physical Description: 1 online resource (135 p.)
Language: english
Creator: Bokhan, Denis
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: density, effective, excited, functional, hyperpolarizabilities, open, optimized, potential, properties, response, shell, states, system, theory
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Ab initio density functional theory (DFT) based on the optimized effective potential (OEP) method is a new approach to study the electronic structure of atomic, molecular and^M solid state systems. It contains elements of both wave function and density functional theories and is free from limitations of conventional DFT because of using orbital-dependent functionals derived from systematic approximations of the wave function theory. Ab initio DFT methods with exchange-correlation functionals based on many-body perturbation theory (MBPT) have been derived and implemented recently. The exchange-correlation potentials derived from MBPT have a complicated structure and their derivation in higher-order of MBPT by the use of the chain-rule for functional differentiation requires significant effort. To facilitate such derivations, I developed a special diagrammatic formalism for taking functional derivatives. An alternative way to construct OEP MBPT exchange-correlation potentials is to use the density condition. It makes it possible to obtain potentials for different partitionings of the full molecular Hamiltonian with MBPT. Using the diagrammatic formalism developed for taking functional derivatives, we show an order-by-order equivalence between the functional derivative and the density condition approaches to OEP MBPT for the case of the Kohn-Sham partitioning of the molecular Hamiltonian. For any other partitionings, different exchange-correlation potentials are produced by the functional derivative and density condition approaches. The time-dependent extension of OEP in the adiabatic approximation with exchange-only potentials and kernels was recently implemented and applied to some molecular systems. The corresponding excitation energies and polarizabilities are in good agreement with time-dependent Hartree-Fock results. However, such results cannot be used for comparison with experimental values due to lack of a description of electron correlation effects. In order to account for correlation effects, a correlation kernel corresponding to the MBPT(2) potentials has been derived. Its structure and properties are described in detail. The first numerical results for excitation energies with pure ab initio potentials and kernels are presented. The recently implemented OEP MBPT(2) method has been extended to the case of spin-polarized open-shell systems. The total energies obtained for several open-shell systems are very close to the corresponding values obtained with the highly-correlated coupled cluster singles and doubles with perturbative triples (CCSD(T)) method. Comparison with results obtained with the OEP MBPT(2) exchange-correlation potentials and some density functionals has shown a qualitatively incorrect shape for some widely used exchange-correlation potentials. Higher-order response properties, such as hyperpolarizabilities, are described very poorly with conventional functionals. The typical relative errors described in the literature are about 100 percent. The reason for the poor description is an incorrect long-range asymptotic behavior and an incomplete cancellation of the coulombic self-interaction. The OEP method is free from those drawbacks and calculated values of static hyperpolarizabilities with the exchange-only potential are close to those derived from Hartree-Fock theory. The second kernel required for calculations of hyperpolarizabilities within the DFT framework has been derived for OEP potentials by using the developed diagrammatic technique. The structure and properties of the second OEP kernel are discussed.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Denis Bokhan.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Bartlett, Rodney J.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021221:00001


This item has the following downloads:


Full Text





AB INITIO DENSITY FUNCTIONAL THEORY FOR OPEN SHELL SYSTEMS,
EXCITED STATES AND RESPONSE PROPERTIES



















By

DENIS BOKHAN


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

2007






























) 2007 Denis Bokhan









ACKNOWLEDGMENTS

I would like to thank professors Henk Monkhorst and So Hirata for helpful discus-

sions. I want to thank also Dr. Igor Schweigert and Dr. Norbert Flocke for the helping me

to write OEP code. My special thanks to Tatyana and Thomas Albert for help with the

preparation of the dissertation text.









TABLE OF CONTENTS
page

ACKNOW LEDGMENTS ................................. 3

LIST OF TABLES ....................... ............. 7

LIST OF FIGURES .................................... 8

A BSTR A CT . . . . . . . . .. . 9

CHAPTER

1 INTRODUCTION ...................... .......... 12

1.1 Ab Initio Wavefunction-Based Methods .................. 13
1.1.1 Hartree-Fock Method ......... .......... .... 13
1.1.2 Electron-Correlation Methods ....... ............ 16
1.2 Kohn-Sham Density Functional Theory ............. .. .. 19
1.2.1 Time-Dependent Density Functional Theory . . 21
1.2.2 Time-Dependent Density Functional Theory Linear Response Theory 23
1.2.3 Problems with Conventional Functionals . . ..... 26
1.2.4 Orbital-Dependent Functionals ................ .. .. 28
1.2.5 Ab Initio Density Functional Theory ................ .. 29

2 INTERCONNECTION BETWEEN FUNCTIONAL DERIVATIVE AND EF-
FECTIVE OPERATOR APPROACHES TO AB INITIO DENSITY FUNC-
TIONAL THEORY ........... ..... ........... 31

2.1 Equations for the Exchange-Correlation Potential in the Functional Deriva-
tive Approach ........ .. ...... ............... .. .. 31
2.2 Equations for the Exchange-Correlation Potential in an Effective Operator
Approach ................... . . .... 34
2.3 Interconnection in Arbitrary Order ............ ...... 37
2.3.1 Diagrammatic Functional Derivatives . . . .... 37
2.3.2 Diagrammatic Functional Derivatives in Second-Order ,M ,iv-Body
Perturbation Theory .................. ....... .. 38
2.3.3 Interconnection in Higher Orders .................. .. 41
2.3.4 Interconnection in Infinite Order .... . .. 45

3 AB INITIO TIME-DEPENDENT DENSITY FUNCTIONAL THEORY EM-
PLOYING SECOND-ORDER MANY-BODY PERTURBATION OPTIMIZED
EFFECTIVE POTENTIAL .................. .......... .. 47

3.1 Diagrammatic Construction of the Exchange-Correlation Kernels . 48
3.1.1 Formalism ................ .... . . 48
3.1.2 An Example: Diagrammatic Derivation of Exchange-Only Kernel 52










3.2 Kernel for the Second-Order Optimized Effective
Perturbation Theory Correlation Potential .
3.3 Properties of the Correlation Kernel .......
3.4 Numerical Testing .. ..............
3.5 Conclusions . . . . .


Potential Many-Body


4 AB INITIO DENSITY FUNCTIONAL THEORY FOR SPIN-POLARIZED
SYSTE M S .... ................... ..............

4.1 Theory ................... ..............
4.2 Results and Discussion .. .....................
4.2.1 Total Energies . . . . . . .
4.2.2 Ionization Potentials .. ..................
4.2.3 Dissociation Energies .. ..................
4.2.4 Singlet-Triplet Separation in Methylene .............
4.3 C conclusions . . . . . . . .


5 EXACT-EXCHANGE TIME-DEPENDENT DENSITY
ORY FOR OPEN-SHELL SYSTEMST ..........

5.1 Exact-Exchange Density Functional Theory .
5.2 Time-Dependent Optimized Effective Potential ..
5.2.1 Theory and Implementation ........
5.2.2 Numerical Results ..............
5.2.3 C(!i ige-Transfer Excited States ......
5.3 Conclusions . . . . . .

6 EXACT EXCHANGE TIME-DEPENDENT DENSITY
ORY FOR HYPERPOLARIZABILITIES .. .....


FUNCTIONAL THE-
..............


FUNCTIONAL THE-
..............


6.1 T heory . . . . . . . . .
6.1.1 Time-Dependent Density Functional Theory Response Properties
6.1.2 Diagrammatic Derivation of the Second Exact-Exchange Kernel
6.1.3 Properties of the Second Exact-Exchange Kernel .. .......
6.2 Num erical Results . . . . . . . .
6.3 C conclusions . . . . . . . . .


APPENDIX


A INTERPRETATION OF DIAGRAMS OF THE SECOND-ORDER MANY-
BODY PERTURBATION THEORY OPTIMIZED EFFECTIVE POTENTIAL
CORRELATION KERNEL .. ........................

B INTERPRETATION OF DIAGRAMS OF EXACT-EXCHANGE SECOND
K E R N E L . . . . . . . . . .

REFERENCES ......................................









BIOGRAPHICAL SKETCH ................... .......... 135









LIST OF TABLES


Table page

3-1 Orbital energies and zero-order approximations to excitation energies . 64

3-2 Excitation energies of Ne atom using OEP-MBPT(2) Kohn-Sham orbital energies 64

3-3 Excitation energies of Ne atom using exchange-only orbital energies ...... ..66

3-4 Excitation energies of Ne atom using orbital energies and orbitals from OEP2(sc).
All equations for TDDFT are the same .................. ...... 66

4-1 Total energies .................. ................. .. 72

4-2 Ionization potentials (in e. v.) .................. ........ .. 74

4-3 Dissociation energies (in kJ/mol) .................. ....... .. 75

4-4 Singlet and triplet energies of methylene .................. ..... 76

5-1 Total (in a. u.) and orbital (in e. v.) energies of Ne atom . ..... 88

5-2 Total and orbital energies of He atom .................. ...... 88

5-3 Excitation energies (V valence state, R Rydberg) .............. ..93

5-4 Orbital energies (in e. v.) of Ne atom .................. ...... 94

5-5 Ionization energies (in e. v.) .................. ......... .. 94

5-6 Static polarizabilities (in a. u.) .................. ........ .. 99

5-7 Isotropic C6 coefficients (in a. u.) .................. ..... .. 99

6-1 Hyperpolarizabilities of several molecules (in a. u.) ............... ..107









LIST OF FIGURES


Figure page

4-1 Exchange and correlation potentials of Li atom (radial part). A) Exchange po-
tential. B) Correlation potential ............... ... 73

4-2 Exchange and correlation potentials of 02 molecule across the molecular axis.
A) Exchange potential. B) Correlation potential .... . ... 73

4-3 LiH potential energy curve. ............... ......... 77

4-4 OH potential energy curve ............... ......... .. 78

4-5 HF potential energy curve. ............... .......... 79

5-1 Exchange potentials of Ne atom, obtained in different basis sets . ... 88

5-2 A 1II charge-transfered excited state of He ... Be ... . .... 96

5-3 LUMO-HOMO orbital energy difference ................ . 96









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

AB INITIO DENSITY FUNCTIONAL THEORY FOR OPEN SHELL SYSTEMS,
EXCITED STATES AND RESPONSE PROPERTIES

By

Denis Bokhan

August 2007

('C! i: Rodney J. Bartlett
Major: Chemistry

Ab initio density functional theory (DFT) based on the optimized effective potential

(OEP) method is a new approach to study the electronic structure of atomic, molecular

and solid state systems. It contains elements of both wave function and density functional

theories and is free from limitations of conventional DFT because of using orbital-

dependent functionals derived from systematic approximations of the wave function

theory.

Ab initio DFT methods with exchange-correlation functionals based on many -

body perturbation theory (MI IPT) have been derived and implemented recently. The

exchange-correlation potentials derived from MBPT have a complicated structure and

their derivation in higher-order of MBPT by the use of the chain-rule for functional

differentiation requires significant effort. To facilitate such derivations, I developed a

special diagrammatic formalism for taking functional derivatives. An alternative way to

construct OEP MBPT exchange-correlation potentials is to use the density condition.

It makes it possible to obtain potentials for different partitionings of the full molecular

Hamiltonian with MBPT. Using the diagrammatic formalism developed for taking

functional derivatives, we show an order-by-order equivalence between the functional

derivative and the density condition approaches to OEP MBPT for the case of the Kohn-

Sham partitioning of the molecular Hamiltonian. For any other partitionings, different









exchange-correlation potentials are produced by the functional derivative and density

condition approaches.

The time-dependent extension of OEP in the adiabatic approximation with

exchange-only potentials and kernels was recently implemented and applied to some

molecular systems. The corresponding excitation energies and polarizabilities are in good

agreement with time-dependent Hartree-Fock results. However, such results cannot be

used for comparison with experimental values due to lack of a description of electron

correlation effects. In order to account for correlation effects, a correlation kernel corre-

sponding to the MBPT(2) potentials has been derived. Its structure and properties are

described in detail. The first numerical results for excitation energies with pure ab initio

potentials and kernels are presented.

The recently implemented OEP MBPT(2) method has been extended to the case

of spin-polarized open-shell systems. The total energies obtained for several open-shell

systems are very close to the corresponding values obtained with the highly-correlated

coupled cluster singles and doubles with perturbative triples (CCSD(T)) method. Com-

parison with results obtained with the OEP MBPT(2) exchange-correlation potentials and

some density functionals has shown a qualitatively incorrect shape for some widely used

exchange-correlation potentials.

Higher-order response properties, such as hyperpolarizabilities, are described

very poorly with conventional functionals. The typical relative errors described in the

literature are about 100 percent. The reason for the poor description is an incorrect

long-range .,i-mptotic behavior and an incomplete cancellation of the coulombic self-

interaction. The OEP method is free from those drawbacks and calculated values of static

hyperpolarizabilities with the exchange-only potential are close to those derived from

Hartree-Fock theory. The second kernel required for calculations of hyperpolarizabilities

within the DFT framework has been derived for OEP potentials by using the developed









diagrammatic technique. The structure and properties of the second OEP kernel are

discussed.









CHAPTER 1
INTRODUCTION

The definition of a chemical reaction as a transformation of one substance to another

at the molecular level means a regrouping of nuclei and electrons. In the adiabatic

approximation, such nuclear regroupings can be presented as a motion in the field of

some potential, also known as a potential energy -, if',.. Potential energy surfaces can be

obtained by solving the eigenvalue problem with the so called electronic Hamiltonian

1 lec elec Nucl ZA elec 1
2 E Iri RA I ri rj
i=1 i=1 A=1 i
Whole nuclear coordinates are treated as a set of parameters. If there are more than two

particles in the system, the eigenvalue problem


H,i(R...RM)(rl... rN, R1...RM) = E(R ...RM)1(rl... rN, R... RM) (1-2)


can not be solved analytically. For the calculation of electronic energy levels, some

approximations have to be used.

Modern quantum chemistry includes three classes of methods for the solution

of the electronic problem. The first one, called wave function 'I' ,., '; uses different

approximations for the calculation of the wave function '(rl...rN, R1...RM) and the

corresponding energy E(R1...RM). The second class, known as /. ,:-.:/;/ functional theory

uses the density as a primary object. Within the DFT approach, the energy is written as

a functional of density, and the construction of the wave functions is not necessary. The

third class of methods uses a whole density matrix. The density matrix renolmalization

group (DMRG) and the density matrix functinal theory (DMFT) are typical methods of

the third class.

Ab initio DFT contains elements of the first two classes. It uses a local multiplicative

potential, typical for the DFT approach. However, the corresponding exchange-correlation









potentials are derived from orbital-dependent energy functionals, taken from wave function

theory.

The main advantage of wave-function based methods is the possibility to obtain

systematically improvable results. This means that with the extension of the basis set, it

will be possible to obtain more accurate energies and wave functions, and in the complete

set it is possible to get the exact solution of the Schr6dinger equation. On the other

hand, rigorous ab initio methods using wave function methods are usually very costly for

computations on large molecules. Ab initio wave function methods can usually be applied

only for systems with ~20-30 atoms in reasonable basis sets. The main DFT advantage

is a small computational cost; it can handle systems with several hundreds of atoms.

However, most of conventional density functionals do not have the capacity to produce

systematically improvable results.

Ab initio DFT is capable of producing systematically improvable results, but it

is computationally more costly than DFT with conventional functionals. Despite its

computational cost, ab initio DFT can be used as a method for the calibration of density-

dependent functionals. An alternative way of providing such information is Quantum

Monte-Carlo (QMC) method, however, QMC results are not available for molecules and

open-shell systems.

1.1 Ab Initio Wavefunction-Based Methods

1.1.1 Hartree-Fock Method

In the Hartree-Fock method, a wave function is considered as a Slater determinant[1]


0 (ri) ...* ci(rN)
HF= : (1-3)

(PN(r1) ... (PN(rN)

The single-electron wave functions 4p(r) (or orbitals) are determined by the condition

that the corresponding determinant minimizes the expectation value of the electronic









Hamiltonian


EHF = min < THFIHI|HF > (14)

subject to the condition of orthonormality on the orbitals < ,; i; >= 6ij. Substituting

the explicit form of the wave function from Equation (1-3) into an expectation value, it is

possible to write the Hartree-Fock energy in terms of orbitals

elec Nucl elec
EHF i 2 ZA j > 1 >)
A RAI
(1-5)

To derive the Hartree-Fock equations it is convenient to minimize the following

functional

I = EHF : ij < > (1-6)
i,j
where Eij are Lagrange multipliers. Minimization of the functional from Equation (1-6)

requires that the functional derivatives of I with respect to the orbitals vanish

61 61
S0, 0 (17)
6oi 6(p,

Taking into account Equation (1-3) condition (1-7) can be presented as


fP=pi, F(1-8)


The operator f has the structure


f V2 + =Vt+ VH + nl (1-9)









where


elec Nucl

>I: Iri RAI
i A

VH(r) -yC f *(Fi)4)(ri) dr p(ri) dr
) rll /r i-- ^1
elec
p(r) = (T) i (r) (1-10)

The Hartree-Fock exchange operator nix is non-local, i. e. it cannot be presented as an

analytical function of spatial variables. However, it is possible to write its action on some

orbital pi

vnlxi(r) ji(/ --)P1 (r)dri (1 11)

Since the Fock operator f is hermitian and vH(r) and vnixi(r) are invariant with respect

to unitary transformations of orbitals, Equation (1-8) can be rewritten in the canonical

form

fPi = Eii (1-12)

This form of the Hartree-Fock method is most common for practical implementations.

The Hartree-Fock method is a system of integro-differential equations, which cannot

be solved analytically. Iterative methods can be used for the approximate solution of

the Hartree-Fock equations with a given accuracy. As a first step some guess of orbitals

should be assumed and substituted into the equations. During the second step the

system of equations is solved and a new guess for the orbitals is obtained. Then those

new orbitals should be substituted into the equations again, until the total energy and

density at successive iterations differs less than a required accuracy. This is known as the

self-consistent field (SCF) method.

For practical purposes molecular orbitals are usually presented as being decomposed

into Gaussian functions, centered on the nuclei. In this case the number of solutions is

much larger than the number of electrons. To decide which orbitals should be included









into the Slater determinant, the A;,nll/r, principle should be used. Consider the increasing

orbital energies in the order


E1 < E2 < 3... < N... < EM (113)


where N number of electrons and M number of solutions. According to the Aufbau

principle, the corresponding first N spin-orbitals should constitute the Slater determinant.

Such orbitals are called occupied orbitals and the rest of the orbitals called virtual orbitals.

1.1.2 Electron-Correlation Methods

The Hartree-Fock method can recover up to 9' 1' of the total electronic energy. Yet,

even the remaining error of 1 is too big from the point of view of chemistry and can lead

to qualitatively wrong predictions of electronic structure.

The difference between the Hartree-Fock and the exact solution is due to electron

- correlation effects. The primary concern of electron-correlation methods is to develop

]n ,iv- body techniques for going beyond the Hartree-Fock approximation and taking into

account the simultaneous electron-electron interactions. These methods, contrary to the

relatively simple Hartree-Fock approximation, can be quite challenging conceptually and

costly, computationally.

The correlation limit can be obtained from the Full Configuration Interaction (FCI)

method. In this approach the total electronic wave function is expressed in the following

way
occ virt occ virt
'FCI HF + K C + >1K C K ... (1-14)
i a ihj ajb
where a, ab etc are Slater determinants, formed by substitution of occupied orbitals i,

j ... by virtual orbitals a, b ... with the corresponding reordering of rows. The expansion

coefficients are found from the variational condition on the expectation value of the

Hamiltonian
< <^FCI\H|FCI>
EFCI mmin - CT,Ci ,... < FCI ltFCI >









However, the number of possible excited determinants grows rapidly with the

number of electrons and basis functions in the system. Therefore, the full CI method

is computationally intractable for any but very small systems. Among the approximate

electron-correlation approaches, the most common are the Coupled Cluster method[2]

and Many Body Perturbation TI. ii [ :]. Any truncated version of the CI method

has a qualitatively wrong behavior of the energies and wave functions while increasing

the number of particles in the system. Therefore, the CI methods with limited level of

excitations cannot be used for highly-correlated systems. The Coupled Cluster method

and Many Body Perturbation theories are free from this lack of extensivity failure and

are very common for the molecular computations. In some cases perturbation theory

can provide an accurate description of electron-correlation effects at a much lower cost

than necessary for the Coupled Cluster method. The second-order Rayleigh-Schr6dinger

perturbation theory is the simplest and least expensive ab initio method for taking into

account electron correlation effects.

In this perturbation theory the solution of Schr6dinger equation


Hf = E' (1-16)


can be found using the Slater determinant as a reference. Generally, such a determinant

may be constructed from the orbitals, generated by some one-electron operator


hop = (1V2 + u)yp = pp (1 17)
2

The first step of any perturbation theory is the partitioning of the Hamiltonian into a

zero-order Ho and perturbation

H = Ho + V (1-18)

where
elec
Ho = -Eo = (hi)K = (- F)K (1-19)
i i









The perturbation operator is usually defined as the difference between the full and the
zero-order Hamiltonians
elec elec elec
V = H Ho = t(r) ) + (1-20)
i i 'i

Introducing A as the small perturbation parameter, the Hamiltonian and the wave
function can be written as

H = Ho + XV (1-21)

S= + A(1) + A2(2) + ... (1-22)

E = Eo + + A2AEE + 2E2 ... (1-23)

These order-by-order corrections can be found by substituting expansions (1-21),

(1-22) and (1-23) into (1-16) and collecting terms with the corresponding order in A

(Eo -H fo)() >= (V E ())) > (1-24)

(Eo Ho)(2) >= (V E(1))(1) > -E(2) > (125)

C'!....-ig the perturbative corrections to be orthogonal to the reference determinant
< T(") I) > 0 (so called intermediate normalization) it is easy to get the expressions for

perturbative corrections at any order. Projecting the equations (1-24) and (1-25) onto the

reference determinant expressions for energy corrections are obtained

E(1) =< |IV| > (1-26)

E(2) =< |IVI|(1) > (1-27)

Expressions for the order-by-order expansion of wave functions can be written using

the resolvent operator[4]

|I(1) >= RoVI) > (1-28)

(2) > 0(V E(1)) () > Ro(V E 1))RoVI) > (1-29)









where

Ro (1-30)
Eo Ho
and Q 1 I >< K is the projector onto the complementary space of |P >. Since Ho is

diagonal in the basis of Slater determinants, it is possible to write

R0 > ti i <1 j y j (1 3 1 )
Ro = Eo E, (-31)
n 0

For the special case of the Hartree-Fock reference determinant


Eo + E(1) =< liHo01 > + < 1V1 >= EHF (1-32)


the second-order correction to the energy is
10CC uOCC .
E (2) 1 oc < zjlab > (< ijIab > < ijlba >)
2 E EY E. -Eb
i,j a,b

For the case of a more general reference determinant, constructed from orbitals of Equa-

tion (1-17), the second-order correction to the energy has the following structure
ocC unoc 12 occ unocc .
E(2) I < i h- fla > 1 1 < ijlab > (< ijab > < i(ba> 34)
i a i,j a,b

1.2 Kohn-Sham Density Functional Theory

Density Functional Theory is an alternative approach to the description of the

electronic structure of molecular and solid-state systems. This method uses electronic

density instead of the wave function as the basic object of theory. The mathematical basis

of DFT is provided by two theorems, introduced by Kohn, Hohenberg and Sham. The

first one, known as Hohenberg-Kohn theorem[5], establishes a one-to-one mapping between

the ground-state electronic density and the external potential. The external potential

defines a particular object (atom, molecule, etc) and, because of the one-to-one mapping,

the density contains all the information about the system. In particular, the ground-state

energy can be written as a functional of the density. To get the ground state energy in









terms of the density, the Kohn-Sham theorem[6] can be used. This theorem states, that

the ground-state energy as a functional of density has a minimum, if the density is exact.

Therefore, given the energy functional, one can obtain the ground-state density and energy

by variational minimization of the functional. The formal definition of DFT does not

tell how to construct such a functional. Several approximate forms have been -ir-i-. -1. I

however, whose accuracy varies greatly for different properties. The kinetic energy of

electronic motion is particularly difficult to approximate as a density functional.

The basic idea of Kohn and Sham was to transform the variational search over the

density into a search over the orbitals that integrate to a given trial density

(- V2 + v(r))P'(r) -= p,(r) (1-35)


Such a transformation does not restrict the variational space, provided that every phys-

ically meaningful density corresponds to a unique set of orbitals (the v-representability

condition). The use of the Kohn-Sham SCF model ensures not only that the variational

density is a fermionic density, but it also provides a good approximation for the kinetic

energy. If orbitals integrate to the true density it is natural to expect that
Sdec
T, < |V21> (136)
i

accounts for a large part of real kinetic energy.

The rest of the unknown terms are grouped into the exchange-correlation functional


Ec[p] = E[p] T, Eet EH (137)

where EH is a Hartree energy and can be easily be calculated from a given set of Kohn-

Sham orbitals. The non-interacting kinetic energy T, is supposed to reproduce a large

part of the exact kinetic energy T. Therefore, Ec is easier to approximate as a density

functional than the total energy E.









The definition of the exchange part of E,, can take into account its definition in

wave-function theory,

E, =< |,IV, > -EH (138)

Kohn-Sham orbitals are defined by an effective local potential vs. Transforming the

variational condition on the energy functional into a condition for the constrained search

over the orbitals, it is possible to write

(r) (E[p] T,) 6(E.Et +EH + Exc)
vs(r) vext + VH + Vc (1-39)
6p 6p

where the exchange-correlation potential is defined as the functional derivative of the

exchange-correlation energy

c(r) 6 (1-40)
6p(r)
Once Exc is approximated and the Kohn-Sham equations are solved, the total energy can

be found from the following expression


E = E drldr2P(1)P(2) drprlxc() + E (1-41)
i= 1 2 Iri r2
The solution of the Kohn-Sham equations is completely analogous to the solution for

the Hartree-Fock case, and it is usually done by an SCF procedure. After self-consistency

is reached, the Kohn-Sham orbitals are guaranteed to reproduce the true density of the

]rn i,--electron system.

Virtually all modern implementations of DFT use the Kohn-Sham scheme. However,

the theory still has open questions as to how to construct the exchange-correlation

functional. Thus, the main challenge for the theoretical development of DFT remains the

construction of accurate exchange-correlation functionals.

1.2.1 Time-Dependent Density Functional Theory

The Kohn-Shame scheme provides the possibility to describe ground state energies

and densities. For the description of excitation energies a time-dependent generalization of

conventional DFT can be used. Ordinary time-dependent DFT is based on the existence









of an exact mapping between densities and external potentials. In the ground state

formalism, the existence proof relies on the Rayleigh-Ritz minimum principle for the

energy. A straightforward extension to the time-dependent domain is not possible since

a minimum principle is not available in this case. The existence proof for the one-to-one

mapping between time-dependent potentials and time-dependent densities, was first given

by Runge and Gross[7].

We can start from the time-dependent Schr6dinger equation

at H( (t)= (t) (1 42)

evolving from a fixed initial many-particle state


((to) -= o (1-43)

under the influence of different external potentials v(r,t). For each fixed initial state To,

the formal solution of the Schr6dinger equation (1-42) defines a map

A: v(r,t) -+ T(t) (1-44)

between the external potential and the corresponding time-dependent r: ,_vi--particle wave

function and a second map

B : (t) p(r, t) =< T (t) &(, t) (t) > (1-45)

Thus, the following mapping can be established:

G: v(r, t) p(r, t) (146)

The Runge-Gross theorem establishes that the G mapping is invertible up to some

additive, time-dependent constant. In other words, two densities p(r, t) and p'(r, t)

evolving from the common ground state To under the influence of potentials v(r,t) and








v'(r,t) are alv--, different, if


v(r,t) v'(r, t) + c(t) (1-47)

After invertibility for the G mapping is established, the action functional can be
introduced

A[p] = dt < W[p](t)i HIW[p](t) > (1-48)

and the variational condition can be applied,

A[p] 0 (149)
6P(r,t)

Using the same manipulations as for the ground state a set of one-particle equations can
be derived
a (1
p(r, t) (- V2 + ,[p(r, t)])i(r,t) (1-50)
at 2
where
vs [p(r, t)] =vt(r,t) + Jdr' p(r +. [,1 (r,t) (1-51)
f r -( )

The great advantage of the time-dependent Kohn-Sham scheme lies in its compu-
tational simplicity compared to other electron-correlation methods. The time-dependent
Kohn-Sham scheme with explicit time-dependence of the density and the potential can
be applied for any type of external, time-dependent potentials. However, when the time-
dependent external potential is small it can be treated with time-dependent perturbation
theory. If the applied perturbation is a periodic electromagnetic field, it is more convenient
to use response theory.
1.2.2 Time-Dependent Density Functional Theory Linear Response Theory
Consider the N-electron system being initially, i. e. at t < to in its ground state. In
this case the initial density po can be calculated from the ordinary ground-state Kohn-
Sham equation

(-1 2 + vo(r) d' 0 0(plr)) 0)) +0) ) (r) (1-52)
( 2 | r' I r/ '| 0 O i i )(









At t = to the perturbation is applied so that the total potential is given by

v(r,t) '= (r) + vi(r,t) (1-53)

where vi(r,t)=0 for t < to.
The objective is to calculate the linear density response pi(r, t) to the perturbation

vi(r, t). Conventionally, pi(r, t) is computed from the full linear response function X as

pi(r,t) dr dt'x(r, t,r',t')vi(r',t') (1-54)

Since the time-dependent Kohn-Sham equations (1-50) provide a formally exact way of
calculating the time-dependent density, it is possible to compute exact density response

pi(r, t) as the response of the non-interacting system

p (r, t) = dr' dtt'XKS (T,, ', r t1) (1-55)

where v~i) (r, t) is the effective time-dependent potential evaluated to first order in the
perturbing potential, i.e.,

v) (r, t) -= v, (r, t) + dr'p t')+ + dr! f dt'f,(r,(t,r', t')pl(r',, t') (1-56)

The exchange-correlation kernel fc, is given by the functional derivative of v,,

f(r, r', t') v[P]) (1-57)
p(r',P')

at p = po.
While the full response function X is very hard to calculate, the non-interacting

XKS can be computed fairly easily. In terms of the static Kohn-Sham orbitals the Fourier
transform of XKS(, t, T t') with respect to (t t') can be expressed as
OCC 11OCC
cKS ( -r) W W1 ()-58)
XKS(F, r', w) limg a w a (158)
St-( a -(i ,









Equations (1-55) and (1-56) are the basis of linear response theory. Since equation (1

55) is not linear with respect to pi the solution should be obtained by some iterative

procedure.

However, for all practical purposes a direct use of equation (1-55) is not convenient.

To get a more convenient form, the following matrix element should be considered


< plpi(w)l\pi >=< plvl(wi)lpi > + dr~cp(r)( I dr' r) >c~(r) +

+ J dr r)(J dr/'f,,(r, r', U)pIWr', U)) r) (1-59)


On the other hand, < o, lpi(r, w) | > can be expressed in terms of a response function
OCC unoCC
P,(r, ) = Pa( api + ai(W)ppi (1-60)
i a
where


Pia() =, i( ( =7 (1 61)
L + (Ea ~- i) (a C- i)
Substituting equation (1-61) into equation (1-59) we have
OCC unoCC
Pia(W) ( + Fa i <) =< aVI(W) \i > + ([< ab > +
j b
< f f (w) PaPb b + [< OiObjOa >j >+ < I fxc(w) >]Pbj)
OCC unoCC
-Pai(W)(Fa )i +i b ([< OiPbOa j >+
j b
< I -| fGc(w) \PaPj >]Pjb + [< Pab iPj > < fc(') >]Pbj) (1-62)

Introducing the notations


Aai,bj = 6ij6ab(Ea Ei)+ < 0aPb > + < f ;.fc(,) PalPb >

Bai,bj < K 1ipObjpaOj > + < f( c-l) (cPapj > (1-63)









equation (1-62) can be rewritten in the following matrix form

A B 1 0 (P v1 (Go)
B A 0 -1 P+ (w)

This equation provides a possibility to calculate the linear response of the density, caused

by the external perturbation vi(r, w). If the frequency of the applied electromagnetic field
is equal to some resonance frequency (i.e. frequency, which correspond to a transition
to some excited state) even an infinitesimal perturbation causes a finite response of the
density. In this case equation (1-64) can be rewritten as [8-11]


~(A BA); (: ( (1-65)
-B -A P+ )P+)

Equation (1-65) is common for the calculation of excitation energies with time-
dependent density functional theory (TDDFT). Most modern implementations of TDDFT
use the frequency-independent exchange-correlation kernels, defined as a simple functional
derivative
JV (r)
f1(,rJ') Jpr' (166)

Such an approach is known as the adiabatic approximation. If the excited state corre-
sponds to the promotion of a single electron, the adiabatic approximation is applicable
and calculated transition frequencies are quite accurate. However, if there is a promotion
of two electrons taking place, the adiabatic approximation works very poorly because the
underlying equations have no solutions, that correspond to the promotion of two particles.
1.2.3 Problems with Conventional Functionals

The conventional approach is to approximate the energy functional by some analyti-
cal expression of the density and its gradients. The effective potential can then be written
in analytical form, and the Kohn-Sham equations can be solved readily. There are several
levels of approximation for exchange-correlation functionals. The starting point for the









construction of approximate functionals is the Local Density Approximation (LDA), where

the energy is given through an integral of a local functional of the density. The second

level is Generalized Gradient Approximation (GGA) functionals, which improve upon

LDA by the addition of the dependence on the gradient of the density. The third level,

known as meta-GGA, contains also a dependence upon the gradients of the Kohn-Sham

orbitals. For the GGA and the meta-GGA there is the possibility to choose a form of a

functional. A number of different forms have been ir.-. -1. I A standard way of choosing

the form of GGA and meta-GGA functionals is to satisfy a set of theorems, which are

known to be satisfied by the exact functional. Some functionals contain a set of parame-

ters to reproduce experimental data. Such functionals are called semiempirical. There are

also functionals which do not contain parameters but satisfy an extended set of theorems

(non-empirical functionals).

Kohn-Sham DFT with conventional functionals can produce very accurate results for

some properties. Values of the total energy, obtained with GGA are usually comparable

with some ab initio correlated results, like MBPT(2) or CCSD. This is a great advantage

of conventional DFT. However, the accuracy of GGA functionals is restricted by its

analytical form. Generalized gradient approximation exchange functionals cannot provide

the exact elimination of coulombic self-interaction of the Hartree energy. Since the

exchange energy is much larger than the correlation energy, an incomplete elimination of

self-interaction can significantly reduce the accuracy of the total energy. Particularly, the

impossibility to describe weakly-interacting systems, bounded mostly by dispersion forces,

arises from an incomplete elimination of self-interaction and the wrong behavior of the

exchange-correlation functionals with the separation of the system onto several fragments.

Although the exchange-correlation energies are quite accurate with GGA functionals,

the corresponding potentials are not nearly accurate, especially in the inter-shell and

.I-i~,nIl ''I ic regions. As a consequence, the accuracy of the density is much lower than the

accuracy for the energy. Furthermore, the qualitatively incorrect potentials reduce the









usefulness of the Kohn-Sham orbitals and orbital energies for the calculation of ground-

state properties. Excitation energies and response properties are also very sensitive to the

form of the potential and their description with GGA are sometimes very inaccurate.

Some of the problems can be fixed without changing the functional form. Several

post-SCF corrections have been -, r.-. -Ii '1 to partially remove the self-interaction error.

For example, after the Kohn-Shan equations have been solved, a dispersion correction

to the energy is introduced to fix the long-range .,-imptotics of the exchange potential.

However, such corrections are very specific for the particular functional and the class of

the systems, and they are incompatible with each other. The extension of GGA to meta-

GGA does not fix all these problems. Their systematic and consistent resolution is needed

to go beyond GGA.

1.2.4 Orbital-Dependent Functionals

An alternative approach to DFT is to completely dismiss the conventional way of

expressing the exchange-correlation energy as an explicit functional of density. Instead,

the exchange-correlation energy is expressed in terms of orbitals and orbital energies,

yet, the potentials remain implicit functionals of the density. In Kohn-Sham DFT there

is a one-to-one mapping between the density and potential. Since orbitals are generated

by an effective potential through Kohn-Sham equations, there is a one-to-one mapping

between the set of orbitals and the density. Therefore, orbitals and orbital energies can be

considered as functionals of density.

In the case of orbital-dependent energy functionals corresponding exchange-

correlation potentials can not be calculated directly as functional derivatives with respect

to density. Instead, the chain rule should be used for the derivation of corresponding

potentials. Such potentials are equivalent to the corresponding potentials from optimized

effective potential (OEP) method. Therefore, the OEP-based methods constitute the basis

of DFT with orbital-dependent functionals.









The Hartree-Fock exchange functional provides an exchange part of orbital-

dependent exchange-correlation potentials. Kohn-Sham DFT with OEP exchange

potentials provides a superior results to all known density-dependent exchange func-

tionals. To make OEP potentials useful for practical calculations it is necessary to have

the correlation potential, which can be combined with it. Conventional GGA correlation

functionals are developed together with the corresponding exchange potentials, which are

very inaccurate. Because of this correlation potential of GGA compensates errors in the

exchange part. In most cases GGA correlation potentials have an opposite sign to a nearly

exact QMC potentials. The correcting terms incorporated into correlation potentials,

so it is really difficult to extract the "real correlation p oii So far, it is not surprising

that combining GGA correlation potentials with OEP exchange potentials the associated

results are inferior to exchange-only cases. Since conventional correlation functionals are

not compatible with OEP, the development of orbital-dependent correlation functionals is

necessary.

Ab initio DFT uses orbital-dependent correlation energy functionals from the

rigorous wave-function methods. Since corresponding correlation potentials are derived

from ab initio methods, they are compatible with OEP exchange potentials. Unlike

conventional ones, ab initio functionals are systematically improvable, since one can ah-i--,

use a higher-level approximation to obtain a more accurate functional. They also have a

well-defined limit represented by the FCI method.

1.2.5 Ab Initio Density Functional Theory

The main advantage of orbital-dependent energy functionals is the fact, that

such functionals are known in analytical form. Particularly, the exchange functional can be

written in following way


E < Vee >-EH < ;.;,; > (167)
i,j









Here orbitals considered as implicit functionals of the density and the corresponding

functional derivative will produce exact-exchange (EXX) potential.

The most important property of EXX functional is that, unlike conventional density

functionals, it completely cancels the self-interaction component of Hartree energy, so the

corresponding EXX potential will completely eliminate the self-interaction terms in Kohn-

Sham equations. Thus, using the EXX functional and potential will avoid many problems,

caused by the self-interaction error in conventional, approximated DFT for both energy

and density. However, orbital dependence of EXX functional causes one complication.

Since there is no explicit dependence upon density, potentials can not be written directly.

Instead, one must rely on the chain rule, which takes into account the implicit dependence

of orbitals on the density, expressing the derivative of interest through the product of

known derivatives.

Ab initio orbital-dependent correlation energy functionals such as MBPT(2) can be

used for the construction of correlation potentials. Such potentials exhibit correct -C/r4

.i-vi|l!l I ic behavior. It was also shown that MBPT(2) potentials are exact in the high-

density limit and also very close to nearly exact QMC potentials. Alternatively, correlation

potentials can be constructed from the so called /. ,J.:'h; condition.









CHAPTER 2
INTERCONNECTION BETWEEN FUNCTIONAL DERIVATIVE AND EFFECTIVE
OPERATOR APPROACHES TO AB INITIO DENSITY FUNCTIONAL THEORY

The use of the Kohn-Sham orbitals and orbital energies to construct an implicit

density-dependent energy functional using,e.g perturbation theory[12, 13] is a straightfor-

ward approach to the ab initio DFT. The corresponding Kohn-Sham potential can then be

obtained by taking the functional derivative of the finite-order energy via the chain rule,

that transforms the derivative with respect to the density into the derivative with respect

to orbitals and orbital energies[14].This leads to the Optimized Effective Potential[13, 15]

equations which can be complicated already in second order[13]

Alternatively, one can determine the first and second order Kohn-Sham potential

by requiring that the corresponding first and second order perturbative corrections to the

reference density vanish.Unlike the functional derivative approach such a condition on

the density can be described with standard rii ii:-body techniques.The recent work[17]

uses diagrams to derive the second-order OEP equation in a systematic and compact

fashion,while a second paper [16] does so algebraically.

Both the functional derivative and density(effective operator) approaches lead to

exactly the same equation in the first order[17].However,the functional derivative of the

second energy involves a certain type of denominators that are not present in the density

condition(effective operator approach).Still the terms involving such denominators can be

transformed to match the density-based equation exactly. This caveat raises the question

of whether the two approaches are equivalent in higher orders,or for different partitionings

of the hamiltonian.

2.1 Equations for the Exchange-Correlation Potential in the Functional
Derivative Approach

To derive the equations in both approaches we first split the full electronic hamilto-

nian into a perturbation and a zero-order hamiltonian:


H Ho + V (21)









Let us first consider the choice of Ho as a sum of Kohn-Sham orbital energies:
K
Ho0 = ,{(aa,} (2-2)
p=1

where K is the number of orbitals.The brackets means that the product of second

quantization operators are written in normal order. In the functional derivative method we

define the exchange-correlation potential of the n-th order as:


V (') (r) = (2-3)
v 6p(r)

where E(') is the energy of the n-th order of many-body perturbation theory. In single-

reference many-body perturbation theory we have the energy of the nth order as an

explicit functional of the orbitals.In the G6rling-Levy approach this functional is consid-

ered to be a functional for the orbital energies also [14] :

E(-) E(-)(v c..... p, 1, .....^ )


Another possible consideration[13],when orbital energies are considered to be a functionals

over corresponding orbitals


E ) = E (')(o .....o C ol [ ]..... ,c[.k ]) (2-4)


is equivalent to the G6rling-Levy[14] approach.

To construct the chain rule we should take into account the fact that the Kohn-Sham

orbitals are functionals of V,:


E(') = E(')(,(oV,)..... n(V)) (2-5)


Then the Kohn-Sham potential can be considered to be a functional of p(r):


E() = E (')[I 1(V(p(r))) ..... ,n(V8(p(r)))] (2-6)








Variation of the energy can be presented in following form:


>uJ=C


dr2 6~p(r2)
6p,(r2)


(2-7)


We can write down variations of the orbitals and the Kohn-Sham potential too:


(2-8)

(2-9)


Using equations (2-8) and (2-9) we can rewrite the variation of the energy in the following
form:


6E ") r= dr2
S 4p(T)


dr6p, (r2)
dr6V(r
Vs(ri)


V (rl)
dr 6p(r)
6p(r)


Now we can write down the chain-rule[13, 14] for the calculation of Vc) :


6E ")r 6E ")
6Jp(r) 6 6p(r2)


6rp,(r2) 6V,(r3)
6Vs(r3) 6p(r3)


S 6E( 6V(r3)
dj (V() ( p(rl)


Taking into account the fact that[14]:


6Vs(ri) C p ,C
p,q p


6Fr
6v,(r)


W6V(r2)
6V(r2)


it is possible to write an explicit expression for V~) (r) [13]


V(")(r) 2( J
p,q p
Using the fact that


dr26E- () dr3 p (r3) q (r3)X- (r, r3)
Op,(r2) Jp Eq


(2-15)


dr2X(r1, r2)X- (r2, r3)


(2-10)


(2-11)


X(r, r2) 2 z ) i( ) c)a(rl)i(r2)>a(r2)


(2-12)


(2-13)

(2-14)


6p(2f d p2 6V/(r) (
6V,(r) dr 6 p(r) 6p(r)
j 6P(r)


S,(r) ,(r)


6(r1 r3)


(2-16)









we can rewrite equation (2-15) as:


Y o.a(ri)o(ri) < t 1Vnloa > > < E(>) I p(F3)tq(3) (2 17)
a i Fpa q p p -
a,i p,qlP

Using the fact that orbitals p and q can be occupied or unoccupied,we can separate the

right-hand site of equation (2-17) into four parts:

Pa(ri)(Pt() < iK(V a > E(n) 3)(> (r3)
a Fai 6o( Fi -- F

a t a Fi 6 o Fi Ft
a,i J 4,i
6E(") ((3)Yb(r3) (2
+ > < I> a b + c.c (2-18)
abea (pa Fa Fb
a,b740

This form of equation ( 2-18) will be the basis for the formulation of the diagrammatic

rules for taking functional derivatives and for making the connection with an effective

operator approach to the OEP-MBPT exchange correlation potential.

2.2 Equations for the Exchange-Correlation Potential in an Effective
Operator Approach

Let us consider a one-particle density operator in second-quantized form:


p(r) = < v(r rl)Pq > a+aq (2-19)
p,q

Using Wick theorem,we rewrite this operator in normal form with the Kohn-Sham

determinant as the Fermi-vacuum:


p(r) = < Pp J(r- rl)Jl q > {a+aq}+ < LKs 6(r rl)\l KS >
p,q
Nocc
< Kp 1 \(r ri)|Pq>< > IP)2+ a a(I)+ (220)
p,q i=1

The first member of the previous expression will be called the /. ,, .:-/ correction:


Ap(r) = < ppJ6(r rl)|, > {aqa,} (2-21)
p,q









Since the converged Kohn-Sham scheme gives an exact density, all corrections to this

density must be equal to zero[17].This means, that if we construct an effective operator of

the density using MBPT, the correction to the density must vanish in any order.This is

the main idea of the effective operator approach. For the exchange-correlation potential of

first order we will have:


< KS l (')+A(r)SKS (Q(l)+Ap(r))Sc + (Ap(r)(l))Sc 0 (2-22)

This condition can be presented by diagram (2-23).
6

2 i a =0

F (1) (2-23)


Using the fact that fpq = p6p+ < #pVHF V(p) Vclq >, we can extract our

desirable exchange potential.The correlation part is excluded to maintain first order, as the

correlation potential contains expressions of second and higher orders. Then the equation

for V, can be diagrammatically represented by diagrams 2 and 3.



i a i a



2 3

For the second-order effective operator for the density correction we will use the

expression (2-24)


< #KS (1)+ + (2)+)Ap()() + 2(2)) l KS >

(2(l)Ap(r))Sc + (Ap(r)S2(l))S +

+(Q(2)+Ap(r))S, + (Ap(r)Q(2))Sc + (Q(l)Ap(r)(l))sc = 0 (2-24)










Hence,the diagrammatic representation for the second-order equation is presented by the

diagrams (3)-(18) of equation(2-25).


2 3 4 4


6 +2


+4



S 1

10


8 +4


k



+2 11
F3

1 1 F a

13 +2 14
13 +2 14


9 +
F



+ 12a

+ 12 +


+2 15 +


F

a
b j


16


b j



+2 18
+2 18


Diagrammatic representation of potential, derived from the density condition will be used

for the establishing interconnection with the functional derivative approach.


(2-25)









2.3 Interconnection in Arbitrary Order

2.3.1 Diagrammatic Functional Derivatives

For formulation of the rules we will use equation (2-18). The 6 function needs to be

introduced according to equation (2-26):


b
aa

6 (rqP(r) + + i+ a (2-26)
p,qPp q

Brackets on the last two diagrams denote denominators. Expressions for the energy

in MBPT consist of linear combinations of terms, which have products of molecular

integrals in the numerators and product of differences of one-particle energies in the

denominators, so it is easy to make a diagrammatic representation for such expressions.To

take the functional derivatives from numerators, according to equation (2-18), all lines

connected to some vertex must be disconnected from the corresponding place of linking,

and when this is done, a new line (corresponding to occupied or unoccupied orbitals) must

be inserted. In the final step contraction with the corresponding 6 function must be

provided. All lines corresponding to denominators are still unchanged. This procedure

must be provided for all vertices,because when we take functional derivatives from

products of functions, we have a sum of products, according to rules for taking derivatives.

If a new diagram changes its sign, a minus sign must be assigned to this diagram.

When functional derivatives from denominators are taken, it is more suitable to use

equation (2-13) The most general form for the denominator can be represented by the

following formula:
Ka Ka
Den= c( fia -V E )) (2-27)
a i=l j=1









Using equation (2-13) it is possible to write down the functional derivatives from some

denominator in the general form:

De K1 K1 Ka Ka
4 (_ | Ii2 1|2 2 )1|jl 2 c" 12 > ia Y ja +.+
V, / I ( occ I uno I J ( occ unocc) ..... +
i=1 j=1 a,a#1 i=1 j=1
K/ K3 Ka Ka
... ...+ ( 2 _- E I2) E E (2-28)
i=l j=1 a,a43p i=l j=1

Now it is possible to write down a term which includes the functional derivative from the

denominator:

Norm Den
(Den)2 6V,
KT(vi3 1<3 2 KB3 2

Socc j noocc j u(2n29)

We now have the possibility to formulate how to take functional derivatives from denom-

inators. To take functional derivatives from denominators, all diagrams, where one of the

lines is doubled and between these lines which arise from doubling, the corresponding

diagonal 6-function is inserted. This procedure must be provided for all horizontal lines on

the diagram and for all the contours the lines cross.

2.3.2 Diagrammatic Functional Derivatives in Second-Order Many-Body
Perturbation Theory

Direct interconnection in first order is clear from diagram's 2 and 3.The left-hand

side of equation (2-17) is equal to diagram 2, the right-hand site is equal to diagram 3

when the functional derivative has been taken from the expression for the exchange energy.

The diagrammatic expression for the second-order energy in the general case has the

form[l]:
F


E(i b a b (
E(2) F +2 + 2 (230)









Consider the second and third diagram in the energy expression. For the sake of simplicity

only non-equivalent diagrams will be given. Taking functional derivatives with respect to

occupied orbitals we will have equation (2-31)




i J bi + (231)


Inserting the lines of unoccupied orbitals and making contractions with the 6 function,

diagrams 4 and 8 of equation (2-25) will be obtained (2-32):

C C


a a

2 4 +2 8 (2-32)


When functional derivatives are taken with respect to unoccupied orbitals and occupied

orbitals are inserted, diagrams 5 and 9 of equation (2-25) will be obtained.When this

procedure is done for the lower vertex, complex conjugate diagrams will be obtained

and we will have the same number of diagrams of this sort as in the effective operator

approach.

When functional derivatives are taken with respect to occupied orbitals for the

upper and lower vertexes and an occupied line is inserted, we have the diagrams, given by

equation (2 33)


k a 1i a k a
a b b jjk a b j k a b j
-22-2 -2 -2 (2-33)


"Bracket-type" denominator means the difference Ei Ek. Summation of the fist two

diagrams according to the Frantz-Mills theorem[18] and the same procedure for the second









two diagrams gives the diagrams of equation (2-34)


Si j J

2 +2 (2-34)

On the two diagrams we impose the restriction i / k. To make a direct correspondence

with diagrams 7 and 11 we need to add diagrams that arise from taking functional

derivatives from denominators (2-35)


k b j k a b /
2 2 (2-35)

After these diagrams are added to the previous two, we will have a direct correspondence

with diagrams 7 and 11. When functional derivatives are taken with respect to the

unoccupied orbitals and an unoccupied line inserted, the same procedure will give a direct

correspondence with diagrams 6 and 10 of equation (2-25).

Now consider the first diagram for the second-order energy expression. The pro-

cedure for taking functional derivatives from the right side of the vertex, which do not

contain a closed F-ring are absolutely the same as in the case of other second and third

diagrams in the energy expression. When such functional derivatives are taken, we have

four diagrams corresponding to diagrams 12-15. To take functional derivatives from the

F-rings, the first diagram in the energy expression must be represented with more detail,

taking into account that f = h + J K,in following the form (2-36)



CF X + X (2 3

F X+ )+ + 2 X +2 +2 (2-36)









When rules for taking functional derivatives diagrammatically are applied to this set of

diagrams, we will have the next set of diagrams (2-37)






2 + 2 +2 F + 2 ((2-37)


The first two diagrams are equal to diagram 16; the third and fourth equal to diagrams

17 and 16 respectively. Since the left-hand site of eq.12 can be represented by diagram

3, we have an exact equivalence between the functional derivative and the effective

operator approaches in the first and second order for the "Kohn-'lI ii partitioning of the

hamiltonian.

2.3.3 Interconnection in Higher Orders

Equations for the effective-operator approach in the n-th order of MBPT have the

form (2-38)


< (KS (1)+.... (n) (+)Ap(r)((1) .... (n))IK > 6(6Q(k)+Q(n-k)) = 0 (2-38)
k-0
This equation tells as that we need to collect and equate to zero the sum of all possible

combinations of wave operators and density correction operators, and these combinations

must be of n-th order. The same result should correspond to taking functional derivatives

from all energy diagrams of n-th order.

Statement 1 : When functional derivatives from some vertex of a diagram are

taken with respect to an occupied (unoccupied) orbital, and a line, which corresponds to

an unoccupied (occupied) orbital inserted, we .i.-- ,i- will have diagrammatic expressions

which correspond to part of k 0(Q(k)+2Q(n-k))sc.

Proof: When functional derivatives are taken with respect to an occupied (un-

occupied) orbital, and a line which corresponds to an unoccupied (occupied) orbital is

inserted, we alv--,v- have a contraction of the delta function with some fully connected









diagrammatic expression. All lines which correspond to denominators are still unchanged,

except there is one extra denominator from the delta-function. If the functional derivative

had been taken with respect to the highest or lowest vertex in the energy diagram, the re-

sulting diagram corresponds to the product 6 Q'), in the nth-order of perturbation theory.

After applying the same procedure to some intermediate vertex, we have an expression

which contains n-1 denominators and one extra denominator from the delta-function. To

transform the resulting diagram into the product of a delta-function and two wave opera-

tors, we need one extra line, which cross all lines at its level. Such a line can be obtained,

using the Frantz-Mills[18] factorization theorem. Since all initial horizontal lines cross all

lines, which go up from lower lying vertexes, it is possible to represent the diagrams, which

appear after applying the Frantz-Mills[18] theorem in the form (2-39)

6 (k)+Q(f-k) (2-39)


Each k corresponds to some time version. When functional derivatives will be taken from

all vertexes in all diagrams, we will have all possible contractions of the density correction

operator, which have lines going up or down, with products of wave operators. So we have

part of E'k (Q(k)+Q(n-k))sc.

For example, in third order we will have:

b b b
d d

S aC k C k C a k
i \ a i a i a
S + (2-40)


The diagram on the left-hand side is the result of taking functional derivatives from

typical energy diagrams of third order. Applying the factorization theorem[18] to the

right-hand side we will have the same result as on the left-hand side of this relation. The









first member of the right-hand side corresponds to 60(3), while the second corresponds to

56(1)+Q (2).

To finally establish the interconnection between the two approaches, we need to

show that it is possible to transform diagrams which contains "bracket-type denominators

to standard diagrams. In a previous subsection this procedure was described for the

case where one horizontal line crosses the contour in the diagram, which corresponds

to the energy in second order MBPT. We need to prove, that energy diagrams, which

contain contours crossed by m lines, after taking the functional derivative which respect to

unoccupied (occupied) orbitals and inserting lines of unoccupied (occupied) orbitals; can

be transformed to linear combination of m diagrams which do not contain "bracket-type"

denominators.

Consider the following diagram (2-41):

< \ 1
2
k



I m
(2-41)


Numbers on lines show the conditional number of lines which are entered for simplicity of

further manipulations. The desirable sum of diagrams has the form (2-42)



2 2 1
k k


m m / / m
S+ + ......... + L k (2-42)


Together with diagrams which correspond to functional derivatives from denominators,

this set of diagrams forms products like Q(k)+S6(1), where k+l equals the order of MBPT.








Analytical expressions which correspond to this set of diagrams have the form (2-43)

Nomr
S(eCk + Ai)(c, + A,)....(eCk + Am)
Nom
(Qc + A,)(Ck + A2)(C + A2).... (k + Am) +
NVom
+ Nom (2-43)
.+ (c, + A,) .... (c + Am)(Ck + Am)

In these expressions Nom means numerator,Ai means the rest of the terms, which are
present in denominators. After multiplication and division by (Ck c1) we have (2-44)

z Nom(Ek El + A1 A1)
(Ck c- )(Cl + A)(Ck + A) ....(ek + Am)
Nom(Ck cl + A2 A2)
(Ck -1 )(C + Ai)(c + A2)(k + A2).... (k + Am)
+^ Nom(Ck Cl + Am Am)
S(ek Q1)(Q + Ai).... (c + Am)(Ck + Am)
SNom
(ek Q)(CQ + A,)(Ck + A2) .... (k + Am)
SNom
(ek Q1)(ek + A,)(ek + A2) .... (Ck + A)
Nom
z (k QC )(CQ + Ai)(eC + A2) (k + A3) .... (k + Am)
Nom
(ek Q)(Q + A,)(ek + A2) .... (k + Am) +
Nom
(ek QC)( + Ai)(c, + A2)(C + A3) (k + A4.... (Ck + Am)
Norm
(ek CQ)( + Ai)(c, + A2)(Ck + A3) .... (k + Am)
Nom
.(ek C1) (c + A1) .... (c + Am)
Nomn
Nom (2-44)
(c Q1)(Q + A)....(eC + Am)

After cancellation of all equivalent terms, we have the expressions (2-45)

SNom Nom'
(k- C )(ck + Ai)....(eCk + Am) (- C)(Cek + Ai)....(eCk + Am)










where Nom' is the numerator, with interchanged indexes k and 1.These expressions can be

represented by the following two diagrams (2-46)

1 1

2
k
k
m* m

<+ <(2-46)


Namely these two diagrams will appear after taking functional derivatives using dia-

grammatic rules for taking functional derivatives. Proof for the case where functional

derivatives are taken with respect to unoccupied orbitals is the complete analog of this

one. Hence, we can formulate a second statement.

Statement 2: When functional derivatives from some vertex of a diagram are

taken with respect to occupied (unoccupied) orbitals, and a line which corresponds to

occupied (unoccupied) orbital inserted, together with diagrams which arise from taking

functional derivatives from denominators; we ahv-- have diagrammatic expressions, which

corresponds to part of EY )((k)+( C-k))s.

This statement is a direct corollary of the above proved statement about "bracket-

type" denominators and statement 1. Since we have correspondence in the second and

n-th order,using the method of mathematical induction, it is possible to prove, that the

correspondence taks place in all orders (finite or infinite).

2.3.4 Interconnection in Infinite Order

It should be noted that summation of perturbation corrections to the energy and the

wave functions up to infinite order provide correspondence with the full CI:

00
E n E) E(FCI)
n0

n O (n) (FCI) (247)
n=0









Using the effective operator approach it is possible to define VC7) in all possible orders.

This procedure for an infinite sum of orders gives Vxc. Inserting this potential into the

Kohn-Sham equations, we can find p("),which corresponds to the one-particle density in

the full CI method.The same density could be used in a method like Zhao-Morrison-Parr

(ZMP)[19] to extract the corresponding exchange-correlation potential. Since the full CI

energy does not depend upon the choice of orbital basis set, the OEP procedure[13, 15]

cannot be used directly for this case. The infinite sum of all energy corrections does not

depend upon choice of orbital basis set, but each term of this sum does depend upon

the choice of orbitals. This fact enables us to consider all term of the infinite sum of

energy corrections as orbital-dependent functionals. After the construction of the set

Vx1) .... t) and summation up to infinity, we will have the exchange-correlation potential

which corresponds to the full CI. Using the equivalence of the functional derivative and

the effective operator approach, it is possible to conclude that after summation of this

set of potentials, we again will have the same result as in the effective operator approach.

Redefined in such a way, the OEP procedure[13, 15] for the full CI energy produces the

same density as the ZMP method[19] would from full CI.









CHAPTER 3
AB INITIO TIME-DEPENDENT DENSITY FUNCTIONAL THEORY EMPLOYING
SECOND-ORDER MANY-BODY PERTURBATION THEORY OPTIMIZED
EFFECTIVE POTENTIAL

The time-dependent OEP exchange-only method in the adiabatic approximation

has been previously derived [20] and implemented for molecules by Hirata et. al [21].

Numerical results in that work show a reasonable description of both valence and Rydberg

excited states, partly due to the correct .,-i-!!, ,l'tic behavior of the exchange potential,

but also due to the elimination of the self-interaction error. In particular, OEP-TDDFT

is superior to standard approaches like TDDFT based upon local density approximation

(LDA) or Becke-Lee-Yang-Parr (BLYP) functionals. Similarly, exchange-only OEP with

exact local exchange (EXX)[22, 23] has been shown to greatly improve band-gaps in

polymers [23]

Another advantage of OEP based methods is that since virtual orbitals in the

exchange-only DFT as well as occupied orbitals are generated by a local potential,

which corresponds to the N-particle system, the differences between orbital energies of

virtual and occupied orbitals offer a good zeroth-order approximation to the excitation

energies[16, 17].This is not possible in the case of Hartree-Fock theory, where occupied

orbitals are generated by an N-l particle potential and the energies of unoccupied orbitals

come from N-electron potential, and thereby, approximate electron affinities.

Once OEP correlation is added [13, 17], the essential new element in the time-

dependent DFT scheme is the exchange-correlation kernel, which in the adiabatic approx-

imation is defined as a functional derivative of the exchange-correlation potential with

respect to the d, i,-il [20]:

6p(r2)
Hence, in this paper we will derive the kernel for the OEP-MBPT(2) correlation potential,

when we use the standard KS Ho, Ho = i hks(i). In prior papers[16, 17] this has

been called OEP Kohn-Sham (KS) to distinguish it from other choices for Ho. It is









also the choice of Gorling and Levy[14, 24]. Only for this choice is there an immediate

correspondence between the functional derivative and KS density condition that a single

determinant provides the exact d, nil-y[24]. The other choices for Ho lead to what is called

ab initio DFT [16, 17] and becomes a better separation of hamiltonian into Ho + V' Its

convergence for orbital-dependent perturbation approximations to the correlation energy

functionals are considerably better behaved compared to the KS choice that frequently

causes divergence. Nevertheless, in this first application we will adhere to the standard KS

separation, where the potential and kernel are functional derivatives.

The traditional way of deriving kernels and potentials is to tediously derive all

terms with the use of the chain-rule for functional differentiation. Yet, even the exchange-

only kernel has a complicated structure in OEP, and its further extension to include

correlation would be almost impossible. To avoid the use of traditional methods, an

effective diagrammatic formalism for taking functional derivatives has been developed and

implemented.

3.1 Diagrammatic Construction of the Exchange-Correlation Kernels

3.1.1 Formalism

The adiabatic (frec -i' l-i--independent) approximation to the kernel of the nth order

is defined as (3-2)


f(n)(r1, r2) V (ri) dr3 xe;(r) X (r3,r2) (3-2)
6p(r2) J Vrs(3)

The exchange-correlation potential of nth order is than defined through the chain-rule by

the following formula[13] (3-3)

Vj(r)} = 2 () (r2Jdp3 pp(r3)P(Fr3) X-(rr3) + c (3-3)
p~~p 6(r2) FP Fq








where the c.c means the complex-conjugate. Inserting equation (3-3) into (3-2) we obtain


: S Jdrz4 (J
p,q p r,sr


d 6E(")
dr.5 s (r5) x
4(,(r5)


x dr6 r X -(r6, r6)+ C.)(r4) x
Sr Es
x dr3 Op(r3 )q- 3X (r2,3) + C.
Sp Eq


This may be rewritten as (3-5)


SJdr3 6 (J dr5 ) (P)(r5) X
p,q4p (r3) Ep(r5)


xJdr6 p (r6)q (r6)X- (rl,r6) + .c)Xl(r2, r3) + c.c
J p c


Equation (3-5) can be made more explicit,


P J 3X (r2,r3
p,q p 6V,(r3


dr\


x dr6 P(6)q(76)-]X- (ri, r6) +c.c) + > J
p p,qp


x f E dr )q (r5)
j JOp(r5)


6E(")
5 q(r5) x
3- p(r5)
dr3X -I(r2, r3) X


d p(r6) q (6) X-1(r, r6)
dF6 + c(r)
Ep Eq 6Vs(r3)


Using the fact that [14]


6X-l(rl,r2)
6V(r3)


dr drX rr5) X(5, r6)X-1(r6,2)
Jdr 6Vs(r3)


it is possible to rewrite equation (3-6) in the following form (3-8)


dr3 dr6X-(ri, r3)h(3,6)X-(r6,2)


Taking into account equation(3-3) we have the explicit expression (3-9) for h(r3, r6)


h(r3, r6) V(r3)Y J
p,q p


Edr() p (r6)q (r6)
d (r5 ) r5) q
6()p (r5) Ep Eq


V) ((r5)6X(r5, r6)
dr5 0 6V(rs)


+ c.c


(3-4)


(3-5)


(3-6)


(3 7)


(3 8)


f ()( r2)


f/(n 1, r2)


/f()(rl, 2)


f(P)(r1,r2)


(3-9)









Equation (3-9) will be used as the basis for the diagrammatic procedure for taking

functional derivatives from the OEP-MBPT exchange-correlation potentials. The first

term of equation (3-9) contains the functional derivative with respect to the potential,

from an expression which already includes the functional derivative of the energy with

respect to the same potential.

Diagrammatic rules for taking functional derivatives from numerators of diagrams

are the following:

Detach the line from the vertex

Insert parts of the 6-function (2-26) to get the fully connected diagram

Repeat the procedure for all vertices in all diagrams which contribute to the

exchange-correlation energy

To take the functional derivative of the denominators, the denominator line should be

doubled and diagonal expressions of the last two parts of the 6-function (2-26) should

be inserted between doubled lines for all lines which are intersected by the denominator

line. This procedure should be done for all possible denominator lines. The rules of

interpretation are the same as for the usual Goldstone diagrams, except the numerical

factor should be taken from the initial diagram when functional derivatives are taken.

After taking the functional derivatives from the exchange-correlation with respect

to the potential using diagrammatic rules, a set of diagrams, containing the 6-function

vertex will appear. To get the first element of equation (3-9), it is necessary to apply

diagrammatic rules to the set of functional derivative diagrams one more time. After that

we will have a set of diagrams containing two 6-functions, which are necessary for the

construction of the exchange-correlation kernels.

Taking into account equation(2-14), the second term of equation (3-9) can be

rewritten in the following way (3-10)

/ V ) (r5s)6X(rs, r6) () 6 ci (r5(Pa( 5)i r(6a(r6)
-I = d,, 17( -2 V,* I I (3-10)
6Vs(r3) Vs(r3) i Fa









The expression, which is suppose be differentiated with respect to the potential on the

right-hand side of equation (3-10) can be represented by diagram (V) (3-11)


i a

V (n)
xc

(V) (3-11)


The functional derivatives from diagram (V) produce the set of diagrams (V1)-(V6) of set

(3-12)

16 6 2 6 1 6 2 6
b 1 b 2 1 2b
I a (n
S 2 1 a 2 V a V_ I
V a V (n) a V (n) I V (n
xc xc xc xc
(VI) (V2) (V3) (V4) (V5) (V6) (3-12)


During the derivation of diagrams (V1)-(V4), diagrams containing "bracket-type" denom-

inators will appear, because one of the steps in the diagrammatic rules of differentiation

requires detaching the unoccupied line from diagram V and inserting the last two parts of

the 6- function (2-26). Such diagrams can be transformed into a set of regular diagrams,

using the diagrammatic relation (313). This relation shows us that for the transformation

to regular diagrams it is necessary to double each of the denominator lines and insert the

last two members of the 6-function (2-26), subject to the restricted summation, I / k.We

use the same procedure for the case of occupied orbitals.

Sa > 1 1
2 2 2> 2
k 2 > k k
+ :k + + + (Dl)

m .
(313)









After the transformation of irregular diagrams and adding the functional derivatives from

the denominator of diagram (V), diagrams (V1)-(V4) will appear. Using diagrammatic

expressions for the first member of equation (3-9) and diagrams (V1)-(V6) it is possible to

build the exchange-correlation kernels to any order.

3.1.2 An Example: Diagrammatic Derivation of Exchange-Only Kernel

The kernel for the exchange-only case was initial derived by G6rling [20], and then

rederived and initial implemented by Hirata et. al [21]. Such a derivation requires much

effort. Here we offer a facile derivation with the diagrammatic formalism.

The exchange energy (Ex=-~E~ j < ij ji >) can be represented by diagram (3-14).

E, (EX) (3-14)

After taking the first functional derivative with respect to V we will have the diagram


(3-15).

a
E,/6V, = 2


(VX) (3 15)

After that the diagrammatic rules must be applied one more time to get the second

functional derivative, as is necessary according to equation (3-9). After taking the second

functional derivatives we will have diagrams (FX1)-(FX8) of the set (3-16)


i a 1 2 1 2 aa 1 1 2 i -
i a b 1 1 i b b 1 1 i a


62 62 4 _
V- k k kj j k J b j
(FX1) (FX2) (FX3) (FX4) (FX5) (FX6) (FX7) 2 (FX8) (3-16)


After the addition of diagrams (V1)-(V6) we will have all the diagrams necessary for

building the exchange-only kernel. During the interpretation of diagrams (V1)-(V6) the









potential vertex should be the matrix element < pp\ Vlpq > and a factor -2 must be added

according to equation (3-10). Interpretation of diagrams (FX1)-(FX8) and (V1)-(V6)

gives us the expression (3-17)


h(r3, 6 -2
i,j,a,b


< ialjb > (i(3)(ao(r3)(j (r6)L(b(r6)
(Ei Ea)(E. Eb)


2 > l< ij ba > i(r3) oa(r3) j (r6) b(r6)
-2
ij, a, -F)(E b)
i,j,a,b
+2 y < jk|ka > [pj(r3) i(r3) i(r6) a(r6) + j (r6)(P(6)(i(r3) o(r3)]
ij,k,a (- Fa)(- Fa)
2 y K ai|j *I > [ao(r3)b(r3)-Eb(r6)aj(6) + ( Pa(r6)b(rF6)(cb(r3)i0(3)1
i,j,a,b ( i a
< iklkj > (PF3)La(r3)3j (r6)ao(r6)
,j,k,a (- )(- a)
-2 y < aj jb > Pi(F3)a(r3) (i 6)b(r6)
(Ei a)(Ei b)
i, j, a,b
-2 C < iy ll > i((3 a 3 j (6 + ( ( 6

S< aV lb > y4(r3) a(r3)Pi(r6)ob(r6)
(Ei Ea)(i b)

S [p(r3)fcj(r3)cj (r6)foa(r6) + 6 i6)6j(r 6)(Oj (r3)yOa(F3)1

< i\ [Va > [o(r3)b(r3) b(Tr6)(6T) + Yoa(r6)Yb(r6) b b(r3)3i(r3)1
(E- Ea)( Eb)
a,b,i


(3-17)


Expression (3-17) for h(r3, r6) exactly corresponds to the expression, obtained by Hirata

et. al, but the diagrammatic derivation requires far less effort and is unambiguous in

terms of signs and numerical factors. Since all the diagrams contain only one contour, it

is possible to make summations in equation (3-17) only over spatial orbitals, and, as a

result f,,(rl, r2) and faa(rl, r2) will appear separately. The exchange-only kernel does

not contain the f,3(rl, r2) part, which is a critical difference between the exchange and

correlation kernels.










3.2 Kernel for the Second-Order Optimized Effective Potential Many-Body
Perturbation Theory Correlation Potential

For the derivation of the second-order exchange-correlation kernel we require

functional derivatives of the second-order correlation energy with respect to V,. These

functional derivatives can be represented by diagrams (VC1)-(VC15) of (3-18).


c ^ k


a b j b j6 a


(VC1) (VC2) (VC3) (VC4)

b k

ba
c b 1 i\ /a b
\7 \/

(VC5) (VC6) (VC7) (VC8)
b i F
CF
j c
F a F a b
a a
a a i a i b j
F FF F
(VC9) (VC10) (VC11) (VC12)(vC13) (VC14) (VC15) (3-18)


Diagrams (VC1), (VC2), (VC5), (VC6) and (VC15) of (3-18) have an external factor 4,

diagrams (VC9) and (VC10) have no factor, while the rest of the diagrams have a factor

of 2. To get the correlation kernel diagrams we need to take functional derivatives with

respect to the potential from diagrams (V1)-(V15). After applying the diagrammatic

rules to diagram (VC1) and the use of relation (3-13), diagrams (VC1-1)-(VC1-16) of sets

(3-19) and (3-20) will appear.

~l _2 /01 2 f1 k1 C1 C1 A0,
c k cd ci
b i Ji j- i b b i i b j

S
(VC1-1) (VC1-2) (VC1-3) (VC1-4) (VC1-5) (VC1-6) (VC1-7) (3-19)














2
(VC18) (V
(VC1-8) (VC1-9)


(VC1-10)


(VC1-11) (VC1-12) (VC1-13)


(VC1-14) (VC1-15)


(VC1-16)


(3-20)


Differentiation of diagram (VC2) produce diagrams (VC2-1)-(VC2-16) of set (3-21). These

diagrams have the same skeleton structure, but different positions of indices.


k ba ja a

i(VC2-) (C2-2) (C2
(VC2-11 (VC2-21 (VC2-3)


k 01 0- 0 1

k ba i b
i( i 6 ( c 1 2 C
2 V 2 2
(VC2-4) (VC2-5) (VC2-6) (VC2-7) (VC2-8)


a(C b ja(V b

(VC2-9) (VC2-10)


(VC2-11) (VC2-12) (VC2-13) (VC2-14) (VC2-15)


(VC2-16) (3-21)


After taking the functional derivatives from diagram (VC3) diagrams (VC3-1) (VC3-14)

(3-22) will appear.


2 2

2 a a bj a
2 i j i
a k 1b1 i1 a
k -tik kk


i a ia
b 6 b 1 ) b\ 2a
k k a k


(VC3-1) (VC3-2) (VC3-3) (VC3-4) (VC3-5) (VC3-6) (VC3-7)
J J J J
1 a a 1 na a 5 a
2 ia i' b b Jb b t b b/ b
k i k i k i V k i k iV k k 2


(VC3-8) (VC3-9) (VC3-10) (VC3-11) (VC3-12) (VC3-13)


2 k


G1c k


cib k
b


&-k c


(VC3-14) (3-22)










When functional derivatives are taken from the diagram (VC4), diagrams (VC4-1)-(VC4-


14) (3-23) will appear.


2 i = b2
ajO


k d

c a
lc iJ c a c


(VC4-1) (VC4-2) (VC4-3) (VC4-4) (VC4-5) (VC4-6) (VC4-7)


_Ljd dTTT dF^T a ^
S 4-8 a a a5
=C 1 1C4-9 CC-
(VC4-8) (VC4-9) (VC4-10) (VC4-11)


(VC4-12) (VC4-13)


Diagram (VC5) produces diagrams (VC5-1)-(VC5-16) (3 24) after differentiation.


c c d
1 d c c ib 7ib c b

bj c b 1 ab a
j b 2i b 2 b k
k (k k 2
(VC5-1) (VC5-2) (VC5-3) (VC5-4) (VC5-5) (VC5-6) (VC5-7) (VC5-8)


S c 2 C 1 2 d 2

S bj b jk ab jkab ja b a b

(V 5-9) (VC5-10) (VC5-) (VC5-12) (VC5-13) (VC5-14)
(VC5-9) (VC5-10) (VC5-11) (VC5-12) (VC5-13) (VC5-14)


(VC5-15)


(VC5-16) (3-24)


Differentiation of diagram (VC6) generate diagrams (VC6-1)-(VC6-16) (3-25),


a b 2c bca ba ba ( b

(VC6-10) (VC6-1 1) (VC6-12) (VC6-13) (VC6-14)


bi\\ 2 k


aic b b

(VC6-15) (VC6-16) (3 25)


k b
1;i


a Fb


(VC4-14)


(3-23)


a Mbh
2
(VC6-9)











while diagram (VC7) gives diagrams (VC7-1)-(VC7-14) (3-26).


2 i 1a

(VC7-1) (VC7-2)


(VC7-3) (VC7-4) (VC7-5) (VC7-6) (VC7-7)


(326)


(VC7-8) (VC7-9) (VC7-10) (VC7-11) (VC7-12) (VC7-13) (VC7-14)


Finally, diagram (VC8) produces diagrams (VC8-1)-(VC8-14) (3-27).


(VC8-1) (VC8-2) (VC8-3) (VC8-4) (VC8-5) (VC8-6) (VC8-7)


(VC8-8) (VC8-9) (VC8-10) (VC8-11) (VC8-12) (VC8-13)


(VC8-14)


(3-27)


In the same way the rest of the diagrams should be differentiated. Since the Fock operator

depends upon occupied orbitals, F-rings on the diagrams (VC9)-(VC15) must also be

differentiated. Diagram (VC9), after taking functional derivatives, produces diagrams

(VC9-1)-(VC9-10) (3-28).


a 6 a
2 k2 az2




(VC9-1) (VC9-2)

k F 6


F F b
(VC9-6) (VC9-7)
(VC9-6) (VC9-7)


F(

b 6


F
(VC9-3)


F



(VC9-8)


(VC9-4) (VC9-5)
b k





22
V 2 a J6
(VC9-4) (VC9-105)


(3-28)











Diagram (VC10) produces diagrams (VC10-1)-(VC10-10) (3-29).


V V2


(VC10-1) (VC10O-2) (VC10O-3)


F c F a ji

2(


(VC1O-6) (VC1O-7)


S a b
(4c 1 8) ( )

(VC10-8) (VC10-9) (VC10-10)


Diagram (VC11), produces diagrams (VC11-1)-(VC11-16) (3-30).


(VC11-1) (VC11-2



a




(VC11-6)


b
n-8
F
J
F
) (VC11-3)



b/ F




(VC 11-7)


lb 7\j a jF
b j k F 1
b 62 2 a U 2a
bk
b k 62
(VCll-10) (VC11-11) (VC11-12) (VC11-13)



kb kb
kb



(VCll-14) -15) (VC-16)
(VC11-14) (VC11-15) (VC11-16)


2


F


28
F -5)
(VC1O-5)


26


(VC10-4)


(3-29)


2b


F
j a
F
(VC11-4)


b

a

(VCll-5)








(VC11-9)


(3-30)










In the same way diagram (VC12) generates diagrams (VC12-1)-(VC12-16) of set (3-31)


ij

a
F
b i
F
(VC12-3)


2 j

2 6

b i
F
(VC12-4)


S b
F
(VC12-5)


1 b

F



c
(VC12-6)


F 1 F



(VC12-7) (VC12-8)


aj 6


/b
(VC12-9)
(VC12-9)


U ij2
b


i a a


l J
,J a 8 CC


(VC12-10) (VC12-11) (VC12-12) (VC12-13)
6 b b



1 Ob
Fb Ca (V7a
(VC12-14) (VC12-15) (VC12-16)


Differentiation of diagram (VC13) gives us diagrams (VC13-1)-(VC13-15) (3-32)


1 2 1 2
b k b k j c j c
82 82 8
j j b b
1 1 a a
a a i a 13
F F F F

(VC13-1) (VC13-2) (VC13-3) (VC13-4)


b l
2 k


a F


j 6
2 C b

a
F


(Cb -



a F i F

(VC13-5) (VC13-6)


0 V1 0i 2 2
b j k i C
b F F
c k
i 82 a 62
a b b
F F 61


(VC13-7) (VC13-8) (VC13-9) (VC13-10) (VC13-11) (VC13-12)

2b1 a -
S61Z 62 b J

b a i b
a(VC13-13) (Vc k a(VC13-15)
082 Nk c
(VC13-13) (VC13-14) (VC13-15)


(3-31)


T,


j 6 2 2


(3-32)










After taking the functional derivative of diagram (VC14), we obtain diagrams (VC14-1)-

(VC14-15) (3-33)


F F
b i Fb 6Z -a b a
b a j a ^.
(VC14-1) (VC14-2) (VC14-3)


(F F(F
a ba b k b


(VC14-7) (VC14-8) (VC14-9)


- 82 b a i j b


(VC14-4) (VC14-5)



a iai

C14-10) (VC14-
(VC14-10) (VC14-11)


(VC14-13) (VC14-14) (VC14-15)


The last set of diagrams (VC15-1)-(VC15-15) can be produced by

diagram (VC15) (3-34)

162 (1 62" kF F 53 C F k & F
(F k F F t k

1 1 a a & 1
1i a 1 ai %a61 -6, a
(VC15-1) (VC15-2) ( 15-3) (VC15-4) (VC1
(VC15-1) (VC15-2) (VC15-3) (VC15-4) (VC15-5) (


(VC15-7)
b k
(VC15-7)


b


(VC15-11)


F 1

b 'k c
(VC15-8) (VC15-9)
2 6 61


C


differentiation of




P a ,






M'b

VC15-10)
VC15-16)


F)8 b, F) bk UF k

b( (VC15-14)Li (C515 b

(VC15-12) (VC15-13) (VC15-14) (VC15-15)
(VC15-12) (VC15 13) (VC15 14) (VC15 15)


Together with diagrams (V1)-(V6), all diagrams presented in this section form a set, which

is necessary to construct the correlation kernel. The interpretation of all diagrams is given

in Appendix A.


(VC14-6)


(VC14-12)


(3-33)


(3-34)









3.3 Properties of the Correlation Kernel

The second-order correlation kernel has different spin components, which can be

represented by the formula


fn(ri, r2) dr3 dr6Xaa(rl,r3)haar3, r6)Xa (r6,r2)

/ (n r2) f dr df6Xa-(ri, r3) 3, 6)X (r6, r2) (3 35)

Diagrams, which contain only one contour cannot produce ha3 components after differenti-

ation. Three of the diagrams (VC5)-(VC8) after differentiation can produce only diagrams

contributing to the haa or h3p parts. The diagrams (VC9-1)-(VC9-7), (VC9-9), (VC9-10),

(VC10-1)-(VC10-7), (VC10-9), (VC10-10), (VC11-1)-(VC11-10), (VC11-13)-(VC11-16),

(VC12-1)-(VC12-10), (VC12-13)-(VC12-16), (VC13-1)-(VC13-12), (VC13-14), (VC13-15),

(VC14-1)-(VC14-12), (VC14-14), (VC14-15) have the same property. Diagrams containing

2 or more contours can produce all spin components. Sets of diagrams (VC1-7)-(VC1-16),

(VC2-7)-(VC2-16), (VC3-7)-(VC3-14), (VC4-7)-(VC414), (VC15-7)-(VC15-12) have two

contours, but both 6-functions present in one of the contours means that these diagrams

make a contribution to haa or hp parts, but have an extra factor of 2. That factor

appears after summation over all spin-orbitals of the second contour. Diagrams (VC1-

1)-(VC1-6), (VC2-1)-(VC2-6), (VC3-1)-(VC3-6), (VC4-1)-(VC4-6), (VC9-8), (VC10-8),

(VC11-11), (VC11-12), (VC12-11), (VC12-12), (VC13-13), (VC14-13), (VC15-1)-(VC15-6),

(VC15-4)-(VC15-15) have 6 functions on different contours, so they contribute to both haa

and h3p parts. Diagram (VC15-13) have an additional factor of 2 and also contribute into

both spin parts. To build all spin parts of the correlation kernel we need to substitute the

above diagrams into equation (3-35)

The next essential property of the exchange and correlation kernels is the symmetry

with respect to permutation of its arguments


S(r1, 2) f(r2, r1) (3-36)









This is possible because of equation (3-35) and


h(ri, r2)- h(r2, r) (3 37)


To establish equation (3-35) we again need to analyze the structure of the diagrams.

Diagrams (VC1-2), (VC1-5), (VC1-7), (VC2-1), (VC2-6), (VC2-7), (VC5-2), (VC5-7),

(VC5-9), (VC6-2), (VC6-7), (VC6-9), (VC11-6), (VC12-6), (VC13-14), (VC14-15), (VC15-

13), (VC15-11), (VC15-12), (V5) and (V6) are symmetric with respect to the permutation

of variables r1 and r2. After consideration of the rest of the diagrams, it is easy to see

that for each diagram there is another diagram, which differs from the initial one by

only the interchanged variables, rl and r2. Such pairs provide invariance with respect to

permutation of variables.

During the construction of the correlation kernel Kohn-Sham (KS) orbitals and

orbital energies are used. This implies the fact that our zero-order hamiltonian is chosen

to be the Kohn-Sham choice

Ho0 = p{a t} (3-38)
P

In all the diagrams containing the diagonal Fock matrix elements,fpp, should be replaced

by fpp p because all the matrix elements of the Kohn-Sham operator equal to zero,

except the diagonal ones.

Like an exchange-correlation potential, the kernel is a local, multiplicative operator.

This gives the possibility to use auxiliary basis sets for decomposition of kernels, what can

be very helpful for implementation.

3.4 Numerical Testing

To calculate excitation energies the TDDFT equation (1-65) should be solved.

In the adiabatic approximation it has the RPA-like form.The critical new quantity is

the exchange-correlation kernel, f( ), which can be seen from the foregoing is a quite

complicated quantity if we insist upon obtaining it rigorously for OEP-MBPT(2). The

diagonal dependence of A on ca c tells us that if we are to get good excitation energies,









this zeroth-order difference should be a decent approximation to the excitation energy, and

is, as shown elsewhere [16] when accurate exchange-correlation potentials are used. In this

sense, the orbital energies in KS-DFT should have a certain meaning.

Furthermore, as pointed out in ref [16], we can also consider this equation to offer

a Koopmans-like approximation to the principal ionization potentials, since, barring

pathological behavior, when we allow an electron to be excited into the continuum, its

orbital a will then have no overlap with the bound orbital matrix elements of hKS and

the kernel. Consequently, we are left with nothing but -ci in the TDDFT equations. So in

this 'sudden', adiabatic approximation, the KS orbital energies should offer an estimate

for each of the principal Ip's, not just the highest-occupied (homo) one. When based upon

the relatively correct V,, obtained from ab initio dft[16, 17] this estimate is superior to

Koopmans' theorem for the homo and the first few valence Ip's, but is inferior for the core

or'.il '1-[21]. See also C('l.: et al [25]. However, the OEP2 semi-canonical (sc) ab initio

dft [16, 17] approximation has the distinct advantage that it uses a much better behaved

unperturbed Hamiltonian than the usual KS choice, Ho = : hKs(i).

To illustrate the evaluation of the kernel and the solution of TDDFT equations, we

consider the Ne atom. All the excited states in Ne correspond to Rydberg excited states.

To obtain reasonable values requires a quite extensive, diffuse basis set. We choose to

start with the ROOS-ATZP atomic natural orbital '. ,--[[27] consisting of (14s9p4dlf)

primitive gaussian functions contracted to a [5s4p3dlf] set. This underlying basis was

then augmented by a set of even-tempered diffuse functions [3s3p3d], with exponential

parameters = ,' a8 = 0.015, a = 0.013, ad = 0.012, b=1/3. The auxiliary basis is

chosen to be the same, but without the p and d diffuse functions, since for the description

of the exchange-correlation potential, diffuse functions are not necessary. The orbital

energy estimates are shown in Table 3-1.

The ionization potential equation of motion coupled-cluster (IP-EOM CC) result

in this basis for the Ne homo Ip is 21.3 eV with the experimental value being 21.5645









Table 3-1. Orbital energies and zero-order approximations to excitation energies
Exchange-only OEP-MBPT(2)
orbital IP[28] Orbital a, FHOMO Orbital Fa FHOMO Excitation
energies energies energies(exp)
2s 48.42 -46.19 -43.25
2p 21.56 -22.51 -20.23
3s -5.15 17.36 -7.58 12.65 16.85
3p -2.59 19.92 -3.72 16.51 18.70
4s -1.95 20.56 -2.66 17.57 18.73
3d -1.49 21.02 -2.05 18.18 18.97


[28] This compares to the OEPx value of 22.51 and OEP-MBPT(2) value of 20.23. The

remaining unoccupied, but negative energy 3s orbital is changed by over 2 eV due to the

MBPT(2) correlation. The OEP2(sc) ab into dft value changes this to 5.18 eV, attesting

to the poor convergence of the standard KS partitioning of the Hamiltonian. Once

OEP2(sc) calculations are done for the 3p, 4s, and 3d states, there is similar agreement

between the OEPx and OEP2(sc) results, contrary to those shown in Table 3-2 for the

standard (ks) choice.

Table 3-2. Excitation energies of Ne atom using OEP-MBPT(2) Kohn-Sham orbital
energies
Term EOM-CCSD TDDFT-exchange-only TDDFT OEP-MBPT(2) Exp[29].
3P 16.5353 17.1748 15.2132
1p 16.7274 17.5209 11.4434 16.847
3 18.8303 19.8681 16.3320
1 18.8822 19.9181 16.9465 18.703
3P 18.9156 19.9134 17.5645
1p 18.9791 19.9134 17.7686 18.725
S 19.5118 19.8811 17.t. ;
S 19.9323 2', '- ; 17.3454 18.965
3P 20.3747 20.4905 18.3455
1p 20.5588 20.6090 18.8788


Despite the differences, at least both the exchange-only and the OEP-MBPT(2) give

qualitatively correct results for the Rydberg series. Helping to ensure this is the fact that

the exchange potential has the correct .,-ii iil '1 i c behavior and the exact cancellation









of self-interaction is assured. The correct .-i-'!.' ictic behavior is achieved using the

Colle-N. -bet algorithm[30].

When we solve the TDDFT equations with the kernel developed in this paper,

we obtain the results for excitation energies shown in Table 3-2. In general, just as we

observed from the orbital energy differences, the OEP-MBPT(2) results tend to fall on the

low side of experiment and the OEPx on the high side. In fact, except for the excitation

energy for the lowest 1P state, where the very low orbital energy of -7.58 biases the

results, an average of the two would seem to be about right. There are, however, still

difficiences in the basis set, as seen by the EOM-CCSD results compared to experiment,

where besides the additional correlation effects introduced by EOM-CCSDT[31], the

further extension of the diffuse functions would remove the 1 eV error in the highest lying

(1S) state. Of course, the dependence of TDDFT on the basis and that for a two-particle

theory like EOM-CC should be quite different.

The great sensitivity of the results to the orbital energies from the underlying KS-

DFT calculation can be further appreciated by simply taking the energies from the OEPx

results and using them in the evaluation of the kernel and the matrix elements in the

TDDFT equations. These results are shown in Table 3-3. The near coincidence of results

for OEPx and OEP-MBPT(2) is apparent, with both now being too high.

A similar experiment can be made where we use orbitals and orbital energies from

OEP2(sc) ab into dft results as described elsewhere[16], to obtain the results in Table

3-4. That is the TDDFT equations and the kernel are assumed to be the same, but we

use orbitals and orbital energies obtained from OEP2(sc). Here we also show the results

from standard TDDFT applications using the LDA and B3LYP functionals for comparison

purposes. Clearly, we have improved results at both the OEPx(sc) and OEP2(sc) levels,

with the same pattern of the former being too high, but less so; and the latter, too low,

but better than before. It is apparent that the proper way to achieve the benefits of









OEP2(sc) is to develop TDDFT from the same principles, which will be pursued in future

work.

Table 3-3. Excitation energies of Ne atom using exchange-only orbital energies
Term TDDFT-exchange-only TDDFT OEP-MBPT(2) Exp[29].
3P 17.1748 17.0438
1p 17.5209 17.1225 16.847
D 19.8681 19.8920
D 19.9181 19.9262 18.703
3P 19.9134 19.9134
1p 19.9134 19.9134 18.725
3S 19.8811 21.2988
1S 2, 21.3050 18.965
3P 20.4905 20.4504
Ip 20.6090 20.4672


Table 3-4. Excitation energies of Ne atom using orbital energies and orbitals from
OEP2(sc). All equations for TDDFT are the same
Term EOM-CCSD B3LYP LDA TDDFT TDDFT- Exp[29].
[32] [32] exchange- OEP
only(sc) MBPT(2)(sc)
3P 16.5353 10.2805 13.3302 17.23 15.09
1p 16.7274 10.2725 13.3512 17.23 15.09 16.847
3 18.8303 14.8961 13.5076 19.74 17.56
D 18.8822 14.8949 13.5067 19.74 17.56 18.703
3P 18.9156 14.3188 13.4675 19.74 17.56
1 18.9791 14.4328 13.4700 19.74 17.56 18.725
3S 19.5118 13.9484 13.5077 19.74 17.57
1S 19.9323 13.9609 13.5057 19.74 17.57 18.965
3P 20.3747 15.0672 13.5077 20.40 18.29
Ip 20.5588 15.0712 13.5431 20.40 18.29


The cornerstone of OEP(sc) for exchange-correlation potentials is that we can

impose the condition that the KS determinant has to give the exact, correlated density.

Then this density condition provides equations that define V,, for a given functional[17].

Since no energy variational condition or functional derivative is used, we can change

the choice of Ho from the KS one to the semi-canonical (sc) one that offers greatly

enhanced convergence of perturbation theory. But then there is no apparent immediate

correspondence to a functional derivative as there is in the standard KS theory. A similar









condition to the density condition that would naturally lead to the analog of the kernel

is necessary to correctly use the OEP2(sc) method. Of course, in infinite order methods,

as opposed to perturbative approximations, there can be no difference between the two

approaches.

3.5 Conclusions

The TDDFT kernel for an OEP method that includes a second-order MBPT(2)

orbital-dependent correlation functional is derived. This derivation is made possible by a

diagrammatic technique familiar from many-body theory. The complexity of the kernel

for MBPT(2) is excessive, resulting in more than 200 diagrams that have to be evaluated.

This is the price paid to retain the one-particle structure of DFT and TDDFT yet

introduce rigorous, orbital-dependent correlation functionals even at second order.

Yet, despite the kernel's complexity it has been evaluated in a large basis set and

illustrated for Ne. As only the KS choice of Ho leads naturally to the kernel as the second

functional derivative of the orbital-dependent functional, ELc, the numerical result can

suffer from the fact that the functional E(0) + E(1) + E(2) is not bounded, and without

modification of Ho as used in [16, 17, 26], the radius of convergence is poor, causing poor

convergence[16] in determining the correlation potential. Nevertheless, the first step in

the underlying framework has now been defined to apply TDDFT with OEP correlation

potentials.

However, the complexity is great, though imposition of additional rigorous conditions

might result in simplification of the kernel. Lacking such simplifications, the question

arises as whether this is a case of diminishing returns to evaluate some 203 diagrams to

retain the attractive, one-particle, correlated structure of DFT? First, the answers for Ne

are not very good at the OEP-MBPT(2) level. Furthermore, the basis set dependence

of OEP methods when done in gaussian basis is severe to even get the V,, right[16].

The failure of most such OEP calculations to satisfy the exact HOMO condition <

HOMOIKIHOMO > < HOMOIVHOMO > is a case in point[16, 26].









Finally, there is no doubt that the standard equation of motion coupled-cluster

methods (EOM-CC)[31]) are less time-consuming than is the present calculation. As long

as the kernel for a rigorous orbital-dependent correlation potential is this complicated,

two-particle wave function theories like EOM-CC are both superior and easier to do. But

before we can further exploit the interface between DFT and wave function theory to the

benefit of both, it is requisite to have the orbital-dependent kernel defined. This, and its

initial evaluation, is what this chapter offers.









CHAPTER 4
AB INITIO DENSITY FUNCTIONAL THEORY FOR SPIN-POLARIZED SYSTEM\ IS

4.1 Theory

Consider a one-particle density operator in second-quantized form


() = < v~pp(r- rl)|lq > ajaq (4-1)
P,q

Using Wick theorem, we rewrite this operator in normal form with the Kohn-Sham

determinant as the Fermi-vacuum:

ptr) < p\(r ri)\pq > {a+aq}+ < K)KSl6(r rl)l)KS >
P,q
Nocc
S< Op, (r rl)lq > {a+aq} + y (r)2 (4 -2)
p,q i 1

The second term is pks, hence, the first term will be called the /. ,'.:1'; correction:


Ap(r) = < pp|6(r rl)|yq > {a+a,} (4-3)
p,q

Since the converged Kohn-Sham scheme gives an exact density, all corrections to this

density must be equal to zero. Hence, if we construct an effective operator of the density

using MBPT, the correction to the density must vanish in any order[17, 24]. The first and

second orders of the MBPT density condition can be written in the following way


< )Ks l (1)+Ap(r)2)K > (Q(1)+Ap())Sc + (Ap(r)21))Sc 0 (4-4)


< )KSI(Q()+ + 2(2)+)Ap (1) + 2(2)) KS

= (Q(A)+p(r))se + (Ap(r)(1))e +

+(-(2)+Ap(r))Sc + (Ap(r)(2))Sc + (Q(1)+Ap()(1))Sc 0 (4-5)

Using the KS partitioning of the hamiltonian[13] and taking into account the relation fpq

p6,pq+ < (|pIVHF V(p) Vclpq >, it is possible to derive the same expressions for










the exchange and correlation potentials as a functional derivative of E" + E(2) [13, 33].

However, this is a numerically hopeless procedure without any resummation of terms: (1)

it suffers from adding a large diagonal term into the perturbation; (2) MBPT(2) is not
(2)
bounded from below and any effort to use the variational condition 6E is likely to suffer

from numerical problems.

Our approach is to correct the above by changing the choice of Ho and focusing on

the expectation of the density operator instead of any variational functional derivation.

The partitioning of the hamiltonian is chosen to be the one-body part as the zeroth-order

approximation

a, 3
Ho0=^ fiI{aIa+, } + Y fbbatabh} + fij{aaa} + f abab aataa} (4-6)
a i b i'j a4b

However, to avoid dealing with non-diagonal Ho, a semi-canonical transformation will be

performed to obtain the more convenient zeroth-order hamiltonian,

a,l3
Ho fpp{ ap} (4 7)
( p

Dropping the ~ for simplicity, with the semi-canonical partitioning of the hamiltonian the

OEP exchange and correlation potentials assume the following form


V(r) -2 dr1 < ijlja > i(rl)a(r)( l(, (4-8)
Sijc fiir faa 1(


a,3
V, f Jdr l(25 < 1,jTlabr > (< aUbTlc,jT > < aobrljTc, >)(Pi(ril)(Pco(rl)
i,j,abc( + fjj faac fbbr)(fia fc)
S i,j,a,b,3
_- K < ij|Ta UbT > (< ijlkbT > < jiklb >)(Pkc-(Fl)oPao(1)
-2y Y < > (< >
Si,j,k,a,b iia jjf r faaa fbbT)(fkkc faaa)

-, K < ij, iab > (< aUbIkj, > < braUIkj >)*kaC. (1)i(r1)
Sijk,,b fii ( + fjj faaa fbbr)(fkk + fjjr faaa fbbr)










V < ij a,7br > (< ijrT Cbr > < jrTi lca >)a o(ri)(c(ri)
Si,j,a,b,c ii + fjjr faa fbbT)(fiia + fjjr fcca fbbr)
S(fibf)ia(fOi(r(1))Oa(r1) Y fajafijaic(rI1) a(rI1)
(fiiu faaa)(fiia fbbu) faaa) [jja -aaa)
a,b,i ij,a
f -A fab (r1)ab(1) V fifaj ,(i)yjc(1i)
a bto faaoi(f)ii fbb-) att faa() fjjc faac)
abi i, j-a
,V f&ja(< '.:l' > + < aab~,>|i >)Otc(1)lo)ac( r1))X1() (4 9)
tj*a b n faao)(fjjjcr fbbc)

where

X(r1,r) 2 c(l)ac(l)ic()ac(F) (4-10)

Exchange and correlation potentials, given by equations (8) and (9) must be substituted

into the Kohn-Sham equations on each iteration, until the fully self-consistent solution is

obtained. The method is implemented, using gaussian basis sets for orbitals and exchange-

correlation potentials as described in references [13, 17].

4.2 Results and Discussion

4.2.1 Total Energies

Ground-state energies are calculated for several systems with results presented in

Table 4-1. The uncontracted ROOS-ATZP basis is used for the Ne atom, while the rest

of the atoms and molecules are calculated using the uncontracted ROOS-ADZP '. -i-[27].

All the molecules and their ions are considered to be in the equilibrium ,.ill. l ii [ :1.] of

the corresponding neutral system, except when potential energy curves are considered.

For ab initio DFT calculations exchange-correlation potentials are used from [13] and

equations (4-8)-(4-9). For the comparison energies obtained from KS DFT with the

Perdew- Burke-Ernzerhof[34] (PBE) exchange-correlation potential and coupled-cluster

with single, doubles and perturbative triple excitation[35] (CCSD(T)) are also shown in

Table 4-1. Total energies, calculated with OEP-MBPT(2), based on the density condition

approach with the semi-canonical partitioning are much closer to CCSD(T) then OEP-

MBPT(2) with the KS partitioning. The latter greatly overestimates correlation energy









and frequently diverges. Since the density condition approach is completely equivalent [37]

to the variational OEP-MBPT(2) for the KS partitioning it has the same problems.

However, in the case of the non-variational, semi-canonical density condition approach, the

unboundness from below is less of a problem. The semi-canonical choice of Ho provides

good approximations to the energy and wave function in MBPT(2) and the corresponding

total energies are much closer to highly-accurate CCSD(T) ones, at least in the chosen

basis. The computational cost of the OEP-MBPT(2) method is comparable with the cost

of MBPT(2), the scaling of both methods is N5, where N is the number of basis functions.

This is more expensive than conventional DFT (scales like N3), but less expensive then

CCSD(T), computational time of which is proportional to N7. The correlation potential,

Table 4-1. Total energies
MP2 OEP-KS OEP-semi PBE CCSD(T)
N2 -109.449149 -109.7,11 11, -109.457779 -109.457071 -109.468618
N+ -108. :-. ;;, no conv -108.905444 -108.890773 -108.900556
N -54.544740 -54.593111 -54.545199 -54.535569 -54.5 !-,4
H20 -76.370003 -76.510744 -76.373092 -76.369991 -76 ;"7
H20+ -75.901155 -75.989525 -75.902958 -75.912146 -75.921765
CN -92.598196 no cony -92.651919 -92.646898 -92.658287
CN- -92.772714 no cony -92.780718 -92.783715 -92.796498
CO -113.228520 -113.511203 -113.237817 -113.239593 -113.251239
CO+ -112.702208 no conv -112.734813 -112.729947 -112.739892
02 -150.205776 no conv -150.221870 -150.255051 -150.221560
Ne -128.859598 -128.951437 -128.861175 -12.> .-.844 -1 '.- -,133
Ne+ -128.066796 no conv -128.067574 -128.068711 1- NI IIII'
Mean error, 58.22 26.74 36.16
kJ/mol


calculated with the semi-canonical OEP for the Ne atom[16] is close to the corresponding

quantum Monte-Carlo potential, and we can expect that for other systems, where QMC is

not available, semi-canonical OEP-MBPT(2) potentials should be a good alternative. For

the open-shell case potentials produced by Kohn-Sham OEP show an overestimation of

the correlation energy, the same situation previously reported for the closed-shell case[16].

(Fig 4-1 and 4-2). The correct -1 long-range ..-imptotic behavior of the OEP-MBPT(2)












exchange potentials (Fig 4-1 and 4-2) is guaranteed by the use of the Colle-N. -1 et seed

potential algorithm[30]. In all our calculations a Slater potential[38] was used as a seed


potential, since the Colle-N. -1 i algorithm requires the seed potential to be as close as


possible to the actual potential. Two separate Slater potentials, one for alpha and the

other for beta, were used. For all the closed-shell systems from Table 4-1, the numerical


- OEP-semi alpha
- OEP-semi beta
- PBE alpha
- PBE beta
-1/r


2 3
R,a. u


Figure 4-1. Exchange and correlation potentials of Li atom (radial part). A) Exchange
potential. B) Correlation potential


R, a. u.


-1 -

-2-

-3-

-4-


R, a. u.


-6 -4 -2 0 2 4
R, a. u.


Figure 4-2. Exchange and correlation potentials of 02 molecule across the molecular axis.
A) Exchange potential. B) Correlation potential



results obtained from OEP-MBPT(2) with the semi-canonical potential and from DFT









with the PBE potential are numerically close, and close to the reference CCSD(T)

energies. This, of course, is part of the power of GGA and meta DFT, even when such

energies are obtained with highly erroneous potentials. Instead, PBE partly benefits from

the cancellation of errors in V, and V, (Fig 1 and 2). For most of the open-shell systems

the energies of the semi-canonical OEP-MBPT(2) are much closer to CCSD(T) then the

corresponding PBE energies, which is also a consequence of the wrong behavior of the

PBE potential.

4.2.2 Ionization Potentials

Orbital energies and the corresponding ionization potentials are much more sen-

sitive to the exchange-correlation potential than the total energies. Vertical ionization

potentials, calculated as energy differences between the neutral and its ionized system are

calculated using the same basis sets and geometries as in the previous section. Results

are presented in Table 4-2. In exact DFT, the energy of the highest occupied molecular

Table 4-2. Ionization potentials (in e. v.)
HOMO, OEP-semi AE, OEP-semi HOMO, PBE AE, PBE Exp
Ne 21.01 21.59 13.35 21.69 21.56
N2 16.89 15.03 10.27 15.41 15.58
CO 13.68 13.69 9.05 13.87 14.01
CN- 4.19 3.51 0.149 3.72 3.86
H20 12.37 12.73 7.08 12.46 12.62
N 15.25 14.52 8.30 14.73 14.53
Li 4.89 5.38 3.23 5.59 5.39


orbital (HOMO) corresponds to the negative of the exact vertical ionization potential. In

the case of the semi-canonical OEP the HOMO energy is close to the corresponding AE

values, what can be explained by the correct shape of the exchange-correlation potential

(Fig 4-1 and 4-2 and ref [16, 17]), while PBE does not exhibit correct behavior of potential

and thus, fails to reproduce the correct HOMO energy values. Failure to reproduce the

correct HOMO energy causes the incorrect integer discontinuity of the PBE (or any other

GGA) exchange-correlation potential[39, 40].









4.2.3 Dissociation Energies

Dissociation energies, calculated with semi-canonical OEP and PBE are presented

in Table 4-3. Potential curves for the LiH, OH and HF molecules are shown on Fig 4-3,

4-4 and 4-5. For all curves the dissociation energies have approximately the same level of

accuracy as in the MBPT(2) case, but the semi-canonical OEP-MBPT(2) improves the

shape of the curves.

Table 4-3. Dissociation energies (in kJ/mol)
PBE OEP-semi Exp.
N2 1013.27 964.56 941.64
02 598.14 549.12 493.59
CO 1122.16 1122.00 1071.80
CN 821.151 794.51 745.01


4.2.4 Singlet-Triplet Separation in Methylene

Results for the extensively studied splitting of singlet and triplet states of methylene

[41, 42] are reported, using the uncontracted ROOS-ADZP basis sets: 13s9p3d for carbon

and 8s4pld for hydrogen. Equilibrium geometries for both states are taken from reference

[41].The energies for the two states, calculated with different methods are presented in

Table 4-4. There are significant differences in the energy separation compared to the

experimental value (8.998 kcal/mol) for all of the presented methods. This is a basis

set issue, but can also be explained by the fact that the singlet state of methylene has

a significant contribution from two-determinants[41], so MBPT(2) is a much poorer

underlying approximation for their difference than infinite-order CCSD or CCSD(T), and

especially the two-determinant CCSD (TD-CCSD) results[41]. Where the absolute values

of PBE energies are not too good for the two states of CH2, the difference is consistent

with MBPT(2) and its OEP-generated ab initio dft form.

4.3 Conclusions

Ab initio dft calculations with OEP-MBPT(2) semi-canonical potentials show

significantly improved results over OEP-MBPT(2) with the Kohn-Sham partitioning of









Table 4-4. Singlet and triplet energies of methylene
3B1 1A1 AE, kcal/mol
CCSD(T) -39.127004 -39.109751 10.826
CCSD -39.123412 -39.104569 11.824
OEP semi -39.103994 -39.079152 15.588
PBE -39.108321 -39.083448 15.610
MP2 -39.102565 -39.077674 15.619


Ho. In all cases, when the Kohn-Sham partitioning scheme fails to converge, the semi-

canonical version had no problem with convergence. The corresponding HOMO energies

are close to the correct ionization potentials for neutral systems and electron affinities

for anions, and they correct the integer discontinuity of standard exchange-correlation

potentials. In the case of open-shell systems, semi-canonical OEP-MBPT(2) also shows

improvement in total energies over PBE results and produce more accurate atomization

energies. Finally, correlation potentials, produced by the semi-canonical OEP-MBPT(2)

method, can be used as the first of a sequence of highly correlated methods that provides

a reference for the construction of density-dependent functionals.

































------- OEP-semi
-7.4- -UHF
----UHF
-...... MBPT(2)
= ...... CCSD
cd -7.6- \> CCSD(T)



-7.8-



-8.0-



-8.2
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
R,A



Figure 4-3. LiH potential energy curve.
































------- OEP-semi
----UHF
...... MBPT(2)
------- CCSD
- CCSD(T)


0.5 1.0 1.5 2.0

R,A


Figure 4-4. OH potential energy curve.


1'
1I
1'


-74.6-

-74.7 -

-74.8-

-74.9-

-75.0-

-75.1-

-75.2-

-75.3-

-75.4-

-75.5-

-75.6-

-75.7-
-7C C
































----UHF
...... MBPT(2)
------- OEP-semi
------- CCSD
-- CCSD(T)


1.0 1.5 2.0
R,A


Figure 4-5. HF potential energy curve.


-99.3 -

-99.4 -

-99.5 -

-99.6 -

-99.7 -

-99.8 -

-99.9 -

-100.0-

-100.1-

-100.2-

-100.3-

-100.4-


2.5
2.5









CHAPTER 5
EXACT-EXCHANGE TIME-DEPENDENT DENSITY FUNCTIONAL THEORY FOR
OPEN-SHELL SYSTEM\ IS

Kohn-Sham (KS) density functional theory[5, 6] (DFT) is a widely used method for

ground-state energies and other properties. The time- dependent extension of Kohn-Sham

DFT, whose rigorous foundation has been questioned [43], is a very popular method

for the description of excited states energies, since it can often provide accurate results

by the diagonalization of a single-excitation dimensional matrix as in time-dependent

Hartree-Fock or mono-excited CI. However KS DFT and TDDFT have their natural

limitations. Standard functionals suffer from an incomplete cancellation of self-interaction

terms in the KS equations and incorrect long-range .,-i-ii!!il ic behavior of the potentials.

Both can be particularly troublesome for excited, particularly Rydberg states, and

ionized states[16, 21, 48], making it difficult to get systematically improvable results from

standard TDDFT.

Exact-exchange density functional theory is based on the optimized effective po-

tential method introduced by Talman and Shadwick[15]. In this method the exchange

potential is defined as the functional derivative of the non-local orbital-dependent ex-

change functional from the Hartree-Fock method (EXX), ie 6Ex/6p(1) = Vx(1). The OEP

method is free from the self-interaction problem, it exhibits the correct ..i-mptotic

behavior and its highest occupied molecular orbital (HOMO) energy satisfies Janak's

theorem[44]

< h|V\h >= < h|Klh > (5-1)

where K is the non-local Hartree-Fock exchange operator. Recently the exchange-only

OEP method has been formulated in several, not alvb-- l consistent, or even formally

correct v- li--[16, 22, 45-47], and these alternative approaches can suffer from numerical

problems. However, when evaluated with a proper treatment of the density-density

response matrix X, most of the numerical problems that have been encountered are

resolved.









Our preference in the development of such orbital dependent functionals and

potentials is to start from the density condition that is fundamental to KS theory, as the

single determinant must provide the correct density for the exchange-correlation problem.

Once this is enforced for a given functional, Vxc is uniquely determined up to a constant.

This is the origin of the perturbation method of Goerling and Levy[14], but essentially

modified by Bartlett et al [16, 17] to avoid the failures of a simple sum of KS one-particle

Hamiltonians as the unperturbed problem, to generate correlation potentials. This is the

cornerstone of ab initio dft. The conceptual difference is that the density condition does

not explicitly use the variational determination, 6Exc/6p(1) = Vxc(1). This difference is

of critical importance in generating correlation potentials from low-orders of perturbation

theory. Without that change, no low-order orbital dependent correlation functional like

that from MBPT2 will generally work, but with those changes, it does very well [16, 48].

Those modifications to the perturbation theory also pertain to the exchange-only case, the

subject of this paper, but for that problem the distinctions are less important[16].

Once following this approach, which starts with the KS choice of Ho, the distinctions

between applying the density condition and using direct functional differentiation is

more conceptual than essential, as there is a correspondence in any order of perturbation

theory[24, 37].

The direct optimization procedure advocated by some, built upon the above varia-

tional determination of the functional derivative, though formally equivalent, differs from

the original, X-based OEP in the details of implementation. However, as was pointed

out by Staroverov et. al [45] under certain combinations of molecular and auxiliary basis

sets, particularly when the latter's dimension is larger than that for the former, the direct

optimization method can give the Hartree-Fock energy and density. The latter can be

viewed as a trivial solution, as it can be shown to correspond to the solution of a weighted

least-squares expression. On the other hand, the X-based OEP with proper handling is









quite stable, and does not suffer from the problems alluded to by Staroverov, et.al, even

when the auxiliary basis has a dimension greater than the MO basis.

In another interesting example, Rohr, et al[49] observes the occurrence of a degen-

eracy in performing OEP calculations, and to investigate this, intentionally construct a

problem where the HOMO and LUMO orbitals of the KS problem are degenerate. Subject

to direct minimization, they proceed to report on the failure of the OEP procedure. On

the other hand, it is shown in this paper that the original X based approach handles this

problem, too.

The exchange-only time-dependent OEP has been considered by Goerling[20] and

implemented by Hirata et. al[21, 50]. However, all results to date are for closed-shell

systems. To further address the exact local (OEPx) exchange versus time-dependent

Hartree-Fock (TDHF), TDOEPx is generalized to treat excited states for open-shell

species. Results from adiabatic TDOEPx are in good agreement with TDHF for both

excitation energies and polarizabilities, however, it is shown that charge-transfer states

cannot be properly described in TDDFT. For standard DFT methods, gradient-corrected,

hybrid, etc. this has been noted, and improved upon[51], but standard methods, unlike

TDOEPx, still suffer from other limitations, like the self-interaction error and the incorrect

long-range behavior of the potentials and kernels. In an approach that gives the 'right

answer for the right reason' these exact conditions are requisite. In TDOEPx there are no

such errors. Hence, failures of TDOEPx compared to TDHF have to be exclusively due to

the local versus non-local exchange operator, which is demonstrated.

Finally, C6 coefficients are obtained for open-shell systems from the TDOEPx

freu iiv.- -dependent polarizability.

5.1 Exact-Exchange Density Functional Theory

The most satisfactory way to introduce exact exchange DFT is to insist that

the Kohn-Sham single determinant provide the exact density, but subject to a local,

multiplicative exchange operator, Vx(1), instead of the usual non-local Hartree-Fock









exchange,


7013 P13
K(1) (1) 3) 3) 1) (3) r1tp(3)d3j(l1) (5-2)
J J

where the {Qj}are the occupied Kohn-Sham spin orbitals, p=i,j,k,l.... The latter are the

solutions to the equations,

..,/( (1) CpOP(1) (5-3)

hsl') = h(1) + J(1) + Vx(l) (5-4)

Ji() E I (2) 1 (2)d2 (5-5)

p(t) d2 (5-6)
I 12
PKS () i () (5-7)

where the unoccupied orbitals are indicated by p=a,b,c,d....

The condition that the density, p(l), be the Hartree-Fock one, is that


l, (HFlh+ J- K aHF) 0 (5-8)

PHF(l) -I *( (1) (5 -9)

which, as is well-known, is correct through first order in correlation measured relative to

the sum of HF one-particle operators, Ho = Y f(i), due to the Moeller-Plesset theorem.

For KS orbitals, fia is not zero. Hence, if the objective were to maximize the similarity

in the KS and HF density, it would then require minimizing a quantity composed of KS

orbitals, related to


min |\(if- 1sa)|2 = min1 (iK + x a)2 (5-10)
i,a i,a

,---.I -:i.-; a kind of least-squares minimization as has been discussed previously[13, 16, 17,

48]. Clearly, this minimization would have the trivial solution that (i K a) = -(i Vx a).









However, there are some other pertinent considerations before doing so. The

condition that the KS single determinant gives the exact density through first order

in its correlation perturbation, obviously imposes the condition that p' 0. This

density condition is very general as it applies to exchange and correlation, and has

been used quite successfully in defining ab initio dft exchange and, in higher orders,

correlation potentials[13, 16, 17, 48], since it also transcends any particular separation of

the Hamiltonian in perturbation theory. In particular, it is not subject to HKS = hs(i),

and thereby alleviates the pathological behavior that gives no convergence when defining

such correlation potentials from MBPT(2), for example [13]. For the exchange only,

however, such effects are less important[16, 17]. Therefore this paper will use the sum of

Sham Hamiltonians, but the principle for derivation remains the same.

Hence, requiring that

0 PKS (KS ARV \KS) + (KKS VR0 I KS) (5-11)

A p(x, x- ) (5-12)
a \ + lab W-ab\
Rl C Ca) 1 + 2 )(CtI + C Ca Cb) ( +... (5-13)
i,a I
I|h)R(h|l + |h2)Rl(h2 +... (5-14)

V= -[ (i)+ (i) (5-15)
i
Single excitations in the resolvent operator are indicated by a) ,and collectively by the row

matrix, |hi),doubles by b), h2),etc. It is easy to show[17] that only the singles contribute

to the first-order density, and


Ps y i I )fia +c 2 k(1l)a(1)iK + x|a)/(i )a) 0 (5-16)
?,a i,a

Then this condition defines the Vx(1) operator up to a constant[52]. Note that this

is a pointwise condition as it should equal zero at all xi, but actually imposes the weaker









condition, discussed in [52]. In addition to its pointwise character represented by the delta
function, it differs from the above least-squares form by the presence of the denominator.

The usual way to write the above equation is to introduce the non-interacting density-

density response function,

X(1,2)- =V (1l). (1)O(2)>P(2)/(e e.a) (5-17)
i,a

which then gives

f 1, 2) [K(2) + (2)]d2 Y 1)a(1) J (2) [K(2) + Vx(2)] (2)d2 OVx
J Ci i 6a

W= i(1) / (2)[K(2) + Vx(2)] (2)d2 OVxi(5-18)


This is the usual optimized-effective potential procedure of Talman and Shadwick[15, 22],

geared toward representing the Vx operator on a numerical grid. The weight factor,

wit(1), is also introduced. Note the matrix elements can be written in configuration space
as


(Vx)a (ajVxli) (5-19)

Vx (hl|Vxl0) (5-20)

(K)ai= Y aj, j (5-21)

K= (h KI0) (5-22)

Also note that the response function can be written as

X(1, 2) (O|A|hil)R(hi|A|O) aiRa < 0 (523)

which makes the point that, for the ground state, X(1,2) is semi-negative definite since R1

is.









To make the transition from an infinite basis to a finite computational basis, one

introduces the matrix representation of the OEP equation. Eqn (5-18) then becomes,


w(Vx + K) 0


(5-24)

(5-25)


w = aiR1


Using weighted least squares


6Vx[) :,ia < ilVxa > +Kia22] 0


(5-26)


gives the OEP equation, Eqn (5-16).

As is customary in practice, the computational form for the OEP equation can be

obtained by twice applying the one-electron projector,


0(1) = nr] >< rn\rl >'< r]1


where | r) defines an auxiliary basis, respectively, for electron 1 and 2. Without restriction

the basis can be assumed to be orthonormal The so-called outer projection[53] of the

response function is then


X,, = (ai )(iav) /(ci ea)


(5-27)


Using this expression and insisting upon the linear independence of I ry) gives the OEP

equations in the form,


< l|X(1,2) >< l|Vx(2) + oK(2))

XVx

Y,


Vx


OVx1

Y


S (pai) (ijla)
S(i ja)

X-1Y


where the o indicates the product of X and K in Eqn (5-18) that results in the Y column

matrix.


(5-31)









In practice, the matrix X is explicitly non-singular in a finite, linearly independent

basis[52], unlike the infinite basis analogue of X(1,2), which has to be singular[52, 54].

The latter follows because the infinite basis cannot account for the arbitrary constant.

However, in a finite basis, X is ill-conditioned because the null-space of functions orthog-

onal to Vx can ah--lv- be added to Vx but with zero weight, and this requires that the

OEP equations produce a computed, numerical zero. This makes the equations difficult to

invert, recommending a singular-value decomposition (SVD) procedure. This SVD should

not be confused with the need to exclude linear dependency in the computational basis,

which can alv--iv- be achieved independently; but that, too, would be accommodated in

practice by the SVD procedure.

The role of X is to impose the point-wise closeness of K and Vx, and, as such, it is

critical to the determination of OEP potentials. In the absence of X, as attempted by

Staroverov, et al.[45] unphysical results can be obtained. According to their procedure

[45], to get HF energies and densities from the OEP method (i. e. to find the trivial

solution of Eqn(5-26) ), the number of auxiliary functions used must be greater or equal

to some number, N1i,, which is the number of non-zero Kia matrix elements. In this case

equation(8) of ref[45] will have one or more solutions.

To illustrate, the use of the response matrix approach, we report in Figure 5-1 the

results for the Ne atom exchange potentials using the three basis sets used by Staroverov,

et al, plus a fourth uncontracted Roos basis that we prefer. The auxiliary bases are

the same as the molecular ones and in all cases the number of auxiliary functions is

larger than the corresponding N1mi, necessary for the solution of equation (8) of ref [45].

The total energy values from Table 5-1 and corresponding potentials from Fig 5-1 are

completely opposite to the results of ref[45] because of the broken pointwise closeness

and the absence of the self-consistent solution. Note that once the pointwise closeness is

imposed via X, and a self-consistent solution obtained, there is no problem for the proper

SVD-based OEP procedure.











1 -



-2 I! ---AUG-CC-PVTZ
S-3 -/.-AUG-CC-PV5Z

-5-
6 -r

-

-9-
CZ
-1-- ....AUG-CC-PV6Z
















HOMO
Basis Ng,,, Nmiz E(OEP) E(ref[45]) OEP HF
AUG-CC-PVTZ 17 9 -128.533191 -128.533273 -33.58 -23.16
AUG-CC-PV5Z 25 15 -128.545887 -128.546786 -14.01 -23.14
AUG-CC-PV6Z 25 18 -128.5!n.-'7 -128.547062 -17.80 -23.14
ROOS-ADZP 61 39 -128.545016 -128.546596 -23.74 -23.14
-4-

-8-




R, a. u.









Inspired by Exchan observation of Jiang[55]e atom obtained HOMO and MO degeneracy arising








in OEP, Rohr et al[49] offered a specific combination of molecular and auxiliary basis,

that showed their effect for He, then using the direct optimization algorithm produces

degenerate HOMO and LUMO orbital energies. We performed OEP calculation for the

He atom using the optimized auxiliary basis from Table 1 of ref[49], which caused their

degeneracy, but instead of using the direct minimization algorithm of Broyder-Fletcher-

Goldfarb-Shanno (BFGS) the solution of the OEP equation is solved by the inversion of

X with the SVD procedure. These results are presented in Table 5-2, demonstrating that
Table 5-2. Total (in a. u.) and orbital (in e. v.) energies of He atom
HOMO














Basis N N OEP HF OE P2-[45 S OEP2EP H
AU--PVTZTotal energy, a. u. -2.86115365 -2.86115365 -2.9052473 -33.58 -23.
AUG-CC-PV5Z 25 15 -128.545887 -128.546786 -14.01 -23.14













AUG-CC-PV6ZHOMO energy, e. v. -24.97028.5 -24.970 -24.529 -24.787
ROMO energy S-ADP 6 3 .591 15.561 28.546596 -23.74 -23.618
Inspired by an observation of Jiang[55] of HOMO and LUMO degeneracy arising

in OEP, Rohr et al[49] offered a specific combination of molecular and auxiliary basis,

that showed their effect for He, then using the direct optimization algorithm produces

degenerate HOMO and LUMO orbital energies. We performed OEP calculation for the

He atom using the optimized auxiliary basis from Table 1 of ref [49], which caused their

degeneracy, but instead of using the direct minimization algorithm of Broyder-Fletcher-

Goldfarb-Shanno (BFGS) the solution of the OEP equation is solved by the inversion of

X with the SVD procedure. These results are presented in Table 5-2, demonstrating that

Table 5-2. Total and orbital energies of He atom
OEP HF OEP2-KS OEP2-sc
Total energy, a. u. -2.86115365 -2.86115365 -2.90524738 -2 '--I I%-C.8
HOMO energy, e. v. -24.970 -24.970 -24.529 -24.787
LUMO energy e. v. 3.591 15.561 3.671 3.618









the X based OEP have no problem even for this situation.

As seen in Table 5-2, even the notoriously unstable second-order OEP procedure,

based on HOK has no particular difficulties, either. Since the direct minimization proce-

dure of Yang and Wu[46] is based on the same equation as the regular OEP method, all

of the problems, described in ref [49], are caused by using the BFGS algorithm instead

of solving the OEP Fredholm equation with the SVD procedure for the X. That might

appear to be surprising since Ch = l, would mean that the resolvent operator is manifestly

singular, clearly not that appropriate to a non-degenerate perturbation theory functional,

which has alv--,i been the intent of such orbital dependent expressions. However, it is also

apparent from the standard, chain-rule differentiation used by Rohr et al[49], namely


Vxc(1) Exc/6p(1) = d2d3[6Exc/5,p(2)][6p(2)/6v,(3)][6vs(3)/6p(1)] (5-32)


that the derivative would not be defined from the beginning as 5Exc/56p(2) has to be

singular when cHOMO = LUMO. For comparison, the OEP2sc result from ab initio dft is

also shown, compared to the normally, poorly converged OEP2 based on the HKS

Other simplifications in the OEP solution are potentially possible. The most obvious

is to make an Unsold, constant energy denominator approximation(CEDA), replacing i -

a C v,9.(This should be contrasted with that of Kreiger-Li-Iafrate (KLI)[44] who made
such an approximation in the Green's function where the denominator consists of Cp Cq

before simplifying the from to the occupied-virtual separation invoked here[16, 54, 57].)

See Gritsenko, et al [56]. Such a constant energy denominator is a relatively painless

approximation here, since Cav appears on both sides of eqn. (18), making it disappear

from the equation for Vx. Once that is done, a resolution of the identity can be invoked to

eliminate the virtual orbitals to give,


(ip la)(a|K + Vxi) ({(ip(K + Vx)i (iplj)(-|K + Vxi) (5-33)
i,a i ij









In this way, there is no dependence on the virtual orbitals as is deemed computationally

or formally important in some DFT circles[59] Of course, using the usual finite basis set

computational tools of quantum chemistry, the distinction is less important. This formula

has recently been applied[58].

Because the OEP equation is a point-wise identity having to be zero formally for all

x in a finite basis set it corresponds to a :r, ii -I-.,-few mapping in the general case. The

dependence upon the delta function makes the OEP procedure sensitive to the auxiliary

basis set used in the calculation. However, the SVD procedure handles the null space

functions plus any potential linear dependency in the basis set in a fairly automatic

v ,i-[;n]. If the X matrix is removed, this is not the case, and some algorithms can result

in unphysical or trivial solutions under certain combinations of auxiliary and molecular

bases. C'!....- i to invoke the average energy denominator and the resolution of the

identity A'. f. the least-squares mminimization, will not give Eqn(24). However, once

we eliminate the delta function that makes the approximation point-wise, we reduce the

problem to the minimization of the variance


min Var min (i(K + Vx)2i)- j iK + Vxi)(jlK + Vxli) (5-34)
i i,j

that can be used to define a Vx but one that is not point-wise, and, consequently, does

not satisfy all the conditions above. This further simplification of the weighted least

squares approach in Eqn (5-26) has also been considered recently [58]. Ultimately, one

primary criteria for the best exchange-only potential should be the satisfaction of the

Janak theorem. As shown in Table 5-1, and pointed out previously[16, 48] this is difficult

to achieve in any normal basis set.









5.2 Time-Dependent Optimized Effective Potential

5.2.1 Theory and Implementation

Within the adiabatic approximation the TDOEP equations have the following matrix

form (5-35)
A B U U
A BA (5-35)
-B -A U+ [U+

The matrices A and B have the following structure (5-36)

AI = 6ijSab6iaaT ia)+ < ab,- liJ > + < anbT fc- li, >

B' = < asb.-i1j. > + < abrlf,,|ijj, > (5-36)

where

6VOEP
f." = 6,, P (5-37)
Jp,
< abl f, ij,> >= 6T < aia > (X1),KA(ga)A,(X-1),, < bjav >(5-38)
K,A,pL,V

with

(Kbaa + VEP) < iaA >< ibp >
iab -Ei E)(. aE Eba)

+2iab
(Kjau + Vf4EP) < boaUA >< ab > KU >
25 ia Faa) i)ja baa)
-2 (Kjia + IfEP)(< baA >< ibp > + < i1bA >< baip >)
,, (ia a)(Eia ba)
+ OEP)
i,j,a ( -i c- o7) (Fj 7- Fa,)

-2 Y (< ijlbeaU > + < ia, ljb, >) < iaA >< jb, > (5 39)
2 Z ^-----( v ------ (5-39)
(Ei7 Eac (Evj Eba)
i,j,a,b (b a)(F bT)

Once the eigenvalue problem (5 35) is solved, we have the set of excitation energies, w.









When we need to calculate polarizabilities at frequency u, the following system of

linear equations (5-40) should be solved

(A A B d"- h
h) (540)
B A + ul d`+ h

where

hpq = / dr(pp,(r)ipyq,(r) (5-41)

Once these equations are solved for d, the dynamic polarizability is readily evaluated from
a, 3
a-(w; +w) 1- Z (h jh d + h,,d) (5 42)
i,a a

For the calculation of the C6 van der Waals coefficients, the system of equations (5-40)

should be solved for the imaginary frequencies iu from which the coefficients can by

calculated by
3 r[-
C/0 \ dwa(iw)a(iw) (5-43)
T Jo
where a(iw) = ((awxx ) + ,yy(iw) + a (iL)). Integration of equation (5-43) was carried

out by Gauss-('l. li--!i. v quadrature.

To ensure the correct ..i- mptotic behavior of the exchange potential, the

Colle-N. -bet algorithm[30] was used. As a seed potential, we used the one proposed by

,i ,, i [: ]

V later(T) jo dr (5 44)
,j P, Ir rll

For the spin-polarized case two different Slater potentials are used for the corresponding

spin component of the local exchange.

5.2.2 Numerical Results

Excitation energies and polarizabilities are calculated for several open-shell systems.

Equilibrium geometries are taken from ref[36]. For all calculations the uncontracted









ROOS-ADZP basis set[27] has been used. Results for excitation energies are presented in

Table 5-3.

Table 5-3. Excitation energies (V valence state, R Rydberg)
TDHF OEP EP TDOEP SVWN CCSD Exp
CN
2I(V) 4.18 8.16 4.91 1.96 1.52 1.32
2Z+(V) 5.47 10.58 5.48 3.19 3.62 3.22
CO+
2II(V) 7.55 9.98 8.23 3.03 3.43 3.26
2Z+(V) 11.14 9.37 10.82 4.99 6.14 5.82
CH3
2A' (R) 6.52 6.53 6.42 5.00 5.88 5.72
2A'2 (R) 7.93 7.97 7.93 5.93 7.18 7.44
N
4P (V) 9.84 11.11 9.88 10.98 10.84 10.35
4P (R) 13.59 13.16 13.18 11.76 1 13.62

As follows from the first two columns of Table 5-3, TDHF and TDOEPx produce

results of approximately the same accuracy, as would be expected by the fact that the

local exchange in TDOEP is meant to be a kind of least-squares fit to the HF non-local

exchange potential. This certainly makes the occupied orbitals quite similar, but the

spectrum of unoccupied orbital eigenvalues is very different, as shown in Table 5-4, since

TDOEPx will generate a Rydberg type series instead of anything like the HF virtual.

As is well known, the latter are determined in a potential of n-electrons, making them

appropriate for electron attached states, while the occupied ones feel a potential of n-

1 electrons. To the contrary, the orbitals obtained in OEPx have the same potential

for an electron in the occupied and unoccupied orbitals, which is why the latter more

nearly simulate Rydberg states, as some of the unoccupied orbitals will have negative

orbital energies. Of course, in a finite basis as long as the space separately spanned by

the occupied and the unoccupied orbitals is the same, there would be no difference in the

results, though the diagonal values (but not the trace) would change. The significant



1 This Rydberg state was not obtained in the EOM CCSD Davidson diagonalization









Table 5-4. Orbital
OEPx
Is -839.67
2s -47.36
2p -23.74
3s(3p) -5.21
3p(3s) -3.11
4p 5.66


energies
HF
-891.78
-52.52
-23.14
3.66
3.94
16.08


(in e. v.) of Ne atom
EOM-CCSD
-871.11
-48.11
-21.28
3.70
3.62
15.66


discrepancies with the experimental data can be explained by the fact that electron

correlation effects are not introduced by either TDHF or TDOEP. To get some measure

of the correlation effect, coupled cluster (CC) results from EOM-CCSD are reported[61],

and the Slater-Vosko-Wilk-Nusiar (SWVN) functional is also used. The latter produces

good results for valence states, but the energies of the Rydberg states are underestimated.

This underestimation is a consequence of the incorrect .,-iii!!ll, ic behavior of the SWVN

potential, for which Janak's theorem is not satisfied (see Table 5-5) and is a well-known

characteristic that standard exchange-correlation potentials do not offer a Koopmans-

type approximation for the orbital energies. However for OEP exchange-only and with

correlation, it has been proven that the orbital energies offer a meaningful Koopmans

approximation to principal IP's for the valence and mid- valence states [16], so Rydberg

states should potentially be better described. Of course, definitive comparison would

require OEP with correlation as presented elsewhere[62], and this is seen in the OEP2-sc

orbital energies[16, 48], but a full TDOEP treatment of excited states[62] requires the

correlation kernel. This does not appear to be a viable route yet.

Table 5-5. Ionization energies (in e. v.)
HOMO AE
HF OEP SVWN HF OEP SVWN Exp
CN 14.16 14.61 9.79 16.22 16.32 14.80 13.60
CH3 10.46 11.19 5.38 8.98 8.99 10.09 9.84
N 15.53 16.14 8.41 13.89 13.89 15.00 14.53

Results for the polarizabilities and C6 coefficients are presented in Tables 5-6 and

5-7. In all cases, the static polarizabilities obtained with TDOEP are very close to the









TDHF ones. For both excitation energies and polarizabilities this closeness arises from the

fact that the OEP potential derived from the HF energy functional and its time-dependent

response properties are similar. Systematic overestimation of polarizabilities is also caused

by the lack of the -1 ..i-mptotic dependence. In most cases this causes an even larger

error than that coming from the absence of correlation effects in TDHF and TDOEP. The

anisotropy of the polarizability, though, is usually reproduced much better by SVWN.

This fact must be due to cancellation of errors, coming from the incorrect .--i,.iinl. l ic and

incomplete cancellation of self-interaction error, plus a proper accounting of correlation

effects.

5.2.3 Charge-Transfer Excited States

Within the framework of the standard TDDFT description of excited states the

transfer of charge between two separated fragments has severe difficulties. Detailed

analysis of this problem has been presented by Tozer[63] and Dreuw et al[51]. According

to the latter analysis [51] even if the exact exchange-correlation functional were known, it

would still be impossible to get a proper description of charge-transfer excitations within

TDDFT. Consider the example of He ... Be. The charge-transfer 'I state calculated with

TDHF and TDOEP, taken with the Tamm-Dancoff (mono-excited CI) approximation for

simplicity. For both atoms the uncontracted ROOS-ADZP basis set was used. Potential

curves of the 'I charge-transfer excited state are presented on Fig 5-2. The potential

curve, calculated with the configuration interaction singles method exhibits the correct

long-range ..i-mptotic behavior, while the TDOEP results have a qualitatively wrong

.i',i!1 ill. ic behavior. This phenomena is caused by the fact that for charge-transfer

excitations all integrals in Eqn. (5-36) are equal to zero and the excitation frequency is

equal to the difference of the LUMO orbital energy of the Be atom and the HOMO energy

of the He atom.

What was not addressed numerically in Dreuw et al[51] was the effect of having

a proper self-interaction and .i-', i-i, ll ic behavior in DFT, which can only be achieved














24-

23-

22-

21-

ai 20-

19-

18-

17-

16-


Charge transferred 'n state of He...Be
CIS
TDOEP-Tamm -Dancoff


R(He...Be), A


Figure 5-2. A 1I charge-transfered excited state of He ... Be



with OEP methods. We report those in Fig 5-3. Obviously, the orbital energy difference


--HF
-- OEPx
OEP2-semi
- EOM-CCSD


/


6
R(He...Be), A


Figure 5-3. LUMO-HOMO orbital energy difference


behaves very differently for OEP and HF since in both cases the HOMO orbital energy


corresponds to an ionization potential approximation, but the LUMO energy corresponds


to a Rydberg state approximation for the OEP method and to an approximate electron









affinity in the case of HF. If one considers the LUMO-HOMO energy difference as a zero-

order approximation to the excitation energy and the rest of matrix elements in Eqn.

(5-36) as corrections, it is apparent that TDDFT has a qualitatively wrong behavior in

each order when dealing with charge-transfer excitations.

If one argues that the occupied orbitals in the HF and OEP calculations are nearly

the same, and they span the occupied space, then the unoccupied space would be the

same, too. However, even if the spaces were the same, the difference between the non-

local K and local V, is sufficient to keep the latter local exchange from ever being able

to describe an electron attached state. Thus, it is not possible to overcome the local

operator's failure even in the full space for the calculations.

5.3 Conclusions

In this article some possible formulations of OEP are stressed. It was argued

and shown numerically that the OEP formulations which adhere to the requirement

of pointwise closeness of local and non-local exchange potentials are far more stable to

variations in the molecular and auxiliary basis sets, and do not suffer from arbitrary

limitations on the choice of basis. However, the OEP problem remains ill-conditioned, as

also stated previously[16, 52], and does require the use of a generalized (SVD) inverse or

the equivalent. It was also shown that under the right conditions the exact HF result can

be obtained by an OEP-like procedure as a trivial solution of a weighted least squares

procedure. Any point-wise condition has to be a many to few mapping in a finite basis set.

The elimination of denominators from working expressions for OEPx [20, 44, 56] can

be done as shown, to obtain the computational advantage of using only the occupied or-

bitals in the OEP procedure.The direct optimization algorithm as implemented, however,

has led to work [45, 49] that has -i -.-. --. I incorrect conclusions.

The TDOEP method was implemented for spin-polarized open-shell systems.

Results for several open-shell systems show very similar time-dependent behavior for

the time-dependent response of the OEP and the HF potentials. The correct long-range









. -i ,! II dll ic behavior, achieve by the Colle-N. -bet algorithm, makes it possible to avoid an

underestimation of Rydberg state energies and overestimation of the polarizabilities. This

overestimation for polymers, particularly for hyperpolarizabilities, has attracted a great

deal of attention[64-66]

The problem of charge-transfer excited states in TDDFT relates to the inability of

the zeroth-order orbital energy approximations, as the HF virtual orbitals and TDOEP

excited orbitals show qualitatively different behavior. If the HF and OEP operators were

the same, a proper treatment using the whole space should be able to overcome this

limitation, even if the usual zeroth-order approximation offered by TDDFT would be

a poor one. However, this difference arises from the form of the operator, local versus

non-local, and, as such, remains a fundamental problem for the DFT method itself, as has

been observed [51]

In conclusion, for all the properties considered in the paper, except for charge-

transfer excited states, there seems to be no significant difference between the results of

TDHF and exchange-only TDOEP, as one might expect. Hence, the focus should remain

on the correlation potential as it is in ab initio dft[16, 17, 48, 62].















Table 5-6. Static polarizabilities (in a. u.)
TDHF TDOEP SVWN CCSD


CN

Xyy
COzz
a
Aa



ayy
co+



azz
a
Aa
CH3

ayy
azz
a
Aa
N


11.98
11.98
18.55
14.17
6.57

6.66
6.66
12.22
8.52
5.54

15.06
14.04
15.06
14.72
1.02


11.91
11.91
18.41
14.08
6.50

6.66
6.66
12.15
8.49
5.50

15.00
13.98
15.00
14.66
1.02


15.63
15.63
25.00
18.75
9.38

9.38
9.38
14.06
10.94
4.69

18.75
18.75
18.75
18.75
0.00


12.50
12.50
23.42
16.16
10.94

9.38
9.38
12.50
10.42
3.13

15.63
15.63
15.63
15.63
0.00


aZ 6.24











Table 5-7. Isotropic (
TDH
CN ... CN 93.A
CH3 ... CH3 94.:
N ... N 19.:


6.23 6.25 5.94











76 coefficients (in a. u.)
[F TDOEP
37 92.06
27 93.40
28 19.21









CHAPTER 6
EXACT-EXCHANGE TIME-DEPENDENT DENSITY FUNCTIONAL THEORY FOR
HYPERPOLARIZABILITIES

The time-dependent OEP exchange-only method (OEPx) has been previously

derived by G6rling[20] and implemented for molecules by Hirata et. al [21, 50]. Numerical

results in the latter papers show a reasonable description of both valence and Rydberg

excited states. Reasonable static and dynamic polarizabilities are also obtained. Good

results of TDOEPx for excited states and properties are obtained partly due to the correct

.i-vinl Il' '1 ic behavior of the exchange potential, but also due to the elimination of the

self-interaction error. In particular, OEP-TDDFTx tends to be superior to standard

approaches like TDDFT based upon the local density approximation (LDA) or Becke-Lee-

Yang-Parr (BLYP) functionals. Similarly, exchange-only OEP with exact local exchange

(EXX) [22, 23] has been shown to greatly improve band-gaps in polymers [23]

Another advantage of OEP based methods is that since virtual orbitals in the

exchange-only DFT as well as the occupied orbitals are generated by a local potential,

which corresponds to the N-l particle system due to the satisfaction of the self-interaction

cancellation, the differences between orbital energies of virtual and occupied orbitals

offer a good zeroth-order approximation to the excitation energies[16, 17]. This is not

possible in the case of Hartree-Fock theory, without adding a VN-1 potential[67, 68],

since occupied orbitals are generated by an N-l particle potential and the energies of

unoccupied orbitals come from N-electron potential, and thereby, approximate electron

affinities.

In Kohn-Sham DFT


hKS = T + Vezt + J + V1e









where


6Ec
Vp(r)

j r r/1

To address second-order properties requires additional effects of the density change given

as the second and higher terms of Taylor's expansion of Ex,. The kernel


f(ri, r2) -
Jp(r2)

is used in [21, 50].However, once our objective is third-order molecular property, the

essential new element in TDDFT is the second exchange-correlation kernel, which in an

adiabatic approximation is defined as the second functional derivative of the exchange-

correlation potential with respect to the density

JV1,(ri)
g(ri, 2, r3) =-p (6-1)
6P(r~)6p(r3)

Hence, in this paper we will derive the second kernel for the exact-exchange DFT, based

on the OEPx approach. This is a complicated quantity. The traditional way of deriving

kernels and potentials is to tediously derive all terms with the use of the chain-rule for

functional differentiation. Yet, even the exchange- only kernel, f, has a complicated

structure in OEP, and its further extension to a second kernel is very difficult. To avoid

the use of traditional methods, an effective diagrammatic formalism for taking functional

derivatives was recently developed[37, 62]. In this paper we use that formalism and

apply it to the exchange second kernel for OEPx based time-dependent DFT. After

discussing some properties of the second kernel, we report numerical results to obtain

hyperpolarizabilities, compared to those from Hartree-Fock and coupled cluster singles and

doubles (CCSD).

There is a long history of the treatment of hyperpolarizabilities and associated non-

linear optical (NLO) properties. See [69] for a review. In particular, issues of the incorrect









behavior of TDDFT hyperpolarizabilities for NLO design for polymers[64, 65] points to

the necessity of an improved theory. A rigorous (ab initio) DFT analogue starts with

exchange-only, TDOEPx.

6.1 Theory

6.1.1 Time-Dependent Density Functional Theory Response Properties

When an external time-dependent electric field is applied, the perturbation can be

written in the form (6-2)

V(t) =- Z- r E(t) (6-2)

For the case of periodic perturbations, the virtual-occupied block of the linear response of

the density can be found from the following system of linear equations


(A + B)U = -h (6-3)


The matrices A, B have the structure, given by Eqn (5-36) and h is


haa =< a, r i, > (6-4)


The calculation of the components of the first, static hyperpolarizability tensor

requires [70]

S=2 (U F U UF U +UF U +

i,p,q,(
!3A~w 2Z(U> F qJ zpU(;Jqa pq?,qia i~pa pqcT qicI
UFP U FU U UP, F, U) +

-2 (U\. c UJ + UX U + U +
i,j,p,u
U + U" U U U ) +

8S< baubc gj i Pib\ j9T\ k > lUa Utjvk (6-5)
a ab,c,i,j,k








The matrix elements Fq, and e0, can be calculated from the relations (6 6)

F\ = h (AP + BRPq) UrA
q, pq\aiu iq a
a,i



6.1.2 Diagrammatic Derivation of the Second Exact-Exchange Kernel
For the derivation of the second kernel it is convenient to use the form (6-8)


g9(ri, r2, r3)


6fxc(r, r2)
6p(r3)


Which can be rewritten as


f(r,=r2) dr4dr5 6X-- (rr4) Qr 5)X (r5,r2)+

Sdr4dr5X- (rI r4) 6Q(4, r5) X- (r5, 2)

Jdr4dr5X l(rr) (4,r5) 6p(r3)

Taking into account the fact that

6X-\(r^_, r2)6X(rs, r4 _
(rl -r2) dr4drf X (r, r) ( 4)X '(r2, 4)
bp(r3) V,(rs)

and doing some simple transformations, it is possible to write


g(rl, r2, r3)


drdr7dr8sf(r2, rs8)X- (r3,6) X(r, r) X (r, r7) +
J 6Vs,(r6)
dr4dr5dr6X-(rl, r4)X- (r3, r6)Q(4,r5)X (r2,r5)
r ir (Vs(r6)
dr6dr7dr8f (ri, r7)X- (r3, r6) 6X(r, r) X-(r2, 8)
j 6bV,(r6)


(6-11)


where


Sdr' p() + V(r)


(6-6)

(6 7)


(6-8)


(6-9)




(6-10)


Vs(r) = ve.t(r) +


(6-12)










The first element of Eqn (6-11) can be written as


/X(r7, 7 ,
dr6dr7dr8 (r2, 8)X- (r3, 6) 6X(r 7, r) X- (r r7)
6Vs(r6)

dr6dr7X- 1(r3, r6)Pi(r2, r6, r7)X- '(ri, r7)


and Pi(r2,r6, r7) can be presented as a sum of diagrams (Fl) (F6) of (6-14)


Sr(r ) (r) /a (r
(r7) a 8(r6) 1 b (r ) 1
f(r2) f(r2) f(r2) b f(r2)
(Fl) (F2) (F3)


5a 7(r1 ) 68(r) 6 (r6)
S(r6 ) b f(r2) 0 6(r2)

(F4) (F5) (F6)


In the same way the third term of eq (16) can be rewritten as

r r dr fci r -X r 6X(r7, rs) 1( r2r
Sdr6dr7dr8f r r7)X-r3, 6) 8)X- (r2, 78)
S6Vs (T6)


dr6dr7X -(r3, r6)P2(ri, r6, r7)X 1(r2, r7)


where P2(rl, r, r7) gives the sum of diagrams (F7) (F12) of (6-16)


6(r )
(r7)
(F7)


S8(r7 )
) (r6 6
fx(r ) 0
(F8)


68(r6)
1 (r7
() fx(r
(F9)


S S(r ) 6a (r6) 6 (r6)
F (r f ) ff(r1)
(F10) (F ( r) (F2)
(F1O) (F11) (F12)


The diagrammatic expression for Q(r6, r7) contains terms (VI) (V6) and (FX1)

of (6-17) and (6-18)[62]


8(r7 ) 8(r6)
S(r6 ) b 8 (r7 )
V V-


(V2)


a 8(r7)
8 6(r6)

V (r7
(V4)


a (r6)
V


8(r 6)
V
r7 x6
(V6)


S8 ( r7 ) (r (r6 )/r
b j. aa
) i / k (/ k
(FX2) (FX3) (FX4)


(6-13)


(6-14)


(6-15)


(6-16)


S(r6)

VVa (r7
x


(FX8)


6(r 6)
a
j b

6(r7
(FX1)


(FX5)


(FX6)


(6-17)










Tr6) (r6)


S8(r7 )5 (r7 )- J(
(FX7) 7(FX8) (6-18)


Functional derivatives from the set (V1)-(V6) can be presented by two types of diagrams.

The first type contains the exchange kernel, diagrams (F13)-(F18) of (6-19)

(r6) 1(r7) 6(r6) (r7) a6(r6) (r6)
S(r7 ) (r6) 1 (r7) ( ) f(r) f(r3)
f(r) (r) (f(r3) b fx(r3) ) b (r b) (r,)
fxrr)x3rx) xr7
(F13) (F14) (F15) (F16) (F17) (F18) (6-19)


The second type contains the exchange potential, giving diagrams (VX1)-(VX8) of (6-20)

/ r4\ r 6 r ) 6r4 r5\ 6 6r4\ r5\ r6) (r\ rT \ 6 )
(r4\ r5\ r6 )a (r4\ r5\ r6 ) (r4\ r \5 r6) 6(r4\5 \ r5 6)
(r4 r5 r6 4 r5 r6 6r5(r4 r \ 6 ) 64r\ r )r6
V V V V
(VX1) (VX2) (VX3) (VX4)
/ 6(r rs rr ) ( r4 r ) 6(r4\ r r6) (r4\ r\ r6 )


Vx ax x x
6(r4\ r5\ r6) a6r\ ( 5 r6) lrrr5\ r6 6 ) ar 64r 54 r 6 )
6(r4\ rs*5 x J ) b'4 s r54\ 17\ r6 ) V ^(4 rs\ r6 )
4 6456 456
(VX5) (VX6) (VX7) (VX8) (6-20)


The notation r4\r5\r6 means that for each particular diagram all six permutations of r4, r5

and r6 should be taken. The functional derivatives from diagrams (FX1)-(FX8) lead to the

set (SX1)-(SX14) of (6-21) and (6-22).

8/ 4 \ r \ r6) 1 (r4\ 5 6) \A(r4\ r\ r6 ) 6 (r \ r )
S5 6 4 5 6 ) 5r4 6 ( \7 (r5\ r \ r6
S8(r 54\ r5\ r6 4 5 6 )r (\ r55\ r6
k c a b 61r4 175 r
8(r4\ r5 6 r ) 8(r4\ r5\ r6 ra
(SX1) (SX2) (SX3) (SX4) (SX5)

A^ ^ ^ ) ar4 6\ r5\ r6
(r4\ r5\ r6 (r\ r5\ r6
k(r\r5\ r6 )\(r\ r5\ r6) -6(r,\ r5\ r6 ) ( rs r 6
6 )
S (r4 5 r 6 a 6a b k
b k 64) 175 6 k4 (5 ) 4 5 6(
(SX6) (SX7) (SX8) (SX9) (6-21)









a(r4 r s5 r6 6(r4 r 5 r 6)- 6(r4 r5 r 6)4 8 (r4\ r\ r) 6(r 4 r 6)-6b
r4 r r6(r 4 rs\ r 6) 4 r56 r 6)
k a 6(r4 rs5 r6)l 6 ) rrr) r5 r 66 )jr r 6 r4 r5 5 r 6
4 5\ r6 )k k k
(SX10) (SX11) (SX12) (SX13) (SX146 -22)


Interpretation of all diagrams which contributes into the second kernel and details of the

implementation are given in Appendix B.

6.1.3 Properties of the Second Exact-Exchange Kernel

All the diagrams contributing to the second exact-exchange kernel have only one

contour and because of this fact only aaa and 333 spin components are present, as

in Goldstone diagrams where Sz has to be conserved at each vertex. The algebraic

expressions for all the diagrams have to be symmetric with respect to permutations of

rl, r2, and r3 variables. In the case of the two-electron spin-unpolarized systems the OEP
exchange potential becomes equivalent to the Slater potential[38]

vSlater(r) -_ i()j() r l) l (623)
Sizj PA IJ r ri(

Taking into account the HOMO condition and considering the expression for the exchange

kernel, it is possible to write for the special case of two electrons


,f EP(p1, 2) (6-24)
ri r2

Diagrams (VX1)-(VX5) and (VX8) of (6-20) will cancel diagrams (SX7), (SX8) and

(SX11)-(SX14) of (6-21) and (6-22) because of the HOMO condition and the behavior of

the Hartree-Fock exchange for the case of two electrons. Diagrams (VX6) and (VX7) will

cancel diagrams (SX1) and (SX2) because of Eqn (6-23). Diagrams (SX5), (SX6), (SX9)

and (SX10) will cancel diagrams (F1)-(F4), (F7)-(F10) and (F13)-(F16) because of Eqn

(6-24). Finally the sum of the diagrams (SX3) and (SX4) cancel the sum of (F5), (F6),

(Fll), (F12), (F17) and (F18). Thus for the special case of two electrons

gOEP (12, 3) 0 (6-25)









The condition (6-25) might be used as a hint for developing new density functionals, as

none of the existing functionals exhibit such behavior.

6.2 Numerical Results

As a numerical test the static hyperpolarizabilities of LiH, H20, CO and CN are

calculated. In all cases uncontracted ROOS-ADZP basis sets[27] are used and equlibrium

geometries are taken from ref[36]. The results calculated with different methods are

presented in Table 6-1. In the first four columns are static hyperpolarizabilities obtained

Table 6-1. Hyperpolarizabilities of several molecules (in a. u.)
HF OEP-HF OEPx LDA CCSD
LiH
3 312.130 314.910 312.126 621.093 691.406
f3. 201.150 207.868 201.641 404.297 204.171
CO
,,, -31.016 -24.864 -29.455 -66.406 27.343
0.. -3.073 -4.660 -3.118 -13.203 5.859
CN
0.. -1.299 0.211 -3.906 -1.953 164.02
A,, -17.651 -15.646 -18.121 -62.500 68.359
H20
03. -6.713 -6.239 -7.812 -13.672 -5.859
j3. -0.497 -0.604 -1.215 -6.138 -3.870
3yy, -10.866 -10.32 -11.847 -23.437 -7.645


with Hartre-Fock non-local exchange and with the two local exchange operators: OEP-HF,

which means equation (6-5) without the second kernel term, and OEPx, which means

all terms in equation (6-5). The hyperpolarizability values calculated with the OEPx

method are in good agreement with the corresponding Hartree-Fock values, and generally

no better. This good agreement can be explained by the fact that OEPx method is free

from the self-interaction error, has the correct long-range ., -ii1! i .I ic behavior, and the

correct HOMO values. As hyperpolarizabilities are response properties of third order their

calculation within the DFT method requires very precise exchange-correlation potentials.

Only potentials which simultaneously satisfy all theorems and conditions pertaining

to the DFT exchange potentials can be expected to reproduce even the Hartree-Fock









hyperpolarizabilities. The large errors of the LDA method are caused by the wrong

long-range .,i-mptotic behavior and the incomplete cancellation of the self-interaction

terms, though LDA does reflect some estimate of correlation. In ref[66] it was pointed out

that none of the existing hybrid and .,-i mptotically corrected functionals are capable of

reproducing Hartree-Fock hyperpolarizabilities. The semi-empirical method, developed

there[66] is the only known way to get reasonable hyperpolarizabilities with conventional

functionals. However, the OEPx potentials and kernels are the only rigorous, purely

ab initio way to obtain high-order properties, within a solely DFT framework of local

potentials and kernels.

For the calculation of hyperpolarizabilities with the Hartree-Fock method[71] the last

term in eq (6-5) should be dropped. Its importance for the DFT scheme can be estimated

from the second column of Table 6-1, where OEP hyperpolarizabilities were calculated

without the last term of equation (6-5). For the example of the LiH molecule the relative

contribution of the last term is small. However, going from closed shell to an open-shell,

like for the CN molecule, the relative contribution of the third kernel becomes much

greater and, sometimes, can even change the sign of hyperpolarizabilities.

The differences between the OEPx and the coupled cluster singles and doubles

(CCSD) method are because in the exchange-only OEP method electron correlation

has not been taken into account. For the calculation of high-order properties, which

can be compared with experimental data, accounting for the electron correlation is

critical [69, 72, 73], and generally CCSD is fairly good, though adding triples offer some

improvement. Nevertheless, the present results show what kind of properties exchange

potentials and kernels must have to be able to describe higher-order properties within the

framework of DFT, and as mentioned previously, none of the current set of functionals for

standard DFT shows this behavior.









6.3 Conclusions

We have presented the analytical expressions and implemented them for the calcula-

tions of static hyperpolarizabilities applicable to general molecular systems with the exact

exchange functional of Kohn-Sham DFT through the TDOEPx formalism, the first step in

the application of ab initio dft[16] It has been shown that the rigorous exchange treatment

in TDOEP and TDHF provide hyperpolarizabilities that closely agree with each other, as

one would expect, although this will not happen for charge-transfer excited states[51, 74].

This clear agreement with TDHF is in contrast to conventional exchange functionals

that severely overestimate hyperpolarizabilities. OEP potentials and kernels are not only

capable of describing excitation energies, but also higher-order properties. Since none of

the standard DFT functionals shows the correct analytic properties of the second kernel,

the OEP formulas and results can be used for the testing and calibration of new density

functionals.









APPENDIX A
INTERPRETATION OF DIAGRAMS OF THE SECOND-ORDER MANY-BODY
PERTURBATION THEORY OPTIMIZED EFFECTIVE POTENTIAL CORRELATION
KERNEL

To build h(r3, r6) we have to provide the summation of the diagrams contributing

into the correlation kernel. The time dependent OEP-MBPT(2) equations are imple-

mented, using a gaussian basis set. Kernels and potentials are decomposed using the

auxiliary basis set


V (r=) ZCA r
A

Xf(,r 2 =r2) X hV )j (rI) (r2) (A




where

< iav >< iaK >
XK, = 2 (-)
Ei Ea
i,aa

< pq7 >= drcp(r)c,(r)}(r) (A-2)




(VC1 1) + (VC2 2)
2 z < ablij >< cjlak > (< icA >< kbp > + < kbA >< ic >)
(A-3)
(E,c,,,i + E Ea Eb)(Ei Ec) (Ek Eb)


(VC1- 2)
< ablij >< cdlab >< icA >< djp > (A
(Ei + Ea Eb)Ei Ec)(Ej Ed)
a,b,c,d,i,j


(VC1 3) + (VC3 1)
2 < ablik >< cjlab > (< icA >< kj* > + < kjA >< ic >)
b,,,j,k (E + Ej a- Eb)( + k a- b)(i c)
a,b,c,i,j,k









(VC1 4) + (VC4 1)
< adlij >< cj|ab > (< icA >< db/ > + < ic/ >< dbA >)
(i + Ej Ea Eb)(i + Ej a Ed)(Ei Ec)


(VC1 5)=

2 < ablid >< cj ab >< icA >< dji >
a,b,c,d,i,j (i + j b)(i c)(j d)


(VC1 6) + (VC2 5)
< ak ij >< cj|ab > (< icA >< kbp > + < kbA >< icp >)
S(i + Ej a Eb)(Ei Ec)( k Eb)


(VC1 7)
< abldj >< cjlab >< icA >< idi >
a,b,c, Ea Eb)(Ei Ec)(Ei Ed)
a,b,c,d,i,j


< kblij >< cjlab >
(Ei + Ej



< abiij >< cj|ab >
(Ei + Ej



< abiij >< cj|ab >
(Ei + Ej a


(VC1 8) + (VC2 8)
(< icA >< ka/ > + < kaA >< icp >)
- Eb)(Ei Ec)(k Ea)


(VC1 9) + (VC1 10)
(< ikA >< kci > + < kcA >< iki >)
- Eb)(i ( c)(k c)


(VC 11) + (VC3 7)
(< ikA >< kci > + < kcA >< iki >)
- Eb)(Ek + Ej a Eb)(Ek Ec)


(VC1 12) + (VC4 8)
< ablij >< cjldb > (< icA >< dap > + < ic/ >< daA >)
(Ei + Ej a b)(Ei + Ej d Eb)(Ei Ec)


(A-10)





(A-ll)





(A-12)





(A-13)


2
ab,c,d,i,j


2 Z
a,b,c,i,j,k


(A-6)





(A-7)





(A-8)


(A-9)


-4 a
a,b,c,i,j,k




-4 E
a,b,c,i,j,k




-4 a
a,b,c,i,j,k




4
ab,c,d,i,j








(VC1 13) + (VC1 14) -
4 < ablij >< cjlab > (< idA >< dc i > + < id, >< dcA >) (A 4)
4 --(A-14)
b,c,d,i,j ( E E Eb)(Ei Ed)(Ei Ec)

(VC1 15) + (VC2 15)
4 < abiij >< cj|kb > (< icA >< kap > + < kaA >< icp >)
ab,c,i,j,k ( i + -a Eb)(Ei Ec)(Ek Ea)
(A-15)


(VC1 16) -
-4 < ablij >< kjlab >< icA >< ckl >
-4 -(A-16)
,bcijk (Ei + Ej E b)(Ei Ec)(Elk c)

(VC2 1)
2 < ablij >< ijkl > (< akA >< bl > (A 7)
a,b,i,j,k,l i + -a -b)(k -a)(l -b)

(VC2 3) + (VC3 2)=
2 < ab il >< ijlkb > (< akA >< jl > + < jlA >< ak >) (A
(Ei E+ Ea Eb)(Ei El Ea Eb)(EFk Ea)

(VC2 4) + (VC4 2)
-2 < aciij >< ijjkb > (< akA >< bcp > + < bcA >< akp >) (A19)
b,,,j,k (C + j- F b)(C + F Ec)(Ek a)

(VC2 6) =
S < allij >< ijlkb > (< akA >< bl > (A20)
,b,,, (i + a Eb)(l Ea )(El Eb)

(VC2 7)=
4' < lblij >< ijkb > (< akA >< alp >(A
a,, (Ek i + EF Eb)(l EF)(El Ea)









(VC2 9) + (VC2 10)
-4 < ablij >< ijlkb > (< ckA >< acp > + < acA >< ck/ >)
(E, E, Ea -+ Eb) (Ek E) (Ek Ea)

(VC2- 11) + (VC4- 7)
-4 < ablij >< ijlkb > (< ckA >< acp > + < acA >< ck/ >)
b,c,,j,k (cE + j a -- b)(i + Cj Eb Ec)(Ek Ec)


(VC2 12) + (VC3 8)
4 < ablij >< ljlkb > (< akA >< ili > + < ilA >< akp >)
,b,i,,, (i + j Ea b)(El + Cj Ea Eb)(Ek Ea)


(VC2 13) + (VC2 14)
4 < ablij >< ijlkb > (< alA >< lki > + < alp >< lkA >)
a,b,i,j,k,l ( + a b)(k a)( a)

(VC2 16)
< ablij >< ijlcb >< akA >< ckp >
a,b,c,i,j,k + a Eb)E E) k c)


< ablik >< ijlcb
(Ei + EF -



< ablik >< ijlcb
(Ei + k a -


(VC3 3) + (VC3 5)
> (< jkA >< ili > + < jkp >< ilA >)
*b)(Ei + k Ea Eb)(El + j E Eb)


(VC3 4) + (VC4 5) =
> (< jkA >< ac/ > + < acA >< jki >)
Eb)(Ei + -k a b)(Ei + j b Ec)


(VC3 6) + (VC4 3)
> (< jkA >< ac/ > + < acA >< jk/ >)
Eb) (i + Ek Ec Eb) (i + j Ec Eb)


(A-22)




(A-23)




(A-24)




(A-25)




(A-26)


S< ablik >< ljlab
a,b,i,j,k,l ( + -


a,b,c,i,j,k




a,b,c,i,j,k


(A-27)




(A-28)




(A-29)









(VC3 9) + (VC 10)
2 < ablik >< ijlab > (< jlA >< Ik > + < jl >< IkA >) (A30)
b,i,k, (i + k EF -- b)(Ei +l Fa -- b)(Ei + j Ea -- b)


(VC3 11) + (VC4 12)
-2 < ablik >< ijlac > (< jkA >< bcl > + < bcA >< jkp >) (A31)
,b,ci,j,k (C k Ea Eb)(CF C -a Ec)(Eci + c- Fa Eb)


(VC3 12) + (VC4 11)
-2 < ablik >< ijlac > (< jkA >< bc > + < bcA >< jk >) (A32)
(E,b,c,,,k i + Ek Ea Eb)(Ei + k -a Ec)(Ei + Ea Ec)
a,b,c,i,j,k

(VC3- 13) + (VC3 14)
< ablik >< ijlab >< jcA >< kcp >
a, + Ek Ea Eb)(i E+ Ea Eb)(Ek Ec)
a,b,c,i,j,k

(VC4- 4) + (VC4 6)
< aciij >< ijtdb > (< cbA >< adp > + < adA >< cbi >) (A34)
ab,c,d,i,j + Ea Ec)(i Ea Eb)(Ei Ed Eb)

(VC4 9) + (VC4 10)
2 < aclij >< ijad > (< cb >< bd > + < cb >< bd >) (A35)
,,c,,, (E + Fa a c)(E + Fa Fb)( + a b ( a d)
a,b,c,d,i, j \ c b\z a )

(VC4 13) + (VC4 14)

(A 36)
-4 < ac ij >< ijjab >< bkA >< kci > (A36)
S (i e+ C a c)(Ei + j a -Eb)(E c)
a,b,c,i,j,k

(VC5- 1) + (VC6 -1)
2 << ablij >< cjlbk > (< icA >< ak> > + < icl >< akA >)
2 (+ )(- ) -(A)37)











-2
a,b,c,d,i,j


(VC5 2)
< ablij >< cdlba >< icA >< djp >
(E + Ej a Eb)(i E,) Ej Ed)


(VC5 3) + (VC8 1)

2 < ablik >< cjba > (< icA >< jk/ > + < jkA>)
ab,c,i,j,k (c( + F- i a -Eb)(Ei+ Ek Ea Eb)(Ei Ec)


(VC5 4) + (VC7 12)
2 < dbjij >< cj|ba > (< icA >< dap > + < daA >< icp >)
~,b,cd(i +C d Ib)( -+ Ea b)(Ei Ec)
a,b,c,d,i,j


(VC5 5) + (VC7 8) =

-2 < abiij >< cj|da > (< icA >< dbi > + < dbA >< icp >)
,c,d,i, (E Ej Ea b)(E Ej -Fa E- Ed) (E Ec)
a,b,c,d,i,j


Z < aklij >< cjlba >
a,b,c,i,j,k ( -


2 Z
a,b,c,d,i,j


< ablid
(Ei + Fj


2 < kblij >< cjlba >
b,,i,j,k (Ei + Ej
a,b,c,i,j,k


(VC5 6) + (VC6 -8)
(< icA >< kb/ > + < ic/ >< kbA >)
- Ea Eb)(i Ec)(Ek Eb)


(VC5 7)
>< cjbba >< icA >< jbp >
- a Eb)(E Ec)(E -? d)


(VC5 8) + (VC6 -6)
(< icA >< kai > + < icp >< kaA >)
- a Eb)(Ei ,c)(k a)


(VC5 9)
>< cba >< ic >< idi >
- E Eb)(i Ec)(Ei d)


(A-38)





(A-39)





(A-40)





(A-41)


(A-42)


(A-43)


(A-44)


< abldj
(Ei + Ej


-2
a,b,c,d,i,j


(A-45)








(VC5 10) + (VC8 10)
2v < ablkj >< cjlba > (< icA >< ikp > + < ic >< ikA >) (A
2 --(A-46)
b, (Ek + -- -- Eb)(Ei E -- E -- Eb)(Fi Ec)


(VC5 11) + (VC5 2)
2 ablkj >< cj|ba > (< icA >< ikl> > + < ici >< ikA >) (A47)
ab,c,i,j,k k E+ ( Ea Eb)(Ei Ec)(Ek Ec)

(VC5 13) + (VC5 14)
-2 < ablij >< cjlba > (< idA >< dci > + < dcA >< idi >) (4
-2 F(A-48)
a,(Ei b,c,d,i,j+ a Eb)(Ei c)(Ei Ed)

(VC5 15) + (VC6 15)
2 < ablij >< cjlka > (< icA >< bkU > + < ic >< bkA >) (A
2 --(A -49)
(E,,, E Ea -- Eb)(Ei Ec) (Ek Eb)
a,b,c,i,j,k

(VC5 16)
S < ablij >< kjba >< icA >< ck > (A50)
a,b,c,i,j,k (i + E Eb) (Ei c) (Ek Ec)

(VC6 2)
2 < ablij >< jilkl >< akA >< lb > (A51)
( abijk E b) (E Fa) (El Fb)


(VC6 3) + (VC7 1) =
2 < acij >< jilkb > (< akA >< cbi > + < aki >< cbA >) (A52)
,c,,, + j Ea Eb)(Ei + j -a Ec)(Ek Ea)
a,b,c,i,j,k

(VC6 -4) + (VC8 8)
-2 < ablij >< jllkb > (< akA >< il > + < ilA >< aki >) (A53)
,,,,, (E + F Eb)(E+ E( E Eb)(E Ea)









(VC6- 5) + (VC8- 9)
2 z < ablij >< likb > (< akA >< jl/ > + < jlA >< akp >)54)
,b,, (E + j j a Eb)( + Fl Fa Eb)(k Ea)


(VC6 7)
-2 < allij >< jilkb >< akA >< lbp > (A55)
a,b,i,j,k, + E Ea b) (E a)(E- b)


(VC6 9)
-2,, < lblij >< ak >< lap >56)
,,k, (Ei + c- Ea- b)(Ek -a E)(El Ea)


(VC6 10) + (VC7 9)=
2 < ablij >< jilkb > (< ckA >< acp > + < ck >< acA >) (A57)
E,, (+ j -- -- Eb)(Ei + E, -- Eb)(Ek Ec)
a,b,c,i,j,k

(VC6- 11) + (VC6- 2)
2 < cb ij >< jilkb > (< akA >< acp > + < akp >< acA >) (A
2 +----b)( k k-) (A-58)
alb,c,i,j,k (Ei E+ Ec Eb)(EFk Ec)(Ek Ea)

(VC6- 13) + (VC6- 14)
-2 < ablij >< jikb > (< klA >< al > + < a >< kl >) (A59)
a,b,i,j,k,l ( E+ F Eb)(l a)(Ek a)

(VC6 16) =
< ablij >< jilcb >< akA >< kcp >
2 -(A60)
,,,. (Ei Fa Eb)(Ek a)(Ek c)
a,b,c,i,j,k

(VC7 2) + (VC7 3)
< aclij >< jildb > (< bcA >< adp > + < adA >< bcp >)
,c,d,i,j (Ei + F Ec)(Ei + j a Eb)(Ei j bd Eb)
a,b,c,d,i,j









(VC7 4) + (VC8 5)
< aclij >< kilab > (< bcA >< kjp > + < bcp >< kjA >) (A62)
a(E E+ Ea EF)(Ei E+ Ea Eb)(EF Ek Ea Eb)



< aclij >< kilab > (< bcA >< kjp > + < bc >< kjA >) (A63)
a,b,c,i,j,k F F F b)

(VC7 6) + (VC8 7)
< aclij >< jklab > (< bcA >< kip > + < bc >< kiA >) (A64)
(A-64)
,ci,j,k (i E, E- )(Ej + k Ea E,)(cj + k -a b)


(VC7 7) + (VC8 4) -
< aclij >< jklab > (< bcA >< kip > + < bc >< kiA >) (A65)
ab,,,j,k ( + aj a )( + j a b)( -j + Ek Ea Eb)
a,b,c,i,j,k

(VC7 ) + (VC7 1)
< aclij >< jilad > (< bcA >< bdp > + < bdA >< bcp >)66)
(A66)
,b,,j Ea Ec)( ( + Ea Eb)(Ei E+ a Ed)


(VC7 13) + (VC7- 14)
< aclij >< jilab >< bkA >< ckp >
(Ec, i + j a Ec)(i + j Fa Fb)(Ek Ec)
a,b,c,i,j,k


(VC8 2) + (VC8 3)
< ablik >< ljdba > (< jkA >< ilp > + < ilA >< jki >)
(A 68)
,,,j,k, (i + k Ea Eb)(i + j -a Eb)(EF +E Ea sb)


(VC8- 1) + (VC8- 2)
< ablik >< ijlba > (< klA >< jli > + < jlA >< kl/ >)
S+ j ) ( + )( + k -(A 6))
a,b,,,i, ja )a









(VC8 13) + (VC8 14)
< ablik >< ijlba >< ckA >< cjp >
(Ei + k Ea Eb)(Ei + sj -a b)(Ej -


(VC9-

2 f faifkj(< ijA >< akp >
i,j,k,a ( a)A (


faifbj(< ijA
i,j,a,b


(VC9-
>< abpi >
- Ea) (Ej -


) (A-70)
c)


1) + (VC11- 8)-
+ < akA >< ijp >)
,a)(Ek a)


2) + (VC10 3)
+ < abA >< ijp >)
a) (Ej Eb)


(A-71)


(A-72)


(VC9
ijA >< abp >
(i F)(i-


3) + (VC10
+< abA ><
E Eb)


-2)
ij >) (A73)
(A-73)


(VC9 4) + (VC12 5)

2 Y faifba(< ijA >< bjp > + < bjA >< ijp >)
i,j,a,b (i Ea)(E E(E Eb)


(VC9 5) + (VC9
fai0fka(< ijA >< kjp > + < kjA ><
,j,,a (E- Fa)(j Fa)(Ek a)
. j, Ak, a


-6)
ijp >)


(VC9- 7)
Sfaifaj < jbp >
( ija) ( a) ( b
i,j,a,b


(VC9 8) + (VC15 3)=
4 fai < jblak > (< ijA >< kbp > + < kbA >< ijp >)
,j,k,a,b ( Ea)(E Ea)(k b)


2
a,b,c,i,j,k


f, fbj(<
i,j,a,b


(A-74)


(A-75)


(A-76)


(A-77)








(VC9 9) + (VC14 4)
2 faj < ablki > (< ijA >< kb > + < kbA >< ij >) (A78
2- (A-78)
(Ei ,a)(Ej Ea)(Ek Eb)

(VC9 10) + (VC13 10)
2 fat < bjlak > (< ijA >< kbp > + < kbA >< ij >) (A 79)
i,j,k,a,b (i a) (E Ea)(Ek b)

(VC10 1) + (VC12- 7)
2 fbifac(< abA >< icp > + < icA >< abi >) (A 80)
,, ( a)(Ei b)(i Ec)

(VC10- 4) + (VC11- 5)
-2 fibfij(< abA >< ajp > + < ajA >< abp >) (A 8)
b (E a)(E Ea)(E{ Eb)

(VC10 5) + (VC10 6)
Sfbific(< abA >< acp > + < acA >< abi >) 82)
,,, (F- a)(i Eb)(Ei c)

(VC10 7)
-2 fifib(< bjA >< ajp > + < ajA >< bjp >) (A83)
i,j,a,b ( a)(E b)(E3 Ea)

(VC10 8) + (VC15 4)
4 fib < iclaj > (< abA >< cjp > + < cjA >< abi >) (A84)
(Ej,,i ,ca)(Ei b),(E c)

(VC10 9) + (VC14 2)
-2 z fib < calij > (< abA >< cjp > + < cjA >< ab >) (A85)
i,j,a,b,(Ei )(Ei b)(Ej c)








(VC10 10) + (VC13 9)
-2 fib < iclja > (< abA >< cjp > + < cjA >< abi >)
-2 (A-86)
(Ei- E,)(E Eb)(Ej ,c)
i,j,a,b,c

(VC11 1) + (VC11 2) -
2 z fijfaj(< akA >< iki > + < ikA >< aki >) (A 87)
,j,k,a (-i a)(j a)(k a)

(VC11 3) + (VC11 4)-
-2 z fifaj(< abA >< ib. > + < ibA >< ab. >) (A-88)
S(Ej a) (Ei Ea) (Ei Eb)
i,j,a,b b a)

(VC11- 6) =
2 fjkfki(< ajA >< aip > + < aiA >< ajp >) (A89)
,j,, (E Fa)(k ,a)(Ej Ea)

(VC11 7) + (VC12 8)
-2 fijfab(< bjA >< aip> + < aiA >< bjp >) (A90)
i,j,a,b (i F)(Ei -b)(E Eb)

(VC11- 9) + (VC12- 9)
2 faifjb(< ajA >< bip > + < biA >< ajp >) (A 91)
,j,a,b (i- Fa)(j a)( b)

(VC11- o10)
-2 faifib < ajA >< bjp > (A92)
(i, i E)( )(j E)(Ej Eb)
i,j,a,b

(VC11 11) + (VC15 5) =
4 fi < ibljk > (< ajA >< kbp > + < kbA >< ajp >) (93)
,j,k,a,b (i a)(j E)(E Eb)








(VC11 12) + (VC15 2) -
S-4 fij < ablik > (< ajA >< kbp > + < kbA >< ajp >) (A94)
,j,k,, (F ,a)(j a)(E Eb)

(VC11 13) + (VC13 11) -
2 fij < akjbi > (< ajA >< kbp > + < kbA >< aj >) (A
2 --(A-95)
i,j,k,a,b (Ei (a j )( )(- b)

(VC11- 14) + (VC4 3)
S fij < abki > (< ajA >< kb > + < kbA >< aj >) (A9
ia,b(i E E a)( Ea)(Ek Eb)

(VC11 5) + (VC14 5)
-fai < kizjb > (< ajA >< kbi > + < kbA >< aji >)
i,j,k,a,b ( a)( a)( b)

(VC11 16) + (VC13 7)
2 fai < bljk > (< ajA >< kbp > + < kbA >< ajp >)
2) (A98)
i,j,k,a,b (Ei Ea)(E Ea)(Ek Eb)

(VC12 1) + (VC12 2) -

2 fabfib(< acA >< ic > + < icA >< acp >) (A99)
a,b,c,i ( F)( b)( )

(VC12 3) + (VC12 4) =

-2 fib ab(< ijA >< ajpd > + < ajA >< ijlp >) (A100)
(Eiji a) (Ei Eb) (E Ea)

(VC12 6) =

02 fabfac < ibA >< icIJ >
ab,c,(Ei Ea)(Ei Eb)(Ei Ec)









(VC12- 10)
S fibfbj(< iaA >< jap >
(Eb Eb)(E~ Ea)(E Ea)
i,j,a,b


(VC12- 11) + (VC15- 6)
4 fib < aclbj > (< iaA >< cjp > + < cjA >< iap >)
ab,c,i,j (Ei Eb)(EFi- Ea) (E Ec)


(VC12 12) + (VC15 1)
4 E fab < bclij > (< iaA >< cjp > + < cjA >< iap >)
a(Eij (- ~b) (Ei Fa) (E' Ec)


2 fab < jb\iC
a,b,c,i,j


2 fab < cblij
a,b,c,i,j



-2 fia < b a
a,b,c,i,j




Sb fia < bcja
a,b,c,i,j


(VC12 13) + (VC13 -
iaA >< cjpi > + < cjA ><
- Eb)(Ei Ea)(Ej Ec)


(VC12 14) + (VC14
iaA >< cjpi > + < cjA ><
- Eb) (Ei Ea)(j Ec)


(VC12 15) + (VC14 -
ibA >< cji > + < cjA ><
- Eb)(Ei Ea)(Ej Ec)


(VC12 16) + (VC13
ibA >< cjp > + < cjA ><
- Eb)(Ei Ea)(Ej Ec)


12)=
iap >)



-1)-
iap >)



10)-
ibp >)



-8)
ibp >)


(VC13 1) + (VC13 2) =
S fai < bilaj > (< bkA >< kjp > + < kjA >< bkp >)
i,j,k,a,b (i a)(E Eb)(E Eb)


(A-105)





(A-106)





(A-107)





(A-108)





(A-109)


(A-102)





(A-103)





(A-104)


> (<
(Ei




> (<
(Ei




(<
(Ei




(<
(Ei








(VC13 3) + (VC13 4)
-2 y fa, < bihaj > (< cjA >< bcp > + < bcA >< cjp >) (A110)
(Ei ,a)(Ej Eb)(Ej Ec)
a,b,c,i,j

(VC13 5) + (VC14 9)
2 fai < kilaj > (< bjA >< bkp > + < bkA >< bjp >) (A
i,j,k,a,b (E -a)(j b)(k b)

(VC13 6) + (VC14 6)
-2 fai < bilac > (< cjA >< bjp > + < bjA >< cjp >) (A 112)
(Ei Ea)(E'j Eb) (E- Ec)
a,b,c,i,j ( )()(

(VC13 13) + (VC15 14) =
4 z < aclik >< bilaj > (< bjA >< ckp > + < ckA >< bjp >) (A
a,,(i E,)(Ej Eb)(Ek Ec)
a,b,c,i,j,k

(VC13 14)
S < aklci >< bilaj >< bjA >< ckp > (A )
2 -(A-114)
ab,c,i,j,k (Ei a)( Eb)(E Ec)

(VC13 15) + (VC14 14)
2 < aclki >< bilaj > (< bjA >< ckp > + < ckA >< bjp >(115)
2 ------~ --- i --- i -- ^------ (A 1 1 5 )
ab,c,i,j,k (C E,)(Ej b)(Ek c)

(VC14 7) + (VC14 8) =
2 faj < ablij > (< bkA >< ikp > + < ikA >< bkp >) (A 6)
i,j,,,b ( a) a(i b)(Ek b)

(VC14 11) + (VC14 12) =
-2 faj < ablij > (< icA >< bcp > + < bcA >< icp >) (A117)
b,c,i,j (Ej Ea) (Ei Eb) (Ei Ec)








(VC14 13) + (VC15 15)
v-4 ^ < cjlka >< ablij > (< ibA >< cky > + < ckA >< ibp >)
*4 -,(A-118)
bcijk (E Eb)(EF Ea) (Ek Ec)

(VC14- 15)
2 < ablji >< kilac >< jbA >< ck > (A 9)
ab,c,i,j,k (i a)( Eb)( c)

(VC15 7) (VC15 8)
8 fa < ablij > (< bkA >< jkp > + < jkA >< bkp >) (A 20)
i,j,k,a,b ( i Ea)(Ei Eb)(E- k Eb)

(VC15 9) + (VC15 10) =
8 fai < ablij > (< cjA >< bcp > + < bcA >< cjp >) (A 2
aL,(Ei E,)(E Ec)(Ej Eb)


(VC15 11)
-8 Y fai < aklij >< jbA >< kb >(A 22)
-8^ -(A-122)
ijkab (Ei Ea)(Ej Eb)(Ek Eb)

(VC15- 12)
8 z fa < ablic >< bjA >< cj > (A 2)
abci,j ( a)( Eb)(EF Ec)

(VC15 13)
16 < ablij >< cika >< bjA >< ck > (A 24)
a,(Eib,c,i,j a)( b)(Ek c)

(V1) + (V2)
-2 < ilVcla > (< abA >< i > + < bA >< ab >) (A 25)
,b ( a)(- b)(A125)
i,a,b









< i|V|a > (< ajA >< ijp > + < ijA >< ajp >)
2jZ-( (F- F)(F Ea)


(VC5)= 2 <
iLj~a


il-lj >< aiA >< ajp >
(Eg j VF- -a)


(VC6) --2 < aVlb >< aiA >< bi >
iab ( a)(Ei b)


(V3) + (V4)


(A-126)


(A-127)


(A-128)









APPENDIX B
INTERPRETATION OF DIAGRAMS OF EXACT-EXCHANGE SECOND KERNEL

For the calculation of last term in equation (6-5)an auxiliary basis set decomposition

is used


< 'Oaa'(ha Oc Iga r- OkO >= < OaOiXA >< 1 TiXp >< 'Pco'CkuXv > g\Pv (B-1)


where X, are the auxiliary functions. The expression for g\,, can be written in the form

(B-2)


gA/\ = Z(X )A(X ),(X )"r + Z(X1)A(X-1), +
Ki7,rl K,r
(X- 1)(X-1),)A( + (X- 1)A(X- ),,o (B 2)


where

X, = 2 < ajA>< a (B-3)
Ei Fa

Two separate g\, should be constructed for each spin component. Expressions for 0,,,,

OXA, and 0,,, can be written in terms of the diagrams (Fl) (F18).

,-,, 2_ f ia,(< 'PaPjK >< ', .'rlI > + < Pa pjrI >< ', -.'.. >)
ija Ea)(Ej Ea)
-2 fiav(< ibbK >< Pa brl > + < bPil >< aIbK >)
-2
a(Eb Ea)(Ei Eb)
i,a,b
+2C fijv < OaOiK >< Oa)jrl > 2 fabv < 2i(4bK >< )Pia > (B4)
( a)( a) ab ( Ea)(E Eb)

where the integrals involving the exchange kernel can be calculated using the formula


fpqv >= < q > (X- 1)AQpX- 1)P/ (B-5)
KAp









Algebraic expression for EE,, cconsist of two parts. The first part is represented by the
sum of diagrams (VX1) (VX8) and (SX7) (SX14).

Dt 2 (<| V|iK I (> +Ka) F1
D1--
Z^~ (E Fa) E Fb)F Ec
a,b,c,i ^z oc)
F1= (< cpicp>< p Ob(l >< ,, > + < Oicc >< bcc >< ,abr > +

< iOc( >< Xb0c >< abK > + < icTl >< ibc >< ', > +

< ic(K> b0 >< bc Pabibrl > + < (b(ccl >< (b(0c >< (a(bK >) (B-6)


D2 2 ( +Kia) F2
a (Ei Ea E3 a) Ek a)
Fi,j,k,a
F2 = (< pckK >< <-, ,. >< ', ,( > + < Pak >K< > ,- > ,- i > +

< O >< -, -, >< .-,, > + < 'Oa X krl >< -,,-, >< r ,, > +

< k( >< >< r -l > + < k >< >< 'r- .ri >) (B-7)


D3 2- (< ciVVa > +Kia) F3
i,j,a,b ( )( )( b b)
F3 (< OK >< p .,, l ><, > + < PiPjK >< >< P ,, >< > +






nD4 2 ( +Kia) *F3
ijab i E)( Ea)( Eb)


D5 2 ( +K ) F4
D5 2
Si,j,a,b ( a)( b) Fa)
F4 = (< LpiOpbK < cpaW bl > + < pipbK X< < > +

< >< (Ca(Obrl >< PjPaK > + < ,Pi brl >< ,PaPb >< ; ; > +

< 6>< Xa(PbK >< J 1; ,r > + < Pibrl >< ',iC, >< (aI >) (B-10)









6 2 (< calV.cb > +Kab) F5
i,j,a,b ( E )(E- b)(j a)
F5 (< pacjK >< CcjT >< > < ai > + >< a >< i > > +

< Pa ( >< i pj T1 >< ib> + < (ij >< K,- -,C > +

< Pja(~C >< (i~(jK >< ibi ab)l > + < (~a(jri >< (jCi(O >< ()(bK >) (B-11)


D7 -2 (< Va V b > +Kab) F6
(,jki Ei EbEi E












< ><(b c i)(j- b)( a)
F6 7 (< kpip-,K >< jpbcl >< Xpipa,( > + < icK: >< bc, >< cica > +

< Pi Pc >< b (Pcr >< i~aK > + < b 7l >< b()O{cK >< r r ,C > +

< pi >< bO >< Ci Cal > + < bOcX >< bO >< ccaK >) (B-12)


( < oi| |o > +Kij) F2
D8 = -2 (B-13)
i Ek Ea) E Ea) (Ek Ea)
2,j,k,a

The second part is represented by the sum of diagrams (SX1) (SX6)


D9-2 < ^Oi|a^fcO > *F7
(,Kjkab K (Ek Eb )(Ej b)(E Ea)
F7 = (< Ob i.' >< OkTl >< a( > + < Wb i-. >< kcOck >< aiOrT > +

< P0bPk( >< (c(jTl >< iaK > + < (j(c >< (~Cc( >< Xia( > +





SD10= 2 < Pi Pa PjPb > F8
ijab,c Ec(- Eb)(Ei Ea)
F8 = (< ;. / >< cWbl >< Oa( > + < (r,-', I r i'< c X Wal > +

< OPjc >< OcC cl >< Oi OaK > + < i jscrl >< PcObK >< Oa( > +

< (PjPcc >< OPcObK >< Oal > + < ijOPcjl >< a rl >< Oi aK >) (B-15)










tb2 b)< ik\j b > *F9
i,j,k,a,b F b) -
F9 = (< ,-,,-, i.> >< pbpjrl >< ia > + < ',, i.K >< pbpj >< Cjac] > +

< ,, > >< b~ >< > + < ',- r, q >< b(6jK >< (OiOa > +

< '- >< j>< b( >< (pi(p > + < 'r, r-,7 q >< ib Pj >< (ri(baK >) (B-16)


D12 --2Z (< Papc \bcj > *F10
D12 -2
(Ei Ec) E( Eb (Ei Ea)
i,j,a,b,c c)( b)(i a)
F10 (< >, ,v-. >< cpipr] >< Cicga( > + < ><, .. >< ci< c >< aGil > +

< .< >< OcrX >< aOiaK > + < >' < ,* K>< c >< a( i(a> +

< ,- ,- >< cj >< ( i(al > + < '- 17 >< ><< >( i(aK >) (B-17)



D13 -2 << =b-a > *F11
D %:(E Ec) E_ Eb_)(i_ E)
i,j,a,b,c ( i
F11 = (< cp >< >< bj( > + < cK ac( >< > bjl > +

< iPc( >< OaOc)l >< Vb(PjK> + < icl >< acK b>< Vj > +

< i< acK >< Xb~Pj1l > + < Ol >< ac( >< Vb(jK >) (B-18)


D14 2 > < ,- ,.'~ ba > *F12
ijkab k Ea) (E Eb)(Ei Ea)
F12 = (< ickKX >< cackl >< ,, .< >+< kK Ci~ca>< O c k >< ',.>+

< ik( Cl > < ',OaOk .i.' > + < OOkl >< OaOk > 'K ..S > +

< >ik( >< OaOkK >< ,-. > + < cick X >< Oa(Ok >< '- .- >) (B-19)


To get final expression for BK all terms D1 D14 should be summed up.









REFERENCES

[1] A. Szabo and N. S. Ostlund, Modern Quantum C'il iii,,.ry, Introduction to Advanced
Electronic Structure Ti, ,.., (\lcGrow-Hill, Inc., New York, 1990).

[2] R. J. Bartlett, in Modern Electronic Structure Th(.. ,; Part I, edited by D. R.
Yarkony (World Scientific Publishing Co., Singapore, 1995).

[3] F. E. Harris, H. J. Monkhorst, and D. L. Freeman, Algebraic and Diagrammatic
Methods in M ri,;-Fermion Theory (Oxford University Press, New York, 1992).

[4] P.-O. L6wdin, J. C'!. in Phys. 19, 1936 (1951).
[5] P. Hohenberg and W. Kohn, Phys. Rev. 136 B864 (1964).

[6] W. Kohn and L. J. Sham, Phys. Rev. A 140 A1133 (1965).

[7] E. Runge and E. K. U. Gross, Phys. Rev. Lett, 52, 997 (1984).
[8] E. K. U. Gross and W. Kohn, Adv. Quantum. ('C!. ii, 21,255 (1990).

[9] M. E. Casida, in Recent Advances in D. -./il; Functional Methods,Part 1, edited by
D. P. C('!..i(World Scientific, Singapore, 1995).

[10] M. Petersilka, U. J. Gossmann, and E. K. U. Gross, Phys. Rev. Lett 76, 1212
(1996).

[11] J. Lindenberg, Y. Ohrn, Propagators in Quantum C'l ii,,..iry (John Wiley & Sons,
Inc, New Jersey, 2004).

[12] S. Ivanov and R. J. Bartlett, J. ('C!. ii Phys. 114, 1952 (2001).

[13] I. Grabowski, S. Hirata, S. Ivanov and R. J. Bartlett, J. Chem. Phys. 116,4415
(2002).

[14] A. G6rling and M. Levy, Phys. Rev. A 50,196 (1994).

[15] J. D. Talman and W. F. Shadwick, Phys. Rev. A 14 36 (1976).

[16] R. J. Bartlett, V. Lotrich and I. Schweigert, J. C'!. iin Phys 123, 062205 (2005).

[17] R. J. Bartlett, I. Grabowski, S. Hirata and S. Ivanov, J. Chem. Phys. 122, 034104
(2005).

[18] L. M. Frantz and R. L. Mills, Nucl. Phys. 15,16 (1960).

[19] Q. Zhao, R. C. Morrison and R. G. Parr, Phys. Rev. A 50, 2138 (1994).
[20] A. G6rling, Phys. Rev. A 57 3433 (1998).

[21] S. Hirata, S. Ivanov, I. Grabowski and R. J. Bartlett, J. Chem. Phys. 116,6468
(2002).









[22] S. Ivanov, S. Hirata and R. J. Bartlett, Phys. Rev. Lett. 83, 5455 (1999).

[23] M.Stadele,J. A. MiI. -1:; P. Vogl and A. Goerling, Phys. Rev. Lett. 79,2089 (1997).

[24] A. G6rling and M. Levy, Int J. Quantum ('C!. 'i Symp 29,93 (1995).

[25] D. P. ('!,.ii, O. V, Gritsenko, E. J. Baerrends, J. C'!. in Phys. 116 1760 (2002).

[26] R. J. Bartlett, I. Schweigert, V. Lotrich, J. Mol. Sruct. (Theochem) 771 1-8, (2206).

[27] P. O. Widmark, B. J. Persson, and B. Roos, Theor. Chim. Acta 79,419 (1991).

[28] K. i., i .1, ,. ESCA applied to free molecules, p. 31, (American Elsevier publishing,
New York, 1969).

[29] C. E. Moore, Atomic energy levels,(U. S. Department of Commerce,Washington,D.
C, 1949) Vol 1.

[30] R. Colle, R. K. N. -Let, J. Phys. B 34 2475 (2001).

[31] S. A. Kucharski, M. Wloch, M. Musial, R. J. Bartlett, J. ('!. i:: Phys, 115 8263
(2001).

[32] T. P. Straatsma, E. Apra, T. L. Windus, E. J. Bylaska, W. de Jong, S. Hirata, M.
Valiev, M. T. Hackler, L. Pollack, R. J. Harrison, M. Dupuis, D. M. A. Smith, J.
Nieplocha, V. Tipparaju, M. Krishnan, A. A. Auer, E. Brown, G. Cisneros, G. I.
Fann, H. Fruchtl, J. Garza, K. Hirao, R. Kendall, J. A. Nichols, K. Tsemekhman,
K. Wolinski, J. Anchell, D. Bernholdt, P. Borowski, T. Clark, D. Clerc, H. Dachsel,
M. Deegan, K. Dyall, D. Elwood, E. Glendening, M. Gutowski, A. Hess, J. Jaffe, B.
Johnson, J. Ju, R. Ko1. -- -hi R. Kutteh, Z. Lin, R. Littlefield, X. Long, B. Meng,
T. N 1,. ii, i S. Niu, M. R._.-i1r G. Sandrone, M. Stave, H. Taylor, G. Thomas,
J. van Lenthe, A. Wong, and Z. Zh Ii.- "NWC'!. i1, A Computational C('!. !!!-l ry
Package for Parallel Computers, Version 4.7" (2005), Pacific Northwest National
Laboratory, Richland, Washington 99352-0999, USA.

[33] H. Jiang and E. Engel, J. ('C!. ii Phys. 123, 224102 (2005).

[34] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77, 3865 (1996).

[35] K. Raghvachari, G. W. Trucks, J. A. Pople and M. Head Gordon, C'!I in Phys.
Lett 157, 479 (1989).

[36] http://webbook.nist.gov

[37] D. Bokhan, I. V. Schweigert, R. J. Bartlett, Mol. Phys,103, 2299 (2005).
[38] J. C. Slater, Phys. Rev. 81, 385 (1951).

[39] D. J. Tozer and N. C. Handy, J. C'!, in Phys. 108, 2545 (1998).









[40] J. P. Perdew, R. G. Parr, M. Levy and J. L. Balduz, Jr., Phys. Rev. Lett. 49, 1691
(1982).
[41] A. Balkova, R. J. Bartlett, J. ('C!. i, Phys. 102,7116 (1995).
[42] Ch. W. Bauschliher, Jr. and P. R. Taylor, J. ('C!. i, Phys. 86, 2844 (1987).

[43] J. Schirmer, A. Dreuw, Phys. Rev. A 75 022513 (2007).
[44] J. B. Krieger, Y. Li and J. Iafrate, Phys. Rev. A 45 101 (1992).

[45] V. N. Staroverov, G. Scuseria, E. R. Davidson, J. C'!. in Phys. 124 141103 (2006).
[46] W. Yang and Q. Wu, Phys. Rev. Lett, 89 143002 (2002).

[47] A. G6rling, Phys. Rev. Lett, 83, 5459 (1999).
[48] R. J. Bartlett, I. V. Schweigert and V. F. Lotrich, J. Mol. Struct: Theochem 771, 1
(2006).

[49] D. Rohr, O. Gritsenko, E. J. Baerends, C'!, in Phys. Lett. 432 336 (2006).
[50] S. Hirata, S. Ivanov, R. J. Bartlett, I. Grabowski, Phys. Rev. A 71, 032507 (2005).
[51] A. Dreuw, J. L. Weisman and M. Head Gordon, J. ('!. in, Phys. 119,2943 (2003).
[52] S. Hirata, S. Ivanov, I. Grabowski, R. J. Bartlett, K. Burke, J. D. Talman, J. ('!, 1'1
Phys. 115, 1635 (2001).

[53] P. O. L6wdin, Phys. Rev. 139, A357 A372 (1965).
[54] S. Ivanov, M. Levy, J. ('C!. in Phys. 119, 7087 (2003).

[55] H. Jiang, E. Engel, J. ('C!. i, Phys. 125 184108 (2006).
[56] 0. V. Gritsenko, E. J. Baerends, Phys. Rev. A, 64 042506 (2001).

[57] S. Ivanov, S. Hirata, I. Grabowskii, R. J. Bartlett, J. ('!., in Phys. 118 461 (2003).
[58] A. F. Izmaylov, V. N. Staroverov, G. E. Scuseria, E. R. Davidson, G. Stoltz and E.
Cances, J. ('!., ii Phys. 126, 084107 (2007).

[59] S. Kiimmel and J. P. Perdew, Phys. Rev. Lett, 90 043004 (2003).
[60] I. Schweigert, V. F. Lotrich and R. J. Bartlett, J. ('C!. ii Phys, 125 104108 (2006); I.
Schweigert, Ph. D. thesis, University of Florida (2005).
[61] R. J. Bartlett and M. Musial, Rev. Mod. Phys, 79 291 (2007).
[62] D. Bokhan, R. J. Bartlett, Phys. Rev. A, 73 022502 (2006).
[63] D. Tozer, J. Chem. Phys. 119 12697 (2003).









[64] S. J. A. van Gisbergen, P. R. T. Schipper, O. V. Gritsenko, E. J. Baerends, B.
C'!l 1,p1. ;iw., and B. Kirtman, Phys. Rev. Lett. 83 697 (1999).

[65] F. A. Bulat, A. Toro-Labbe, B. C'!i ip ,1. i B. Kirtman, W. Yang, J. C'!. in Phys.
123 014319 (2005).

[66] M. Kamiya, H. Sekino, T. Tsuneda and K. Hirao, J. C'!. in Phys 122 234111 (2005).

[67] R. J. Bartlett, D. M. Silver, Phys. Rev. A 10 1927 (1974).

[68] H. P. Kelly, Adv. C'., in Phys 12 314 (1963).

[69] R. J. Bartlett, H. Sekino, ACS Symposium Series 628: Nonlinear Optical Materials
23 57 (1996).

[70] A. M. Lee and S. M. Colwell, J. Chem. Phys. 101 9704 (1994).

[71] S. P. Karna and M. Dupuis, J. Comp. ('!I, ii 12 487 (1991).

[72] H. Sekino and R. J. Bartlett, J. ('!,. ii Phys 85 976 (1986).

[73] H. Sekino and R. J. Bartlett, J. ('!,. ii Phys 98 3022 (1993).

[74] D. Bokhan, R. J. Bartlett, J. ('C!. ii Phys. (submitted).









BIOGRAPHICAL SKETCH

I was born in Avdeevka, a small town located in the southeastern part of the

Ukraine in the Donetsk region. I grew up in the city of Gorlovka, a working-class mining

community. When I was in ninth grade, I acquired an interest in chemistry and decided

that I would become a chemist. During my last year of high school, I was one of the win-

ners of the Ukrainian ('!, iii-I ry Olympiad. As a result, I was admitted to the prestigious

Moscow State University without the otherwise necessary entrance examinations. During

my sophomore year of college, I decided to specialize in quantum chemistry. After I de-

fended my undergraduate thesis and obtained my degree in 2002, I began to work toward

joining Dr. Rodney Bartlett's research group at the University of Florida. After working

as a system administrator for several companies in Moscow for 16 months, I obtained the

necessary funds to l i, for transportation to the United States as well as for the necessary

exams, which included the Graduate Record Examination (GRE) and the Test of English

as a Foreign Language (TOEFL). Finally, I joined Dr. Bartlett's group at the University

of Florida on January 21st, 2004.





PAGE 1

1

PAGE 2

2

PAGE 3

IwouldliketothankprofessorsHenkMonkhorstandSoHirataforhelpfulldiscus-sions.IwanttothankalsoDr.IgorSchweigertandDr.NorbertFlockeforthehelpingmetowriteOEPcode.MyspecialthankstoTatyanaandThomasAlbertforhelpwiththepreparationofthedissertationtext. 3

PAGE 4

page ACKNOWLEDGMENTS ................................. 3 LISTOFTABLES ..................................... 7 LISTOFFIGURES .................................... 8 ABSTRACT ........................................ 9 CHAPTER 1INTRODUCTION .................................. 12 1.1AbInitioWavefunction-BasedMethods .................... 13 1.1.1Hartree-FockMethod .......................... 13 1.1.2Electron-CorrelationMethods ..................... 16 1.2Kohn-ShamDensityFunctionalTheory .................... 19 1.2.1Time-DependentDensityFunctionalTheory ............. 21 1.2.2Time-DependentDensityFunctionalTheoryLinearResponseTheory 23 1.2.3ProblemswithConventionalFunctionals ............... 26 1.2.4Orbital-DependentFunctionals ..................... 28 1.2.5AbInitioDensityFunctionalTheory ................. 29 2INTERCONNECTIONBETWEENFUNCTIONALDERIVATIVEANDEF-FECTIVEOPERATORAPPROACHESTOABINITIODENSITYFUNC-TIONALTHEORY .................................. 31 2.1EquationsfortheExchange-CorrelationPotentialintheFunctionalDeriva-tiveApproach .................................. 31 2.2EquationsfortheExchange-CorrelationPotentialinanEectiveOperatorApproach .................................... 34 2.3InterconnectioninArbitraryOrder ...................... 37 2.3.1DiagrammaticFunctionalDerivatives ................. 37 2.3.2DiagrammaticFunctionalDerivativesinSecond-OrderMany-BodyPerturbationTheory .......................... 38 2.3.3InterconnectioninHigherOrders ................... 41 2.3.4InterconnectioninInniteOrder .................... 45 3ABINITIOTIME-DEPENDENTDENSITYFUNCTIONALTHEORYEM-PLOYINGSECOND-ORDERMANY-BODYPERTURBATIONOPTIMIZEDEFFECTIVEPOTENTIAL ............................. 47 3.1DiagrammaticConstructionoftheExchange-CorrelationKernels ...... 48 3.1.1Formalism ................................ 48 3.1.2AnExample:DiagrammaticDerivationofExchange-OnlyKernel .. 52 4

PAGE 5

.................. 54 3.3PropertiesoftheCorrelationKernel ...................... 61 3.4NumericalTesting ................................ 62 3.5Conclusions ................................... 67 4ABINITIODENSITYFUNCTIONALTHEORYFORSPIN-POLARIZEDSYSTEMS ....................................... 69 4.1Theory ...................................... 69 4.2ResultsandDiscussion ............................. 71 4.2.1TotalEnergies .............................. 71 4.2.2IonizationPotentials .......................... 74 4.2.3DissociationEnergies .......................... 75 4.2.4Singlet-TripletSeparationinMethylene ................ 75 4.3Conclusions ................................... 75 5EXACT-EXCHANGETIME-DEPENDENTDENSITYFUNCTIONALTHE-ORYFOROPEN-SHELLSYSTEMS ........................ 80 5.1Exact-ExchangeDensityFunctionalTheory ................. 82 5.2Time-DependentOptimizedEectivePotential ................ 91 5.2.1TheoryandImplementation ...................... 91 5.2.2NumericalResults ............................ 92 5.2.3Charge-TransferExcitedStates .................... 95 5.3Conclusions ................................... 97 6EXACTEXCHANGETIME-DEPENDENTDENSITYFUNCTIONALTHE-ORYFORHYPERPOLARIZABILITIES ..................... 100 6.1Theory ...................................... 102 6.1.1Time-DependentDensityFunctionalTheoryResponseProperties .. 102 6.1.2DiagrammaticDerivationoftheSecondExact-ExchangeKernel .. 103 6.1.3PropertiesoftheSecondExact-ExchangeKernel ........... 106 6.2NumericalResults ................................ 107 6.3Conclusions ................................... 109 APPENDIX AINTERPRETATIONOFDIAGRAMSOFTHESECOND-ORDERMANY-BODYPERTURBATIONTHEORYOPTIMIZEDEFFECTIVEPOTENTIALCORRELATIONKERNEL ............................. 110 BINTERPRETATIONOFDIAGRAMSOFEXACT-EXCHANGESECONDKERNEL ....................................... 127 REFERENCES ....................................... 131 5

PAGE 6

................................ 135 6

PAGE 7

Table page 3-1Orbitalenergiesandzero-orderapproximationstoexcitationenergies ...... 64 3-2ExcitationenergiesofNeatomusingOEP-MBPT(2)Kohn-Shamorbitalenergies 64 3-3ExcitationenergiesofNeatomusingexchange-onlyorbitalenergies ....... 66 3-4ExcitationenergiesofNeatomusingorbitalenergiesandorbitalsfromOEP2(sc).AllequationsforTDDFTarethesame ....................... 66 4-1Totalenergies .................................... 72 4-2Ionizationpotentials(ine.v.) ........................... 74 4-3Dissociationenergies(inkJ/mol) .......................... 75 4-4Singletandtripletenergiesofmethylene ...................... 76 5-1Total(ina.u.)andorbital(ine.v.)energiesofNeatom ............ 88 5-2TotalandorbitalenergiesofHeatom ....................... 88 5-3Excitationenergies(V-valencestate,R-Rydberg) ............... 93 5-4Orbitalenergies(ine.v.)ofNeatom ....................... 94 5-5Ionizationenergies(ine.v.) ............................ 94 5-6Staticpolarizabilities(ina.u.) ........................... 99 5-7IsotropicC6coecients(ina.u.) ......................... 99 6-1Hyperpolarizabilitiesofseveralmolecules(ina.u.) ................ 107 7

PAGE 8

Figure page 4-1ExchangeandcorrelationpotentialsofLiatom(radialpart).A)Exchangepo-tential.B)Correlationpotential ........................... 73 4-2ExchangeandcorrelationpotentialsofO2moleculeacrossthemolecularaxis.A)Exchangepotential.B)Correlationpotential .................. 73 4-3LiHpotentialenergycurve. ............................. 77 4-4OHpotentialenergycurve. .............................. 78 4-5HFpotentialenergycurve. .............................. 79 5-1ExchangepotentialsofNeatom,obtainedindierendbasissets ......... 88 5-2A1charge-transferedexcitedstateofHe...Be ................. 96 5-3LUMO-HOMOorbitalenergydierence ...................... 96 8

PAGE 9

9

PAGE 10

10

PAGE 11

11

PAGE 12

^Hel=1 2elecXi=1r2ielecXi=1NuclXA=1ZA ^Hel(R1:::RM)(r1:::rN;R1:::RM)=E(R1:::RM)(r1:::rN;R1:::RM)(1{2)cannotbesolvedanalytically.Forthecalculationofelectronicenergylevels,someapproximationshavetobeused.Modernquantumchemistryincludesthreeclassesofmethodsforthesolutionoftheelectronicproblem.Therstone,calledwavefunctiontheory,usesdierentapproximationsforthecalculationofthewavefunction(r1:::rN;R1:::RM)andthecorrespondingenergyE(R1:::RM).Thesecondclass,knownasdensityfunctionaltheoryusesthedensityasaprimaryobject.WithintheDFTapproach,theenergyiswrittenasafunctionalofdensity,andtheconstructionofthewavefunctionsisnotnecessary.Thethirdclassofmethodsusesawholedensitymatrix.Thedensitymatrixrenolmalizationgroup(DMRG)andthedensitymatrixfunctinaltheory(DMFT)aretypicalmethodsofthethirdclass.AbinitioDFTcontainselementsofthersttwoclasses.Itusesalocalmultiplicativepotential,typicalfortheDFTapproach.However,thecorrespondingexchange-correlation 12

PAGE 13

1.1.1Hartree-FockMethodIntheHartree-Fockmethod,awavefunctionisconsideredasaSlaterdeterminant[ 1 ] HF='1(r1)'1(rN)......'N(r1)'N(rN)(1{3)Thesingle-electronwavefunctions'i(r)(ororbitals)aredeterminedbytheconditionthatthecorrespondingdeterminantminimizestheexpectationvalueoftheelectronic 13

PAGE 14

1{3 )intoanexpectationvalue,itispossibletowritetheHartree-Fockenergyintermsoforbitals 2r2NuclXAZA 2elecXi;j(<'i'jj'i'j><'i'jj'j'i>)(1{5)ToderivetheHartree-Fockequationsitisconvenienttominimizethefollowingfunctional 1{6 )requiresthatthefunctionalderivativesofIwithrespecttotheorbitalsvanish 'i=0;I 'i=0(1{7)TakingintoaccountEquation( 1{3 )condition( 1{7 )canbepresentedas 2r2+^vext+^vH+^vnlx(1{9) 14

PAGE 15

^vext(r)=elecXiNuclXAZA (1{10) TheHartree-Fockexchangeoperator^vnlxisnon-local,i.e.itcannotbepresentedasananalyticalfunctionofspatialvariables.However,itispossibletowriteitsactiononsomeorbital'i 1{8 )canberewritteninthecanonicalform 15

PAGE 16

FCI=HF+occXivirtXaCaiai+occXi6=jvirtXa6=bCabijabij+:::(1{14)whereai,abij,etcareSlaterdeterminants,formedbysubstitutionofoccupiedorbitalsi,j...byvirtualorbitalsa,b...withthecorrespondingreorderingofrows.TheexpansioncoecientsarefoundfromthevariationalconditionontheexpectationvalueoftheHamiltonian (1{15) 16

PAGE 17

2 ]andMany-BodyPerturbationTheory[ 3 ].AnytruncatedversionoftheCImethodhasaqualitativelywrongbehavioroftheenergiesandwavefunctionswhileincreasingthenumberofparticlesinthesystem.Therefore,theCImethodswithlimitedlevelofexcitationscannotbeusedforhighly-correlatedsystems.TheCoupledClustermethodandMany-BodyPerturbationtheoriesarefreefromthislackofextensivityfailureandareverycommonforthemolecularcomputations.Insomecasesperturbationtheorycanprovideanaccuratedescriptionofelectron-correlationeectsatamuchlowercostthannecessaryfortheCoupledClustermethod.Thesecond-orderRayleigh-Schrodingerperturbationtheoryisthesimplestandleastexpensiveabinitiomethodfortakingintoaccountelectroncorrelationeects.InthisperturbationtheorythesolutionofSchrodingerequation ^H=E(1{16)canbefoundusingtheSlaterdeterminantasareference.Generally,suchadeterminantmaybeconstructedfromtheorbitals,generatedbysomeone-electronoperator ^h'p=(1 2r2+^u)'p="p'p(1{17)TherststepofanyperturbationtheoryisthepartitioningoftheHamiltonianintoazero-orderH0andperturbation ^H=^H0+^V(1{18)where ^H0=E0=Xi(hi)=(elecXi"i)(1{19) 17

PAGE 18

^V=^H^H0=elecXi^vext(ri)elecXi^u(ri)+elecXi6=j1 ^H=^H0+^V =+(1)+2(2)+::: Theseorder-by-ordercorrectionscanbefoundbysubstitutingexpansions( 1{21 ),( 1{22 )and( 1{23 )into( 1{16 )andcollectingtermswiththecorrespondingorderin (E0^H0)j(2)>=(^VE(1))j(1)>E(2)j> Choosingtheperturbativecorrectionstobeorthogonaltothereferencedeterminant<(n)j>=0(socalledintermediatenormalization)itiseasytogettheexpressionsforperturbativecorrectionsatanyorder.Projectingtheequations( 1{24 )and( 1{25 )ontothereferencedeterminantexpressionsforenergycorrectionsareobtained Expressionsfortheorder-by-orderexpansionofwavefunctionscanbewrittenusingtheresolventoperator[ 4 ] 18

PAGE 19

^R0=^Q E0^H0(1{30)and^Q=1-j>.Since^H0isdiagonalinthebasisofSlaterdeterminants,itispossibletowrite ^R0=Xn6=0jn>() 1{17 ),thesecond-ordercorrectiontotheenergyhasthefollowingstructure 2occXi;junoccXa;b() 5 ],establishesaone-to-onemappingbetweentheground-stateelectronicdensityandtheexternalpotential.Theexternalpotentialdenesaparticularobject(atom,molecule,etc)and,becauseoftheone-to-onemapping,thedensitycontainsalltheinformationaboutthesystem.Inparticular,theground-stateenergycanbewrittenasafunctionalofthedensity.Togetthegroundstateenergyin 19

PAGE 20

6 ]canbeused.Thistheoremstates,thattheground-stateenergyasafunctionalofdensityhasaminimum,ifthedensityisexact.Therefore,giventheenergyfunctional,onecanobtaintheground-statedensityandenergybyvariationalminimizationofthefunctional.TheformaldenitionofDFTdoesnottellhowtoconstructsuchafunctional.Severalapproximateformshavebeensuggested,however,whoseaccuracyvariesgreatlyfordierendproperties.Thekineticenergyofelectronicmotionisparticularlydiculttoapproximateasadensityfunctional.ThebasicideaofKohnandShamwastotransformthevariationalsearchoverthedensityintoasearchovertheorbitalsthatintegratetoagiventrialdensity (1 2r2+vs(r))'p(r)="p'p(r)(1{35)Suchatransformationdoesnotrestrictthevariationalspace,providedthateveryphys-icallymeaningfuldensitycorrespondstoauniquesetoforbitals(thev-representabilitycondition).TheuseoftheKohn-ShamSCFmodelensuresnotonlythatthevariationaldensityisafermionicdensity,butitalsoprovidesagoodapproximationforthekineticenergy.Iforbitalsintegratetothetruedensityitisnaturalytoexpectthat 2elecXi<'ijr2j'i>(1{36)accountsforalargepartofrealkineticenergy.Therestoftheunknowntermsaregroupedintotheexchange-correlationfunctional 20

PAGE 21

2Zdr1dr2(r1)(r2) 21

PAGE 22

7 ].Wecanstartfromthetime-dependentSchrodingerequation (t0)=0(1{43)undertheinuenceofdierentexternalpotentialsv(r,t).Foreachxedinitialstate0,theformalsolutionoftheSchrodingerequation( 1{42 )denesamap 22

PAGE 23

@t^Hj[](t)>(1{48)andthevariationalconditioncanbeapplied, @t'i(r;t)=(1 2r2+vs[(r;t)])'i(r;t)(1{50)where (1 2r2+v0(r)+Zdr00(r0) 23

PAGE 24

1{50 )provideaformallyexactwayofcalculatingthetime-dependentdensity,itispossibletocomputeexactdensityresponse1(r;t)astheresponseofthenon-interactingsystem 24

PAGE 25

1{55 )and( 1{56 )arethebasisoflinearresponsetheory.Sinceequation( 1{55 )isnotlinearwithrespectto1thesolutionshouldbeobtainedbysomeiterativeprocedure.However,forallpracticalpurposesadirectuseofequation( 1{55 )isnotconvenient.Togetamoreconvenientform,thefollowingmatrixelementshouldbeconsidered (1{59) Ontheotherhand,<'aj1(r;!)j'i>canbeexpressedintermsofaresponsefunction !+("a"i);Pai(!)=<'ijv1(!)j'a> 1{61 )intoequation( 1{59 )wehave (1{62) Introducingthenotations 25

PAGE 26

1{62 )canberewritteninthefollowingmatrixform 1{64 )canberewrittenas[ 8 { 11 ] 1{65 )iscommonforthecalculationofexcitationenergieswithtime-dependentdensityfunctionaltheory(TDDFT).MostmodernimplementationsofTDDFTusethefrequency-independentexchange-correlationkernels,denedasasimplefunctionalderivative 26

PAGE 27

27

PAGE 28

28

PAGE 29

2Xi;j<'i'jj'j'i>(1{67) 29

PAGE 30

30

PAGE 31

12 13 ]isastraightfor-wardapproachtotheabinitioDFT.ThecorrespondingKohn-Shampotentialcanthenbeobtainedbytakingthefunctionalderivativeofthenite-orderenergyviathechainrule,thattransformsthederivativewithrespecttothedensityintothederivativewithrespecttoorbitalsandorbitalenergies[ 14 ].ThisleadstotheOptimizedEectivePotential[ 13 15 ]equationswhichcanbecomplicatedalreadyinsecondorder[ 13 ]Alternatively,onecandeterminetherstandsecondorderKohn-Shampotentialbyrequiringthatthecorrespondingrstandsecondorderperturbativecorrectionstothereferencedensityvanish.Unlikethefunctionalderivativeapproachsuchaconditiononthedensitycanbedescribedwithstandardmany-bodytechniques.Therecentwork[ 17 ]usesdiagramstoderivethesecond-orderOEPequationinasystematicandcompactfashion,whileasecondpaper[ 16 ]doessoalgebraically.Boththefunctionalderivativeanddensity(eectiveoperator)approachesleadtoexactlythesameequationintherstorder[ 17 ].However,thefunctionalderivativeofthesecondenergyinvolvesacertaintypeofdenominatorsthatarenotpresentinthedensitycondition(eectiveoperatorapproach).Stillthetermsinvolvingsuchdenominatorscanbetransformedtomatchthedensity-basedequationexactly.Thiscaveatraisesthequestionofwhetherthetwoapproachesareequivalentinhigherorders,orfordierentpartitioningsofthehamiltonian. 31

PAGE 32

14 ]:E(n)=E(n)('1:::::'n;1:::::n)Anotherpossibleconsideration[ 13 ],whenorbitalenergiesareconsideredtobeafunctionalsovercorrespondingorbitals 14 ]approach.ToconstructthechainruleweshouldtakeintoaccountthefactthattheKohn-ShamorbitalsarefunctionalsofVs: 32

PAGE 33

(2{8) (2{9) Usingequations( 2{8 )and( 2{9 )wecanrewritethevariationoftheenergyinthefollowingform: 13 14 ]forthecalculationofV(n)xc: 14 ]: 13 ] 33

PAGE 34

2{15 )as: "i"a=Xp;q6=p "i"a=Xa;ifa+paqg(2{21) 34

PAGE 35

17 ].Thismeans,thatifweconstructaneectiveoperatorofthedensityusingMBPT,thecorrectiontothedensitymustvanishinanyorder.Thisisthemainideaoftheeectiveoperatorapproach.Fortheexchange-correlationpotentialofrstorderwewillhave: 2{23 ). (1)(2{23)Usingthefactthatfpq="ppq+<'pjVHFxVx()Vcj'q>,wecanextractourdesirableexchangepotential.Thecorrelationpartisexcludedtomaintainrstorder,asthecorrelationpotentialcontainsexpressionsofsecondandhigherorders.ThentheequationforVxcanbediagrammaticallyrepresentedbydiagrams2and3. Forthesecond-ordereectiveoperatorforthedensitycorrectionwewillusetheexpression( 2{24 ) (2{24) 35

PAGE 36

2{25 ). 2 =4 +4 ++2 +2 ++4 +4 +2 +2 + + +2 +2 ++4 +2 +2 (2{25) Diagrammaticrepresentationofpotential,derivedfromthedensityconditionwillbeusedfortheestablishinginterconnectionwiththefunctionalderivativeapproach. 36

PAGE 37

2.3.1DiagrammaticFunctionalDerivativesForformulationoftheruleswewilluseequation( 2{18 ).The-functionneedstobeintroducedaccordingtoequation( 2{26 ): + + + (2{26)Bracketsonthelasttwodiagramsdenotedenominators.ExpressionsfortheenergyinMBPTconsistoflinearcombinationsofterms,whichhaveproductsofmolecularintegralsinthenumeratorsandproductofdierencesofone-particleenergiesinthedenominators,soitiseasytomakeadiagrammaticrepresentationforsuchexpressions.Totakethefunctionalderivativesfromnumerators,accordingtoequation( 2{18 ),alllinesconnectedtosomevertexmustbedisconnectedfromthecorrespondingplaceoflinking,andwhenthisisdone,anewline(correspondingtooccupiedorunoccupiedorbitals)mustbeinserted.Inthenalstepcontractionwiththecorresponding-functionmustbeprovided.Alllinescorrespondingtodenominatorsarestillunchanged.Thisproceduremustbeprovidedforallvertices,becausewhenwetakefunctionalderivativesfromproductsoffunctions,wehaveasumofproducts,accordingtorulesfortakingderivatives.Ifanewdiagramchangesitssign,aminussignmustbeassignedtothisdiagram.Whenfunctionalderivativesfromdenominatorsaretaken,itismoresuitabletouseequation( 2{13 ).Themostgeneralformforthedenominatorcanberepresentedbythefollowingformula: 37

PAGE 38

2{13 )itispossibletowritedownthefunctionalderivativesfromsomedenominatorinthegeneralform: Vs=(K1Xi=1j'i1occj2K1Xj=1j'j1unoccj2)Y;6=1(KXi=1ioccKXj=1junocc)+:::::+:::::+(KXi=1j'ioccj2KXj=1j'junoccj2)Y;6=(KXi=1ioccKXj=1junocc) (2{28) Nowitispossibletowritedownatermwhichincludesthefunctionalderivativefromthedenominator: Vs==XNom(PKi=1j'ioccj2PKj=1j'junoccj2) (PKi=1ioccPKj=1junocc)2Q;6=(PKi=1ioccPKj=1junocc) (2{29) Wenowhavethepossibilitytoformulatehowtotakefunctionalderivativesfromdenom-inators.Totakefunctionalderivativesfromdenominators,alldiagrams,whereoneofthelinesisdoubledandbetweentheselineswhicharisefromdoubling,thecorrespondingdiagonal-functionisinserted.Thisproceduremustbeprovidedforallhorizontallinesonthediagramandforallthecontoursthelinescross. 2{17 )isequaltodiagram2,theright-handsiteisequaltodiagram3whenthefunctionalderivativehasbeentakenfromtheexpressionfortheexchangeenergy.Thediagrammaticexpressionforthesecond-orderenergyinthegeneralcasehastheform[ 1 ]: +2 +2 (2{30) 38

PAGE 39

2{31 ) + (2{31)Insertingthelinesofunoccupiedorbitalsandmakingcontractionswiththe-function,diagrams4and8ofequation( 2{25 )willbeobtained( 2{32 ): 2 +2 (2{32)Whenfunctionalderivativesaretakenwithrespecttounoccupiedorbitalsandoccupiedorbitalsareinserted,diagrams5and9ofequation( 2{25 )willbeobtained.Whenthisprocedureisdoneforthelowervertex,complexconjugatediagramswillbeobtainedandwewillhavethesamenumberofdiagramsofthissortasintheeectiveoperatorapproach.Whenfunctionalderivativesaretakenwithrespecttooccupiedorbitalsfortheupperandlowervertexesandanoccupiedlineisinserted,wehavethediagrams,givenbyequation( 2{33 ) (2{33)"Bracket-type"denominatormeansthedierence"i"k.SummationofthesttwodiagramsaccordingtotheFrantz-Millstheorem[ 18 ]andthesameprocedureforthesecond 39

PAGE 40

2{34 ) 2 +2 (2{34)Onthetwodiagramsweimposetherestrictioni6=k.Tomakeadirectcorrespondencewithdiagrams7and11weneedtoadddiagramsthatarisefromtakingfunctionalderivativesfromdenominators( 2{35 ) 2 2 (2{35)Afterthesediagramsareaddedtotheprevioustwo,wewillhaveadirectcorrespondencewithdiagrams7and11.Whenfunctionalderivativesaretakenwithrespecttotheunoccupiedorbitalsandanunoccupiedlineinserted,thesameprocedurewillgiveadirectcorrespondencewithdiagrams6and10ofequation( 2{25 ).Nowconsidertherstdiagramforthesecond-orderenergyexpression.Thepro-cedurefortakingfunctionalderivativesfromtherightsideofthevertex,whichdonotcontainaclosedF-ringareabsolutelythesameasinthecaseofothersecondandthirddiagramsintheenergyexpression.Whensuchfunctionalderivativesaretaken,wehavefourdiagramscorrespondingtodiagrams12-15.TotakefunctionalderivativesfromtheF-rings,therstdiagramintheenergyexpressionmustberepresentedwithmoredetail,takingintoaccountthatf=h+J-K,infollowingtheform( 2{36 ) = + + +2 +2 +2 (2{36) 40

PAGE 41

2{37 ) 2 +2 +2 +2 (2{37)Thersttwodiagramsareequaltodiagram16;thethirdandfourthequaltodiagrams17and16respectively.Sincetheleft-handsiteofeq.12canberepresentedbydiagram3,wehaveanexactequivalencebetweenthefunctionalderivativeandtheeectiveoperatorapproachesintherstandsecondorderforthe"Kohn-Sham"partitioningofthehamiltonian. 2{38 ) 41

PAGE 42

18 ]factorizationtheorem.Sinceallinitialhorizontallinescrossalllines,whichgoupfromlowerlyingvertexes,itispossibletorepresentthediagrams,whichappearafterapplyingtheFrantz-Mills[ 18 ]theoremintheform( 2{39 ) = + (2{40)Thediagramontheleft-handsideistheresultoftakingfunctionalderivativesfromtypicalenergydiagramsofthirdorder.Applyingthefactorizationtheorem[ 18 ]totheright-handsidewewillhavethesameresultasontheleft-handsideofthisrelation.The 42

PAGE 43

2{41 ): (2{41)Numbersonlinesshowtheconditionalnumberoflineswhichareenteredforsimplicityoffurthermanipulations.Thedesirablesumofdiagramshastheform( 2{42 ) + +:::::::::+ (2{42)Togetherwithdiagramswhichcorrespondtofunctionalderivativesfromdenominators,thissetofdiagramsformsproductslike(k)+(l),wherek+lequalstheorderofMBPT. 43

PAGE 44

2{43 ) (2{43) IntheseexpressionsNommeansnumerator,Aimeanstherestoftheterms,whicharepresentindenominators.Aftermultiplicationanddivisionby(kl)wehave( 2{44 ) (kl)(l+A1)(k+A1)::::(k+Am)+XNom(kl+A2A2) (kl)(l+A1)(l+A2)(k+A2)::::(k+Am)+::::::::+XNom(kl+AmAm) (kl)(l+A1)::::(l+Am)(k+Am)==XNom (2{44) Aftercancellationofallequivalentterms,wehavetheexpressions( 2{45 ) 44

PAGE 45

2{46 ) + (2{46)Namelythesetwodiagramswillappearaftertakingfunctionalderivativesusingdia-grammaticrulesfortakingfunctionalderivatives.Proofforthecasewherefunctionalderivativesaretakenwithrespecttounoccupiedorbitalsisthecompleteanalogofthisone.Hence,wecanformulateasecondstatement.Statement2:Whenfunctionalderivativesfromsomevertexofadiagramaretakenwithrespecttooccupied(unoccupied)orbitals,andalinewhichcorrespondstooccupied(unoccupied)orbitalinserted,togetherwithdiagramswhicharisefromtakingfunctionalderivativesfromdenominators;wealwayshavediagrammaticexpressions,whichcorrespondstopartofPnk=0((k)+(nk))Sc.Thisstatementisadirectcorollaryoftheaboveprovedstatementabout"bracket-type"denominatorsandstatement1.Sincewehavecorrespondenceinthesecondandn-thorder,usingthemethodofmathematicalinduction,itispossibletoprove,thatthecorrespondencetaksplaceinallorders(niteorinnite). 45

PAGE 46

19 ]toexctractthecorrespondingexchange-correlationpotential.SincethefullCIenergydoesnotdependuponthechoiceoforbitalbasisset,theOEPprocedure[ 13 15 ]cannotbeuseddirectlyforthiscase.Theinnitesumofallenergycorrectionsdoesnotdependuponchoiceoforbitalbasisset,buteachtermofthissumdoesdependuponthechoiceoforbitals.Thisfactenablesustoconsideralltermoftheinnitesumofenergycorrectionsasorbital-dependentfunctionals.AftertheconstructionofthesetV(1)xc::::V(1)xcandsummationuptoinnity,wewillhavetheexchange-correlationpotentialwhichcorrespondstothefullCI.Usingtheequivalenceofthefunctionalderivativeandtheeectiveoperatorapproach,itispossibletoconcludethataftersummationofthissetofpotentials,weagainwillhavethesameresultasintheeectiveoperatorapproach.Redenedinsuchaway,theOEPprocedure[ 13 15 ]forthefullCIenergyproducesthesamedensityastheZMPmethod[ 19 ]wouldfromfullCI. 46

PAGE 47

20 ]andimplementedformoleculesbyHirataet.al[ 21 ].NumericalresultsinthatworkshowareasonabledescriptionofbothvalenceandRydbergexcitedstates,partlyduetothecorrectasymptoticbehavioroftheexchangepotential,butalsoduetotheeliminationoftheself-interactionerror.Inparticular,OEP-TDDFTissuperiortostandardapproacheslikeTDDFTbaseduponlocaldensityapproximation(LDA)orBecke-Lee-Yang-Parr(BLYP)functionals.Similarly,exchange-onlyOEPwithexactlocalexchange(EXX)[ 22 23 ]hasbeenshowntogreatlyimproveband-gapsinpolymers[ 23 ]AnotheradvantageofOEPbasedmethodsisthatsincevirtualorbitalsintheexchange-onlyDFTaswellasoccupiedorbitalsaregeneratedbyalocalpotential,whichcorrespondstotheN-particlesystem,thedierencesbetweenorbitalenergiesofvirtualandoccupiedorbitalsoeragoodzeroth-orderapproximationtotheexcitationenergies[ 16 17 ].ThisisnotpossibleinthecaseofHartree-Focktheory,whereoccupiedorbitalsaregeneratedbyanN-1particlepotentialandtheenergiesofunoccupiedorbitalscomefromN-electronpotential,andthereby,approximateelectronanities.OnceOEPcorrelationisadded[ 13 17 ],theessentialnewelementinthetime-dependentDFTschemeistheexchange-correlationkernel,whichintheadiabaticapprox-imationisdenedasafunctionalderivativeoftheexchange-correlationpotentialwithrespecttothedensity[ 20 ]: 16 17 ]thishasbeencalledOEPKohn-Sham(KS)todistinguishitfromotherchoicesforH0.Itis 47

PAGE 48

14 24 ].OnlyforthischoiceisthereanimmediatecorrespondencebetweenthefunctionalderivativeandKSdensityconditionthatasingledeterminantprovidestheexactdensity[ 24 ].TheotherchoicesforH0leadtowhatiscalledabinitioDFT[ 16 17 ]andbecomesabetterseparationofhamiltonianintoH0+V0.Itsconvergencefororbital-dependentperturbationapproximationstothecorrelationenergyfunctionalsareconsiderablybetterbehavedcomparedtotheKSchoicethatfrequentlycausesdivergence.Nevertheless,inthisrstapplicationwewilladheretothestandardKSseparation,wherethepotentialandkernelarefunctionalderivatives.Thetraditionalwayofderivingkernelsandpotentialsistotediouslyderivealltermswiththeuseofthechain-ruleforfunctionaldierentiation.Yet,eventheexchange-onlykernelhasacomplicatedstructureinOEP,anditsfurtherextensiontoincludecorrelationwouldbealmostimpossible.Toavoidtheuseoftraditionalmethods,aneectivediagrammaticformalismfortakingfunctionalderivativeshasbeendevelopedandimplemented. 3.1.1FormalismTheadiabatic(frequency-independent)approximationtothekernelofthenthorderisdenedas( 3{2 ) 13 ]( 3{3 ) 48

PAGE 49

3{3 )into( 3{2 )weobtain 'p(r4)(Zdr5E(n) Thismayberewrittenas( 3{5 ) Vs(r3)(Zdr5E(n) Equation( 3{5 )canbemademoreexplicit, Vs(r3)[Zdr5E(n) Usingthefactthat[ 14 ] 3{6 )inthefollowingform( 3{8 ) 3{3 )wehavetheexplicitexpression( 3{9 )forh(r3;r6) Vs(r3)[Xp;q6=pZdr5E(n) 49

PAGE 50

3{9 )willbeusedasthebasisforthediagrammaticprocedurefortakingfunctionalderivativesfromtheOEP-MBPTexchange-correlationpotentials.Thersttermofequation( 3{9 )containsthefunctionalderivativewithrespecttothepotential,fromanexpressionwhichalreadyincludesthefunctionalderivativeoftheenergywithrespecttothesamepotential.Diagrammaticrulesfortakingfunctionalderivativesfromnumeratorsofdiagramsarethefollowing: 2{26 )togetthefullyconnecteddiagram 2{26 )shouldbeinsertedbetweendoubledlinesforalllineswhichareintersectedbythedenominatorline.Thisprocedureshouldbedoneforallpossibledenominatorlines.TherulesofinterpretationarethesameasfortheusualGoldstonediagrams,exceptthenumericalfactorshouldbetakenfromtheinitialdiagramwhenfunctionalderivativesaretaken.Aftertakingthefunctionalderivativesfromtheexchange-correlationwithrespecttothepotentialusingdiagrammaticrules,asetofdiagrams,containingthe^-functionvertexwillappear.Togettherstelementofequation( 3{9 ),itisnecessarytoapplydiagrammaticrulestothesetoffunctionalderivativediagramsonemoretime.Afterthatwewillhaveasetofdiagramscontainingtwo^-functions,whicharenecessaryfortheconstructionoftheexchange-correlationkernels.Takingintoaccountequation( 2{14 ),thesecondtermofequation( 3{9 )canberewritteninthefollowingway( 3{10 ) Vs(r3)['i(r5'a(r5)'i(r6)'a(r6) 50

PAGE 51

3{10 )canberepresentedbydiagram(V)( 3{11 ) (3{11)Thefunctionalderivativesfromdiagram(V)producethesetofdiagrams(V1)-(V6)ofset( 3{12 ) (3{12)Duringthederivationofdiagrams(V1)-(V4),diagramscontaining"bracket-type"denom-inatorswillappear,becauseoneofthestepsinthediagrammaticrulesofdierentiationrequiresdetachingtheunoccupiedlinefromdiagramVandinsertingthelasttwopartsofthe^-function( 2{26 ).Suchdiagramscanbetransformedintoasetofregulardiagrams,usingthediagrammaticrelation( 3{13 ).Thisrelationshowsusthatforthetransformationtoregulardiagramsitisnecessarytodoubleeachofthedenominatorlinesandinsertthelasttwomembersofthe^-function( 2{26 ),subjecttotherestrictedsummation,l6=k.Weusethesameprocedureforthecaseofoccupiedorbitals. + = + +...+ (D1)(3{13) 51

PAGE 52

3{9 )anddiagrams(V1)-(V6)itispossibletobuildtheexchange-correlationkernelstoanyorder. 20 ],andthenrederivedandinitialyimplementedbyHirataet.al[ 21 ].Suchaderivationrequiresmucheort.Hereweoerafacilederivationwiththediagrammaticformalism.Theexchangeenergy(Ex=-1 2Pi;j)canberepresentedbydiagram( 3{14 ). (EX)(3{14)AftertakingtherstfunctionalderivativewithrespecttoVswewillhavethediagram( 3{15 ). (3{15)Afterthatthediagrammaticrulesmustbeappliedonemoretimetogetthesecondfunctionalderivative,asisnecessaryaccordingtoequation( 3{9 ).Aftertakingthesecondfunctionalderivativeswewillhavediagrams(FX1)-(FX8)oftheset( 3{16 ) (3{16)Aftertheadditionofdiagrams(V1)-(V6)wewillhaveallthediagramsnecessaryforbuildingtheexchange-onlykernel.Duringtheinterpretationofdiagrams(V1)-(V6)the 52

PAGE 53

3{10 ).Interpretationofdiagrams(FX1)-(FX8)and(V1)-(V6)givesustheexpression( 3{17 ) ("i"a)("j"b)2Xi;j;a;b'i(r3)'a(r3)'j(r6)'b(r6) ("i"a)("j"b)+2Xi;j;k;a['j(r3)'i(r3)'i(r6)'a(r6)+'j(r6)'i(r6)'i(r3)'a(r3)] ("i"a)("j"a)2Xi;j;a;b['a(r3)'b(r3)'b(r6)'i(r6)+'a(r6)'b(r6)'b(r3)'i(r3)] ("i"a)("i"b)+2Xi;j;k;a'i(r3)'a(r3)'j(r6)'a(r6) ("i"a)("j"a)2Xi;j;a;b'i(r3)'a(r3)'i(r6)'b(r6) ("i"a)("i"b)+2Xi;j;a'i(r3)'a(r3)'j(r6)'a(r6) ("i"a)("j"a)2Xa;b;i'i(r3)'a(r3)'i(r6)'b(r6) ("i"a)("i"b)+2Xi;j;a['i(r3)'j(r3)'j(r6)'a(r6)+'i(r6)'j(r6)'j(r3)'a(r3)] ("i"a)("j"a)2Xa;b;i['a(r3)'b(r3)'b(r6)'i(r6)+'a(r6)'b(r6)'b(r3)'i(r3)] ("i"a)("i"b) (3{17) Expression( 3{17 )forh(r3;r6)exactlycorrespondstotheexpression,obtainedbyHirataet.al,butthediagrammaticderivationrequiresfarlesseortandisunambiguousintermsofsignsandnumericalfactors.Sinceallthediagramscontainonlyonecontour,itispossibletomakesummationsinequation( 3{17 )onlyoverspatialorbitals,and,asaresultf(r1;r2)andf(r1;r2)willappearseparately.Theexchange-onlykerneldoesnotcontainthef(r1;r2)part,whichisacriticaldierencebetweentheexchangeandcorrelationkernels. 53

PAGE 54

3{18 ). (3{18) Diagrams(VC1),(VC2),(VC5),(VC6)and(VC15)of( 3{18 )haveanexternalfactor4,diagrams(VC9)and(VC10)havenofactor,whiletherestofthediagramshaveafactorof2.Togetthecorrelationkerneldiagramsweneedtotakefunctionalderivativeswithrespecttothepotentialfromdiagrams(V1)-(V15).Afterapplyingthediagrammaticrulestodiagram(VC1)andtheuseofrelation( 3{13 ),diagrams(VC1-1)-(VC1-16)ofsets( 3{19 )and( 3{20 )willappear. (3{19) 54

PAGE 55

Dierentiationofdiagram(VC2)producediagrams(VC2-1)-(VC2-16)ofset( 3{21 ).Thesediagramshavethesameskeletonstructure,butdierentpositionsofindices. (3{21) Aftertakingthefunctionalderivativesfromdiagram(VC3)diagrams(VC3-1)-(VC3-14)( 3{22 )willappear. (3{22) 55

PAGE 56

3{23 )willappear. (3{23) Diagram(VC5)producesdiagrams(VC5-1)-(VC5-16)( 3{24 )afterdierentiation. (3{24) Dierentiationofdiagram(VC6)generatediagrams(VC6-1)-(VC6-16)( 3{25 ), (3{25) 56

PAGE 57

3{26 ). (3{26) Finally,diagram(VC8)producesdiagrams(VC8-1)-(VC8-14)( 3{27 ). (3{27) Inthesamewaytherestofthediagramsshouldbedierentiated.SincetheFockoperatordependsuponoccupiedorbitals,F-ringsonthediagrams(VC9)-(VC15)mustalsobedierentiated.Diagram(VC9),aftertakingfunctionalderivatives,producesdiagrams(VC9-1)-(VC9-10)( 3{28 ). (3{28) 57

PAGE 58

3{29 ). (3{29) Diagram(VC11),producesdiagrams(VC11-1)-(VC11-16)( 3{30 ). (3{30) 58

PAGE 59

3{31 ) (3{31) Dierentiationofdiagram(VC13)givesusdiagrams(VC13-1)-(VC13-15)( 3{32 ) (3{32) 59

PAGE 60

3{33 ) (3{33) Thelastsetofdiagrams(VC15-1)-(VC15-15)canbeproducedbydierentiationofdiagram(VC15)( 3{34 ) (3{34) Togetherwithdiagrams(V1)-(V6),alldiagramspresentedinthissectionformaset,whichisnesessarytoconstructthecorrelationkernel.TheinterpretationofalldiagramsisgiveninAppendix A 60

PAGE 61

(3{35) Diagrams,whichcontainonlyonecontourcannotproducehcomponentsafterdierenti-ation.Threeofthediagrams(VC5)-(VC8)afterdierentiationcanproduceonlydiagramscontributingtothehorhparts.Thediagrams(VC9-1)-(VC9-7),(VC9-9),(VC9-10),(VC10-1)-(VC10-7),(VC10-9),(VC10-10),(VC11-1)-(VC11-10),(VC11-13)-(VC11-16),(VC12-1)-(VC12-10),(VC12-13)-(VC12-16),(VC13-1)-(VC13-12),(VC13-14),(VC13-15),(VC14-1)-(VC14-12),(VC14-14),(VC14-15)havethesameproperty.Diagramscontaining2ormorecontourscanproduceallspincomponents.Setsofdiagrams(VC1-7)-(VC1-16),(VC2-7)-(VC2-16),(VC3-7)-(VC3-14),(VC4-7)-(VC414),(VC15-7)-(VC15-12)havetwocontours,butboth^-functionspresentinoneofthecontoursmeansthatthesediagramsmakeacontributiontohorhparts,buthaveanextrafactorof2.Thatfactorappearsaftersummationoverallspin-orbitalsofthesecondcontour.Diagrams(VC1-1)-(VC1-6),(VC2-1)-(VC2-6),(VC3-1)-(VC3-6),(VC4-1)-(VC4-6),(VC9-8),(VC10-8),(VC11-11),(VC11-12),(VC12-11),(VC12-12),(VC13-13),(VC14-13),(VC15-1)-(VC15-6),(VC15-4)-(VC15-15)have^functionsondierentcontours,sotheycontributetobothhandhparts.Diagram(VC15-13)haveanadditionalfactorof2andalsocontributeintobothspinparts.Tobuildallspinpartsofthecorrelationkernelweneedtosubstitutetheabovediagramsintoequation( 3{35 )Thenextessentialpropertyoftheexchangeandcorrelationkernelsisthesymmetrywithrespecttopermutationofitsarguments 61

PAGE 62

3{35 )and 3{35 )weagainneedtoanalyzethestructureofthediagrams.Diagrams(VC1-2),(VC1-5),(VC1-7),(VC2-1),(VC2-6),(VC2-7),(VC5-2),(VC5-7),(VC5-9),(VC6-2),(VC6-7),(VC6-9),(VC11-6),(VC12-6),(VC13-14),(VC14-15),(VC15-13),(VC15-11),(VC15-12),(V5)and(V6)aresymmetricwithrespecttothepermutationofvariablesr1andr2.Afterconsiderationoftherestofthediagrams,itiseasytoseethatforeachdiagramthereisanotherdiagram,whichdiersfromtheinitialonebyonlytheinterchangedvariables,r1andr2.Suchpairsprovideinvariancewithrespecttopermutationofvariables.DuringtheconstructionofthecorrelationkernelKohn-Sham(KS)orbitalsandorbitalenergiesareused.Thisimpliesthefactthatourzero-orderhamiltonianischosentobetheKohn-Shamchoice 1{65 )shouldbesolved.IntheadiabaticapproximationithastheRPA-likeform.Thecriticalnewquantityistheexchange-correlationkernel,f(xc);whichcanbeseenfromtheforegoingisaquitecomplicatedquantityifweinsistuponobtainingitrigorouslyforOEP-MBPT(2).ThediagonaldependenceofAonaitellsusthatifwearetogetgoodexcitationenergies, 62

PAGE 63

16 ]whenaccurateexchange-correlationpotentialsareused.Inthissense,theorbitalenergiesinKS-DFTshouldhaveacertainmeaning.Furthermore,aspointedoutinref[ 16 ],wecanalsoconsiderthisequationtooeraKoopmans-likeapproximationtotheprincipalionizationpotentials,since,barringpathologicalbehavior,whenweallowanelectrontobeexcitedintothecontinuum,itsorbitalawillthenhavenooverlapwiththeboundorbitalmatrixelementsofhKSandthekernel.Consequently,weareleftwithnothingbut-iintheTDDFTequations.Sointhis'sudden',adiabaticapproximation,theKSorbitalenergiesshouldoeranestimateforeachoftheprincipalIp's,notjustthehighest-occupied(homo)one.WhenbasedupontherelativelycorrectVxcobtainedfromabinitiodft[ 16 17 ]thisestimateissuperiortoKoopmans'theoremforthehomoandtherstfewvalenceIp's,butisinferiorforthecoreorbitals[ 26 ].SeealsoChong,etal[ 25 ].However,theOEP2semi-canonical(sc)abinitiodft[ 16 17 ]approximationhasthedistinctadvantagethatitusesamuchbetterbehavedunperturbedHamiltonianthantheusualKSchoice,H0=PihKS(i).ToillustratetheevaluationofthekernelandthesolutionofTDDFTequations,weconsidertheNeatom.AlltheexcitedstatesinNecorrespondtoRydbergexcitedstates.Toobtainreasonablevaluesrequiresaquiteextensive,diusebasisset.WechoosetostartwiththeROOS-ATZPatomicnaturalorbitalbasis[ 27 ]consistingof(14s9p4d1f)primitivegaussianfunctionscontractedtoa[5s4p3d1f]set.Thisunderlyingbasiswasthenaugmentedbyasetofeven-tempereddiusefunctions[3s3p3d],withexponentialparameters=abn;as=0:015;ap=0:013;ad=0:012;b=1/3.Theauxiliarybasisischosentobethesame,butwithoutthepandddiusefunctions,sinceforthedescriptionoftheexchange-correlationpotential,diusefunctionsarenotnecessary.TheorbitalenergyestimatesareshowninTable 3-1 .Theionizationpotentialequationofmotioncoupled-cluster(IP-EOMCC)resultinthisbasisfortheNehomoIpis21.3eVwiththeexperimentalvaluebeing21.5645 63

PAGE 64

Orbitalenergiesandzero-orderapproximationstoexcitationenergies Exchange-onlyOEP-MBPT(2)orbitalIP[ 28 ]Orbital"a"HOMOOrbital"a"HOMOExcitationenergiesenergiesenergies(exp) 2s48.42-46.19-43.252p21.56-22.51-20.233s-5.1517.36-7.5812.6516.853p-2.5919.92-3.7216.5118.704s-1.9520.56-2.6617.5718.733d-1.4921.02-2.0518.1818.97 [ 28 ].ThiscomparestotheOEPxvalueof22.51andOEP-MBPT(2)valueof20.23.Theremainingunoccupied,butnegativeenergy3sorbitalischangedbyover2eVduetotheMBPT(2)correlation.TheOEP2(sc)abinitodftvaluechangesthisto5.18eV,attestingtothepoorconvergenceofthestandardKSpartitioningoftheHamiltonian.OnceOEP2(sc)calculationsaredoneforthe3p,4s,and3dstates,thereissimilaragreementbetweentheOEPxandOEP2(sc)results,contrarytothoseshowninTable 3-2 forthestandard(ks)choice. Table3-2. ExcitationenergiesofNeatomusingOEP-MBPT(2)Kohn-Shamorbitalenergies TermEOM-CCSDTDDFT-exchange-onlyTDDFTOEP-MBPT(2)Exp[ 29 ]. Despitethedierences,atleastboththeexchange-onlyandtheOEP-MBPT(2)givequalitativelycorrectresultsfortheRydbergseries.Helpingtoensurethisisthefactthattheexchangepotentialhasthecorrectasymptoticbehaviorandtheexactcancellation 64

PAGE 65

30 ].WhenwesolvetheTDDFTequationswiththekerneldevelopedinthispaper,weobtaintheresultsforexcitationenergiesshowninTable 3-2 .Ingeneral,justasweobservedfromtheorbitalenergydierences,theOEP-MBPT(2)resultstendtofallonthelowsideofexperimentandtheOEPxonthehighside.Infact,exceptfortheexcitationenergyforthelowest1Pstate,wheretheveryloworbitalenergyof-7.58biasestheresults,anaverageofthetwowouldseemtobeaboutright.Thereare,however,stilldiciencesinthebasisset,asseenbytheEOM-CCSDresultscomparedtoexperiment,wherebesidestheadditionalcorrelationeectsintroducedbyEOM-CCSDT[ 31 ],thefurtherextensionofthediusefunctionswouldremovethe~1eVerrorinthehighestlying(1S)state.Ofcourse,thedependenceofTDDFTonthebasisandthatforatwo-particletheorylikeEOM-CCshouldbequitedierent.ThegreatsensitivityoftheresultstotheorbitalenergiesfromtheunderlyingKS-DFTcalculationcanbefurtherappreciatedbysimplytakingtheenergiesfromtheOEPxresultsandusingthemintheevaluationofthekernelandthematrixelementsintheTDDFTequations.TheseresultsareshowninTable 3-3 .ThenearcoincidenceofresultsforOEPxandOEP-MBPT(2)isapparent,withbothnowbeingtoohigh.AsimilarexperimentcanbemadewhereweuseorbitalsandorbitalenergiesfromOEP2(sc)abinitodftresultsasdescribedelsewhere[ 16 ],toobtaintheresultsinTable 3-4 .ThatistheTDDFTequationsandthekernelareassumedtobethesame,butweuseorbitalsandorbitalenergiesobtainedfromOEP2(sc).HerewealsoshowtheresultsfromstandardTDDFTapplicationsusingtheLDAandB3LYPfunctionalsforcomparisonpurposes.Clearly,wehaveimprovedresultsatboththeOEPx(sc)andOEP2(sc)levels,withthesamepatternoftheformerbeingtoohigh,butlessso;andthelatter,toolow,butbetterthanbefore.Itisapparentthattheproperwaytoachievethebenetsof 65

PAGE 66

Table3-3. ExcitationenergiesofNeatomusingexchange-onlyorbitalenergies TermTDDFT-exchange-onlyTDDFTOEP-MBPT(2)Exp[ 29 ]. Table3-4. ExcitationenergiesofNeatomusingorbitalenergiesandorbitalsfromOEP2(sc).AllequationsforTDDFTarethesame TermEOM-CCSDB3LYPLDATDDFT-TDDFT-Exp[ 29 ].[ 32 ][ 32 ]exchange-OEPonly(sc)MBPT(2)(sc) ThecornerstoneofOEP(sc)forexchange-correlationpotentialsisthatwecanimposetheconditionthattheKSdeterminanthastogivetheexact,correlateddensity.ThenthisdensityconditionprovidesequationsthatdeneVxcforagivenfunctional[ 17 ].Sincenoenergyvariationalconditionorfunctionalderivativeisused,wecanchangethechoiceofH0fromtheKSonetothesemi-canonical(sc)onethatoersgreatlyenhancedconvergenceofperturbationtheory.ButthenthereisnoapparentimmediatecorrespondencetoafunctionalderivativeasthereisinthestandardKStheory.Asimilar 66

PAGE 67

16 17 26 ],theradiusofconvergenceispoor,causingpoorconvergence[ 16 ]indeterminingthecorrelationpotential.Nevertheless,therststepintheunderlyingframeworkhasnowbeendenedtoapplyTDDFTwithOEPcorrelationpotentials.However,thecomplexityisgreat,thoughimpositionofadditionalrigorousconditionsmightresultinsimplicationofthekernel.Lackingsuchsimplications,thequestionarisesaswhetherthisisacaseofdiminishingreturnstoevaluatesome203diagramstoretaintheattractive,one-particle,correlatedstructureofDFT?First,theanswersforNearenotverygoodattheOEP-MBPT(2)level.Furthermore,thebasissetdependenceofOEPmethodswhendoneingaussianbasisisseveretoevengettheVxcright[ 16 ].ThefailureofmostsuchOEPcalculationstosatisfytheexactHOMOcondition=isacaseinpoint[ 16 26 ]. 67

PAGE 68

31 ])arelesstime-consumingthanisthepresentcalculation.Aslongasthekernelforarigorousorbital-dependentcorrelationpotentialisthiscomplicated,two-particlewavefunctiontheorieslikeEOM-CCarebothsuperiorandeasiertodo.ButbeforewecanfurtherexploittheinterfacebetweenDFTandwavefunctiontheorytothebenetofboth,itisrequisitetohavetheorbital-dependentkerneldened.This,anditsinitialevaluation,iswhatthischapteroers. 68

PAGE 69

^(r)=Xp;q<'pj(rr1)j'q>a+paq(4{1)UsingWicktheorem,werewritethisoperatorinnormalformwiththeKohn-ShamdeterminantastheFermi-vacuum: Thesecondtermisks,hence,thersttermwillbecalledthedensitycorrection: (r)=Xp;q<'pj(rr1)j'q>fa+paqg(4{3)SincetheconvergedKohn-Shamschemegivesanexactdensity,allcorrectionstothisdensitymustbeequaltozero.Hence,ifweconstructaneectiveoperatorofthedensityusingMBPT,thecorrectiontothedensitymustvanishinanyorder[ 17 24 ].TherstandsecondordersoftheMBPTdensityconditioncanbewritteninthefollowingway (4{5) UsingtheKSpartitioningofthehamiltonian[ 13 ]andtakingintoaccounttherelationfpq="ppq+<'pjVHFxVx()Vcj'q>,itispossibletoderivethesameexpressionsfor 69

PAGE 70

13 33 ].However,thisisanumericallyhopelessprocedurewithoutanyresummationofterms:(1)itsuersfromaddingalargediagonaltermintotheperturbation;(2)MBPT(2)isnotboundedfrombelowandanyeorttousethevariationalconditionE(2)c However,toavoiddealingwithnon-diagonalH0,asemi-canonicaltransformationwillbeperformedtoobtainthemoreconvenientzeroth-orderhamiltonian, (fii+fjjfaafbb)(fifc)2;XXi;j;k;a;b()'k(r1)'a(r1) (fii+fjjfaafbb)(fkkfaa);XXi;j;k;a;b()'k(r1)'i(r1) (fii+fjjfaafbb)(fkk+fjjfaafbb) 70

PAGE 71

(fii+fjjfaafbb)(fii+fjjfccfbb)+Xa;b;ifibfia'i(r1)'a(r1) (fiifaa)(fiifbb)Xi;j;afajfij'i(r1)'a(r1) (fiifaa)(fjjfaa)+Xa;b;ifiafib'a(r1)'b(r1) (fiifaa)(fiifbb)Xi;j;afaifaj'i(r1)'j(r1) (fiifaa)(fjjfaa)Xi;j;a;bfbj(+)'i(r1)'a(r1) (fiifaa)(fjjfbb))X1(r1;r) (4{9) where 13 17 ]. 4.2.1TotalEnergiesGround-stateenergiesarecalculatedforseveralsystemswithresultspresentedinTable 4-1 .TheuncontractedROOS-ATZPbasisisusedfortheNeatom,whiletherestoftheatomsandmoleculesarecalculatedusingtheuncontractedROOS-ADZPbasis[ 27 ].Allthemoleculesandtheirionsareconsideredtobeintheequilibriumgeometry[ 36 ]ofthecorrespondingneutralsystem,exceptwhenpotentialenergycurvesareconsidered.ForabinitioDFTcalculationsexchange-correlationpotentialsareusedfrom[ 13 ]andequations( 4{8 )-( 4{9 ).ForthecomparisonenergiesobtainedfromKSDFTwiththePerdew-Burke-Ernzerhof[ 34 ](PBE)exchange-correlationpotentialandcoupled-clusterwithsingle,doublesandperturbativetripleexcitation[ 35 ](CCSD(T))arealsoshowninTable 4-1 .Totalenergies,calculatedwithOEP-MBPT(2),basedonthedensityconditionapproachwiththesemi-canonicalpartitioningaremuchclosertoCCSD(T)thenOEP-MBPT(2)withtheKSpartitioning.Thelattergreatlyoverestimatescorrelationenergy 71

PAGE 72

37 ]tothevariationalOEP-MBPT(2)fortheKSpartitioningithasthesameproblems.However,inthecaseofthenon-variational,semi-canonicaldensityconditionapproach,theunboundnessfrombelowislessofaproblem.Thesemi-canonicalchoiceofH0providesgoodapproximationstotheenergyandwavefunctioninMBPT(2)andthecorrespondingtotalenergiesaremuchclosertohighly-accurateCCSD(T)ones,atleastinthechosenbasis.ThecomputationalcostoftheOEP-MBPT(2)methodiscomparablewiththecostofMBPT(2),thescalingofbothmethodsisN5,whereNisthenumberofbasisfunctions.ThisismoreexpensivethanconventionalDFT(scaleslikeN3),butlessexpensivethenCCSD(T),computationaltimeofwhichisproportionaltoN7.Thecorrelationpotential, Table4-1. Totalenergies MP2OEP-KSOEP-semiPBECCSD(T) calculatedwiththesemi-canonicalOEPfortheNeatom[ 16 ]isclosetothecorrespondingquantumMonte-Carlopotential,andwecanexpectthatforothersystems,whereQMCisnotavailable,semi-canonicalOEP-MBPT(2)potentialsshouldbeagoodalternative.Fortheopen-shellcasepotentialsproducedbyKohn-ShamOEPshowanoverestimationofthecorrelationenergy,thesamesituationpreviouslyreportedfortheclosed-shellcase[ 16 ].(Fig 4-1 and 4-2 ).Thecorrect1 72

PAGE 73

4-1 and 4-2 )isguaranteedbytheuseoftheColle-Nesbetseedpotentialalgorithm[ 30 ].InallourcalculationsaSlaterpotential[ 38 ]wasusedasaseedpotential,sincetheColle-Nesbetalgorithmrequirestheseedpotentialtobeascloseaspossibletotheactualpotential.TwoseparateSlaterpotentials,oneforalphaandtheotherforbeta,wereused.Foralltheclosed-shellsystemsfromTable 4-1 ,thenumerical Figure4-1. ExchangeandcorrelationpotentialsofLiatom(radialpart).A)Exchangepotential.B)Correlationpotential Figure4-2. ExchangeandcorrelationpotentialsofO2moleculeacrossthemolecularaxis.A)Exchangepotential.B)Correlationpotential resultsobtainedfromOEP-MBPT(2)withthesemi-canonicalpotentialandfromDFT 73

PAGE 74

4-2 .InexactDFT,theenergyofthehighestoccupiedmolecular Table4-2. Ionizationpotentials(ine.v.) HOMO,OEP-semiE,OEP-semiHOMO,PBEE,PBEExp Ne21.0121.5913.3521.6921.56N216.8915.0310.2715.4115.58CO13.6813.699.0513.8714.01CN4.193.510.1493.723.86H2O12.3712.737.0812.4612.62N15.2514.528.3014.7314.53Li4.895.383.235.595.39 orbital(HOMO)correspondstothenegativeoftheexactverticalionizationpotential.Inthecaseofthesemi-canonicalOEPtheHOMOenergyisclosetothecorrespondingEvalues,whatcanbeexplainedbythecorrectshapeoftheexchange-correlationpotential(Fig 4-1 and 4-2 andref[ 16 17 ]),whilePBEdoesnotexhibitcorrectbehaviorofpotentialandthus,failstoreproducethecorrectHOMOenergyvalues.FailuretoreproducethecorrectHOMOenergycausestheincorrectintegerdiscontinuityofthePBE(oranyotherGGA)exchange-correlationpotential[ 39 40 ]. 74

PAGE 75

4-3 .PotentialcurvesfortheLiH,OHandHFmoleculesareshownonFig 4-3 4-4 and 4-5 .ForallcurvesthedissociationenergieshaveapproximatelythesamelevelofaccuracyasintheMBPT(2)case,butthesemi-canonicalOEP-MBPT(2)improvestheshapeofthecurves. Table4-3. Dissociationenergies(inkJ/mol) PBEOEP-semiExp. 41 42 ]arereported,usingtheuncontractedROOS-ADZPbasissets:13s9p3dforcarbonand8s4p1dforhydrogen.Equilibriumgeometriesforbothstatesaretakenfromreference[ 41 ].Theenergiesforthetwostates,calculatedwithdierentmethodsarepresentedinTable 4-4 .Therearesignicantdierencesintheenergyseparationcomparedtotheexperimentalvalue(8.998kcal/mol)forallofthepresentedmethods.Thisisabasissetissue,butcanalsobeexplainedbythefactthatthesingletstateofmethylenehasasignicantcontributionfromtwo-determinants[ 41 ],soMBPT(2)isamuchpoorerunderlyingapproximationfortheirdierencethaninnite-orderCCSDorCCSD(T),andespeciallythetwo-determinantCCSD(TD-CCSD)results[ 41 ].WheretheabsolutevaluesofPBEenergiesarenottoogoodforthetwostatesofCH2,thedierenceisconsistentwithMBPT(2)anditsOEP-generatedabinitiodftform. 75

PAGE 76

Singletandtripletenergiesofmethylene CCSD(T)-39.127004-39.10975110.826CCSD-39.123412-39.10456911.824OEP-semi-39.103994-39.07915215.588PBE-39.108321-39.08344815.610MP2-39.102565-39.07767415.619 76

PAGE 77

LiHpotentialenergycurve. 77

PAGE 78

OHpotentialenergycurve. 78

PAGE 79

HFpotentialenergycurve. 79

PAGE 80

5 6 ](DFT)isawidelyusedmethodforground-stateenergiesandotherproperties.Thetime-dependentextensionofKohn-ShamDFT,whoserigorousfoundationhasbeenquestioned[ 43 ],isaverypopularmethodforthedescriptionofexcitedstatesenergies,sinceitcanoftenprovideaccurateresultsbythediagonalizationofasingle-excitationdimensionalmatrixasintime-dependentHartree-Fockormono-excitedCI.HoweverKSDFTandTDDFThavetheirnaturallimitations.Standardfunctionalssuerfromanincompletecancellationofself-interactiontermsintheKSequationsandincorrectlong-rangeasymptoticbehaviorofthepotentials.Bothcanbeparticularlytroublesomeforexcited,particularlyRydbergstates,andionizedstates[ 16 21 48 ],makingitdiculttogetsystematicallyimprovableresultsfromstandardTDDFT.Exact-exchangedensityfunctionaltheoryisbasedontheoptimizedeectivepo-tentialmethodintroducedbyTalmanandShadwick[ 15 ].Inthismethodtheexchangepotentialisdenedasthefunctionalderivativeofthenon-localorbital-dependentex-changefunctionalfromtheHartree-Fockmethod(EXX),ieEX=(1)=VX(1):TheOEPmethodisfreefromtheself-interactionproblem,itexhibitsthecorrect1 44 ] 16 22 45 { 47 ],andthesealternativeapproachescansuerfromnumericalproblems.However,whenevaluatedwithapropertreatmentofthedensity-densityresponsematrixX,mostofthenumericalproblemsthathavebeenencounteredareresolved. 80

PAGE 81

14 ],butessentiallymodiedbyBartlettetal[ 16 17 ]toavoidthefailuresofasimplesumofKSone-particleHamiltoniansastheunperturbedproblem,togeneratecorrelationpotentials.Thisisthecornerstoneofabinitiodft.Theconceptualdierenceisthatthedensityconditiondoesnotexplicitlyusethevariationaldetermination,EXC=(1)=VXC(1).Thisdierenceisofcriticalimportanceingeneratingcorrelationpotentialsfromlow-ordersofperturbationtheory.Withoutthatchange,nolow-orderorbitaldependentcorrelationfunctionallikethatfromMBPT2willgenerallywork,butwiththosechanges,itdoesverywell[ 16 48 ].Thosemodicationstotheperturbationtheoryalsopertaintotheexchange-onlycase,thesubjectofthispaper,butforthatproblemthedistinctionsarelessimportant[ 16 ].Oncefollowingthisapproach,whichstartswiththeKSchoiceofH0;thedistinctionsbetweenapplyingthedensityconditionandusingdirectfunctionaldierentiationismoreconceptualthanessential,asthereisacorrespondenceinanyorderofperturbationtheory[ 24 37 ].Thedirectoptimizationprocedureadvocatedbysome,builtupontheabovevaria-tionaldeterminationofthefunctionalderivative,thoughformallyequivalent,diersfromtheoriginal,X-basedOEPinthedetailsofimplementation.However,aswaspointedoutbyStaroverovet:al[ 45 ]undercertaincombinationsofmolecularandauxiliarybasissets,particularlywhenthelatter'sdimensionislargerthanthatfortheformer,thedirectoptimizationmethodcangivetheHartree-Fockenergyanddensity.Thelattercanbeviewedasatrivialsolution,asitcanbeshowntocorrespondtothesolutionofaweightedleast-squaresexpression.Ontheotherhand,theX-basedOEPwithproperhandlingis 81

PAGE 82

49 ]observestheoccurrenceofadegen-eracyinperformingOEPcalculations,andtoinvestigatethis,intentionallyconstructaproblemwheretheHOMOandLUMOorbitalsoftheKSproblemaredegenerate.Subjecttodirectminimization,theyproceedtoreportonthefailureoftheOEPprocedure.Ontheotherhand,itisshowninthispaperthattheoriginalXbasedapproachhandlesthisproblem,too.Theexchange-onlytime-dependentOEPhasbeenconsideredbyGoerling[ 20 ]andimplementedbyHirataet.al[ 21 50 ].However,allresultstodateareforclosed-shellsystems.Tofurtheraddresstheexactlocal(OEPx)exchangeversustime-dependentHartree-Fock(TDHF),TDOEPxisgeneralizedtotreatexcitedstatesforopen-shellspecies.ResultsfromadiabaticTDOEPxareingoodagreementwithTDHFforbothexcitationenergiesandpolarizabilities,however,itisshownthatcharge-transferstatescannotbeproperlydescribedinTDDFT.ForstandardDFTmethods,gradient-corrected,hybrid,etc.thishasbeennoted,andimprovedupon[ 51 ],butstandardmethods,unlikeTDOEPx,stillsuerfromotherlimitations,liketheself-interactionerrorandtheincorrectlong-rangebehaviorofthepotentialsandkernels.Inanapproachthatgivesthe'rightanswerfortherightreason'theseexactconditionsarerequisite.InTDOEPxtherearenosucherrors.Hence,failuresofTDOEPxcomparedtoTDHFhavetobeexclusivelyduetothelocalversusnon-localexchangeoperator,whichisdemonstrated.Finally,C6coecientsareobtainedforopen-shellsystemsfromtheTDOEPxfrequency-dependentpolarizability. 82

PAGE 83

(5{3)bhS(1)=bh(1)+bJ(1)+bVX(1) (5{4)bJ(1)=XjZj(2)1 (5{5)=Z(1) (5{6)KS(1)=Xjj(1)j(1) (5{7)wheretheunoccupiedorbitalsareindicatedbyp=a,b,c,d....Theconditionthatthedensity,(1);betheHartree-Fockone,isthatfia=hHFijbh+bJbKjHFai=0 (5{8)HF(1)=XjHFj(1)HFj(1) (5{9)which,asiswell-known,iscorrectthroughrstorderincorrelationmeasuredrelativetothesumofHFone-particleoperators,H0=Pif(i);duetotheMoeller-Plessettheorem.ForKSorbitals,fiaisnotzero.Hence,iftheobjectiveweretomaximizethesimilarityintheKSandHFdensity,itwouldthenrequireminimizingaquantitycomposedofKSorbitals,relatedto 13 16 17 48 ].Clearly,thisminimizationwouldhavethetrivialsolutionthathijKjai=hijVXjai. 83

PAGE 84

13 16 17 48 ],sinceitalsotranscendsanyparticularseparationoftheHamiltonianinperturbationtheory.Inparticular,itisnotsubjecttoHKS0=PihS(i);andtherebyalleviatesthepathologicalbehaviorthatgivesnoconvergencewhendeningsuchcorrelationpotentialsfromMBPT(2),forexample[ 13 ].Fortheexchangeonly,however,sucheectsarelessimportant[ 16 17 ].ThereforethispaperwillusethesumofShamHamiltonians,buttheprincipleforderivationremainsthesame.Hence,requiringthat0=(1)KS=hKSjbR0bVjKSi+hKSjbVR0bjKSi (5{12)R0=Xi;ajaii(ia)1haij+XI
PAGE 85

52 ].Inadditiontoitspointwisecharacterrepresentedbythedeltafunction,itdiersfromtheaboveleast-squaresformbythepresenceofthedenominator.Theusualwaytowritetheaboveequationistointroducethenon-interactingdensity-densityresponsefunction, Thisistheusualoptimized-eectivepotentialprocedureofTalmanandShadwick[ 15 22 ],gearedtowardrepresentingtheVXoperatoronanumericalgrid.Theweightfactor,wia(1);isalsointroduced.Notethematrixelementscanbewrittenincongurationspaceas(VX)ai=hajbVXjii 85

PAGE 86

5{18 )thenbecomes, (5{24) Usingweightedleastsquares 5{16 ).Asiscustomaryinpractice,thecomputationalformfortheOEPequationcanbeobtainedbytwiceapplyingtheone-electronprojector, 53 ]oftheresponsefunctionisthen 5{18 )thatresultsintheYcolumnmatrix. 86

PAGE 87

52 ],unliketheinnitebasisanalogueofX(1,2),whichhastobesingular[ 52 54 ].Thelatterfollowsbecausetheinnitebasiscannotaccountforthearbitraryconstant.However,inanitebasis,Xisill-conditionedbecausethenull-spaceoffunctionsorthog-onaltoVXcanalwaysbeaddedtoVX,butwithzeroweight,andthisrequiresthattheOEPequationsproduceacomputed,numericalzero.Thismakestheequationsdiculttoinvert,recommendingasingular-valuedecomposition(SVD)procedure.ThisSVDshouldnotbeconfusedwiththeneedtoexcludelineardependencyinthecomputationalbasis,whichcanalwaysbeachievedindependently;butthat,too,wouldbeaccommodatedinpracticebytheSVDprocedure.TheroleofXistoimposethepoint-wiseclosenessof^Kand^VX,and,assuch,itiscriticaltothedeterminationofOEPpotentials.IntheabsenceofX,asattemptedbyStaroverov,etal.[ 45 ]unphysicalresultscanbeobtained.Accordingtotheirprocedure[ 45 ],togetHFenergiesanddensitiesfromtheOEPmethod(i.e.tondthetrivialsolutionofEqn( 5{26 )),thenumberofauxiliaryfunctionsusedmustbegreaterorequaltosomenumber,Nmin,whichisthenumberofnon-zeroKiamatrixelements.Inthiscaseequation(8)ofref[ 45 ]willhaveoneormoresolutions.Toillustrate,theuseoftheresponsematrixapproach,wereportinFigure 5-1 theresultsfortheNeatomexchangepotentialsusingthethreebasissetsusedbyStaroverov,etal,plusafourthuncontractedRoosbasisthatweprefer.TheauxiliarybasesarethesameasthemolecularonesandinallcasesthenumberofauxiliaryfunctionsislargerthanthecorrespondingNmin,necessaryforthesolutionofequation(8)ofref[ 45 ].ThetotalenergyvaluesfromTable 5-1 andcorrespondingpotentialsfromFig 5-1 arecompletelyoppositetotheresultsofref[ 45 ]becauseofthebrokenpointwiseclosenessandtheabsenceoftheself-consistentsolution.NotethatoncethepointwiseclosenessisimposedviaX,andaself-consistentsolutionobtained,thereisnoproblemfortheproperSVD-basedOEPprocedure. 87

PAGE 88

ExchangepotentialsofNeatom,obtainedindierendbasissets Table5-1. Total(ina.u.)andorbital(ine.v.)energiesofNeatom HOMOBasisNauxNminE(OEP)E(ref[ 45 ])OEPHF AUG-CC-PVTZ179-128.533191-128.533273-33.58-23.16AUG-CC-PV5Z2515-128.545887-128.546786-14.01-23.14AUG-CC-PV6Z2518-128.546287-128.547062-17.80-23.14ROOS-ADZP6139-128.545016-128.546596-23.74-23.14 InspiredbyanobservationofJiang[ 55 ]ofHOMOandLUMOdegeneracyarisinginOEP,Rohretal[ 49 ]oeredaspeciccombinationofmolecularandauxiliarybasis,thatshowedtheireectforHe,thenusingthedirectoptimizationalgorithmproducesdegenerateHOMOandLUMOorbitalenergies.WeperformedOEPcalculationfortheHeatomusingtheoptimizedauxiliarybasisfromTable1ofref[ 49 ],whichcausedtheirdegeneracy,butinsteadofusingthedirectminimizationalgorithmofBroyder-Fletcher-Goldfarb-Shanno(BFGS)thesolutionoftheOEPequationissolvedbytheinversionofXwiththeSVDprocedure.TheseresultsarepresentedinTable 5-2 ,demonstratingthat Table5-2. TotalandorbitalenergiesofHeatom OEPHFOEP2-KSOEP2-sc Totalenergy,a.u.-2.86115365-2.86115365-2.90524738-2.89488588HOMOenergy,e.v.-24.970-24.970-24.529-24.787LUMOenergy,e.v.3.59115.5613.6713.618 88

PAGE 89

5-2 ,eventhenotoriouslyunstablesecond-orderOEPprocedure,basedonHKS0hasnoparticulardiculties,either.Sincethedirectminimizationproce-dureofYangandWu[ 46 ]isbasedonthesameequationastheregularOEPmethod,alloftheproblems,describedinref[ 49 ],arecausedbyusingtheBFGSalgorithminsteadofsolvingtheOEPFredholmequationwiththeSVDprocedurefortheX.Thatmightappeartobesurprisingsinceh=l,wouldmeanthattheresolventoperatorismanifestlysingular,clearlynotthatappropriatetoanon-degenerateperturbationtheoryfunctional,whichhasalwaysbeentheintentofsuchorbitaldependentexpressions.However,itisalsoapparentfromthestandard,chain-ruledierentiationusedbyRohretal[ 49 ],namely 44 ]whomadesuchanapproximationintheGreen'sfunctionwherethedenominatorconsistsofpqbeforesimplifyingthefromtotheoccupied-virtualseparationinvokedhere[ 16 54 57 ].)SeeGritsenko,etal[ 56 ].Suchaconstantenergydenominatorisarelativelypainlessapproximationhere,sinceavgappearsonbothsidesofeqn.(18),makingitdisappearfromtheequationforVX:Oncethatisdone,aresolutionoftheidentitycanbeinvokedtoeliminatethevirtualorbitalstogive, 89

PAGE 90

59 ]Ofcourse,usingtheusualnitebasissetcomputationaltoolsofquantumchemistry,thedistinctionislessimportant.Thisformulahasrecentlybeenapplied[ 58 ].BecausetheOEPequationisapoint-wiseidentityhavingtobezeroformallyforallxinanitebasissetitcorrespondstoamany-to-fewmappinginthegeneralcase.ThedependenceuponthedeltafunctionmakestheOEPproceduresensitivetotheauxiliarybasissetusedinthecalculation.However,theSVDprocedurehandlesthenullspacefunctionsplusanypotentiallineardependencyinthebasissetinafairlyautomaticway[ 60 ].IftheXmatrixisremoved,thisisnotthecase,andsomealgorithmscanresultinunphysicalortrivialsolutionsundercertaincombinationsofauxiliaryandmolecularbases.Choosingtoinvoketheaverageenergydenominatorandtheresolutionoftheidentitybeforetheleast-squaresmminimization,willnotgiveEqn(24).However,onceweeliminatethedeltafunctionthatmakestheapproximationpoint-wise,wereducetheproblemtotheminimizationofthevariance minVar=minXihij(bK+bVX)2jiiXi;jhijbK+bVXjjihjjbK+bVXjii(5{34)thatcanbeusedtodeneaVX,butonethatisnotpoint-wise,and,consequently,doesnotsatisfyalltheconditionsabove.ThisfurthersimplicationoftheweightedleastsquaresapproachinEqn( 5{26 )hasalsobeenconsideredrecently[ 58 ].Ultimately,oneprimarycriteriaforthebestexchange-onlypotentialshouldbethesatisfactionoftheJanaktheorem.AsshowninTable 5-1 ,andpointedoutpreviously[ 16 48 ]thisisdiculttoachieveinanynormalbasisset. 90

PAGE 91

5.2.1TheoryandImplementationWithintheadiabaticapproximationtheTDOEPequationshavethefollowingmatrixform( 5{35 ) 5{36 ) where with (g)=2Xi;a;b(Kba+VOEPba) ("i"a)("i"b)+2Xi;j;a(Kia+VOEPia)(+) ("i"a)("j"a)2Xi;j;a;b(+) (5{39) Oncetheeigenvalueproblem( 5{35 )issolved,wehavethesetofexcitationenergies,!. 91

PAGE 92

5{40 )shouldbesolved 5{40 )shouldbesolvedfortheimaginaryfrequenciesi!fromwhichthecoecientscanbycalculatedby 3(xx(i!)+yy(i!)+zz(i!)).Integrationofequation( 5{43 )wascarriedoutbyGauss-Chebyshevquadrature.Toensurethecorrect1 30 ]wasused.Asaseedpotential,weusedtheoneproposedbySlater[ 38 ] 36 ].Forallcalculationstheuncontracted 92

PAGE 93

27 ]hasbeenused.ResultsforexcitationenergiesarepresentedinTable 5-3 Table5-3. Excitationenergies(V-valencestate,R-Rydberg) TDHF"OEPi"OEPaTDOEPSVWNCCSDExp CN2(V)4.188.164.911.961.521.322+(V)5.4710.585.483.193.623.22CO+2(V)7.559.988.233.033.433.262+(V)11.149.3710.824.996.145.82CH32A01(R)6.526.536.425.005.885.722A002(R)7.937.977.935.937.187.44N4P(V)9.8411.119.8810.9810.8410.354P(R)13.5913.1613.1811.76 AsfollowsfromthersttwocolumnsofTable 5-3 ,TDHFandTDOEPxproduceresultsofapproximatelythesameaccuracy,aswouldbeexpectedbythefactthatthelocalexchangeinTDOEPismeanttobeakindofleast-squaresttotheHFnon-localexchangepotential.Thiscertainlymakestheoccupiedorbitalsquitesimilar,butthespectrumofunoccupiedorbitaleigenvaluesisverydierent,asshowninTable 5-4 ,sinceTDOEPxwillgenerateaRydbergtypeseriesinsteadofanythingliketheHFvirtuals.Asiswellknown,thelatteraredeterminedinapotentialofn-electrons,makingthemappropriateforelectronattachedstates,whiletheoccupiedonesfeelapotentialofn-1electrons.Tothecontrary,theorbitalsobtainedinOEPxhavethesamepotentialforanelectronintheoccupiedandunoccupiedorbitals,whichiswhythelattermorenearlysimulateRydbergstates,assomeoftheunoccupiedorbitalswillhavenegativeorbitalenergies.Ofcourse,inanitebasisaslongasthespaceseparatelyspannedbytheoccupiedandtheunoccupiedorbitalsisthesame,therewouldbenodierenceintheresults,thoughthediagonalvalues(butnotthetrace)wouldchange.Thesignicant 93

PAGE 94

Orbitalenergies(ine.v.)ofNeatom OEPxHFEOM-CCSD 1s-839.67-891.78-871.112s-47.36-52.52-48.112p-23.74-23.14-21.283s(3p)-5.213.663.703p(3s)-3.113.943.624p5.6616.0815.66 discrepancieswiththeexperimentaldatacanbeexplainedbythefactthatelectroncorrelationeectsarenotintroducedbyeitherTDHForTDOEP.Togetsomemeasureofthecorrelationeect,coupledcluster(CC)resultsfromEOM-CCSDarereported[ 61 ],andtheSlater-Vosko-Wilk-Nusiar(SWVN)functionalisalsoused.Thelatterproducesgoodresultsforvalencestates,buttheenergiesoftheRydbergstatesareunderestimated.ThisunderestimationisaconsequenceoftheincorrectasymptoticbehavioroftheSWVNpotential,forwhichJanak'stheoremisnotsatised(seeTable 5-5 )andisawell-knowncharacteristicthatstandardexchange-correlationpotentialsdonotoeraKoopmans-typeapproximationfortheorbitalenergies.HoweverforOEPexchange-onlyandwithcorrelation,ithasbeenproventhattheorbitalenergiesoerameaningfulKoopmansapproximationtoprincipalIP'sforthevalenceandmid-valencestates[ 16 ],soRydbergstatesshouldpotentiallybebetterdescribed.Ofcourse,denitivecomparisonwouldrequireOEPwithcorrelationaspresentedelsewhere[ 62 ],andthisisseenintheOEP2-scorbitalenergies[ 16 48 ],butafullTDOEPtreatmentofexcitedstates[ 62 ]requiresthecorrelationkernel.Thisdoesnotappeartobeaviablerouteyet. Table5-5. Ionizationenergies(ine.v.) HOMOEHFOEPSVWNHFOEPSVWNExp CN14.1614.619.7916.2216.3214.8013.60CH310.4611.195.388.988.9910.099.84N15.5316.148.4113.8913.8915.0014.53 ResultsforthepolarizabilitiesandC6coecientsarepresentedinTables 5-6 and 5-7 .Inallcases,thestaticpolarizabilitiesobtainedwithTDOEPareveryclosetothe 94

PAGE 95

63 ]andDreuwetal[ 51 ].Accordingtothelatteranalysis[ 51 ]eveniftheexactexchange-correlationfunctionalwereknown,itwouldstillbeimpossibletogetaproperdescriptionofcharge-transferexcitationswithinTDDFT.ConsidertheexampleofHe...Be.Thecharge-transfer1statecalculatedwithTDHFandTDOEP,takenwiththeTamm-Danco(mono-excitedCI)approximationforsimplicity.ForbothatomstheuncontractedROOS-ADZPbasissetwasused.Potentialcurvesofthe1charge-transferexcitedstatearepresentedonFig 5-2 .Thepotentialcurve,calculatedwiththecongurationinteractionsinglesmethodexhibitsthecorrect1 5{36 )areequaltozeroandtheexcitationfrequencyisequaltothedierenceoftheLUMOorbitalenergyoftheBeatomandtheHOMOenergyoftheHeatom.WhatwasnotaddressednumericallyinDreuwetal[ 51 ]wastheeectofhavingaproperself-interactionandasymptoticbehaviorinDFT,whichcanonlybeachieved 95

PAGE 96

A1charge-transferedexcitedstateofHe...Be withOEPmethods.WereportthoseinFig 5-3 .Obviously,theorbitalenergydierence Figure5-3. LUMO-HOMOorbitalenergydierence behavesverydierentlyforOEPandHFsinceinbothcasestheHOMOorbitalenergycorrespondstoanionizationpotentialapproximation,buttheLUMOenergycorrespondstoaRydbergstateapproximationfortheOEPmethodandtoanapproximateelectron 96

PAGE 97

5{36 )ascorrections,itisapparentthatTDDFThasaqualitativelywrongbehaviorineachorderwhendealingwithcharge-transferexcitations.IfonearguesthattheoccupiedorbitalsintheHFandOEPcalculationsarenearlythesame,andtheyspantheoccupiedspace,thentheunoccupiedspacewouldbethesame,too.However,evenifthespaceswerethesame,thedierencebetweenthenon-local^Kandlocal^Vxissucienttokeepthelatterlocalexchangefromeverbeingabletodescribeanelectronattachedstate.Thus,itisnotpossibletoovercomethelocaloperator'sfailureeveninthefullspaceforthecalculations. 16 52 ],anddoesrequiretheuseofageneralized(SVD)inverseortheequivalent.ItwasalsoshownthatundertherightconditionstheexactHFresultcanbeobtainedbyanOEP-likeprocedureasatrivialsolutionofaweightedleastsquaresprocedure.Anypoint-wiseconditionhastobeamanytofewmappinginanitebasisset.TheeliminationofdenominatorsfromworkingexpressionsforOEPx[ 20 44 56 ]canbedoneasshown,toobtainthecomputationaladvantageofusingonlytheoccupiedor-bitalsintheOEPprocedure.Thedirectoptimizationalgorithmasimplemented,however,hasledtowork[ 45 49 ]thathassuggestedincorrectconclusions.TheTDOEPmethodwasimplementedforspin-polarizedopen-shellsystems.Resultsforseveralopen-shellsystemsshowverysimilartime-dependentbehaviorforthetime-dependentresponseoftheOEPandtheHFpotentials.Thecorrectlong-range 97

PAGE 98

64 { 66 ]Theproblemofcharge-transferexcitedstatesinTDDFTrelatestotheinabilityofthezeroth-orderorbitalenergyapproximations,astheHFvirtualorbitalsandTDOEPexcitedorbitalsshowqualitativelydierentbehavior.IftheHFandOEPoperatorswerethesame,apropertreatmentusingthewholespaceshouldbeabletoovercomethislimitation,eveniftheusualzeroth-orderapproximationoeredbyTDDFTwouldbeapoorone.However,thisdierencearisesfromtheformoftheoperator,localversusnon-local,and,assuch,remainsafundamentalproblemfortheDFTmethoditself,ashasbeenobserved[ 51 ]Inconclusion,forallthepropertiesconsideredinthepaper,exceptforcharge-transferexcitedstates,thereseemstobenosignicantdierencebetweentheresultsofTDHFandexchange-onlyTDOEP,asonemightexpect.Hence,thefocusshouldremainonthecorrelationpotentialasitisinabinitiodft[ 16 17 48 62 ]. 98

PAGE 99

Staticpolarizabilities(ina.u.) TDHFTDOEPSVWNCCSD CNxx11.9811.9115.6312.50yy11.9811.9115.6312.50zz18.5518.4125.0023.4214.1714.0818.7516.166.576.509.3810.94CO+xx6.666.669.389.38yy6.666.669.389.38zz12.2212.1514.0612.508.528.4910.9410.425.545.504.693.13CH3xx15.0615.0018.7515.63yy14.0413.9818.7515.63zz15.0615.0018.7515.6314.7214.6618.7515.631.021.020.000.00Nzz6.246.236.255.94 Table5-7. IsotropicC6coecients(ina.u.) TDHFTDOEP CN...CN93.9792.06CH3...CH394.2793.40N...N19.2819.21 99

PAGE 100

20 ]andimplementedformoleculesbyHirataet.al[ 21 50 ].NumericalresultsinthelatterpapersshowareasonabledescriptionofbothvalenceandRydbergexcitedstates.Reasonablestaticanddynamicpolarizabilitiesarealsoobtained.GoodresultsofTDOEPxforexcitedstatesandpropertiesareobtainedpartlyduetothecorrectasymptoticbehavioroftheexchangepotential,butalsoduetotheeliminationoftheself-interactionerror.Inparticular,OEP-TDDFTxtendstobesuperiortostandardapproacheslikeTDDFTbaseduponthelocaldensityapproximation(LDA)orBecke-Lee-Yang-Parr(BLYP)functionals.Similarly,exchange-onlyOEPwithexactlocalexchange(EXX)[ 22 23 ]hasbeenshowntogreatlyimproveband-gapsinpolymers[ 23 ]AnotheradvantageofOEPbasedmethodsisthatsincevirtualorbitalsintheexchange-onlyDFTaswellastheoccupiedorbitalsaregeneratedbyalocalpotential,whichcorrespondstotheN-1particlesystemduetothesatisfactionoftheself-interactioncancellation,thedierencesbetweenorbitalenergiesofvirtualandoccupiedorbitalsoeragoodzeroth-orderapproximationtotheexcitationenergies[ 16 17 ].ThisisnotpossibleinthecaseofHartree-Focktheory,withoutaddingaVN1potential[ 67 68 ],sinceoccupiedorbitalsaregeneratedbyanN-1particlepotentialandtheenergiesofunoccupiedorbitalscomefromN-electronpotential,andthereby,approximateelectronanities.InKohn-ShamDFT

PAGE 101

^Vxc(r)=Exc 21 50 ].However,onceourobjectiveisthird-ordermolecularproperty,theessentialnewelementinTDDFTisthesecondexchange-correlationkernel,whichinanadiabaticapproximationisdenedasthesecondfunctionalderivativeoftheexchange-correlationpotentialwithrespecttothedensity 37 62 ].InthispaperweusethatformalismandapplyittotheexchangesecondkernelforOEPxbasedtime-dependentDFT.Afterdiscussingsomepropertiesofthesecondkernel,wereportnumericalresultstoobtainhyperpolarizabilities,comparedtothosefromHartree-Fockandcoupledclustersinglesanddoubles(CCSD).Thereisalonghistoryofthetreatmentofhyperpolarizabilitiesandassociatednon-linearoptical(NLO)properties.See[ 69 ]forareview.Inparticular,issuesoftheincorrect 101

PAGE 102

64 65 ]pointstothenecessityofanimprovedtheory.Arigorous(abinitio)DFTanaloguestartswithexchange-only,TDOEPx. 6.1.1Time-DependentDensityFunctionalTheoryResponsePropertiesWhenanexternaltime-dependentelectriceldisapplied,theperturbationcanbewrittenintheform( 6{2 ) (A+B)U=h(6{3)ThematricesA,Bhavethestructure,givenbyEqn( 5{36 )andhis 70 ] 102

PAGE 103

6{6 ) Fpq=hpq+Xa;i(Apqai+Bpqai)Uai ij=Fij 6{8 ) (6{9) Takingintoaccountthefactthat (6{11) where 103

PAGE 104

6{11 )canbewrittenas (6{13) andP1(r2;r6;r7)canbepresentedasasumofdiagrams(F1)-(F6)of( 6{14 ) (6{14)Inthesamewaythethirdtermofeq(16)canberewrittenas (6{15) whereP2(r1;r6;r7)givesthesumofdiagrams(F7)-(F12)of( 6{16 ) (6{16)ThediagrammaticexpressionforQ(r6;r7)containsterms(V1)-(V6)and(FX1)-(FX8)of( 6{17 )and( 6{18 )[ 62 ] (6{17) 104

PAGE 105

Functionalderivativesfromtheset(V1)-(V6)canbepresentedbytwotypesofdiagrams.Thersttypecontainstheexchangekernel,diagrams(F13)-(F18)of( 6{19 ) (6{19)Thesecondtypecontainstheexchangepotential,givingdiagrams(VX1)-(VX8)of( 6{20 ) (6{20) Thenotationr4nr5nr6meansthatforeachparticulardiagramallsixpermutationsofr4,r5andr6shouldbetaken.Thefunctionalderivativesfromdiagrams(FX1)-(FX8)leadtotheset(SX1)-(SX14)of( 6{21 )and( 6{22 ). (6{21) 105

PAGE 106

InterpretationofalldiagramswhichcontributesintothesecondkernelanddetailsoftheimplementationaregiveninAppendix B 38 ] 6{20 )willcanceldiagrams(SX7),(SX8)and(SX11)-(SX14)of( 6{21 )and( 6{22 )becauseoftheHOMOconditionandthebehavioroftheHartree-Fockexchangeforthecaseoftwoelectrons.Diagrams(VX6)and(VX7)willcanceldiagrams(SX1)and(SX2)becauseofEqn( 6{23 ).Diagrams(SX5),(SX6),(SX9)and(SX10)willcanceldiagrams(F1)-(F4),(F7)-(F10)and(F13)-(F16)becauseofEqn( 6{24 ).Finallythesumofthediagrams(SX3)and(SX4)cancelthesumof(F5),(F6),(F11),(F12),(F17)and(F18).Thusforthespecialcaseoftwoelectrons 106

PAGE 107

6{25 )mightbeusedasahintfordevelopingnewdensityfunctionals,asnoneoftheexistingfunctionalsexhibitsuchbehavior. 27 ]areusedandequlibriumgeometriesaretakenfromref[ 36 ].TheresultscalculatedwithdierentmethodsarepresentedinTable 6-1 .Intherstfourcolumnsarestatichyperpolarizabilitiesobtained Table6-1. Hyperpolarizabilitiesofseveralmolecules(ina.u.) HFOEP-HFOEPxLDACCSD LiHzzz312.130314.910312.126621.093691.406xxz201.150207.868201.641404.297204.171COzzz-31.016-24.864-29.455-66.40627.343xxz-3.073-4.660-3.118-13.2035.859CNzzz-1.2990.211-3.906-1.953164.02xxz-17.651-15.646-18.121-62.50068.359H2Ozzz-6.713-6.239-7.812-13.672-5.859xxz-0.497-0.604-1.215-6.138-3.870yyz-10.866-10.32-11.847-23.437-7.645 withHartre-Focknon-localexchangeandwiththetwolocalexchangeoperators:OEP-HF,whichmeansequation( 6{5 )withoutthesecondkernelterm,andOEPx,whichmeansalltermsinequation( 6{5 ).ThehyperpolarizabilityvaluescalculatedwiththeOEPxmethodareingoodagreementwiththecorrespondingHartree-Fockvalues,andgenerallynobetter.ThisgoodagreementcanbeexplainedbythefactthatOEPxmethodisfreefromtheself-interactionerror,hasthecorrectlong-rangeasymptoticbehavior,andthecorrectHOMOvalues.AshyperpolarizabilitiesareresponsepropertiesofthirdordertheircalculationwithintheDFTmethodrequiresverypreciseexchange-correlationpotentials.OnlypotentialswhichsimultaneouslysatisfyalltheoremsandconditionspertainingtotheDFTexchangepotentialscanbeexpectedtoreproduceeventheHartree-Fock 107

PAGE 108

66 ]itwaspointedoutthatnoneoftheexistinghybridandasymptoticallycorrectedfunctionalsarecapableofreproducingHartree-Fockhyperpolarizabilities.Thesemi-empiricalmethod,developedthere[ 66 ]istheonlyknownwaytogetreasonablehyperpolarizabilitieswithconventionalfunctionals.However,theOEPxpotentialsandkernelsaretheonlyrigorous,purelyabinitiowaytoobtainhigh-orderproperties,withinasolelyDFTframeworkoflocalpotentialsandkernels.ForthecalculationofhyperpolarizabilitieswiththeHartree-Fockmethod[ 71 ]thelasttermineq( 6{5 )shouldbedropped.ItsimportancefortheDFTschemecanbeestimatedfromthesecondcolumnofTable 6-1 ,whereOEPhyperpolarizabilitieswerecalculatedwithoutthelasttermofequation( 6{5 ).FortheexampleoftheLiHmoleculetherelativecontributionofthelasttermissmall.However,goingfromclosedshelltoanopen-shell,likefortheCNmolecule,therelativecontributionofthethirdkernelbecomesmuchgreaterand,sometimes,canevenchangethesignofhyperpolarizabilities.ThedierencesbetweentheOEPxandthecoupledclustersinglesanddoubles(CCSD)methodarebecauseintheexchange-onlyOEPmethodelectroncorrelationhasnotbeentakenintoaccount.Forthecalculationofhigh-orderproperties,whichcanbecomparedwithexperimentaldata,accountingfortheelectroncorrelationiscritical[ 69 72 73 ],andgenerallyCCSDisfairlygood,thoughaddingtriplesoersomeimprovement.Nevertheless,thepresentresultsshowwhatkindofpropertiesexchangepotentialsandkernelsmusthavetobeabletodescribehigher-orderpropertieswithintheframeworkofDFT,andasmentionedpreviously,noneofthecurrentsetoffunctionalsforstandardDFTshowsthisbehavior. 108

PAGE 109

16 ]IthasbeenshownthattherigorousexchangetreatmentinTDOEPandTDHFprovidehyperpolarizabilitiesthatcloselyagreewitheachother,asonewouldexpect,althoughthiswillnothappenforcharge-transferexcitedstates[ 51 74 ].ThisclearagreementwithTDHFisincontrasttoconventionalexchangefunctionalsthatseverelyoverestimatehyperpolarizabilities.OEPpotentialsandkernelsarenotonlycapableofdescribingexcitationenergies,butalsohigher-orderproperties.SincenoneofthestandardDFTfunctionalsshowsthecorrectanalyticpropertiesofthesecondkernel,theOEPformulasandresultscanbeusedforthetestingandcalibrationofnewdensityfunctionals. 109

PAGE 110

(A{1) where "i"a)=Zdr'p(r)'q(r)(r) (A{2) (VC11)+(VC22)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"b) (A{3) (VC12)=2Xa;b;c;d;i;j (A{4) (VC13)+(VC31)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"k"a"b)("i"c) (A{5) 110

PAGE 111

("i+"j"a"b)("i+"j"a"d)("i"c) (A{6) (VC15)=2Xa;b;c;d;i;j (A{7) (VC16)+(VC25)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"b) (A{8) (VC17)=4Xa;b;c;d;i;j (A{9) (VC18)+(VC28)=4Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"a) (A{10) (VC19)+(VC110)=4Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"c) (A{11) (VC111)+(VC37)=4Xa;b;c;i;j;k(+) ("i+"j"a"b)("k+"j"a"b)("k"c) (A{12) (VC112)+(VC48)=4Xa;b;c;d;i;j(+) ("i+"j"a"b)("i+"j"d"b)("i"c) (A{13) 111

PAGE 112

("i+"j"a"b)("i"d)("i"c) (A{14) (VC115)+(VC215)=4Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"a) (A{15) (VC116)=4Xa;b;c;i;j;k (A{16) (VC21)=2Xa;b;i;j;k;l( (A{17) (VC23)+(VC32)=2Xa;b;i;j;k;l(+) ("i+"j"a"b)("i+"l"a"b)("k"a) (A{18) (VC24)+(VC42)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"j"a"c)("k"a) (A{19) (VC26)=2Xa;b;i;j;k;l( (A{20) (VC27)=4Xa;b;i;j;k;l( (A{21) 112

PAGE 113

("i+"j"a"b)("k"c)("k"a) (A{22) (VC211)+(VC47)=4Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"j"b"c)("k"c) (A{23) (VC212)+(VC38)=4Xa;b;i;j;k;l(+) ("i+"j"a"b)("l+"j"a"b)("k"a) (A{24) (VC213)+(VC214)=4Xa;b;i;j;k;l(+) ("i+"j"a"b)("k"a)("l"a) (A{25) (VC216)=4Xa;b;c;i;j;k (A{26) (VC33)+(VC35)=Xa;b;i;j;k;l(+) ("i+"j"a"b)("i+"k"a"b)("l+"j"a"b) (A{27) (VC34)+(VC45)=Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"k"a"b)("i+"j"b"c) (A{28) (VC36)+(VC43)=Xa;b;c;i;j;k(+) ("i+"k"a"b)("i+"k"c"b)("i+"j"c"b) (A{29) 113

PAGE 114

("i+"k"a"b)("i+"l"a"b)("i+"j"a"b) (A{30) (VC311)+(VC412)=2Xa;b;c;i;j;k(+) ("i+"k"a"b)("i+"j"a"c)("i+"j"a"b) (A{31) (VC312)+(VC411)=2Xa;b;c;i;j;k(+) ("i+"k"a"b)("i+"k"a"c)("i+"j"a"c) (A{32) (VC313)+(VC314)=4Xa;b;c;i;j;k (A{33) (VC44)+(VC46)=Xa;b;c;d;i;j(+) ("i+"j"a"c)("i+"j"a"b)("i+"j"d"b) (A{34) (VC49)+(VC410)=2Xa;b;c;d;i;j(+) ("i+"j"a"c)("i+"j"a"b)("i+"j"a"d) (A{35) (VC413)+(VC414)=4Xa;b;c;i;j;k (A{36) (VC51)+(VC61)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"a) (A{37) 114

PAGE 115

(A{38) (VC53)+(VC81)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"k"a"b)("i"c) (A{39) (VC54)+(VC712)=2Xa;b;c;d;i;j(+) ("i+"j"d"b)("i+"j"a"b)("i"c) (A{40) (VC55)+(VC78)=2Xa;b;c;d;i;j(+) ("i+"j"a"b)("i+"j"a"d)("i"c) (A{41) (VC56)+(VC68)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"b) (A{42) (VC57)=2Xa;b;c;d;i;j (A{43) (VC58)+(VC66)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"a) (A{44) (VC59)=2Xa;b;c;d;i;j (A{45) 115

PAGE 116

("k+"j"a"b)("i+"j"a"b)("i"c) (A{46) (VC511)+(VC512)=2Xa;b;c;i;j;k(+) ("k+"j"a"b)("i"c)("k"c) (A{47) (VC513)+(VC514)=2Xa;b;c;d;i;j(+) ("i+"j"a"b)("i"c)("i"d) (A{48) (VC515)+(VC615)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i"c)("k"b) (A{49) (VC516)=2Xa;b;c;i;j;k (A{50) (VC62)=2Xa;b;i;j;k;l (A{51) (VC63)+(VC71)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"j"a"c)("k"a) (A{52) (VC64)+(VC88)=2Xa;b;i;j;k;l(+) ("i+"j"a"b)("l+"j"a"b)("k"a) (A{53) 116

PAGE 117

("i+"j"a"b)("i+"l"a"b)("k"a) (A{54) (VC67)=2Xa;b;i;j;k;l (A{55) (VC69)=2Xa;b;i;j;k;l (A{56) (VC610)+(VC79)=2Xa;b;c;i;j;k(+) ("i+"j"a"b)("i+"j"c"b)("k"c) (A{57) (VC611)+(VC612)=2Xa;b;c;i;j;k(+) ("i+"j"c"b)("k"c)("k"a) (A{58) (VC613)+(VC614)=2Xa;b;i;j;k;l(+) ("i+"j"a"b)("l"a)("k"a) (A{59) (VC616)=2Xa;b;c;i;j;k (A{60) (VC72)+(VC73)=Xa;b;c;d;i;j(+) ("i+"j"a"c)("i+"j"a"b)("i+"j"d"b) (A{61) 117

PAGE 118

("i+"j"a"c)("i+"j"a"b)("i+"k"a"b) (A{62) (VC75)+(VC86)=Xa;b;c;i;j;k(+) ("i+"j"a"c)("i+"k"a"c)("i+"k"a"b) (A{63) (VC76)+(VC87)=Xa;b;c;i;j;k(+) ("i+"j"a"c)("j+"k"a"c)("j+"k"a"b) (A{64) (VC77)+(VC84)=Xa;b;c;i;j;k(+) ("i+"j"a"c)("i+"j"a"b)("j+"k"a"b) (A{65) (VC710)+(VC711)=Xa;b;c;d;i;j(+) ("i+"j"a"c)("i+"j"a"b)("i+"j"a"d) (A{66) (VC713)+(VC714)=2Xa;b;c;i;j;k (A{67) (VC82)+(VC83)=Xa;b;i;j;k;l(+) ("i+"k"a"b)("l+"j"a"b)("i+"j"a"b) (A{68) (VC811)+(VC812)=Xa;b;i;j;k;l(+) ("i+"j"a"b)("i+"l"a"b)("i+"k"a"b) (A{69) 118

PAGE 119

(A{70) (VC91)+(VC118)=2Xi;j;k;afaifkj(+) ("i"a)("j"a)("k"a) (A{71) (VC92)+(VC103)=Xi;j;a;bfaifbj(+) ("i"a)("j"a)("j"b) (A{72) (VC93)+(VC102)=Xi;j;a;bfaifbj(+) ("i"a)("i"b)("j"b) (A{73) (VC94)+(VC125)=2Xi;j;a;bfaifba(+) ("i"a)("j"a)("j"b) (A{74) (VC95)+(VC96)=Xi;j;k;afaifka(+) ("i"a)("j"a)("k"a) (A{75) (VC97)=2Xi;j;a;bfaifaj (A{76) (VC98)+(VC153)=4Xi;j;k;a;bfai(+) ("i"a)("j"a)("k"b) (A{77) 119

PAGE 120

("i"a)("j"a)("k"b) (A{78) (VC910)+(VC1310)=2Xi;j;k;a;bfai(+) ("i"a)("j"a)("k"b) (A{79) (VC101)+(VC127)=2Xi;a;b;cfbifac(+) ("i"a)("i"b)("i"c) (A{80) (VC104)+(VC115)=2Xi;j;a;bfibfij(+) ("j"a)("i"a)("i"b) (A{81) (VC105)+(VC106)=Xi;a;b;cfbific(+) ("i"a)("i"b)("i"c) (A{82) (VC107)=2Xi;j;a;bfiafib(+) ("i"a)("i"b)("j"a) (A{83) (VC108)+(VC154)=4Xi;j;a;b;cfib(+) ("i"a)("i"b)("j"c) (A{84) (VC109)+(VC142)=2Xi;j;a;b;cfib(+) ("i"a)("i"b)("j"c) (A{85) 120

PAGE 121

("i"a)("i"b)("j"c) (A{86) (VC111)+(VC112)=2Xi;j;k;afijfaj(+) ("i"a)("j"a)("k"a) (A{87) (VC113)+(VC114)=2Xi;j;a;bfijfaj(+) ("j"a)("i"a)("i"b) (A{88) (VC116)=2Xi;j;k;afjkfki(+) ("i"a)("k"a)("j"a) (A{89) (VC117)+(VC128)=2Xi;j;a;bfijfab(+) ("i"a)("i"b)("j"b) (A{90) (VC119)+(VC129)=2Xi;j;a;bfaifjb(+) ("i"a)("j"a)("i"b) (A{91) (VC1110)=2Xi;j;a;bfaifib (A{92) (VC1111)+(VC155)=4Xi;j;k;a;bfai(+) ("i"a)("j"a)("k"b) (A{93) 121

PAGE 122

("i"a)("j"a)("k"b) (A{94) (VC1113)+(VC1311)=2Xi;j;k;a;bfij(+) ("i"a)("j"a)("k"b) (A{95) (VC1114)+(VC143)=2Xi;j;k;a;bfij(+) ("i"a)("j"a)("k"b) (A{96) (VC1115)+(VC145)=2Xi;j;k;a;bfai(+) ("i"a)("j"a)("k"b) (A{97) (VC1116)+(VC137)=2Xi;j;k;a;bfai(+) ("i"a)("j"a)("k"b) (A{98) (VC121)+(VC122)=2Xa;b;c;ifabfib(+) ("i"a)("i"b)("i"c) (A{99) (VC123)+(VC124)=2Xi;j;a;bfibfab(+) ("i"a)("i"b)("j"a) (A{100) (VC126)=2Xa;b;c;ifabfac (A{101) 122

PAGE 123

(A{102) (VC1211)+(VC156)=4Xa;b;c;i;jfib(+) ("i"b)("i"a)("j"c) (A{103) (VC1212)+(VC151)=4Xa;b;c;i;jfab(+) ("i"b)("i"a)("j"c) (A{104) (VC1213)+(VC1312)=2Xa;b;c;i;jfab(+) ("i"b)("i"a)("j"c) (A{105) (VC1214)+(VC141)=2Xa;b;c;i;jfab(+) ("i"b)("i"a)("j"c) (A{106) (VC1215)+(VC1410)=2Xa;b;c;i;jfia(+) ("i"b)("i"a)("j"c) (A{107) (VC1216)+(VC138)=2Xa;b;c;i;jfia(+) ("i"b)("i"a)("j"c) (A{108) (VC131)+(VC132)=2Xi;j;k;a;bfai(+) ("i"a)("j"b)("k"b) (A{109) 123

PAGE 124

("i"a)("j"b)("j"c) (A{110) (VC135)+(VC149)=2Xi;j;k;a;bfai(+) ("i"a)("j"b)("k"b) (A{111) (VC136)+(VC146)=2Xa;b;c;i;jfai(+) ("i"a)("j"b)("j"c) (A{112) (VC1313)+(VC1514)=4Xa;b;c;i;j;k(+) ("i"a)("j"b)("k"c) (A{113) (VC1314)=2Xa;b;c;i;j;k (A{114) (VC1315)+(VC1414)=2Xa;b;c;i;j;k(+ (A{115) (VC147)+(VC148)=2Xi;j;k;a;bfaj(+) ("j"a)("i"b)("k"b) (A{116) (VC1411)+(VC1412)=2Xa;b;c;i;jfaj(+) ("j"a)("i"b)("i"c) (A{117) 124

PAGE 125

("i"b)("j"a)("k"c) (A{118) (VC1415)=2Xa;b;c;i;j;k (A{119) (VC157)(VC158)=8Xi;j;k;a;bfai(+) ("i"a)("j"b)("k"b) (A{120) (VC159)+(VC1510)=8Xa;b;c;i;jfai(+) ("i"a)("j"c)("j"b) (A{121) (VC1511)=8Xi;j;k;a;bfai (A{122) (VC1512)=8Xa;b;c;i;jfai (A{123) (VC1513)=16Xa;b;c;i;j (A{124) (V1)+(V2)=2Xi;a;b(+) ("i"a)("i"b) (A{125) 125

PAGE 126

("i"a)("j"a) (A{126) (VC5)=2Xi;j;a (A{127) (VC6)=2Xi;a;b (A{128) 126

PAGE 127

6{5 )anauxiliarybasissetdecompositionisused B{2 ) where "i"a(B{3)Twoseparategshouldbeconstructedforeachspincomponent.Expressionsfor,andcanbewrittenintermsofthediagrams(F1)-(F18). =2Xi;j;afia(<'a'j><'i'j>+<'a'j><'i'j>) ("i"a)("j"a)2Xi;a;bfia(<'i'b><'a'b>+<'i'b><'a'b>) ("i"a)("i"b)+2Xi;j;afij<'a'i><'a'j> (B{4) wheretheintegralsinvolvingtheexchangekernelcanbecalculatedusingtheformula 127

PAGE 128

("i"a)("i"b)("i"c)F1=(<'i'c><'b'c><'a'b>+<'i'c><'b'c><'a'b>+<'i'c><'b'c><'a'b>+<'i'c><'b'c><'a'b>+<'i'c><'b'c><'a'b>+<'i'c><'b'c><'a'b>) (B{6) ("i"a)("j"a)("k"a)F2=(<'a'k><'k'j><'i'j>+<'a'k><'k'j><'i'j>+<'a'k><'k'j><'i'j>+<'a'k><'k'j><'i'j>+<'a'k><'k'j><'i'j>+<'a'k><'k'j><'i'j>) (B{7) ("i"a)("i"b)("j"b)F3=(<'i'j><'j'b><'a'b>+<'i'j><'j'b><'a'b>+<'i'j><'j'b><'a'b>+<'i'j><'j'b><'a'b>+<'i'j><'j'b><'a'b>+<'i'j><'j'b><'a'b>) (B{8) ("i"a)("j"a)("j"b) (B{9) ("i"a)("i"b)("j"a)F4=(<'i'b><'a'b><'j'a>+<'i'b><'a'b><'j'a>+<'i'b><'a'b><'j'a>+<'i'b><'a'b><'j'a>+<'i'b><'a'b><'j'a>+<'i'b><'a'b><'j'a>) (B{10) 128

PAGE 129

("i"a)("i"b)("j"a)F5=(<'a'j><'i'j><'i'b>+<'a'j><'i'j><'i'b>+<'a'j><'i'j><'i'b>+<'a'j><'i'j><'i'b>+<'a'j><'i'j><'i'b>+<'a'j><'i'j><'i'b>) (B{11) ("i"a)("i"b)("i"c)F6=(<'i'c><'b'c><'i'a>+<'i'c><'b'c><'i'a>+<'i'c><'b'c><'i'a>+<'i'c><'b'c><'i'a>+<'i'c><'b'c><'i'a>+<'i'c><'b'c><'i'a>) (B{12) ("i"a)("j"a)("k"a) (B{13) Thesecondpartisrepresentedbythesumofdiagrams(SX1)-(SX6) ("k"b)("j"b)("i"a)F7=(<'b'k><'k'j><'i'a>+<'b'k><'k'j><'i'a>+<'b'k><'k'j><'i'a>+<'b'k><'k'j><'i'a>+<'b'k><'k'j><'i'a>+<'b'k><'k'j><'i'a>) (B{14) ("j"c)("j"b)("i"a)F8=(<'j'c><'c'b><'i'a>+<'j'c><'c'b><'i'a>+<'j'c><'c'b><'i'a>+<'j'c><'c'b><'i'a>+<'j'c><'c'b><'i'a>+<'j'c><'c'b><'i'a>) (B{15) 129

PAGE 130

("k"a)("j"b)("i"a)F9=(<'a'k><'b'j><'i'a>+<'a'k><'b'j><'i'a>+<'a'k><'b'j><'i'a>+<'a'k><'b'j><'i'a>+<'a'k><'b'j><'i'a>+<'a'k><'b'j><'i'a>) (B{16) ("i"c)("j"b)("i"a)F10=(<'b'j><'i'c><'i'a>+<'b'j><'i'c><'i'a>+<'b'j><'i'c><'i'a>+<'b'j><'i'c><'i'a>+<'b'j><'i'c><'i'a>+<'b'j><'i'c><'i'a>) (B{17) ("i"c)("j"b)("i"a)F11=(<'i'c><'a'c><'b'j>+<'i'c><'a'c><'b'j>+<'i'c><'a'c><'b'j>+<'i'c><'a'c><'b'j>+<'i'c><'a'c><'b'j>+<'i'c><'a'c><'b'j>) (B{18) ("k"a)("j"b)("i"a)F12=(<'i'k><'a'k><'b'j>+<'i'k><'a'k><'b'j>+<'i'k><'a'k><'b'j>+<'i'k><'a'k><'b'j>+<'i'k><'a'k><'b'j>+<'i'k><'a'k><'b'j>) (B{19) TogetnalexpressionforalltermsD1-D14shouldbesummedup. 130

PAGE 131

[1] A.SzaboandN.S.Ostlund,ModernQuantumChemistry,IntroductiontoAdvancedElectronicStructureTheory(McGrow-Hill,Inc.,NewYork,1990). [2] R.J.Bartlett,inModernElectronicStructureTheory,PartI,editedbyD.R.Yarkony(WorldScienticPublishingCo.,Singapore,1995). [3] F.E.Harris,H.J.Monkhorst,andD.L.Freeman,AlgebraicandDiagrammaticMethodsinMany-FermionTheory(OxfordUniversityPress,NewYork,1992). [4] P.-O.Lowdin,J.Chem.Phys.19,1936(1951). [5] P.HohenbergandW.Kohn,Phys.Rev.136B864(1964). [6] W.KohnandL.J.Sham,Phys.Rev.A140A1133(1965). [7] E.RungeandE.K.U.Gross,Phys.Rev.Lett,52,997(1984). [8] E.K.U.GrossandW.Kohn,Adv.Quantum.Chem.21,255(1990). [9] M.E.Casida,inRecentAdvancesinDensityFunctionalMethods,Part1,editedbyD.P.Chong(WorldScientic,Singapore,1995). [10] M.Petersilka,U.J.Gossmann,andE.K.U.Gross,Phys.Rev.Lett76,1212(1996). [11] J.Lindenberg,Y.Ohrn,PropagatorsinQuantumChemistry(JohnWiley&Sons,Inc,NewJersey,2004). [12] S.IvanovandR.J.Bartlett,J.Chem.Phys.114,1952(2001). [13] I.Grabowski,S.Hirata,S.IvanovandR.J.Bartlett,J.Chem.Phys.116,4415(2002). [14] A.GorlingandM.Levy,Phys.Rev.A50,196(1994). [15] J.D.TalmanandW.F.Shadwick,Phys.Rev.A1436(1976). [16] R.J.Bartlett,V.LotrichandI.Schweigert,J.Chem.Phys123,062205(2005). [17] R.J.Bartlett,I.Grabowski,S.HirataandS.Ivanov,J.Chem.Phys.122,034104(2005). [18] L.M.FrantzandR.L.Mills,Nucl.Phys.15,16(1960). [19] Q.Zhao,R.C.MorrisonandR.G.Parr,Phys.Rev.A50,2138(1994). [20] A.Gorling,Phys.Rev.A573433(1998). [21] S.Hirata,S.Ivanov,I.GrabowskiandR.J.Bartlett,J.Chem.Phys.116,6468(2002). 131

PAGE 132

S.Ivanov,S.HirataandR.J.Bartlett,Phys.Rev.Lett.83,5455(1999). [23] M.Stadele,J.A.Majewski,P.VoglandA.Goerling,Phys.Rev.Lett.79,2089(1997). [24] A.GorlingandM.Levy,IntJ.QuantumChem.Symp29,93(1995). [25] D.P.Chong,O.V,Gritsenko,E.J.Baerrends,J.Chem.Phys.1161760(2002). [26] R.J.Bartlett,I.Schweigert,V.Lotrich,J.Mol.Sruct.(Theochem)7711-8,(2206). [27] P.O.Widmark,B.J.Persson,andB.Roos,Theor.Chim.Acta79,419(1991). [28] K.Siegbang,ESCAappliedtofreemolecules,p.31,(AmericanElsevierpublishing,NewYork,1969). [29] C.E.Moore,Atomicenergylevels,(U.S.DepartmentofCommerce,Washington,D.C,1949)Vol1. [30] R.Colle,R.K.Nesbet,J.Phys.B342475(2001). [31] S.A.Kucharski,M.Wloch,M.Musial,R.J.Bartlett,J.Chem.Phys,1158263(2001). [32] T.P.Straatsma,E.Apra,T.L.Windus,E.J.Bylaska,W.deJong,S.Hirata,M.Valiev,M.T.Hackler,L.Pollack,R.J.Harrison,M.Dupuis,D.M.A.Smith,J.Nieplocha,V.Tipparaju,M.Krishnan,A.A.Auer,E.Brown,G.Cisneros,G.I.Fann,H.Fruchtl,J.Garza,K.Hirao,R.Kendall,J.A.Nichols,K.Tsemekhman,K.Wolinski,J.Anchell,D.Bernholdt,P.Borowski,T.Clark,D.Clerc,H.Dachsel,M.Deegan,K.Dyall,D.Elwood,E.Glendening,M.Gutowski,A.Hess,J.Jae,B.Johnson,J.Ju,R.Kobayashi,R.Kutteh,Z.Lin,R.Littleeld,X.Long,B.Meng,T.Nakajima,S.Niu,M.Rosing,G.Sandrone,M.Stave,H.Taylor,G.Thomas,J.vanLenthe,A.Wong,andZ.Zhang,"NWChem,AComputationalChemistryPackageforParallelComputers,Version4.7"(2005),PacicNorthwestNationalLaboratory,Richland,Washington99352-0999,USA. [33] H.JiangandE.Engel,J.Chem.Phys.123,224102(2005). [34] J.P.Perdew,K.Burke,M.Ernzerhof,Phys.Rev.Lett.77,3865(1996). [35] K.Raghvachari,G.W.Trucks,J.A.PopleandM.Head-Gordon,Chem.Phys.Lett157,479(1989). [36] D.Bokhan,I.V.Schweigert,R.J.Bartlett,Mol.Phys,103,2299(2005). [38] J.C.Slater,Phys.Rev.81,385(1951). [39] D.J.TozerandN.C.Handy,J.Chem.Phys.108,2545(1998). 132

PAGE 133

J.P.Perdew,R.G.Parr,M.LevyandJ.L.Balduz,Jr.,Phys.Rev.Lett.49,1691(1982). [41] A.Balkova,R.J.Bartlett,J.Chem.Phys.102,7116(1995). [42] Ch.W.Bauschliher,Jr.andP.R.Taylor,J.Chem.Phys.86,2844(1987). [43] J.Schirmer,A.Dreuw,Phys.Rev.A75022513(2007). [44] J.B.Krieger,Y.LiandJ.Iafrate,Phys.Rev.A45101(1992). [45] V.N.Staroverov,G.Scuseria,E.R.Davidson,J.Chem.Phys.124141103(2006). [46] W.YangandQ.Wu,Phys.Rev.Lett,89143002(2002). [47] A.Gorling,Phys.Rev.Lett,83,5459(1999). [48] R.J.Bartlett,I.V.SchweigertandV.F.Lotrich,J.Mol.Struct:Theochem771,1(2006). [49] D.Rohr,O.Gritsenko,E.J.Baerends,Chem.Phys.Lett.432336(2006). [50] S.Hirata,S.Ivanov,R.J.Bartlett,I.Grabowski,Phys.Rev.A71,032507(2005). [51] A.Dreuw,J.L.WeismanandM.Head-Gordon,J.Chem.Phys.119,2943(2003). [52] S.Hirata,S.Ivanov,I.Grabowski,R.J.Bartlett,K.Burke,J.D.Talman,J.Chem.Phys.115,1635(2001). [53] P.O.Lowdin,Phys.Rev.139,A357-A372(1965). [54] S.Ivanov,M.Levy,J.Chem.Phys.119,7087(2003). [55] H.Jiang,E.Engel,J.Chem.Phys.125184108(2006). [56] O.V.Gritsenko,E.J.Baerends,Phys.Rev.A,64042506(2001). [57] S.Ivanov,S.Hirata,I.Grabowskii,R.J.Bartlett,J.Chem.Phys.118461(2003). [58] A.F.Izmaylov,V.N.Staroverov,G.E.Scuseria,E.R.Davidson,G.StoltzandE.Cances,J.Chem.Phys.126,084107(2007). [59] S.KummelandJ.P.Perdew,Phys.Rev.Lett,90043004(2003). [60] I.Schweigert,V.F.LotrichandR.J.Bartlett,J.Chem.Phys,125104108(2006);I.Schweigert,Ph.D.thesis,UniversityofFlorida(2005). [61] R.J.BartlettandM.Musial,Rev.Mod.Phys,79291(2007). [62] D.Bokhan,R.J.Bartlett,Phys.Rev.A,73022502(2006). [63] D.Tozer,J.Chem.Phys.11912697(2003). 133

PAGE 134

S.J.A.vanGisbergen,P.R.T.Schipper,O.V.Gritsenko,E.J.Baerends,B.ChampagneandB.Kirtman,Phys.Rev.Lett.83697(1999). [65] F.A.Bulat,A.Toro-Labbe,B.Champagne,B.Kirtman,W.Yang,J.Chem.Phys.123014319(2005). [66] M.Kamiya,H.Sekino,T.TsunedaandK.Hirao,J.Chem.Phys122234111(2005). [67] R.J.Bartlett,D.M.Silver,Phys.Rev.A101927(1974). [68] H.P.Kelly,Adv.Chem.Phys12314(1963). [69] R.J.Bartlett,H.Sekino,ACSSymposiumSeries628:NonlinearOpticalMaterials23-57(1996). [70] A.M.LeeandS.M.Colwell,J.Chem.Phys.1019704(1994). [71] S.P.KarnaandM.Dupuis,J.Comp.Chem.12487(1991). [72] H.SekinoandR.J.Bartlett,J.Chem.Phys85976(1986). [73] H.SekinoandR.J.Bartlett,J.Chem.Phys983022(1993). [74] D.Bokhan,R.J.Bartlett,J.Chem.Phys.(submitted). 134

PAGE 135

IwasborninAvdeevka,asmalltownlocatedinthesoutheasternpartoftheUkraineintheDonetskregion.IgrewupinthecityofGorlovka,aworking-classminingcommunity.WhenIwasinninthgrade,IacquiredaninterestinchemistryanddecidedthatIwouldbecomeachemist.Duringmylastyearofhighschool,Iwasoneofthewin-nersoftheUkrainianChemistryOlympiad.Asaresult,IwasadmittedtotheprestigiousMoscowStateUniversitywithouttheotherwisenecessaryentranceexaminations.Duringmysophomoreyearofcollege,Idecidedtospecializeinquantumchemistry.AfterIde-fendedmyundergraduatethesisandobtainedmydegreein2002,IbegantoworktowardjoiningDr.RodneyBartlett'sresearchgroupattheUniversityofFlorida.AfterworkingasasystemadministratorforseveralcompaniesinMoscowfor16months,IobtainedthenecessaryfundstopayfortransportationtotheUnitedStatesaswellasforthenecessaryexams,whichincludedtheGraduateRecordExamination(GRE)andtheTestofEnglishasaForeignLanguage(TOEFL).Finally,IjoinedDr.Bartlett'sgroupattheUniversityofFloridaonJanuary21st,2004. 135


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101211_AAAAAA INGEST_TIME 2010-12-11T06:34:07Z PACKAGE UFE0021221_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 1051985 DFID F20101211_AAAAML ORIGIN DEPOSITOR PATH bokhan_d_Page_094.jp2 GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
95ebe7ccd205aa018a45a451fa31b2d5
SHA-1
0b97983f4f3703c3b7050ce88f5d354ba72e22b0
61260 F20101211_AAAALW bokhan_d_Page_076.jp2
cdd88879199b0846d55f8019d8445e2c
67b26227d9e260f9f6624ec44ec460351cda55e5
18002 F20101211_AAAALX bokhan_d_Page_077.jp2
c4d0bc172cab227a14394f733cd91a28
b30d3841a1e9085eb34fda862d07e3147901a1c6
55827 F20101211_AAAANA bokhan_d_Page_112.jp2
fd2f76aa9d76406638f36ed0b65a1ab6
597db3cc67280124ff33d7798de9a3f397392723
1051908 F20101211_AAAAMM bokhan_d_Page_095.jp2
f38223210a6d16b91666f23fe53ef194
cfa010ff073a5c3137db43f76d79fe17e03985ea
20265 F20101211_AAAALY bokhan_d_Page_079.jp2
222aaa9d5993415adf5f0bc2a2f092fb
b6520cd1b2cbf79f690a7a48ecb89ad541ff8315
62971 F20101211_AAAANB bokhan_d_Page_114.jp2
08c5179496cb1aa669aee8903fa8ed5e
27d35177116b91fa51e3ec73fd40a3317cc2686f
489288 F20101211_AAAAMN bokhan_d_Page_096.jp2
dffab8db91af3323021b162269644084
79f1033342e94171df4e8e26ae48e740874547a6
1051980 F20101211_AAAALZ bokhan_d_Page_080.jp2
b1b212e12c7dae0aa43b879cc062e5be
e9169ca69f5a2c67a9e012d461617dc518242ce0
57968 F20101211_AAAANC bokhan_d_Page_115.jp2
a85c01a260d20cb5b4afce5937df1bee
39cfb5e062e036a805cf95631154c12051ee62ab
1051920 F20101211_AAAAMO bokhan_d_Page_097.jp2
9800d174a981dc0ef5659181fbe3951c
641df2f380d2fbc19c16cbdb3739c06a6f082d2b
59793 F20101211_AAAAND bokhan_d_Page_116.jp2
a6197540f3c3dd172f2c1bf4d7a719f6
ea738a46f2c38a51f2be0ce673de9fe8d49b2cf1
746209 F20101211_AAAAMP bokhan_d_Page_098.jp2
1156789dcd167cf13482f9298b4f37f3
3bb9a3b34adc26c5b0960d0038f377b2da786176
58394 F20101211_AAAANE bokhan_d_Page_117.jp2
74fffd75e2d71995f100f8a0314fae34
2b75dbc9687a351017399b7acf7929f137d11449
40855 F20101211_AAAAMQ bokhan_d_Page_099.jp2
a7ec37b80c02dc27ce00fa6f7a23f3ff
6e1bc7205f913e72bf2ff2db083a11a438278d39
64223 F20101211_AAAANF bokhan_d_Page_118.jp2
caeecc28282b7f99e2143b2980aea77b
6b3c47ba6befebf93871accb38428e8cc35c7ac6
54936 F20101211_AAAANG bokhan_d_Page_120.jp2
0a1fcf28cd72f67ebefd2cc17c3f48ca
93d3e187b4aada54c2d083da5afccba8467405ea
1051932 F20101211_AAAAMR bokhan_d_Page_100.jp2
79e65821a5a493339a643a9b6edde019
6748421cd2a93c56b7868834b6536e3eedda43b0
52272 F20101211_AAAANH bokhan_d_Page_121.jp2
8f4bc4d9e7c1cd66c4f99ab24957ebb7
3771046c919883bf9eb1c253493de8ec7f7d6be4
994100 F20101211_AAAAMS bokhan_d_Page_101.jp2
ddeabb282499f7c03056b2050ea45c2b
4b88476ea6342d54c34151992e1a93e783a42976
54362 F20101211_AAAANI bokhan_d_Page_122.jp2
13f384a0475d72e530e0d38e1aa291cb
a404e7eca4818f41b6480991fe411a4a7a4eb9b2
683353 F20101211_AAAAMT bokhan_d_Page_102.jp2
def74807aa8107df2791b90b7794ad1c
68d8a39fe78ff1aeb07e6ac5d5789654d5bacc5d
56636 F20101211_AAAANJ bokhan_d_Page_124.jp2
d3c21d575b935ea795acb765c64323f3
6bfb03e588bf71a097279c9f3262b05de8d867d6
613316 F20101211_AAAAMU bokhan_d_Page_103.jp2
a02d4621b2a7e1c8441adc2786e6d25b
9b29d41b1351d9154e807d5471c6b44e213d967b
51082 F20101211_AAAANK bokhan_d_Page_125.jp2
093df12f45f0a068c66110cfec18f7a3
7f5b914733cc04eec79b6fc6d8a6a31e1d555953
856178 F20101211_AAAAMV bokhan_d_Page_105.jp2
949a55a8a9da95e4fd5f3e4d0581a91f
23bc6c544dbf4270194129520e0ebb3d8489fe55
17227 F20101211_AAAANL bokhan_d_Page_126.jp2
89d631c6f2402f96ef8f11be673b239c
cd4ce894e48f7762fa7c9ad353b7823af1f2b5fb
1051978 F20101211_AAAAMW bokhan_d_Page_106.jp2
330e296f28bd4821bfdfd966d01a6964
d3b9ada1817dc843032ef8004ab99b18af426519
25271604 F20101211_AAAAOA bokhan_d_Page_013.tif
90d169362bcd33e989166558dbd62cab
2e785b93ce576e658a0ada140671a526b81fdd77
78473 F20101211_AAAANM bokhan_d_Page_128.jp2
a8da78333b9dcd8dd9b9082a8d3f5903
ccfd4da9b324dd5f599b907998c3ef7c4099af7e
1051981 F20101211_AAAAMX bokhan_d_Page_107.jp2
cc8367b4a103ae41bdbb7a78e121abb8
f31621de8acc19897d10046c2d06487adb4aa567
F20101211_AAAAOB bokhan_d_Page_015.tif
dc3d7d0e0fc634df0c9bc32680c45b6c
d9579616a08ae5ecaf1eab3500030eb70ba1dd95
74872 F20101211_AAAANN bokhan_d_Page_129.jp2
a70e6f511d8d370018d9e8e1149bf729
55574f9772fb47f7cba99e73d45ecf479bfe882d
607447 F20101211_AAAAMY bokhan_d_Page_109.jp2
77e24460a3bbf81e8471bb687826f7e6
3b70a5a5865dfe1ce9bb2994feb86d3d7266d6e1
1053954 F20101211_AAAAOC bokhan_d_Page_016.tif
3a2081545639dfb0dd73cf7c2cf09134
dc9dfc12cba9f3e5e0d1cdac28a1817936c6f7b1
1051979 F20101211_AAAANO bokhan_d_Page_132.jp2
d46056a5de0b851c9187d7990c3ee201
d9905521f170e30dbd5ded16365a88143d701cb8
58260 F20101211_AAAAMZ bokhan_d_Page_111.jp2
0becb822c0d1dd2f541740ff2b419f60
6c4a8a2e7331900d51007df831816022b3424a0b
F20101211_AAAAOD bokhan_d_Page_017.tif
0f954c0157f10481e170acff5d7e0794
aee8754a0ad86bbbabcf3c8e50d48c4ac379a6b4
102505 F20101211_AAAANP bokhan_d_Page_133.jp2
ee0b5144d9de2ec8fe8ffdc88f9f60e7
6482490c9287b38b8700fff12d1e7bf88498c3ff
F20101211_AAAAOE bokhan_d_Page_018.tif
f1e1e472ff4a16bf1850f731acea6ce1
59411afa121dd24a66b85fcbdf2f0d2b92e4c14d
49950 F20101211_AAAANQ bokhan_d_Page_134.jp2
4b554be23bcdd58759d2b739b5151c7c
92bfc412b6a2bb6bae71acd6ce5a5b01a22c4621
F20101211_AAAAOF bokhan_d_Page_019.tif
b0677ab5e6ff7c39bbd9eae5d36786bb
6613c0fb9ed20d927e3cebed81fc0d22828d6409
F20101211_AAAANR bokhan_d_Page_001.tif
fcca221cfc865c6d3f9388f202063610
dbc701c56a3903c3ba164110edf30898500169d7
F20101211_AAAAOG bokhan_d_Page_021.tif
08994d073be98ae25fb647f434fcc0b9
cfbce0cd082a894d15153b5028026e3da84979f8
F20101211_AAAAOH bokhan_d_Page_023.tif
0a03b0ae93428d6c5e4b5fe7c1571205
9ec1bdd7a88e871352a552e81fd21be9f599cdae
F20101211_AAAANS bokhan_d_Page_002.tif
72bf94eadc7a46e10328fccb658ba351
31120f313bdb72b8d26700d0dc7150bc44905b63
F20101211_AAAAOI bokhan_d_Page_024.tif
ffc3bd2991a7d658157cf3bfd06b8757
dacbb4948ee7cc28916dc72b3d2d9c1df4d5489e
F20101211_AAAANT bokhan_d_Page_003.tif
b2153aca049a816198658411412dbfbd
85097fcc67380dce3f4ed81397854d712dd9a9d6
F20101211_AAAAOJ bokhan_d_Page_025.tif
57f176d80638938b1b527808ec47e6d5
76bd361d247b2510366657a1dcd5d8127773301f
F20101211_AAAANU bokhan_d_Page_005.tif
cf8f5c5cf70b4839d3fccbc47c4770b5
f655e1bc37982e109fa1226200f3a670ff2cd8de
F20101211_AAAAOK bokhan_d_Page_026.tif
9a70662433b8a61b5fdc3745dfdb82bf
022906a5cc19ab5884ff13a2309b32dc3533da1e
F20101211_AAAANV bokhan_d_Page_006.tif
cade2c8ba6207ce62a884591df7f0f93
72a3fe7fa316c82da79876262cc1980cbaf69f6f
F20101211_AAAAOL bokhan_d_Page_027.tif
792e9c5ed2caa363163e7f9d6cf4e2d6
63b37ab2c0fbb3b2d76997b03dc5b281f47c5fdc
F20101211_AAAANW bokhan_d_Page_008.tif
35526525db5024624e79c9f00b4581ff
a67f3fc9adc3d7fabd353b5337244828b0554589
F20101211_AAAAOM bokhan_d_Page_028.tif
18633fc5f6568d45d8ea4eb8749a3ec5
5b726ae17c3927f350512cf9468631e3a4b2e3ab
F20101211_AAAANX bokhan_d_Page_009.tif
196a64fd2ddee64dff029a80efa8b15a
30a1fff32d8481f09ff838b167aaf4de87c574c7
F20101211_AAAAPA bokhan_d_Page_045.tif
0064c0abb2d31acaffc39633e236ad41
472189d329b22fd1a82462880ba34640e2f0860f
F20101211_AAAAON bokhan_d_Page_029.tif
1aa85c04c0c62f9fad61b50aae0ced29
63a50ebe00b6824c5da39c82b56fb43e6f82b5ad
F20101211_AAAANY bokhan_d_Page_011.tif
825034f7e04cbee03782e32345807fb9
2c17b757d053351d32d895d97a9194074911866f
F20101211_AAAAPB bokhan_d_Page_046.tif
976335ef4e5dfd00f8c5fe2e87983920
8635a128a926310bbf3417bac10676097f80b592
F20101211_AAAAOO bokhan_d_Page_030.tif
7768ab06f3816ebfd9692c5c5a5b9c1a
6ed79c2e2aef4ca8473ef4412ba6e5234dbd155a
F20101211_AAAANZ bokhan_d_Page_012.tif
22c9aab6357173e32687e0e947bf8a5e
c173cae832ba5407e695a3573e89da4f5956fd60
F20101211_AAAAPC bokhan_d_Page_047.tif
07a18271cd6d01d23fa20daa63a2b225
6efc27e1e75dc8a64523c685070ea4e2101ddb41
F20101211_AAAAOP bokhan_d_Page_031.tif
fa9dd31ff467eb6e00cb3da19cb437ad
61a64bbf82b05ea22a5574e4c1c3ac7621b4e0ea
F20101211_AAAAPD bokhan_d_Page_048.tif
414139a0206079c54574eeeed72eac87
717067292069cf65e689f4664c445e85bd0c616f
F20101211_AAAAOQ bokhan_d_Page_033.tif
27eadbe642838843d1932efee5ed6bb5
e9316ed66213d37545fdd99ce26a1b7e9519cf69
F20101211_AAAAPE bokhan_d_Page_049.tif
527add9dad6f9a041e88e18ed5c0f897
dd46c869bf8ca936fe2e8f4175870ea6e078a5a1
F20101211_AAAAOR bokhan_d_Page_034.tif
52f74673b8f3c265202ad1166112df29
fb9ac6dc94ff0842873f0862d2da24fd01c4bc5a
F20101211_AAAAPF bokhan_d_Page_050.tif
8c04c8af169367b9b063c68dab0dac51
72b720292cf88ed9265d9289dae81031ba06eec3
F20101211_AAAAOS bokhan_d_Page_035.tif
7464f77fb2882fedb8e12dc29e8c270c
2e6d017c2b93f6a06c3a8cd3d18241a1ba5f6777
F20101211_AAAAPG bokhan_d_Page_051.tif
1c0aaa9b744e2ab67ceb17e35a9b46a5
1644fbdc5e72d678da0876a69af8acab07bfb9ee
F20101211_AAAAPH bokhan_d_Page_052.tif
b6f202ac30522c8f61c777efd21a9559
13cbbaa818bafd9479444dd53523c993d63d0fe8
F20101211_AAAAOT bokhan_d_Page_037.tif
8c26dcff37ea0da55e85f207b244863b
fa0f4fd49fa8cd729fb1a31e1249fad3ea5a4c33
F20101211_AAAAPI bokhan_d_Page_053.tif
51be9230efb8215ef210ca5934dd492e
21a438a5bb8a97a84855b6214711774f6c04511b
F20101211_AAAAOU bokhan_d_Page_038.tif
d1d23a015a84391317ac55ef51ee6872
b6213bad4aa274db89ced63e94921c253fef09d5
F20101211_AAAAPJ bokhan_d_Page_055.tif
0c08c180bdbec392917fe428bc920259
950a847f33b7f92d40332a98013c8b4703e0618f
F20101211_AAAAOV bokhan_d_Page_039.tif
800def1f2243859f6ddb8604108f6a07
1d50582b03caf168381bb278c6159041b7b13656
F20101211_AAAAPK bokhan_d_Page_056.tif
1ae1aea0823858c91f1acb53f97944ea
e2c5f47e9b59f29e78a066c7dc1eb5e0a684629d
F20101211_AAAAOW bokhan_d_Page_040.tif
96b85d5fdc1faf974d84e2a053144eb0
2b1492845e5dd1e2862eca1a0d6e8892865978b8
F20101211_AAAAPL bokhan_d_Page_057.tif
76c55980dd4a60db3854f2d9367737b1
aee1a310f6170b58bcebaefe0b85843512b05b24
F20101211_AAAAOX bokhan_d_Page_042.tif
1149ce48317afdca104120437a28f452
7e6bbaf86bad102672f16248fd932c2accb978f9
F20101211_AAAAQA bokhan_d_Page_072.tif
7842c323135274f17c774d5cd0054e17
16463704a2fd45f94dfd98f21970196a3442ef3a
F20101211_AAAAPM bokhan_d_Page_058.tif
6a37e03d9f1985f7337d6da66e7261f3
d6f17086feb13f5e0bca401cc010a9b5ad24fb5d
F20101211_AAAAOY bokhan_d_Page_043.tif
81735c0197ba1e624fbd07d0952ce009
a4c8d3af9ba1a9ff79ee91e87b1fc5e8a8f888e5
F20101211_AAAAQB bokhan_d_Page_073.tif
11ffa433b89a46bd301c9c65b3befb8f
ccda0eab9753e2d28bcf1b8f952f2c90adb4e4b5
F20101211_AAAAPN bokhan_d_Page_059.tif
abfa708416ba2596c90a36b58f7379d3
3298b5d9dd72e8afb7a9ade910417160a5ca4db6
F20101211_AAAAOZ bokhan_d_Page_044.tif
e675e8d834951823877d3a0a3b26769e
60c71214d06359735049b4e34db57a95b28cb9fc
F20101211_AAAAQC bokhan_d_Page_074.tif
b923588e5c6d1b02220c6f16c8b80090
162c8ad08a37e74dd63f04a3d725ccb00684fe20
F20101211_AAAAPO bokhan_d_Page_060.tif
675e44026c80f8cc336fa872a1ee8073
d715fcafdb905814e133c726b6851b9d48f85ab7
F20101211_AAAAQD bokhan_d_Page_075.tif
746a84b20f01c21b29fb412f5d5ed6b7
8dc2c3ffd6a223470f5882fe85c5e3c02c3e4df7
F20101211_AAAAPP bokhan_d_Page_061.tif
ae2ca540accbd1a45d428553fbb6cbb3
e63e408ddba59324f6ffa1b407332e71351406ca
F20101211_AAAAQE bokhan_d_Page_076.tif
06fefd5c12d834ee0584208e23c0d00e
58830d814cf77fa72d5085782cafb7d0c2133506
F20101211_AAAAPQ bokhan_d_Page_062.tif
8dafb228e7ed262358611d3bbdb8e6be
75a36a0e23d12c6d2d5057dee0e2895539644ac3
F20101211_AAAAQF bokhan_d_Page_077.tif
74b782307511381a3e784b12cc33bc79
27aaf3fefdc4bfd010def43fdf7d99472613cb2c
F20101211_AAAAPR bokhan_d_Page_063.tif
fd854a1a408e658cdfaf9300caf59a27
43b8c00ffd7cde178bd659e6543646260f6d6ce9
F20101211_AAAAQG bokhan_d_Page_079.tif
b79ce94fb62d7cb9267932d6ba736c5a
fed073c6a2b461990736c1725f584ead8e8f563e
F20101211_AAAAPS bokhan_d_Page_064.tif
190505b57e275775a92e98186a7f8a3b
29422fb5a9759a7ce2066d69da5b0ce08c55a68a
F20101211_AAAAQH bokhan_d_Page_080.tif
38d40021fae38ff6dcd068ecf69c61f5
342deea8fcbeca4d24a56d301672a64a74f95925
F20101211_AAAAPT bokhan_d_Page_065.tif
727cccbc9f20ad42f219d2861442264d
2397e7792d0f2c5adf43c12be04e2e15669562cb
F20101211_AAAAQI bokhan_d_Page_081.tif
ac1d6a4e2c4ae867c0cc4c2485d1809c
906dd48f21aeaf24b26fedb4050ce99031035d31
F20101211_AAAAQJ bokhan_d_Page_083.tif
968e16e1d347bd4c8228eba520b98a2e
2aefed0980c989429236db0088bdfc2542371152
F20101211_AAAAPU bokhan_d_Page_066.tif
dc4ce874a99b21a85ca13ec0d910c9b0
4a05758d9ec31fec38d2dd719944e01f81ed348f
F20101211_AAAAQK bokhan_d_Page_084.tif
3cc1a38dc9578348610000fe229ef0ee
392310bb187abadecabe98fc33bee6b24098ad93
F20101211_AAAAPV bokhan_d_Page_067.tif
4590d31f2fb947383c51272b818c0b35
3f11d222b006a01e9e7c92a2ee1f142013bbadb2
F20101211_AAAAQL bokhan_d_Page_085.tif
873eabb391789db1b49e256c69145494
2ecc3b0376ec845e7f4651b8bbae512e9892587e
F20101211_AAAAPW bokhan_d_Page_068.tif
8c0596285b0d0441d80b3e7c7c4fd47b
4c0653d6acb0fbcb33dabc70bca91668409ffe32
F20101211_AAAARA bokhan_d_Page_102.tif
5b9a9142a9fdd9851edcbba061cd919f
4dab95934baf847e64a41765e2f650951cd327a9
F20101211_AAAAQM bokhan_d_Page_086.tif
bb938173942f536bedba48ade6709a21
0851a88fafb17a1d4f8e147b94047002c167242b
F20101211_AAAAPX bokhan_d_Page_069.tif
ed2ed1d3c0737fa32e6d2d13790b471e
a54f80d241537a4413c7b2af1441c0aa8fc7001b
F20101211_AAAARB bokhan_d_Page_103.tif
7e6f418cfae5e5bfdb1a9f914ff78ef2
66b4608059ff1d0607abde6f95afafcbd461cc5d
F20101211_AAAAQN bokhan_d_Page_087.tif
87929bc69e66f01c468a4a6a2f733456
70f86cd8f79fefd05dc4f77e7509c8dd40d17c64
F20101211_AAAAPY bokhan_d_Page_070.tif
d381c42fae5d755c3776f1a0b0cfce29
299092a1ccf32eeef571ce81e843bcf04553878f
F20101211_AAAARC bokhan_d_Page_105.tif
f9ae416a4b4ad40d24b96e9a69261138
918f371e30b45b13116b66655d66e030c955545f
F20101211_AAAAQO bokhan_d_Page_088.tif
21023320f7e9de6bc1b064a7ab9d85fb
b25105e37c9c9c8f8385eea6678f37045de49aaf
F20101211_AAAAPZ bokhan_d_Page_071.tif
34001dff646bad8a22cd378a5e62e19c
3f0ef34e69c2126bcce88c26dfb49884e45b60c4
F20101211_AAAARD bokhan_d_Page_106.tif
64019d868b647f3c1ed65c7c6bc3b84e
1558fd23bd8a8f129ce5fe25dbadb086e526d5a2
F20101211_AAAAQP bokhan_d_Page_089.tif
0f8db43e31e42be843f957f9a9f7eb70
28315b5fd00ed85493d8816a527f3159e0998dd9
F20101211_AAAARE bokhan_d_Page_107.tif
0538ae9449fe9f11c87cf082b4f509c3
7435458d5376fe90371a0321b15dd91a2487e0ae
F20101211_AAAAQQ bokhan_d_Page_092.tif
2c55d0eea7ce302cce6bfd6104f3c950
29121a4d49cd6c4b9c4a21d4e23704a1d88cdc83
F20101211_AAAARF bokhan_d_Page_108.tif
32c6700879a51272030e88be084272b7
54b95471f3e2c53afbf40e0dfbdee40d894416c6
F20101211_AAAAQR bokhan_d_Page_093.tif
9c70502fa86dcf12d6cea7e369e3aaff
09c7b352b5d72cc38dcd0fc80fbd722d24ad60b3
F20101211_AAAARG bokhan_d_Page_109.tif
25d101bee89f6f5784c5ee9624baeb7f
b1283df89c890ecffdcf7c502c153225a3e555dd
F20101211_AAAAQS bokhan_d_Page_094.tif
d6b9097445d6d91297ec77ded2b594a8
b58745556ca505cec068de6d9cf7efd7025b1aa5
F20101211_AAAARH bokhan_d_Page_110.tif
7879d4ca772ecc3a80bd1ab997e5ca71
79754ed908c29fef9e6f4cc31ca0fb002e635636
F20101211_AAAAQT bokhan_d_Page_095.tif
36f6702fc6b0eb24956d22f108cf1c0c
a52701297af7bc595958e9137bd91884f853404e
F20101211_AAAARI bokhan_d_Page_111.tif
58a6a3df7d2ea9c3b5658b1464c3a295
6ed6b7310509613fd3db7d5077c8041a5b5faa06
F20101211_AAAAQU bokhan_d_Page_096.tif
a159b701f8ecc25576921809fd3c07ec
69fbcdf9f3e9b916aef40da8d60296cb0f2dd271
F20101211_AAAARJ bokhan_d_Page_112.tif
174d762e53e2650d7423b29f6cce47f0
bcd443ade56d278c2e05afd8e0eb1533e3302571
F20101211_AAAAQV bokhan_d_Page_097.tif
d279328039fbf5d66ee1645f501b2045
34a8b879d308e675b2a32064803bd9fe0c2e6674
F20101211_AAAARK bokhan_d_Page_113.tif
87434a035fac1bc39d0aa7bdd5b723a8
474cb13727585851e25a1250e46e1a6945d53290
F20101211_AAAAQW bokhan_d_Page_098.tif
d0bd8873fa36a7c763470bb7888ae51e
f96f7bcfff3b12dcf13820521bfb1371f5d24290
F20101211_AAAARL bokhan_d_Page_114.tif
0ea52529b2e56986f1b704ba25bf36ad
0ea1d18714e021ed704bb573d11081ed6de9685e
F20101211_AAAAQX bokhan_d_Page_099.tif
7af6701be5ff99b82107f9360f9d4b14
0d3c6a6a41bad3ad24f0f26b0415e4f63f4ab35b
F20101211_AAAASA bokhan_d_Page_132.tif
26b971c0680a4d62d496487246971dff
b03b1ece1cb603c0251dcd426fb6c3de3c49b777
F20101211_AAAARM bokhan_d_Page_115.tif
184538a2fa4d4badc10cfe07243d0b03
695943dd1610e648b2a90712b56be0f2eb43416f
F20101211_AAAAQY bokhan_d_Page_100.tif
41357d8cf129bcb132b6f7cd01b24f2b
f2f290a79d7112958f150f347998db9144e08515
F20101211_AAAASB bokhan_d_Page_133.tif
09ded1427a49303e46d4abbb96e179a1
801461e604f8b62aa585144fb226c562f06b105c
F20101211_AAAARN bokhan_d_Page_116.tif
895d7e4543d327b746cc263885e2acf7
a1cefede30d442f84122e8a10a9986f6f39c81bc
F20101211_AAAAQZ bokhan_d_Page_101.tif
e5f21e8214b3cb94b4b3a3cdefe31a95
dace24ac946f1e4732dbbe9337ddd1a044b23a50
F20101211_AAAASC bokhan_d_Page_134.tif
71bbb6fbf6dd78c8aee210d63d8cafe3
6b912bc4e86e7e7d4e42b126bcac8791cecd2c81
F20101211_AAAARO bokhan_d_Page_117.tif
bbbfc3c39d29d57215ea130cfeb27d34
de30232d3b107e4d31f020c894ff05535b833020
F20101211_AAAASD bokhan_d_Page_135.tif
2f477520b805d05feea067fc3067a287
1ed3b2b58029ae8dabd4124aba33c2dd175e36ab
F20101211_AAAARP bokhan_d_Page_118.tif
41b444db3e48cf9b6ec8074dbc027df7
4b6ffa25735ed889c666592d3f089509b1399cdb
8711 F20101211_AAAASE bokhan_d_Page_001.pro
939a65603994b15b87d1709955948d64
68cf64561c1e05ca72713858ae642ff8359993c9
F20101211_AAAARQ bokhan_d_Page_119.tif
99494709451e133baf2f3e3af7d5ee2b
02e038a7af20dec2d03b72833a7f05f3a20db887
781 F20101211_AAAASF bokhan_d_Page_002.pro
af8eb51918547309ebab20d8ac717acc
22118a8dc3a261fdb4e1b56f8ffbf7b4513663bc
F20101211_AAAARR bokhan_d_Page_120.tif
c37add5efc6ec3233eca3e018db6eec9
5778318af90e98967e4e613b058180e4a964a97a
8381 F20101211_AAAASG bokhan_d_Page_003.pro
432b2f50b9ccd7c07a9f51e78db05192
fd26688d1929a46ccf919011848849834de0113d
F20101211_AAAARS bokhan_d_Page_121.tif
d406ccad3e55df3a17af9b4566c5975a
7e81f0f71d7cff16b9b0d6250b92143ff81d0039
61464 F20101211_AAAASH bokhan_d_Page_004.pro
6c62a248da51ed3d15545f32d4134541
39881a20be7cb6714722b00f156de0de9c4ff7e3
F20101211_AAAART bokhan_d_Page_122.tif
64baedd6423906fa038823957a2d6978
1e84485333d97752b97f767a634c65b0fefb50f2
58186 F20101211_AAAASI bokhan_d_Page_005.pro
66f339a90723ac760846da5cc7dab61e
7094bae810c9eb521906ff441a0c23b1186cf6cc
F20101211_AAAARU bokhan_d_Page_123.tif
d0f26ccca5439ee2538c9266af9e1b4d
9145f071d60cecee00d8b957324e9d91dc06a98a
1823 F20101211_AAAASJ bokhan_d_Page_006.pro
e16abffbc7575a538f01c6c23d41911b
0edec6b07bc88570a36426dcee9bb4a07f2a46dc
F20101211_AAAARV bokhan_d_Page_124.tif
be9f5dee29645ee3c8213222e634c49e
f1652e2f5d08e9bd571d1b9f1d4eeca3742282e8
23370 F20101211_AAAASK bokhan_d_Page_008.pro
4a291a330836e1c90bd17956c389926b
4305cb7100b1ac03ec751c37d7a4b86dd3f7b51d
47984 F20101211_AAAASL bokhan_d_Page_009.pro
9288be5467ffd574933029b61cca29e3
d815463217e7a28b9b40f5ed57a2e2549a33379b
F20101211_AAAARW bokhan_d_Page_125.tif
a5f5710a804b092f0c5be46857174b6e
b700a6c65ea42a04d0ae16a69cce45c3014c4fac
37229 F20101211_AAAATA bokhan_d_Page_024.pro
80491d2248a4c74e5c9d35a6f535d108
dc28b0c8bf2796d3427d0c03d96c2df383e151df
55027 F20101211_AAAASM bokhan_d_Page_010.pro
0eb55e7b8fdf9972faa31f66b78193bd
fd97b0d135867c322835ba8ec43ddd0d9b5d9150
F20101211_AAAARX bokhan_d_Page_127.tif
adadaea25697bbced3d75fdf50cdcd0e
ff455a87d036ed6136a288dc3beda60d6f126c81
32973 F20101211_AAAATB bokhan_d_Page_025.pro
4c912feaac66b557b806db2de78c2a2b
14a4de99beb004762b90939b93d329f5c8eab7fa
2694 F20101211_AAAASN bokhan_d_Page_011.pro
264758cb79da699aa43179ed71ffb93a
effd9954f6ffc1ba7c4b386b1cd3b0b13d5a2103
F20101211_AAAARY bokhan_d_Page_130.tif
e69103f85d966393c83004a0193800b3
d1083ac90a36e2302109310c371a75aa05938dab
47472 F20101211_AAAATC bokhan_d_Page_026.pro
bc0a39c0fa0dd2c45512fe1d18ad7de3
75c4516c6e3d0e053da4f7ed49ff7b57f8143a68
46860 F20101211_AAAASO bokhan_d_Page_012.pro
ea616a9bde12315cb60a1f507d00a0a4
59dd05ec4554337142f4a5f18ecb8b86f186cc7e
F20101211_AAAARZ bokhan_d_Page_131.tif
e7499c2290000459aa16926838006307
4cbd96306ec713dd572fbcf0b372ea5da45a345a
61491 F20101211_AAAATD bokhan_d_Page_027.pro
10393dfbab14520137758167bde43f11
46693a63a898eb9be3dfaa9279c715592d888e19
46666 F20101211_AAAASP bokhan_d_Page_013.pro
320085fd26f04616efd60bd573ac91ed
5430dcd3bd0b88ce62515a68146373d6497056bd
54882 F20101211_AAAATE bokhan_d_Page_028.pro
2d1c7fc9234ef1f794ffbd829019205b
41566391c378c1b9571f13bd9271882f8c018eb2
24441 F20101211_AAAASQ bokhan_d_Page_014.pro
22f175cf98e7d3e4d0f2a8f3be18f1b6
882dd4ab6687f752f01b4a2fb154f7dfd44debe2
50622 F20101211_AAAATF bokhan_d_Page_029.pro
03f2b55440c1b26e9068cc4f1c784af1
c15ecf694d0a785184d9fdc400ac4b66a8ab84b7
42949 F20101211_AAAASR bokhan_d_Page_015.pro
e1f637655e499df41a5aaf38dda44557
468aee11e2747315eaac217ffa7d1309bba382c7
36746 F20101211_AAAATG bokhan_d_Page_030.pro
e30075e11bc6a142d04751c0270b8423
b122b746d87d7353404abdbf956f02dde10ca5f0
48257 F20101211_AAAASS bokhan_d_Page_016.pro
bda0a8274c900c23982776902d8a368c
c375d6f50972dde86b1d2a8a5d00292f4482c720
53953 F20101211_AAAATH bokhan_d_Page_031.pro
c8243df34a506f25f7fcdd0b58c00351
fb33cd658c211f3b596a994be16242d53b91d1eb
44658 F20101211_AAAAST bokhan_d_Page_017.pro
1d75ceacc334b6181f635fe2fe3db0eb
71cb78345e6a62a12a4803b9746e401847ce5b52
34826 F20101211_AAAATI bokhan_d_Page_032.pro
8784baa03290de6d75cfea2220e4d34b
a740e239901464269ea1212fd97491ffb19738e7
35588 F20101211_AAAASU bokhan_d_Page_018.pro
0219cee5d76cd93ab6611bc9272b35c5
f8dbf56e22de1695a3fd27786db0276d6d656db9
28957 F20101211_AAAATJ bokhan_d_Page_033.pro
f61625b699e8c4fe2d97bc522b81fba5
3f9e00ab18f837af846fc279d0433c263d141b7c
45464 F20101211_AAAASV bokhan_d_Page_019.pro
6ba3f762b31bdf336f48022b3a6fce40
14aadda9a1e31c4d1cab368f6f744d94760706bc
37161 F20101211_AAAATK bokhan_d_Page_034.pro
9bd9e6df5bd14146d6dc5e9e377bd721
4db0eecc07b6724d8922c614bfb020a312516fdd
47598 F20101211_AAAASW bokhan_d_Page_020.pro
97da4b2f822ce3ca67a13ab8701acd79
1159d2824509a033b4846712cc450eb8201216d3
30910 F20101211_AAAATL bokhan_d_Page_035.pro
136d4fdd08f0d97248dcdff0041a8b9a
b44d79330a3a61382a4b959deb431b774393dc61
13981 F20101211_AAAATM bokhan_d_Page_036.pro
32216d77c819348095488d41ec68781a
55ff829ff25dd19502a016fc7c335af6e9901c23
46164 F20101211_AAAASX bokhan_d_Page_021.pro
757a814fc524e02340c0ffa582fadaa0
78ee089e3bde80d3fae698bea829acadcf37d0cd
36941 F20101211_AAAAUA bokhan_d_Page_051.pro
ec7543bdc361eb318de63ec286d071f2
120a5f8b5600a5438004800717d36be627121c58
41556 F20101211_AAAATN bokhan_d_Page_037.pro
203179e96ee06382d887db06ef957a51
41e20b3d57721c4c8311132a1cb1489e3f154249
36317 F20101211_AAAASY bokhan_d_Page_022.pro
5624c201abcc81666aea5e36a98d2908
096f81983a8e2d1a6acd8fc4df21178326ba0479
41045 F20101211_AAAAUB bokhan_d_Page_053.pro
c114dd3762e8968f28d553ce058ca6a3
0605ed003660429f345ab8248d434422e8db0305
35763 F20101211_AAAATO bokhan_d_Page_039.pro
33e6889d552c941195ee6816325db84e
a96b3d8dbc29061179f6a8fd11b3e97deb1166cd
39834 F20101211_AAAASZ bokhan_d_Page_023.pro
e1d6728be996cc269776b88457b03b75
db843dc48d774a67d651eb4e859311cbc768a680
35839 F20101211_AAAAUC bokhan_d_Page_054.pro
596ca4d6193931d002dc8b3c536fa7f5
f82fd7fa05d51eaaf6cc9dd1114a5c4c8ed1954e
36807 F20101211_AAAATP bokhan_d_Page_040.pro
c3794f7dc23574e7c6c949fd9b09d673
480102973e1dca895a678d5f11ca0c13370c3261
31251 F20101211_AAAAUD bokhan_d_Page_055.pro
1cf2f1e00b90c00e323eafd4ce4b0f87
fcc8a2ea40137f8cc344a8c36aa7effffc898ace
44013 F20101211_AAAATQ bokhan_d_Page_041.pro
b455cec483ddef2d2c0373f01814f27e
582e50cb35ec8e077ea91d482c8bcb6c2ff1c929
30332 F20101211_AAAAUE bokhan_d_Page_056.pro
2a8ac91ca8f75a436999111d16634ee5
1728043f7d5f6e8d1d81c7bae29db45fe6b7c0cc
46997 F20101211_AAAATR bokhan_d_Page_042.pro
aae74717ea4065dc4bdd7fb2f38d02ef
4698a618bc43f8505e63a4f085220955d2826246
2159 F20101211_AAABAA bokhan_d_Page_089.txt
e818e745fe9d2ae65e8685805b1d2513
a53e85759e3e6cc4c4595b59476d029793809a63
25114 F20101211_AAAAUF bokhan_d_Page_057.pro
bc249320ea491f7c62f8dae1fc9b44fd
0fd8e3deb11cf07e748c64a6e4fe000263937e8d
35268 F20101211_AAAATS bokhan_d_Page_043.pro
6570bab26e5de0fd70411df5bab15f36
bbcdfa2e7ef1369f72b1151829fed6c49334b664
1926 F20101211_AAABAB bokhan_d_Page_090.txt
cc7e24dbeccbc722e986ed82fe6a3bd8
eeca2dc5adf065f6ad925b9341a818a6ea9b383b
17793 F20101211_AAAAUG bokhan_d_Page_058.pro
a6e626401f860fe10852fddcd6d9b6d8
b9a92c4a450cf90eb1ed7999380b8648ddf7d57b
32797 F20101211_AAAATT bokhan_d_Page_044.pro
63676e8b301c3fcc124a324974f9ad6e
4a193866b5d26859d78fb6e292d2d5b899fd97e8
1699 F20101211_AAABAC bokhan_d_Page_091.txt
77dee579a8f4ce01eb7f83f639b83e28
5dc984016f9184b6170dcb5f0c2ca69ba297139b
28206 F20101211_AAAAUH bokhan_d_Page_059.pro
258b58f1dcd80419024a34363d4e90f9
ef9029e6c9b271a94f83655a7597dc80dc7e4c15
38649 F20101211_AAAATU bokhan_d_Page_045.pro
49de34b191c2494332936308d7f14e3a
b0819bc6ab14b79723ea52fb6eab567b9876a5ac
1958 F20101211_AAABAD bokhan_d_Page_092.txt
2f64182c890080637d9ab2516200e656
52abd180f25c6fcae3cdae5d015526fb31f36cb0
31148 F20101211_AAAAUI bokhan_d_Page_060.pro
fa26a47d3cdbc7f9d5dc6452de931ea9
8a9df37136528a1f00600440b03f5e2c9a6c6a3e
36527 F20101211_AAAATV bokhan_d_Page_046.pro
4736cbeeb874f21bfd8cd31bfd2e25fd
11f2dc98fea0c1db9c40baabaca5c0b38692bed3
49198 F20101211_AAAAUJ bokhan_d_Page_061.pro
2e33417bc3e33972901835ad08c7c694
ffb2ba320e1e0795eb2908c1ea411b574044cfc5
52878 F20101211_AAAATW bokhan_d_Page_047.pro
e1f6f8783b9c9f58575e2d4de558b9de
2f32d7cb13179eb39e5ff0dd0d6b494803d602e1
2177 F20101211_AAABAE bokhan_d_Page_093.txt
72226781e3065031c58f6aee28d7b2a1
1158955a41c3512b8895864ca5cc976a814e750e
48880 F20101211_AAAAUK bokhan_d_Page_062.pro
179ba805509bd6b0ad596d3193391ebc
07dc3ac3dbeffa51b6811f565172eee93d69b315
43311 F20101211_AAAATX bokhan_d_Page_048.pro
8ee34006fb5459d5d0b3a0b5882454ac
ed06988eb405c824eb2d64c5d7ea4e205e68e2bd
2216 F20101211_AAABAF bokhan_d_Page_094.txt
3ea65809abeb914abdd57a192587b2e6
b31ecbaf94013c265410cbb7e5a063367aeb5195
60851 F20101211_AAAAUL bokhan_d_Page_063.pro
ba54d0a4d66b619240fdc39100f0baa5
693ad00db779ced1cc3ed10a06980a8a53dc011c
2271 F20101211_AAABAG bokhan_d_Page_095.txt
986af6e7caece8fceb1139d68ec758ec
b525933e0e2b2d629b96cac3647a6f733d6b358f
6990 F20101211_AAAAVA bokhan_d_Page_079.pro
f93038b0cd3849b878889f3d182d0dbd
9179f15b6933a4a66293ea5987f76ae9242a7772
52320 F20101211_AAAAUM bokhan_d_Page_064.pro
df8d7ebf13652f379ee9908b1bb1be2f
38a92b74a655cfeea72171cdf99be894c6e0fde7
33978 F20101211_AAAATY bokhan_d_Page_049.pro
c65410df0658d6d7ef75ddacb57b47fb
8915ae699c333dcf8c63af62f2c1193792fe86ff
793 F20101211_AAABAH bokhan_d_Page_096.txt
c1ca161058fae24827de446c84b40f90
37fe87e0a32fe62e093c38056b91dc6568589e40
52747 F20101211_AAAAVB bokhan_d_Page_080.pro
0512bbb18ff58e5cec37a34b119ee899
41453b29dccbfab5ed1f6498a3882da742d8ca46
56686 F20101211_AAAAUN bokhan_d_Page_065.pro
2e4d67b63ed7a1739b1c4e950f0ba412
1ad28537ca88a9295ab7477a2698ee41089c11a5
52613 F20101211_AAAATZ bokhan_d_Page_050.pro
a6042c8b561a52f022cceaec4a035941
f6aed9bf9606f12a900b3993e8a90460876fb6a0
2327 F20101211_AAABAI bokhan_d_Page_097.txt
1be22c4e89f33e3a6796e187536e919a
76e0d8338b59ade5089aa37521d9701c5949e857
59200 F20101211_AAAAVC bokhan_d_Page_081.pro
69689e4c9be2a49a17f7c9ac689a6e67
73dfbe885e5aeb11bbe885dadd3c0d7d72b00435
52995 F20101211_AAAAUO bokhan_d_Page_066.pro
6dcc081fd93fdfa64c15557b8f96caf1
ef9fb7bdb0306699966f5c75c41481ca5229e474
1323 F20101211_AAABAJ bokhan_d_Page_098.txt
695d7e7bf3ca33f3126d4c44e549fb3c
556981a5b45eb8c4c374140ed0c444307985d7d5
56201 F20101211_AAAAVD bokhan_d_Page_082.pro
216aba3ed25f7e2cc3cbe00bd9d11a45
f6c35d2ccd5c5537e327ac1bc379d1db8e8435c6
55726 F20101211_AAAAUP bokhan_d_Page_067.pro
e6d26a1ca74b872c3be10180a744ff41
141bf2834f33a51aa0c1e62e43fcef849efc3a55
931 F20101211_AAABAK bokhan_d_Page_099.txt
9712e780b836f8a6d345c18b9453fad5
64bc9c8f222a967539a86d7e45bc391429b6e511
34978 F20101211_AAAAVE bokhan_d_Page_083.pro
be08536e18e0454055feaa51e10fc84a
c2ebd0d0c168989b18fecce22b32566827752133
15301 F20101211_AAAAUQ bokhan_d_Page_068.pro
5c9afef0cf9be3e8058667cac5558c10
6d4d899118eac57eb1719400f4e9909311c56bd1
1877 F20101211_AAABAL bokhan_d_Page_100.txt
1e6ba9b6e1ee86efe2189d9e50b403eb
41d9bbeaa67714389cba9610bd51d64ec6743933
47616 F20101211_AAAAVF bokhan_d_Page_084.pro
15f369c6d866bddba95f42e8a54d7cde
95954dbe313dfaf79400fa2acfe25a881c2cd9a3
35617 F20101211_AAAAUR bokhan_d_Page_069.pro
e27f09357e0e84db45537e9a63051050
f59c58f1748a44b43239a7be273769e3f30b330f
1671 F20101211_AAABBA bokhan_d_Page_116.txt
6d2426a5f9e5d619411f85131a694283
55ce906064c87498c15df8f444bff17c1ed391e8
2109 F20101211_AAABAM bokhan_d_Page_101.txt
97535fda53ba4e2ea1d954514959398f
3345fb084bbe30c3f41d2aa6af934fbee93f456e
30564 F20101211_AAAAVG bokhan_d_Page_085.pro
d4aef453139bf69fd9f7bf485b0f7bdc
ad00311b4c4a243b10d14054baeac53eaff24c99
43854 F20101211_AAAAUS bokhan_d_Page_070.pro
7acb75f30ebf0bfc45d1541aece6eb00
4891b0170c8f6bf15a68dfd80375e195c0372372
1597 F20101211_AAABBB bokhan_d_Page_117.txt
8b3e641514fb2bf1033c4d901be6db7a
a3e5540bf26681e6a8af5ffa6e2309fee5ba742d
1583 F20101211_AAABAN bokhan_d_Page_102.txt
1270a90faf56beae4f9859f64ef8a667
b7f4b15a782b91c390b0de2021c5254c0afb1e65
28267 F20101211_AAAAVH bokhan_d_Page_086.pro
e4eb15ab55af624922ef5e9152b972f1
93d93a38449d87def32438fbf44f2b7869c57be2
52057 F20101211_AAAAUT bokhan_d_Page_071.pro
533fecb6381f99178fa6675a8d267fc6
76eea3962f737f400c204d25c1311974b48b2da2
1644 F20101211_AAABBC bokhan_d_Page_118.txt
445f9c1e0738eae5dd536a5a2db7cf59
bbfe7cab6f3c43a8bfa267d447d8d848f3cc1b95
1406 F20101211_AAABAO bokhan_d_Page_103.txt
9f817106ddeab4c2bd2bd4b7058ed811
505b1f5668c3915d8a1363d41164ca433b1d9a2b
59173 F20101211_AAAAVI bokhan_d_Page_087.pro
e869f9166ffde658cb098ce2a3ab8fb9
d2e0b86756eff7eb0afcfa3c128c1a828a2ecaef
61296 F20101211_AAAAUU bokhan_d_Page_072.pro
3f0fa3d985080b6eeb8534594938889c
faa0721e592cc22fb6cbb48d46553e1cf285ef78
1193 F20101211_AAABBD bokhan_d_Page_119.txt
43db40e7ced4fba63ad09161afd02f02
940f9da79b4a88dc9a33be231e937128319d61e2
1725 F20101211_AAABAP bokhan_d_Page_104.txt
b2f28b2eb0c4afa5e4faaa06240fdc2b
bda60ef98676ebb91f81c89dca17cc46d24d9116
66163 F20101211_AAAAVJ bokhan_d_Page_088.pro
a93e95aefbc537e38a221409801a1fdd
88a4907510233d231d3146a762fe6c6536ab90ef
26425 F20101211_AAAAUV bokhan_d_Page_073.pro
e7cdff145f03045a64eb397ea7b18715
3ee4621546fa408c309c4e626ff06800b37a0da7
1371 F20101211_AAABBE bokhan_d_Page_121.txt
f7f6ec31f24e317debc8f06c69a9b5ab
0c1951c69b2a11ddecc75c0df35d23a1099638a6
2350 F20101211_AAABAQ bokhan_d_Page_105.txt
cd8a51e63e0023409744228767699ba1
e2c5ca88c6b13dc16d23687ab139141594fdc923
47233 F20101211_AAAAVK bokhan_d_Page_090.pro
a8ebefc8c2f87dc1feec22ab37c99e3d
c99c0333601b0a87bc4d295048b3ede9051bcb95
48314 F20101211_AAAAUW bokhan_d_Page_075.pro
16da19e977f3f5291ed2e5c9c4ea8897
94ea4b2eb51e08c5eb641b46cd9d4f0755ca44b3
2230 F20101211_AAABAR bokhan_d_Page_106.txt
0d5cd0a959234bf08c7a5e85455ec08f
b12e8ff4f3163f49e5c2ced89150ebcde69a3cc4
30094 F20101211_AAAAVL bokhan_d_Page_091.pro
10bb8a696b6191a5e5278139e1f7b82e
e2e9c2454b7e9389b9426c78d057d036ff85bbfd
27779 F20101211_AAAAUX bokhan_d_Page_076.pro
bc1ad7fda5da487459b39f55f93c2540
b9d05f5511dfb86cb3676fd7a0940ffc6d4be63c
1246 F20101211_AAABBF bokhan_d_Page_122.txt
bfa5e3dcd1630aba2a41a76af21910f0
88a69916620c66f70d6b7f4560a6e2a5510cfa54
26494 F20101211_AAAAWA bokhan_d_Page_109.pro
fb6d05336785819db8b753952ec339bd
dcd64b62d50930149927355256feedda688fb338
2113 F20101211_AAABAS bokhan_d_Page_107.txt
e66fc77be4f545151563c0106e1b4fe1
1ae9e2a2f2bade60ecd731ed80d3598d93689f46
36569 F20101211_AAAAVM bokhan_d_Page_092.pro
91f14de26cf839c8b22fa1bf1af42b17
9e5c71dbbb86b120f7722dd0d8d25baab66a8dd1
7226 F20101211_AAAAUY bokhan_d_Page_077.pro
32498c810ab4aef7d4051f7510e87a71
bd8367d74a6e35454afef772307d9f9f34590ad8
1407 F20101211_AAABBG bokhan_d_Page_123.txt
5e87d33b79c3b0db61d677270e2ce3bd
2c173638a4097ccf5428e4c437b2b4465cb119bd
25136 F20101211_AAAAWB bokhan_d_Page_110.pro
72e6c5ea959c0c545d88f1e979197f55
e88978dd3de8aff5b7e155f557ed4ae9491ff08f
2134 F20101211_AAABAT bokhan_d_Page_108.txt
e43c4e1bc5e19b22c6f8edc23beee70c
24f1eb80c2b00194b2af3bab3ccca88fada724f9
51331 F20101211_AAAAVN bokhan_d_Page_093.pro
7b6e48ce7ce342325133bc7bef96263a
b61e24261629550a471e30981a84a27c11154f63
1571 F20101211_AAABBH bokhan_d_Page_124.txt
647a0073d1013ba2f49d73f48750827b
8c364f9724f80a423fe2dadff68370ecb2630160
28738 F20101211_AAAAWC bokhan_d_Page_111.pro
1f2c9b310fe02a1d9fc6bdadf65fcdea
34b3799dced08a67a82b830c1933adc8317ad7e4
1093 F20101211_AAABAU bokhan_d_Page_109.txt
903efffed7e38d1eb796da6adad0eb7b
7df5dd406f619abf7abb769df564642d8f2176bf
52078 F20101211_AAAAVO bokhan_d_Page_094.pro
fa0eaa0ff67dd51807f0318e983453e7
6027fa4de6e99af2426577566064e62be7aac347
7089 F20101211_AAAAUZ bokhan_d_Page_078.pro
b2dfb9ecca21fd9b155ae23e9b527ee4
8031f2f581b7688a984831a94c692028dca4bcc3
1457 F20101211_AAABBI bokhan_d_Page_125.txt
749cc5ffd2d99d51813fecdb43317cbb
b4cf04730c45c67bc904b8e2ba07954203e2deec
26787 F20101211_AAAAWD bokhan_d_Page_112.pro
fd00bac1d8244c5f791f5fab53e72004
a5feacbd9c778b1021676f86056ecb2fbd28c454
1643 F20101211_AAABAV bokhan_d_Page_110.txt
0da96bfccbffbeaa4bb7b714f2f51cd6
2d2be5434a8252dcb27b45bd466cb8b8cda754b2
57685 F20101211_AAAAVP bokhan_d_Page_095.pro
3a6ddaba4751ba3030cbd6211793e5f9
59083a6c73883c3a77e6457e9fad8d44c03a0b3a
305 F20101211_AAABBJ bokhan_d_Page_126.txt
95c761c046efe4bbb44beb04906a13a1
619c99baa7c9ef2378567ba9064a33bde27e778a
29276 F20101211_AAAAWE bokhan_d_Page_113.pro
f8b53e85747c14709610093ea6c0c40c
6c590801af28b91470a03e6874fe1e8c85d18c33
1471 F20101211_AAABAW bokhan_d_Page_111.txt
9f4ed994f20d2494edeaabd58fc9196b
fecfc32a617021a0ff18c7ec658be4b699e7375e
18020 F20101211_AAAAVQ bokhan_d_Page_096.pro
f606b0fe8627e56cfdd1a78aa73d0ae8
485b214d5c60b3018ac26063c1e8ced708a840cc
1590 F20101211_AAABBK bokhan_d_Page_127.txt
6f2574b3183814b3cd1e226470f25300
3a997abbfd13e23f9f03c74459731b705d92baef
31724 F20101211_AAAAWF bokhan_d_Page_114.pro
901d78355ba859aad00e864cff4f8c9e
53c306c700925612069b5d8be3933304cd257340
1505 F20101211_AAABAX bokhan_d_Page_113.txt
32e0ba8358ae0ab3ad05e81321cb7f8c
e90ffa994934d1f8e85e434cc4d899c191440040
58107 F20101211_AAAAVR bokhan_d_Page_097.pro
c778cc9cba2312bf1f17f5c833abb890
9f825b2ffaeb89d2a4f52a062ace92ed67ca2f54
24282 F20101211_AAABCA bokhan_d_Page_005.QC.jpg
2dc63c89e908a3f14f8e8bd7f56cf12e
f387878f252e234e4824775539b38fe52bba4664
1669 F20101211_AAABBL bokhan_d_Page_128.txt
2019e13440610a3cdfcb1cd07375fe8d
1cfe352d2482fe65a134035bc37958f8f46d5f8e
F20101211_AAABAY bokhan_d_Page_114.txt
b17a3587979752c41081e579a32c9c5f
b18c8d7676d2090dbdae525184f85f8caf429aec
18711 F20101211_AAAAVS bokhan_d_Page_099.pro
08fb0da0e5dcb997ae852145d738c4f5
9e16085654bcd37c53e5113641c8b91e455b83cb
6222 F20101211_AAABCB bokhan_d_Page_005thm.jpg
be100910ec11b221bfc2d0dfd49e7e5a
5d7a338e2c9208190b07a08a47465235fbf31c2f
1787 F20101211_AAABBM bokhan_d_Page_129.txt
e90fd905358d016e276be9c4cdea0333
641a6befcd6123e94aa50b730f14ef9fd5690097
28953 F20101211_AAAAWG bokhan_d_Page_115.pro
2992383457949e828e6a64a68237ebef
6e77e5a23401831d8f44a95d45e140ee151f0950
1409 F20101211_AAABAZ bokhan_d_Page_115.txt
47e77a9c22bac44f99213366b3ad0841
5d8aa589646348bb9382574780db85ab06818e4e
44523 F20101211_AAAAVT bokhan_d_Page_100.pro
46363a1ae5bfd6e4db7bd15b7dce26c5
e178805efb1f6f7f938c2e0714d1abfba76cdab1
3780 F20101211_AAABCC bokhan_d_Page_006.QC.jpg
41705f58ae05b12e443152ac4775b005
4b17f9c86dc4d12e4e9aa90df26d8f8d1f0168e1
1570 F20101211_AAABBN bokhan_d_Page_130.txt
0653607a0e59b48c9b6aa6bbc45f51d9
ec28c1990fae862bbde9706c844453b633750c17
29212 F20101211_AAAAWH bokhan_d_Page_117.pro
bd7ed6be37a16c2272e557b312f39805
4320082ce1462e3c9b2285c27ddc94f64c3fa962
29290 F20101211_AAAAVU bokhan_d_Page_102.pro
6a176721fb764ae8241fe6b3b5557b68
c4786d661c8836371f238272cb63241b0e2f9762
1419 F20101211_AAABCD bokhan_d_Page_006thm.jpg
17b47a4aae7d2e1ff63450f30c302344
7dd310524de9de0dab2b5b13a9bf6b3a038fcc8d
2010 F20101211_AAABBO bokhan_d_Page_131.txt
2696b78765beb5d6d99e9008b51e97d9
231728a2583a35ef0ac1736cdd173e7be6572850
29341 F20101211_AAAAWI bokhan_d_Page_118.pro
aacfaf09961c9d6ebc02509f36cf69f5
29aeb2e91aa6c5a254473546d7d4489b4aff1462
27226 F20101211_AAAAVV bokhan_d_Page_103.pro
56d6c7a6a093b1fe81eb448f1c3a0826
7148d59276a79e7fc26575198cfa56885b65fdd4
15978 F20101211_AAABCE bokhan_d_Page_007.QC.jpg
80b47077302d671b389072ac44e9acaa
789c7463b4ea04097932f6882242faa827674a09
2335 F20101211_AAABBP bokhan_d_Page_132.txt
1a52e8dce11dbabc95cd0ea741fcedc0
805357c822cf6359137a9b11aca9af80d8d05148
F20101211_AAAAWJ bokhan_d_Page_119.pro
5c19d6640c6ae354f94871117114a08d
9f259a31fd0314e83947735aaa4bae2d2cecc683
37784 F20101211_AAAAVW bokhan_d_Page_104.pro
91ae5d56345cacec894f9a704fb2e96e
bfa2d9f6de7d1a0b3a6d26f2041cc3841badd51e
4335 F20101211_AAABCF bokhan_d_Page_007thm.jpg
363348651b1eca0ed1d38e99685fa605
49e3fcd1a54d72ff7de616f05d96d1ee17467744
1994 F20101211_AAABBQ bokhan_d_Page_133.txt
3673c3b32710dfed55c9d1aa750350c6
663680f6ff5cb29398321cf7fc6def86083189b8
25201 F20101211_AAAAWK bokhan_d_Page_121.pro
e70e4b7379553f5dbba67ae8338e0cbf
cb6f3fcbddc7b2e3d341663ae80f374275036635
50121 F20101211_AAAAVX bokhan_d_Page_105.pro
967dd326a494646d7e9610582f3002b9
9e0d0e23bda5daec7e4a7435fa203f874bdfcca8
1282 F20101211_AAABBR bokhan_d_Page_135.txt
62567e1519692864a63ce31dabd0df0f
5d8f9e90668ad6186112262672ec5f73511041c9
27306 F20101211_AAAAWL bokhan_d_Page_122.pro
5bd64e2e350c59e0f49de6458b336f51
d28197e6877414f195f7ada192c9437f201f16ba
49263 F20101211_AAAAVY bokhan_d_Page_106.pro
4ec07ede4ace6199f01e603c217cb150
d906bbc3bddb9c01aa562c5dc0aba0eb9e7d15a1
10898 F20101211_AAABCG bokhan_d_Page_008.QC.jpg
32e6d6fbafaef6f27ad72bf7202dcba3
f654f72b9757132f0a6c002c400776ce8aa72c4b
78 F20101211_AAAAXA bokhan_d_Page_002.txt
7c46e709fdd45a7c99b7b1082b63e130
50a5673e671e77a816841ac96ad7f4b31947d189
1414568 F20101211_AAABBS bokhan_d.pdf
ce39179666211b66dfc8000c55658209
ba7b69bc84f582c67e70c0511cd1828e644c2e37
26970 F20101211_AAAAWM bokhan_d_Page_123.pro
91efd2b92a3adfcd0df070e413ed4f93
e332431a8a1b9f6b3c5b443e591a72a58c823481
53988 F20101211_AAAAVZ bokhan_d_Page_108.pro
2ebd6271893eb70fa237e45f6d7c65f9
2b4b10a12eacaca784f0535edfc5b61aa795b95b
3093 F20101211_AAABCH bokhan_d_Page_008thm.jpg
d50f735143a78bd890e60bf3e854961d
cb189c7f06c49bfc03f5e7bc7fcb663ecafa049c
371 F20101211_AAAAXB bokhan_d_Page_003.txt
4b9ade299fce49b983d6410669914bcf
c5e4b58bf520b8cd52c71b8989d7a7892df886f4
7518 F20101211_AAABBT bokhan_d_Page_001.QC.jpg
41a99a6236e98627a0cbb6a146231add
c19d10abda9fd06075ef6b9ee0da4d7e0ee80102
29649 F20101211_AAAAWN bokhan_d_Page_124.pro
49cc1472f4dff14be7e9e4420774e914
2fa2457dd21e26373e4245c43a6777b0252cb25e
21131 F20101211_AAABCI bokhan_d_Page_009.QC.jpg
e0bfff91d44727ebe62b921c95973080
41486bc2688b1faab71e278a6f5d959f2d33eec6
2700 F20101211_AAAAXC bokhan_d_Page_004.txt
a6f482a0acc390dccc7b1e1f2931deca
e5e6d280cbef39440a4bea34feb9464bf68d44ec
2253 F20101211_AAABBU bokhan_d_Page_001thm.jpg
3323a7c8a8a4f519617b9bc8b5a3dc41
21341dbe89c621da8cb7f05be9ed48a7d86a7e65
24734 F20101211_AAAAWO bokhan_d_Page_125.pro
4ce3580737f2bb5cfe8522a6dc009d51
8b2d66a994362fbd26211082e2e0ebcac363f039
5595 F20101211_AAABCJ bokhan_d_Page_009thm.jpg
b296ea7bd39c99717c6b9b9f7673173a
96833521a9d8cd2f8c89e84817a0269694b6c874
2379 F20101211_AAAAXD bokhan_d_Page_005.txt
95f24fa0e5f409e2fb54b4ddcd1f1ef4
3bf1393a81b25876308e349797c8d6456fc374b5
3151 F20101211_AAABBV bokhan_d_Page_002.QC.jpg
f761ec1edd64031e4cfdb19c31e770d3
6a26c4affddcb6136d97ad633eee96f6e888af90
6587 F20101211_AAAAWP bokhan_d_Page_126.pro
926daf42cb89283fa3bd5ea7ac0be25e
9f1096c4350e634c9e63d6b194bc6207d447e50d
23668 F20101211_AAABCK bokhan_d_Page_010.QC.jpg
e67574edafe7aada98c2882879646951
e4c8d46fd9b4afa6738e358b8d5fbe33bfc6653d
76 F20101211_AAAAXE bokhan_d_Page_006.txt
aad97d1cc9f85027c684bca3ce27784f
087d66d6d9273753bfa2c9695e1311b66a536f91
1339 F20101211_AAABBW bokhan_d_Page_002thm.jpg
c41690b167db53b51f80990aeff9bdbb
18fa42a9ef6c233649971a85c63ca53609a5be19
29417 F20101211_AAAAWQ bokhan_d_Page_127.pro
672675a2ac75555c0c052933c35a969e
810dc707c24fab1a29c0c520e6bc66381a718022
22634 F20101211_AAABDA bokhan_d_Page_020.QC.jpg
963b01f67b83a10c08bcbf5d3ce73db5
34366a54b637335c81f1454e230417af422f779d
6024 F20101211_AAABCL bokhan_d_Page_010thm.jpg
55d54db3f1a8ceeaefcbd71570675175
22d4f7bdaf83937a157e6b55189babc87d359528
1617 F20101211_AAAAXF bokhan_d_Page_007.txt
a053e4e41cd0aff62e663886148acfb5
782021edddc05cd3840812c63b12237777c6bcd5
6105 F20101211_AAABBX bokhan_d_Page_003.QC.jpg
3e8ded0c560e1e7ab09fb411a2e33663
685411bde5585a359383c72eb2e4e8d29018ad0e
28529 F20101211_AAAAWR bokhan_d_Page_128.pro
a745a56b4470a7725d14d55b973c9705
84d9fd4d6b3478571cc834025c1d227bbf0f2949
20616 F20101211_AAABDB bokhan_d_Page_021.QC.jpg
3c0d45fb7447be6715edf06493f979f8
1a679f2d46d5b63cfc1c2e20489297e556be0289
1508 F20101211_AAABCM bokhan_d_Page_011thm.jpg
fa23c67cd3d79b577e58ab107070a8a7
5b5064c0f09f7e5669609fa647f00153bc6e591c
1002 F20101211_AAAAXG bokhan_d_Page_008.txt
c5e16afcf3988f5d9dfac16d56d0fa84
4df07482629fb3875d78dbd10cbd55feb105bc4a
23568 F20101211_AAABBY bokhan_d_Page_004.QC.jpg
ccb24593cf31f8bcae7684b6c2b055b4
75231e5be0851f64ea7957a56bffc870b8c31d1e
30817 F20101211_AAAAWS bokhan_d_Page_129.pro
d076c94c10d165374372a7e86312acdd
1890742294d400541ba475ea437de7f4a15ac358
17247 F20101211_AAABDC bokhan_d_Page_022.QC.jpg
d14edf67a3890aa161db38a904995ebc
cde8c5af18a10a274e2f6f6e1b7cc5f1fb23e572
19505 F20101211_AAABCN bokhan_d_Page_012.QC.jpg
4ddc942218cbb81e3ff56b0edf4412c7
e4a4c97d23083f6a33245b20da2b58e148a925e1
2074 F20101211_AAAAXH bokhan_d_Page_009.txt
599d590a51fd466e5b6538128f7fc82b
f1d8cc8987bda6ed5910f00e157cc608cded9659
5661 F20101211_AAABBZ bokhan_d_Page_004thm.jpg
d8f90dc6a1fc9ffb724e44bb0e84fb30
e0c0fb71228b181ebc00e25ed6ec732187b839b5
30036 F20101211_AAAAWT bokhan_d_Page_130.pro
7ed47d589158051d1c858e4bf26a7cc8
afb035d280b8dd5db72750fa147cb509a11e7780
5025 F20101211_AAABDD bokhan_d_Page_022thm.jpg
dfd1ae3598c747b9bdb85d030f48df64
8cea3ca25611cf7d87fbc1af42692cfa56dfd3a1
5692 F20101211_AAABCO bokhan_d_Page_012thm.jpg
83737cc4b70b50d75b0680736926706d
599df9fb3a28e4d61a4c83935ed273c954c101a0
2171 F20101211_AAAAXI bokhan_d_Page_010.txt
ee3a3700edb583d3c97dda972fbaa4cd
bb88c0de661be9dec4d5f7057eeae423ed475e4b
49403 F20101211_AAAAWU bokhan_d_Page_131.pro
a4fd067596c6478a6ebdd0e616bff20d
97c20dc1323b0c6550de558eeec21d3571d99c35
23173 F20101211_AAABCP bokhan_d_Page_013.QC.jpg
d48d83faf4b237b39f3a56674a9a9e35
8c9aff3057783a0e31bb5cfdd2c3b086f0117ca0
112 F20101211_AAAAXJ bokhan_d_Page_011.txt
fff389e7f86422b16abad5aef4c19984
6f509e5a6c608a133048dfe690b9c51cd11d0b17
58353 F20101211_AAAAWV bokhan_d_Page_132.pro
aa42b885494b0ca80472f2bc73998c78
2e17be1da42269d1b515113577241b77f8300bb8
16643 F20101211_AAABDE bokhan_d_Page_023.QC.jpg
31303687c28502cf42e0aa29b27315df
deb210d93e769ef24aeb56517bd062282e11c20a
12259 F20101211_AAABCQ bokhan_d_Page_014.QC.jpg
650f52402ab200e87a4f7c814aafb93a
d0f799a137f20ce9d6641a426e39aa162468bc27
2046 F20101211_AAAAXK bokhan_d_Page_012.txt
03e46fe7fe399c652578e628e91979c7
6538a509db79ed026a59ca17a52cf4e81e8256a4
51104 F20101211_AAAAWW bokhan_d_Page_133.pro
e8147777cbc09418293aa14e281c5f11
a5b1697fba3287cac9421b248d5d68312f8e0cad
5089 F20101211_AAABDF bokhan_d_Page_023thm.jpg
00d46cf16e776987e9e743ce1401b3ff
9f1f5f53842111bfe4e64e95d8913aa027bb3613
3943 F20101211_AAABCR bokhan_d_Page_014thm.jpg
6c2d624d5c477670116906dc06c4db69
80d500834f27551a3f69c1dcbcb2fc0ae04845c6
2004 F20101211_AAAAXL bokhan_d_Page_013.txt
150c2557b23cc36200d7e57baed0c8f3
c01d0b6267b597630dd201b5ad868737cb6936ad
23680 F20101211_AAAAWX bokhan_d_Page_134.pro
226ed844a06cfae5c9d89d1c31e77d47
f0f0fb4a56607d97d1e84dee73b2e1d0d10d76d8
16949 F20101211_AAABDG bokhan_d_Page_025.QC.jpg
4caf6a91b53bdc97b6d52bf9cb291fef
7b397bb60cf09b429f4a6deae49a3d1d99113380
1468 F20101211_AAAAYA bokhan_d_Page_030.txt
bd9ce8424a0713fa379ebed84e6809bb
a283493fc36a1b4ffe6d114b9dec5e6cbd779d0e
20414 F20101211_AAABCS bokhan_d_Page_015.QC.jpg
74ac765bdd6135670c8acc19f379b3b4
3cac73a0660f834828d90348edaf359b2908dd7a
1504 F20101211_AAAAXM bokhan_d_Page_014.txt
4df228276bd751aa267e12b482649283
d7668a546d0828fca9dbe72d4b9b2bededa24c96
31416 F20101211_AAAAWY bokhan_d_Page_135.pro
ef6ae19df5cb3a803b83c4a0accb0711
1bcb7c5853778b46a8094b614678421e0f489335
2275 F20101211_AAAAYB bokhan_d_Page_031.txt
4f7f1f5da4b2e2cd8a1c61fa89780f32
095e94c7d75323f259bd27c71758f238717a5327
5536 F20101211_AAABCT bokhan_d_Page_015thm.jpg
e036b8c2336c04d39a2c07585b8f544f
8e117099c64720830309ebaaa29feebe2934fdb4
2094 F20101211_AAAAXN bokhan_d_Page_015.txt
703899f796e6604092cb195100d5ea06
038efb280c50ce406a33142582976cb7a958867d
491 F20101211_AAAAWZ bokhan_d_Page_001.txt
b6506896865618488379a97b220dc69f
29a1535b965ae27b40e8426b357d170f1b592652
5048 F20101211_AAABDH bokhan_d_Page_025thm.jpg
cec54d1f0cb1fbd0380a58a98d9b32b7
87b965433892b6bfe1b50d54d3a31b3b207590bb
1741 F20101211_AAAAYC bokhan_d_Page_032.txt
218520fbedd784a2f4bef5e238d09cbc
d7369146ff7126036db218dc08c1dc1a175c1fa3
20557 F20101211_AAABCU bokhan_d_Page_016.QC.jpg
3cd97ed04d9a7599e99a11906cf55d7a
2bbf3f8d5a819ce10e76739912023a395c1f5631
2199 F20101211_AAAAXO bokhan_d_Page_016.txt
9a89067b2a9b2989f52576e1359069ed
f6ce16ce759fc9954a0b3586b6f22c201bdac764
23943 F20101211_AAABDI bokhan_d_Page_026.QC.jpg
4a565e804dd63041d7a144cdf34b24e8
63b1cd8668071a05dc1f2432016be910aa64473c
1357 F20101211_AAAAYD bokhan_d_Page_033.txt
d08d666809d7322f1ce424bd87129555
a63872f9914f0465e3ca009235a5537f7b95de84
5553 F20101211_AAABCV bokhan_d_Page_016thm.jpg
f3fdd067cfa5e2140f642c7f2efe7eb2
123c85ad22c7c8d9ce8c413b3574a5747606c32e
2104 F20101211_AAAAXP bokhan_d_Page_017.txt
9fe8b847026fbc1e33a58443dd17ba57
9618803439ad7a5f33c0c424556acd48730e2f75
26025 F20101211_AAABDJ bokhan_d_Page_027.QC.jpg
93d3d841eeba2ee1a681c9e1594910d8
2c9f2e6a9ae466e3f55b2df6c67b39dfd3d657db
2112 F20101211_AAAAYE bokhan_d_Page_034.txt
91dd9b8adf7e9fce556542bed6d07d86
b4888b8c4582d41746ca535896d0aa52ebb50bfe
21220 F20101211_AAABCW bokhan_d_Page_017.QC.jpg
60f3f9d0123c33186b458bdb0682921d
4312f0ce008a9d854e37a6c0cc13dfaa185bb812
2213 F20101211_AAAAXQ bokhan_d_Page_019.txt
83478dd0898c07a82bfc569d8611b8f6
1fc03c545430c12bca233387f165c0db20f09835
51300 F20101211_AAAAAI bokhan_d_Page_107.pro
fd088adcca1f0cb515c8443b2c696ec1
8af451d1b4457d2315294155b45a69f2cec1d2a3
6345 F20101211_AAABDK bokhan_d_Page_027thm.jpg
8f778658c3e3b0e61800b653fe384d19
4899b7f3d27f29b05c59b86bd453aba448bcb9db
1545 F20101211_AAAAYF bokhan_d_Page_035.txt
49a39b9b299261bf7236a9beebc128e3
bc853f7a8948c7a143d8527ef69a89028b4e6e4f
5808 F20101211_AAABCX bokhan_d_Page_017thm.jpg
69f0bbda50a6f5baa913fc9df9f04465
c5ac192cadd0be5262c203faec1a5f939010520c
2085 F20101211_AAAAXR bokhan_d_Page_020.txt
6d64947e7f581027e84888b363b7af9d
f3d650dc8b1a1bbb5d2762134b306516239a975f
4788 F20101211_AAABEA bokhan_d_Page_035thm.jpg
61a966f63759120d0adddda2e6ae1154
176ffe19f9f218574ad0207007f63b548ac3650f
12111 F20101211_AAAAAJ bokhan_d_Page_123.QC.jpg
875a1b853aef29c5022be7b3553d6091
91a68f04b83d5c0603a169aa8b7a415fd6f7fc9d
23955 F20101211_AAABDL bokhan_d_Page_028.QC.jpg
4c93fb65baf762e1ba0ee4b70e34168d
7559cf4acf25e240b8680337326e2fbdd0e46ef8
697 F20101211_AAAAYG bokhan_d_Page_036.txt
e5ebdfca6f65d15eaada95000676c9c3
a8c2b43f0444baa43976e0a4f684b9fc758d7117
17276 F20101211_AAABCY bokhan_d_Page_018.QC.jpg
45c40dca79dad4b27492574e40d4d4e8
c3d5bb3c6273a23fc86ff1daa0b242916c8b30ef
1661 F20101211_AAAAXS bokhan_d_Page_022.txt
2a24e10bd1eaa45d56768ba0e6c3b818
eb299e03670e53d7515640d70a13ba6f85761658
4347 F20101211_AAABEB bokhan_d_Page_036thm.jpg
f179709d7e770b64f1dfca77eda73055
8e14c5103c2e211af561fbc5010654d9d9a7078c
F20101211_AAAAAK bokhan_d_Page_022.tif
e7cfabddca4853a3c6f67482cd10d7f6
efe4a4aa7a1a14a4f8c066e7f3a60a7437b3fe52
5959 F20101211_AAABDM bokhan_d_Page_028thm.jpg
64e9dfb36b613a20baff56d83004325b
0968116170e390b940846579e8a0bedcb184eef7
1965 F20101211_AAAAYH bokhan_d_Page_037.txt
1a1391648305bba7f43c0ca628d6865c
91aed83be3aa6cc120bba153bb6a0acccc2e3eac
21195 F20101211_AAABCZ bokhan_d_Page_019.QC.jpg
9700e12677257e0b150a6f4e719b10f3
7fc0f831f4d4c10eefb0e679d5d05ad50185d410
1925 F20101211_AAAAXT bokhan_d_Page_023.txt
73f825dd9a1ca4551435ec12a6bbc832
cd9d978528edabb9584e743713f93203cf99ee19
3769 F20101211_AAAABA bokhan_d_Page_011.QC.jpg
f60a8b6ae6534c48f78b24dcf76cf0ef
d241dbbd62e28e87ee1c8573dfb6052528ff6a96
5765 F20101211_AAABEC bokhan_d_Page_037thm.jpg
6c2ee7d936e0f4a4ae2792c7639f0b4e
4be9e485d1255e99c21215276b42097d631adff0
4036 F20101211_AAAAAL bokhan_d_Page_124thm.jpg
57a14480bf8f02e44a8979e6414a036e
1301f8a9eae75b99d674a703a55cc1874c60bac8
22128 F20101211_AAABDN bokhan_d_Page_029.QC.jpg
fda68bda6ed6b2fa3e81b6602fdd89f9
6756b4a2a8a8f52efe9e9fa0533a88ef27d3094e
1675 F20101211_AAAAYI bokhan_d_Page_039.txt
3cf6a6a36989308491bccc1031162147
6cfc7e0bacef7113898568fee63e7fd0b7fa607b
1816 F20101211_AAAAXU bokhan_d_Page_024.txt
6f1dfae7820bd26bf81486833a6ff389
ad16a2881869e08056f410cec890700d6edd0d5f
53704 F20101211_AAAABB bokhan_d_Page_127.jpg
903870b98fbdbce5a476c9368b35bb85
3a52565ef5b43c3da05c6b2e27d449e99151abb4
20610 F20101211_AAABED bokhan_d_Page_038.QC.jpg
ed749b94b3a1c7714d850903e3869e1b
0171661fdce1c9a759217dfc39f374c3bd450920
70326 F20101211_AAAAAM bokhan_d_Page_133.jpg
0ef386b984303bdfc531f896d6dd5585
8bd4a0fb006c72b81533697e705c1566f009ef1d
5693 F20101211_AAABDO bokhan_d_Page_029thm.jpg
49fe017ccb8e0ebae5a7e470accaa241
456c854eed0b9e61958e174e9a42df45c0d4cd69
1707 F20101211_AAAAYJ bokhan_d_Page_040.txt
68bbd96766e34d16221cd5e8cf2dd810
be27ab1773dd7e7dad262d616684f6cebc2335ee
1845 F20101211_AAAAXV bokhan_d_Page_025.txt
85c0fb2db876b70e42b2c6d542b2f43f
ac4b92e6a60e513a900ccc6841f181984d4c90cb
6852 F20101211_AAAABC bokhan_d_Page_081thm.jpg
e88d13d72a729b93cf41b377805749aa
f321970129d0796d8754b6df61e6cf101f4c29a0
5583 F20101211_AAABEE bokhan_d_Page_038thm.jpg
84564edbed2295f96aa87d99e1259cfd
bbf47cd760bc2e86d39d642e8cb89d9d2f2ad332
82974 F20101211_AAAAAN bokhan_d_Page_031.jpg
bfa487e6e9de4dcc454bcaf23829b9cf
5c675b2544600d2a27ddf3d8f4b5a933ac332bd5
16631 F20101211_AAABDP bokhan_d_Page_030.QC.jpg
cde001089e9a6019e777153d72c9e0c5
bdc02ea71e73e80fede7241103a4a4aad4ba0a5c
1854 F20101211_AAAAYK bokhan_d_Page_041.txt
f63d022fafb449f6da1dd9c3c883eb3d
de2e596c8a9fb460d12f325ec5d6e01ec7cd0d33
2082 F20101211_AAAAXW bokhan_d_Page_026.txt
a96a306fb7a37618adec31746e612107
71fccfb567bc5deda9920771c520fd50927ae8c7
53332 F20101211_AAAABD bokhan_d_Page_089.pro
79a42dfb8220c18a42c4f354915ac9e4
1c7df673bb85ef1a205d535c0ac87c13f392a103
19003 F20101211_AAABEF bokhan_d_Page_039.QC.jpg
d060f57d27e57984769a0c26f09c1275
bddb90a0aa0e1bea735f077e81f002e9d99da15c
24834 F20101211_AAAAAO bokhan_d_Page_094.QC.jpg
d9d5735cd98543c09df4aa0a0ae6f7b9
d8662d43c89c40a10744b837c4354ebb9e3a0f8f
4404 F20101211_AAABDQ bokhan_d_Page_030thm.jpg
4f4e2b6e1fe6e3f99fc06e7ecb148bb4
ed5123d3823381bd61e9b148bb757884ea996f0e
F20101211_AAAAYL bokhan_d_Page_042.txt
6021304af4653a4a8c474d7d14e199ab
7e967e5f26d98b6445f58bd05e7d10c8cdd80631
2410 F20101211_AAAAXX bokhan_d_Page_027.txt
ad8001330c32cf3dfd96528ac2857fbd
52b2d280091da1ec552f3ecfdb91372456a2cd2c
F20101211_AAAABE bokhan_d_Page_090.tif
801f2a61f451eb49e8858d3d422a7338
5b497c85323f4e756c92913c245924e5e214f2a7
5370 F20101211_AAABEG bokhan_d_Page_039thm.jpg
27887d7303c816beeac00d716f35888b
0417166f2aa73c5f0ad41248ed45d55532ffdc2d
1051925 F20101211_AAAAAP bokhan_d_Page_089.jp2
b37d4fcd1c991e5a831cae0eb806fa0d
92ee3f6eed4ef53542e7a41b9a61f37c45d0247c
25908 F20101211_AAABDR bokhan_d_Page_031.QC.jpg
26809510d775f37823c822d9c1a6cca5
1c5a48681ae8f151702fc4d9612438e2a70ae576
2330 F20101211_AAAAYM bokhan_d_Page_044.txt
00a4885f19df9e2b9af6024f2d4f0994
2ceba71e1fda9a1c005d7c56767fd1f11b599669
2169 F20101211_AAAAXY bokhan_d_Page_028.txt
e15ff9ff4e06924224342cad44d4eac9
1619dff4619476bb3b8d96421b05bbb1be6fd7fb
80887 F20101211_AAAABF bokhan_d_Page_094.jpg
897dc4a33930de40383975678532815d
dff63258a8e211073a96f4cf1a718141476c0463
20367 F20101211_AAABEH bokhan_d_Page_040.QC.jpg
bfa87dcc7a7dcb879b7dc167782e8f9c
0bbc1a2006128ff684758dc32ef5dc94f4db712b
85431 F20101211_AAAAAQ bokhan_d_Page_082.jpg
fcdc07333c73dddd7a70adbcda7676e3
824aabc2058f7ef794daa3f31bdaf10b9139ab59
2043 F20101211_AAAAZA bokhan_d_Page_061.txt
f8104d2d88aca272eb00ebe3b479ada9
5a9584dc14157ab6f0e0c16bc641364703e8b00f
6834 F20101211_AAABDS bokhan_d_Page_031thm.jpg
8d3bd6ef40afe6a3d6e3ab03ceb634b8
2b209f8fb5a6f247fce83cd8c382966f005a6c44
2002 F20101211_AAAAYN bokhan_d_Page_045.txt
d03e40b7d809bdb74354cfc2a29d23dc
706a2eb254b36721ba08338d5e037608698596f5
F20101211_AAAAXZ bokhan_d_Page_029.txt
661402348426622d47595d24a5f3c6af
7794aba6d3715f986c8dddd0750d21e657faff67
21347 F20101211_AAAAAR bokhan_d_Page_052.QC.jpg
606473d2827eabd9391e5f28c72de4e7
8fbbc994ebbeae596ba6bb8a57c582f1762a12e6
F20101211_AAAAZB bokhan_d_Page_062.txt
f2f6e534ab2732d7bd877b2e01b4130c
f465388acc70443c90e3371e924bdb745d379d85
16919 F20101211_AAABDT bokhan_d_Page_032.QC.jpg
27909291b3e84a9da6310df05fc22271
311fcda14ffd3ae69a21021bd9ec772da9891066
2204 F20101211_AAAAYO bokhan_d_Page_047.txt
8bf0ebeae1b30f747cd48675cfbd9e0f
6168f769acab4b56d3a9b879a416e85da52e967c
41391 F20101211_AAAABG bokhan_d_Page_117.jpg
ccd9ba2193b8c3bb6b792a50cc0d332b
c58252ea3219acb7d9ff1e095c474341202aac79
5364 F20101211_AAABEI bokhan_d_Page_040thm.jpg
3316d26193d6716a08c35eda6bac961d
aaad526f70811577d12dc2cae26b38910770db9e
27475 F20101211_AAAAAS bokhan_d_Page_087.QC.jpg
41a01136d2f8864b05ce34e4a7142394
355b64188c341b63846079a0b38656852c55ab85
2396 F20101211_AAAAZC bokhan_d_Page_063.txt
729a76cd31bd66976d8422eced9df678
885b33fc07ab005ec9bfb758853ada438ec9d622
4994 F20101211_AAABDU bokhan_d_Page_032thm.jpg
423f3687fdf4c9c5d2efac8c052b5667
5c0624aef6ebddb80f12d6137c21e6b2fc0b98e0
1838 F20101211_AAAAYP bokhan_d_Page_048.txt
2230d34549645362d56fc270e55cb70a
13a30f5678ecee143c08caa7d9b328028092b13e
2187 F20101211_AAAABH bokhan_d_Page_070.txt
826ceb9e42a0aadb74ce791b7b2cc3d4
fad837e13d23acca4b034b066ea81372b87b91ee
22645 F20101211_AAABEJ bokhan_d_Page_041.QC.jpg
7f34c030a96a55b22c167c117951bfb2
56f4243e217b404dc9d6c6325b39d1cffb73db55
5810 F20101211_AAAAAT bokhan_d_Page_100thm.jpg
b7e03d807ca6de617c2e777923b08faa
7c1e0e0daa22ddee27c858c1f4fa89d020801bc7
2450 F20101211_AAAAZD bokhan_d_Page_064.txt
282e5bbdcddc174173ca2821b5a8e41b
118c77389972960f9539a378f132a77de38534f3
16159 F20101211_AAABDV bokhan_d_Page_033.QC.jpg
9386b93576950059f32d778551968c7f
0009ffaaddc8012074d52cdc500ea01272d7b9c9
1558 F20101211_AAAAYQ bokhan_d_Page_049.txt
4a61b6cf899847ba7daf892edd574961
7abc0c61496f136bc71956d8c7f8d6ffe2b835e3
69554 F20101211_AAAABI bokhan_d_Page_130.jp2
c0ec39adc30c76743ef796bfbfd23e11
b07eb4aca4ad8b616a0a478b8e58f893354faba9
6054 F20101211_AAABEK bokhan_d_Page_041thm.jpg
804360a011c6ca114e1eabe097b87f46
bc723ce503c1a602d47fe0101b8b1cf72a24f5c4
F20101211_AAAAAU bokhan_d_Page_014.tif
55ddfa5ee7e82bc053dee083c215df46
6e59e11dbff5f72cffe4c3a41cd49f25f0e32d4b
2235 F20101211_AAAAZE bokhan_d_Page_065.txt
b3ba765cbcaaec218f2cfde13b0c3149
f5cdd14f451d43fa337279e3cc5ec9d1c984e9ab
4810 F20101211_AAABDW bokhan_d_Page_033thm.jpg
1e2e1ef3d757b907bf8aba81910f37b8
da97a7696681742b21e6f44068aba755717a227c
6260 F20101211_AAABFA bokhan_d_Page_050thm.jpg
85049a71c78eedbd21f42f44f2153567
d9006e7a581c80dac87efd392b699c798ec7d9bf
2182 F20101211_AAAAYR bokhan_d_Page_050.txt
4c4fd8bd56857fd5cbea27addcb7364f
f6faf9fd4d364dc03bd6d67d68d36e55864177f6
F20101211_AAAABJ bokhan_d_Page_126.tif
a8931a38cc6e76cb0ad38f98de918385
e73520b0ec1c0770367de10771b2993689cba14f
23613 F20101211_AAABEL bokhan_d_Page_042.QC.jpg
38c4b3af564bf11681541c3779a07e6e
9137e389c556e5ba55ff200c0a3cf2fca042def4
31316 F20101211_AAAAAV bokhan_d_Page_116.pro
ae56fc1e16b0f218e66fb311fa3e23a4
0fa2029498c0313c4a5c93090bce5ae141cc16bb
2510 F20101211_AAAAZF bokhan_d_Page_066.txt
1155e5669f92f360a8d2b853679da6e1
3faa430fb4dc3c875e408e2c28eef60ea64a17b1
19175 F20101211_AAABDX bokhan_d_Page_034.QC.jpg
9a64ca26bae6ff2191ebd08ebfcf5d0b
25f2f57abadb03025780907a7143687f78f7161d
18688 F20101211_AAABFB bokhan_d_Page_051.QC.jpg
746e2ed873af23164bbfefac50d86e9f
653fea7445410d7118aafdebc829d5e8156884bf
2050 F20101211_AAAAYS bokhan_d_Page_051.txt
9e48466782e85f52e71ada2229f72916
b595e48b2a15fafed0ef36a923d1e8125d223c36
41734 F20101211_AAAABK bokhan_d_Page_052.pro
750b5da2dbe8bb534c01c2ebbc8892f2
d2efeb7f8b8121b0d4514e93211927bb053bf98c
6185 F20101211_AAABEM bokhan_d_Page_042thm.jpg
fb77cb9e422fa03b9152a2f5ae59927c
1091cde6be687109e03e4232c82d50e54f68915f
47249 F20101211_AAAAAW bokhan_d_Page_135.jpg
cd5562537b2e1f88fe75c28fcdd7a2b9
5806776d738f20ae75abf353f4abe201f9deb92b
614 F20101211_AAAAZG bokhan_d_Page_068.txt
ff0496b8e4e1d985eeabfdcbe4d3b3d0
0fb35edf73f45e13b7e4f0bdc5456a27f8c5267e
5470 F20101211_AAABDY bokhan_d_Page_034thm.jpg
e2a0d8d86f32c131558cff59e97c4298
8e0097396aaac3904c74b5face1fb360e11f528e
5487 F20101211_AAABFC bokhan_d_Page_051thm.jpg
ea1551712e5a20154a7a48cd69e0689c
d10f0343cbc999d87a76786ab97ca5d95d9f0338
1923 F20101211_AAAAYT bokhan_d_Page_052.txt
063ed49a525ea412c278097f57bb27ad
23cf2387504204fdf9bfb07e42829c6bb4398cf4
18626 F20101211_AAAACA bokhan_d_Page_059.QC.jpg
0ac8b5dc1ff2e1c40e39b0cacd06a1a8
03702dca89d0266bbe88bef2acc2d30796beb79e
F20101211_AAAABL bokhan_d_Page_082.tif
4fc7c0827540195de16a48a098109b55
f7239def1e0018f6a69f5441838e4163505e1c19
18847 F20101211_AAABEN bokhan_d_Page_043.QC.jpg
b99d282e2d2279e1e6bff269c8388f77
eb6282e7fa520de8858dbc94d478165eacf1558c
1276 F20101211_AAAAAX bokhan_d_Page_120.txt
9367d90e8e408fa8830eb1c6cdb357e2
5f12efd45af34d796fba4e838d4a48447dfd82ef
1873 F20101211_AAAAZH bokhan_d_Page_069.txt
e8e94445c76539aa0a5e42d84e5d38b0
169174f406deb45bff431f90c9029bcaf1e61599
16780 F20101211_AAABDZ bokhan_d_Page_035.QC.jpg
8247c6298f564b0cdc6c4d36a4dfc4bd
02dd5181efb1f47773dbc27579ba056a75cf6d4d
5833 F20101211_AAABFD bokhan_d_Page_052thm.jpg
b6be3739aee336089fce3508b9a71c46
bb63debb69f0d24d264e089b4caf78239c4e78c7
5252 F20101211_AAABEO bokhan_d_Page_043thm.jpg
0ec5b557593e84e6d348df285d6de095
10d92132b9817a928f8ff7a3987edab65e927595
2075 F20101211_AAAAYU bokhan_d_Page_053.txt
866bb79e7087c0fb10a6acfc29f23b30
b88567d6fa9aee9b3696ca0e7346a1dac0b8900a
1015917 F20101211_AAAACB bokhan_d_Page_020.jp2
3f3ce746810135c75c2b9e80659a5e12
3a56e9b845419c1b5653d81282b3973059228914
55322 F20101211_AAAABM bokhan_d_Page_123.jp2
cf04f3882652efae83ad545cd76d9ec0
806c30edf4c109ebc82b1221c9941ac58239b8b0
2233 F20101211_AAAAAY bokhan_d_Page_067.txt
8f84a6c8adff77b02c26c67e10079d05
0fc87065908a03cb281e6645ca0704f88c6e7ff6
2336 F20101211_AAAAZI bokhan_d_Page_071.txt
d25a65a7e1908ad8be1d013970751180
df9ab73da5f4177a638f828c017225d0ff7ab26e
20591 F20101211_AAABFE bokhan_d_Page_053.QC.jpg
19170a29473b26bc10468b5d5e0bcd0c
e6b6d5d5b4335c30fd662a8dc46cf945fd385efd
16707 F20101211_AAABEP bokhan_d_Page_044.QC.jpg
e3169205264d35c86e4dba3949626ae3
f46732518645e5775ce1529809b4bdba5537ec23
1949 F20101211_AAAAYV bokhan_d_Page_054.txt
b0113771f7bd848adf1f96a12c7b1232
c80624e4861ac27b3cbaa9346d50f68477f49f72
52202 F20101211_AAAACC bokhan_d_Page_083.jpg
847527076b36e2156eaa23d92690ea70
eb909b335034b5f1b95fca5db2f98426bd91205b
1029956 F20101211_AAAABN bokhan_d_Page_013.jp2
0d8648fda581d00b48ff9c6e9f591118
a16f929c754bf60bebf034f432f3b54c637c022b
1051913 F20101211_AAAAAZ bokhan_d_Page_108.jp2
8bb61a411d728fd92e817c25546ea9df
5e607931661a6a94afea5d9cc752e2030fdc1f70
2576 F20101211_AAAAZJ bokhan_d_Page_072.txt
7cfcb9bbdcdacda9d705104d4df2acdf
951b860af81898eef2034db7b883f0bf5ff2f78a
5454 F20101211_AAABFF bokhan_d_Page_053thm.jpg
d164c88be54e49ede8d9792f0dff9fc5
a490f4d493bc8a22f9a43c2b234ef8c326605e35
19303 F20101211_AAABEQ bokhan_d_Page_045.QC.jpg
c9a05510fd20cc588d810aea080f7955
d37a946e73bb7df2c076a97d09255c772d3fe072
1451 F20101211_AAAAYW bokhan_d_Page_056.txt
64b823b355130022987784d30b7b93ca
4c6f814907fa80b186e023746b70f0b198302568
73460 F20101211_AAAACD bokhan_d_Page_013.jpg
4aad98ebce67d747fa22403a1b06e3cf
1dc7409237d5fd7e50953fb0edc24ad185b4852a
38291 F20101211_AAAABO bokhan_d_Page_007.pro
f48e1da3c471f304c81e7430e159bb23
ae7ed80b969140954533de68ee21ca78c7d60469
F20101211_AAAAZK bokhan_d_Page_073.txt
e2aaaccd5bb54c2d5bacd28e3063547d
2143d61f622830333c9535ef870f5f40613f8c22
21551 F20101211_AAABFG bokhan_d_Page_054.QC.jpg
7325b18bee7f0c22bbc100d1a7025bf0
dd89db971399c5e9d58e4e74a1aeb5d81c7b23e1
18331 F20101211_AAABER bokhan_d_Page_046.QC.jpg
3decc1b2d077cada1af1fa4598146ee1
981778e84b7536fc8d4d848a9d980bf3bf0023b1
948 F20101211_AAAAYX bokhan_d_Page_058.txt
741f7405fc07facf4631245fdbdf0ae0
cd4bc93b798675b9eaf08c51b5e476c7f31353cf
25681 F20101211_AAAACE bokhan_d_Page_120.pro
a16364e76ec743ade4587044269e2c33
2dc0a2cb4a277703c13b5af42396a203680374cf
2461 F20101211_AAAABP bokhan_d_Page_078thm.jpg
aac0cdfdf28f08b7f34a6c92139a4db1
bd65fed8e312648284e806c4b22e72b9bb51f209
2310 F20101211_AAAAZL bokhan_d_Page_074.txt
48bdfb15f21e357dbe7c90703b958116
fcad2404cf7d1ac09e68204b97058d4ba9ff16b3
19229 F20101211_AAABFH bokhan_d_Page_055.QC.jpg
4a3e15e6802d8ddc6c14d4a7a8a07ea6
4fffd4812cd63890034ea1dbbf61ab26b31e7030
4768 F20101211_AAABES bokhan_d_Page_046thm.jpg
ebce86018d94594b2d100f16143d5aea
f1a458b0b39a2572d1cf8f0976614db5581f8350
1547 F20101211_AAAAYY bokhan_d_Page_059.txt
260947462eb9bce4f6b248a528ad0425
82a251620652cb7624cd04906d1835b0a52b4c2d
53730 F20101211_AAAACF bokhan_d_Page_098.jpg
170c1311e16ba865cfa4a1624d9c2c6a
e153814a26bbe835db1b7355ec843b0b03de4a9a
65213 F20101211_AAAABQ bokhan_d_Page_038.jpg
9bce02c2c569b9288f32a4cd4c1ac2af
7c437111939f63ac87fcdaddfa1bd93c3480b0c2
1977 F20101211_AAAAZM bokhan_d_Page_075.txt
1409b65aa60a5fe4d08a91a85f97f034
15edf7dd6220f4af16385e4747b712fae2c5f36c
5334 F20101211_AAABFI bokhan_d_Page_055thm.jpg
8d5a451953738dc8423f42ddaf52a49e
864e3f8adf88119ada4a0b89b13a6beb7970d353
26711 F20101211_AAABET bokhan_d_Page_047.QC.jpg
8f0436ec0190dc8e142b526c5366e694
865d254c2342c5f4db948706a56edcfa8ac3f41e
1563 F20101211_AAAAYZ bokhan_d_Page_060.txt
4bafc5eb7be14a37a8d19e7e8fab34fb
9629ecd85cae0dafc83096518d5c267d09728e05
F20101211_AAAACG bokhan_d_Page_091.tif
bb505dd7710e862fe9e74bac52f5dfde
b544187e47c33c5d418a8701924b842b96b7d830
91284 F20101211_AAAABR bokhan_d_Page_005.jpg
260e21485b7abb75f81deb4d48e146c0
dc0e8eda18a32a204f1681687a1644b017b9888e
1175 F20101211_AAAAZN bokhan_d_Page_076.txt
46f4340460c54dc4d0219d6069d37446
058a794956eabf7d565ce501508587cc67ccb9d1
6937 F20101211_AAABEU bokhan_d_Page_047thm.jpg
19d37eacd9795b10dfed37ccefa5cb15
e10424cbdd22c71181e6a976324feb76dc18eb57
F20101211_AAAABS bokhan_d_Page_054.tif
1feb66321af2fa359d4dfc80c0a7a6df
8cc629ad46455618b8a663abead9082dce18eae5
641 F20101211_AAAAZO bokhan_d_Page_077.txt
ff2ecd1cfcd98b4c09421caf25b2c962
ac1cf8e163ecfaaeede5ce465068912e5ee37af7
23385 F20101211_AAABFJ bokhan_d_Page_056.QC.jpg
0bc8edc8ad32b2fe1164d0c66a0523c1
81a06a9381f21b608893dcf1d944c12cb0fbd331
21768 F20101211_AAABEV bokhan_d_Page_048.QC.jpg
59b6163c86ee1bf6681e52cc24270d35
5b2d6ea18d9ebc43f1d7a286d6645a93f10d414d
F20101211_AAAACH bokhan_d_Page_036.tif
a7d2624351cf8cb86c8ffc1009b9efa4
7d97e01173c565e3a8d7b2e5b8c75eeefc2e9a1b
1915 F20101211_AAAABT bokhan_d_Page_018.txt
687702560b9118d64d067933af0d3b0e
d9d5f68e2eb065c903e296e0f93144a82d0ce8a6
381 F20101211_AAAAZP bokhan_d_Page_078.txt
fa05228973af334112066d5391147959
5d2a47c0de24fa5b250a597eca758d8201015570
6186 F20101211_AAABFK bokhan_d_Page_056thm.jpg
fecfc94fa4fd13e956a3a756343d0a59
e5ccf83720c33e5b7d546051f1ac2a45f1608ab1
5918 F20101211_AAABEW bokhan_d_Page_048thm.jpg
289cf676da7c237519587217f87ffaa3
d9c1ea26c473ad4e2f995959a403f97b91910e0b
21769 F20101211_AAAACI bokhan_d_Page_003.jp2
e35d2e4e69c92c9a31925b700c682f84
25e60015d4d000f160d8e185e73ff9aeacd43a28
F20101211_AAAABU bokhan_d_Page_041.tif
010b15a536591f4cf145b1144bdfc28f
7926124e071b44a09928f300860dba847e628ec6
364 F20101211_AAAAZQ bokhan_d_Page_079.txt
be76d406d063e61422cd5074528790b1
73309b642a9c028b2e1a8c273cda80dced1d0ae8
6751 F20101211_AAABGA bokhan_d_Page_065thm.jpg
596234f4a1712ab52b3ec880eba3afcb
1a2230d0af7b538dbacb3deb235d3511e4dc23cf
20471 F20101211_AAABFL bokhan_d_Page_057.QC.jpg
18262599caaa454792227a74f8a6c305
c91236c19fe1a4ad4bd72c467e495295d0c9cecc
17189 F20101211_AAABEX bokhan_d_Page_049.QC.jpg
a19748d5cc48a29918f54ab610b455e6
b51dc72d615cfe1cb9da1e4497408a522d558cee
F20101211_AAAACJ bokhan_d_Page_020.tif
d59614ddd3f79db71d3e25b6ddb3cfe1
0e1553dab655301107473b97dd2b2c7ab5a721be
704753 F20101211_AAAABV bokhan_d_Page_044.jp2
da6febc06899cada4608a60eb30881e1
3f40f7da3e5a08287a109e050f807ba9d19192de
2189 F20101211_AAAAZR bokhan_d_Page_080.txt
02d98f30a948a81b2124506097ebf909
94200a8212cb1dc68b50f0904e9bcb1fce083a3c
22366 F20101211_AAABGB bokhan_d_Page_066.QC.jpg
772f8ff0979ada0f953f60835d03f6c7
6b065304adecb944135bf8b4b1115a7058e7af24
5490 F20101211_AAABFM bokhan_d_Page_057thm.jpg
c87233db6e32ebcb805e8d95d8bd4330
5556bdce902fc561e0d5d656de5d784125595ea2
5154 F20101211_AAABEY bokhan_d_Page_049thm.jpg
9caae30855de1fb4cffa921668bb4a4c
bf90d99cbf5b3275556bcbb7224bae21b1975355
5878 F20101211_AAAACK bokhan_d_Page_054thm.jpg
144eb81fd328c6a420f761f0f9ca689f
3c7e67ffc98d25f686e344a42e6a44407977429a
5864 F20101211_AAAABW bokhan_d_Page_020thm.jpg
913101ecc42c464e478f068bb02223df
76851129963a22f6e1caf65d62e4944292483cb9
2331 F20101211_AAAAZS bokhan_d_Page_081.txt
9921a7f79d05d018e11fa9de6ab181e1
b1e8bcf88ec76f4f232af8f282c53caa1564427e
5913 F20101211_AAABGC bokhan_d_Page_066thm.jpg
1cc3bb3ff47fbc4f7c7f033dabbe55cd
7e3ebb3dcabdb3e9bd6afb53decf2cbcace2a274
13582 F20101211_AAABFN bokhan_d_Page_058.QC.jpg
482361d680792af8c019fc21b7f431de
3e664bdf1bd56175dc7236794d0d95ee6e54b173
24789 F20101211_AAABEZ bokhan_d_Page_050.QC.jpg
3690638afb4e176cd8fa011d11fc22fa
0d2e7fc1ae65d109ab8946c2f97096e60387bc35
14343 F20101211_AAAACL bokhan_d_Page_109.QC.jpg
8ab50c64bf033fb20eb397c029951a78
7b82a80f96227569ae81efd89bd450c377fa696a
12800 F20101211_AAAABX bokhan_d_Page_116.QC.jpg
1cd5f51a2bb7e71fb95dbed3a716013f
346c440fbeff2231e05a5fd335649faa89f9e29d
2241 F20101211_AAAAZT bokhan_d_Page_082.txt
2dccdf13033695eee39c748e1dcc4f43
b6c74a4e80d8f5f18e128ea48495076c34bc3a76
2126 F20101211_AAAADA bokhan_d_Page_021.txt
8523333fccb144176cfa9ed0c44e9185
59b26eb2ad23450093b26bf0bf9b88533d491fbc
26664 F20101211_AAABGD bokhan_d_Page_067.QC.jpg
c5330e891a694d7b80061bb0a7f5c743
22f8544411998c164d20cb09e0e786633d79d05f
4010 F20101211_AAABFO bokhan_d_Page_058thm.jpg
5cc616f8c9d3abb4236c3d39fbd2dc94
909d691122813a8364546c3036200955f0e65350
72579 F20101211_AAAACM bokhan_d_Page_057.jpg
43b7154bb0b3c5e46bdcaf977da8ab86
1e6118fe138eae777fbd5efbf684ceb457c99c76
18571 F20101211_AAAABY bokhan_d_Page_105.QC.jpg
ebfb9e072eb36a1892fef659f6b1ff0f
be1ff88738cea37e98c94f3450dc0ad992ce2a59
1887 F20101211_AAAAZU bokhan_d_Page_083.txt
70db61032a3ece3f16cf9d86c4a7e7f8
7b9f1635fc5e2ed865d93a8650e87e4f000fd743
55110 F20101211_AAAADB bokhan_d_Page_007.jpg
6e6e53d090b07f144942ce4371e8baa2
14bdf0c47fe573a385e87a9aaf981932903f6778
6478 F20101211_AAABGE bokhan_d_Page_067thm.jpg
5b98ce36a04af9a6da4cdecebcd46af6
15769a12f44c6f6d75597cfdcb2a97eca03914dc
17305 F20101211_AAABFP bokhan_d_Page_060.QC.jpg
9d7c42b26b5ac48d8d3ea075d3fed089
ab181a203042f9a4cad2385131d2d361c25d01d3
1447 F20101211_AAAACN bokhan_d_Page_046.txt
737f5c4aa623851781cb7c569fe2675f
83857def3b1a7f6a71d4f7f5205f53853d6460bf
77157 F20101211_AAAABZ bokhan_d_Page_026.jpg
52bcf61a522b97ecd4fa39dd3cc1e078
fdc37c90e016363e9a2d052d4d3fb8c9f5e61579
2165 F20101211_AAAAZV bokhan_d_Page_084.txt
084a5177c13f5d8421542da2fda2d1e1
cc43a610adbf0bb486151e8e364acd866083cf5b
1032915 F20101211_AAAADC bokhan_d_Page_064.jp2
4592ff274af7272d3bc4351e15c42af5
70e47b219f539a0774f7c8cd6d40e1c3e8fcc87b
9323 F20101211_AAABGF bokhan_d_Page_068.QC.jpg
682767ba5ceba9aee87250c60d05d606
2215aedf1799af58dc1cb86661c849f1615518a8
4964 F20101211_AAABFQ bokhan_d_Page_060thm.jpg
623c10db2eaefe85698eb920da63d5d2
b2903662a0dea16a17514808659222fa59355c1a
F20101211_AAAACO bokhan_d_Page_010.tif
4c77f28bd38fc699ec4ec807b160eb5e
6e430e2dc1ad64fb7442ffac586b661a0f0e7f9a
1586 F20101211_AAAAZW bokhan_d_Page_085.txt
d5ab666b05c059aa3095f5bfd1b33173
ce40c128685ea98a670494058cfa40b91985ab90
1554 F20101211_AAAADD bokhan_d_Page_112.txt
71b220f4e97f78e1c8eb1bca114eb06d
9ab49c0cfb4d98eb731a728c695254f58a315df5
2722 F20101211_AAABGG bokhan_d_Page_068thm.jpg
5498776c943c69ceefc33657af31fb1f
72d0a0b4d86357f7750f4ec3c2ddde071ead7365
24453 F20101211_AAABFR bokhan_d_Page_061.QC.jpg
b5005fc191158a7bdaf2c9b83eec077e
fa9af432a156247d8b8f176f5eef897c80290a51
F20101211_AAAACP bokhan_d_Page_078.tif
325c284fec5051b25ba5d2a1db443d9e
44590e56e76ab33552be214adb20ac3ea8fef003
1296 F20101211_AAAAZX bokhan_d_Page_086.txt
c5d4f427ac8e83197675acdb594130f4
e8e94eda49fc44dc1101f77f9aa0f64fabc3e21d
1071 F20101211_AAAADE bokhan_d_Page_057.txt
c0ac4ad0a1f1fce34f29cac9ea404095
97961d104ef007e11fe45ea5b5952d2e479ad7de
18510 F20101211_AAABGH bokhan_d_Page_069.QC.jpg
1255fcb5452dbac990931b244ae2fff9
90b4a723aa5fbc376f11997b03d1dd3e560b032c
6499 F20101211_AAABFS bokhan_d_Page_061thm.jpg
83b8beb1ae223504c6b1fa09ea6e20b2
83044e391eb50b448c808f576cbb5c93c991550e
57898 F20101211_AAAACQ bokhan_d_Page_051.jpg
00920bdc179ff9fb84bf9c0f28bb3775
a1c3ca3469ba62733d48901b6303c9783d3dbf33
F20101211_AAAAZY bokhan_d_Page_087.txt
8a2f8d58d244d8d549cb6482b6773370
75588fd6f9d173ca2eb1533acbad1564cf18c811
5621 F20101211_AAAADF bokhan_d_Page_021thm.jpg
23b5fa894d32d2cd655a348c6c998109
2f14d090780c698c01eb0de1d9d25c7a864a5919
5579 F20101211_AAABGI bokhan_d_Page_069thm.jpg
2cb8083777557a6f91d62f8b0c254857
30473208db97970835d139631907f5fdea2b1c2b
23445 F20101211_AAABFT bokhan_d_Page_062.QC.jpg
8b42e06df1c1067caaa9e55075a5b482
d952f07030ff244d61fb782bbf23858ac7de61c6
3253 F20101211_AAAAZZ bokhan_d_Page_088.txt
4ea2e4bef2110228469fb83183d951b8
1dae94d62445122c2fb336be729592f3a83c97d1
21054 F20101211_AAAADG bokhan_d_Page_078.jp2
15ebb41775d63508ddf49ccbc97fc59d
d42b503c67621328b94ba748197e557166fccb78
5280 F20101211_AAAACR bokhan_d_Page_024thm.jpg
90898e5571c70062a9dfc33afc7694b1
e16816eade108a7d34d80c859072492f26de0ebb
21441 F20101211_AAABGJ bokhan_d_Page_070.QC.jpg
09e065a89893804c8805e9fe2cec263a
86e23e07ada83e88ee2f525d3aec9a2ac0f941d4
6418 F20101211_AAABFU bokhan_d_Page_062thm.jpg
ac0963eedde90530ed99bec82cb7a64f
dc573aee3b8340e4f8779216226c0a2f9e02c45e
14002 F20101211_AAAADH bokhan_d_Page_036.QC.jpg
44a4eae3776da0527796700716641fcc
d47ed936e146483d8df28678119628f75e3c5074
F20101211_AAAACS bokhan_d_Page_129.tif
9f69bef8473cac441b42456960e2357a
d1fc2d8e9740873697741c65c92f267ce213b9bb
28499 F20101211_AAABFV bokhan_d_Page_063.QC.jpg
83ea1e3b061a7218534322885ab70079
2377ddff061b99d2b3abe5b115654f401501b9d1
693438 F20101211_AAAACT bokhan_d_Page_127.jp2
175ac554f8c15ad7d23f5a784519a683
6108f0eea44bbfc2f05b649f98cbadbd7d7ddc6f
5705 F20101211_AAABGK bokhan_d_Page_070thm.jpg
c797fc091533afdd2efef245ecb6f6a1
2a310f010e0531ea39b97bc4c9a0868565cebcfa
6965 F20101211_AAABFW bokhan_d_Page_063thm.jpg
45f30c7f88f24b11da72b79d2c1b97d4
8a687bff4c4f2cc5f8a2b2d346288493c273849d
38155 F20101211_AAAADI bokhan_d_Page_096.jpg
8902b4630559a5a14eb9a29f6ebb9db3
19f43f9ff9fbcc675464b47c108c9a1097722b51
21451 F20101211_AAAACU bokhan_d_Page_037.QC.jpg
ee13b574814bf057e055fe06e1a5db87
11ecf18869429408efe9e1eacc94fef9da2fdd5d
2526 F20101211_AAABHA bokhan_d_Page_079thm.jpg
ef7840f29f4fdfbb7e4cc0f05f3d624e
f9938921cb394b922093fc820b8181ec87e30de1
23993 F20101211_AAABGL bokhan_d_Page_071.QC.jpg
8900e2c18ccfca1dd82424b5d89f7cd8
b8e07d659916265b4c2ebf19285198dd0b96acd7
22091 F20101211_AAABFX bokhan_d_Page_064.QC.jpg
e20ed9e3fb038caf47c967a9f61c203b
64170ae286981775dc174eb53e8924df46951904
78362 F20101211_AAAADJ bokhan_d_Page_093.jpg
f6cf1dd11006f310a7f162b66154ee71
f040a730db903cc5a966335a6fcb2621d9f9048c
18108 F20101211_AAAACV bokhan_d_Page_024.QC.jpg
4624f2ffad8f6d0b6fe04e5b91a177c7
722b7d897e7e1b76655fa87a5159ad43df5d4874
25415 F20101211_AAABHB bokhan_d_Page_080.QC.jpg
d348cadc381f5e3ff8cfea06220c8b16
9587cc09a457af2397127e0a023253259ff2be5c
6326 F20101211_AAABGM bokhan_d_Page_071thm.jpg
e211811445c7031bd1399a76c3fbec24
69b6ec165d3058e31db804b1b0ff24319f2a5bb8
6046 F20101211_AAABFY bokhan_d_Page_064thm.jpg
552718aa9fc60f17c049fbb3a7b63580
36642e818c548753a698da771524e0a1eafbbf06
1829 F20101211_AAAADK bokhan_d_Page_043.txt
b4988b58bda3bbcebbca001437ea4420
0b64e310a29967ad640a5be20c8f56f2d7a88539
33015 F20101211_AAAACW bokhan_d_Page_098.pro
3c8d8746db92f68f9c54ac8c54e9b514
7669ecd91ce90be73ac9fd0a0a6e07bcbd66a08d
6617 F20101211_AAABHC bokhan_d_Page_080thm.jpg
4dfc772528c1b3a332d9ffa38c52a38e
581efada6fb508ee1f422e8926e97b7425f47e1e
28826 F20101211_AAABGN bokhan_d_Page_072.QC.jpg
6d397e2846f234e547a8a36b89f00efa
c9c3c4f756b71fb44a1b8d0fd6a03d286e809133
26973 F20101211_AAABFZ bokhan_d_Page_065.QC.jpg
a404c9cfafb86aa139aee2d6cb77d2da
c5c91b9698bdd93f7415f0ddb399028abedb63e3
104995 F20101211_AAAAEA bokhan_d_Page_131.jp2
a10fe25133215b471ce62500dbc1ae43
711c55975b4473128d26a9a11f1003e61a23b7fd
F20101211_AAAADL bokhan_d_Page_004.tif
b436bdf70929de59ff0f414703ee246b
1b8f34e80881d3cf38377a7b2efd46c31781e633
954 F20101211_AAAACX bokhan_d_Page_134.txt
1a6c72797365d6c84e5a5ab00195fa59
cca15ba4136cc3abf50e807d4ec3dfb513292fd0
27760 F20101211_AAABHD bokhan_d_Page_081.QC.jpg
bdffbc653e7e0c6136ed8bad04de60c2
b90b2d675714137d2b9c0d60ccc66a1f86c52fe1
6992 F20101211_AAABGO bokhan_d_Page_072thm.jpg
bf8153c4ebda44d67282fc96fd639aa7
3982af4d3ef86565f72ac1f4c667d4d872104f33
12374 F20101211_AAAAEB bokhan_d_Page_124.QC.jpg
5a934241adc848b314e586a90f708266
3159eb8413d8b5a0d47946f6680cc546c6abb5ae
F20101211_AAAADM bokhan_d_Page_128.tif
156e69dbae60adf24fda3879ace8bbf8
a2e3dcb943eb0022816dc382f98b2e7219668a31
782798 F20101211_AAAACY bokhan_d_Page_104.jp2
9f76387973f48ed1f068442432303596
a6ba3481ab5eec7820a465d5fcb644529238a46e
6699 F20101211_AAABHE bokhan_d_Page_082thm.jpg
874c5ea9283d09e1de8b5857a8bf550a
3382b4c71952480ca9eaf9003ca36c4a2e103dcb
20239 F20101211_AAABGP bokhan_d_Page_073.QC.jpg
47162e5d80fa9428cfe3d6da69be8405
57e05438cc440a94af50d8566a6e107a93781232
5006 F20101211_AAAAEC bokhan_d_Page_018thm.jpg
f0a0ad5ab9956a6c7b37ce9d9255f993
8d33db83ef2393992aaa17eeac770986876749b0
11512 F20101211_AAAADN bokhan_d_Page_121.QC.jpg
674f9e666495057310283658a13583e8
e76442334ef45739349cefdf717a05c42d3781a1
1051983 F20101211_AAAACZ bokhan_d_Page_087.jp2
e5d10a7a3cd0bb15a54d4b11fdfbfe18
2514a9eb21407b0d4c6091a0da46bc85ab79c9aa
16390 F20101211_AAABHF bokhan_d_Page_083.QC.jpg
9ae06b0dc4b6db45236dfd83648bb5ba
3225f4ce7d4f63560e9d8ae313102d10d0d73907
5509 F20101211_AAABGQ bokhan_d_Page_073thm.jpg
72cf68877044d9e3d85e458a995ef1ef
ef04e6510d50f12938be93d58c72a7b3e2c9c407
88435 F20101211_AAAAED bokhan_d_Page_095.jpg
aa7deca01bd447b43f59d43a66025c12
9698f487b738c859350e7f7377e0352a895e0923
5201 F20101211_AAAADO bokhan_d_Page_059thm.jpg
52920f817c0061ec6d91ef3610526f09
6fc0af03cf7d6dbec25bf3e0103c1535b89b5bd4
4918 F20101211_AAABHG bokhan_d_Page_083thm.jpg
71aec088d7cce3608d49021b05744290
3e41f2757e95a4d6eacb51b1527f8a169165c691
24523 F20101211_AAABGR bokhan_d_Page_074.QC.jpg
f14b5ddd9727dfdb26a0dc966a8b6e00
1b56f1d4418318b680c2f0856eaebd7112602bc1
F20101211_AAAAEE bokhan_d_Page_038.txt
3606ead5e7b3705ef25869e25c0d6a2a
083a6db122141901499f72c3de77ce602b00722d
6266 F20101211_AAAADP bokhan_d_Page_013thm.jpg
47a65138c49dcacf628e6984aa2a24e3
30fad8516468385c556a7df76f95396d0533c1f7
22334 F20101211_AAABHH bokhan_d_Page_084.QC.jpg
4c7cd1aaa1989246069c7a5ffa5e86a2
20d61a8c0e08bbc2be5a7f3022f7611f95fb336b
6502 F20101211_AAABGS bokhan_d_Page_074thm.jpg
3f7383c3cf8873e769ce48a7997d45d6
46f6f9e82b1fff2a12fc91ddf54d9be1adead4d9
1385 F20101211_AAAAEF bokhan_d_Page_055.txt
d338068ce856922fe1a043e85efe08c3
1e8f9488338d5547ac7c186f2d7900f4dfbadfba
44708 F20101211_AAAADQ bokhan_d_Page_101.pro
47b7acf10961db5dcd4905f9b4107b5b
037e29cc558df8c88f3aeadd0c35639b3d86c8df
5952 F20101211_AAABHI bokhan_d_Page_084thm.jpg
eb3b525645ada9cdc17957db1290a7ae
75341b0b068acad9066b80b9ef2ada0a98ee3014
24004 F20101211_AAABGT bokhan_d_Page_075.QC.jpg
1b688ecffe42ebd9c3340e6f48ec2877
36e80e8a29e49e537a56f31d3d2130105e6143e2
54809 F20101211_AAAAEG bokhan_d_Page_074.pro
af0e6a9df0445ad0992445901386127d
e76bd9ff7ab4eb9ff180f76e20504107c70070ea
86047 F20101211_AAAADR bokhan_d_Page_089.jpg
6279b39e9902dbbaef1957bf63391f57
5c97a8397ddce0b4a1e275097c0cb685325c4f0e
4661 F20101211_AAABHJ bokhan_d_Page_085thm.jpg
058a6e661d6784aa12ec026c4a9845f9
2105a3484eab01a92f948d158c5094bea46e6195
6191 F20101211_AAABGU bokhan_d_Page_075thm.jpg
9448922227d206b6446b3990474a1ec4
80532cdb748981b7d3d15ad57d1cee7bf20a5175
15294 F20101211_AAAAEH bokhan_d_Page_085.QC.jpg
1f31ea6acb2f5d00fcec7f978f974b48
1862bb3ea6ed4f1599e7bb689b9a5d4c8bc326ab
52392 F20101211_AAAADS bokhan_d_Page_119.jp2
9ae9cdb3f917e6caf2b66542d60abaf2
695c87c418a1ef8ff3ddd94bce23ceba79697078
15981 F20101211_AAABHK bokhan_d_Page_086.QC.jpg
177989da28a65216d5a6e43a67b17a87
cee37970a74cfe49aea33881eab6870c2ff3cee4
3650 F20101211_AAABGV bokhan_d_Page_076thm.jpg
138f68b25eb9e41d263f0fedfd22b28b
5eac98b01419c890f3a9a1c539b98b074c8f18ed
5658 F20101211_AAAAEI bokhan_d_Page_019thm.jpg
18b11fefd9d14493ce6c4023297c9e47
b6792c94c8ec8cc95a7a01a15a9d453c75c477f6
F20101211_AAAADT bokhan_d_Page_032.tif
3bfc875dfb903ab536c27b127678bf7c
763f6ad6d8363ef8431de02f81ebedd37f60920f
6428 F20101211_AAABGW bokhan_d_Page_077.QC.jpg
4d4e66f4e171827c4f6a7b818f82bd6d
94093c8637950c143050a758f30dbdba36624fa7
F20101211_AAAADU bokhan_d_Page_050.jp2
f46030d6b4dd05223c11b19f2078c845
96eb131a2d30e045753e5abc0a6d5aa14b974751
27965 F20101211_AAABIA bokhan_d_Page_095.QC.jpg
2b5b76cf013d83c480e56ccce82caa9e
0df8e30186ebd2698fee638eaa8f3013ddfbb71d
4864 F20101211_AAABHL bokhan_d_Page_086thm.jpg
3ef7163cdadfee64dd1014358415fec5
27eb2fb36f321390202c8a8dd104698985ec57cb
2339 F20101211_AAABGX bokhan_d_Page_077thm.jpg
01f7a37e1201a9f2c6771c00372ba4cc
454022eb3e2e828be60df16df808527e4a462ad0
6503 F20101211_AAAAEJ bokhan_d_Page_026thm.jpg
b4fcd93c0a47406f155a8cbd93a258dc
7f5efdddb5b4c3cfedf95ca6d71ae218754bf3bb
F20101211_AAAADV bokhan_d_Page_045thm.jpg
e52b94641893e8a39d99fe9af2047a75
3791b061a160668f064d70ccdc20a6c1edfd0fa7
7049 F20101211_AAABIB bokhan_d_Page_095thm.jpg
f1789349b16c5d1c89b7adb920958672
904eb7dc80ee32743b3ebf6b5b649f97b2fad3ce
6916 F20101211_AAABHM bokhan_d_Page_087thm.jpg
3c8341d4be142c1188ece34f4b96292d
ff4ae49044f5e7c0dd46041f338c43519025c919
6756 F20101211_AAABGY bokhan_d_Page_078.QC.jpg
520db4c782fa4337ec185b194162f579
20b77be3b48bccd71abd94a3e9c8b29a32d705af
61281 F20101211_AAAAEK bokhan_d_Page_110.jp2
4dbdc50fef2c59fc2ed02dc75780bc41
2e430926df78ead64099c7557913679fdd46882a
40350 F20101211_AAAADW bokhan_d_Page_038.pro
2303517afeb36433c98e52bf6dd4328c
14202566f4ca43889c8d90863224cada7c0719d8
11536 F20101211_AAABIC bokhan_d_Page_096.QC.jpg
98616edd150b495e37f23da62f90da91
4db6d9a5d711250dfbbaa4df6b1b050226d9d766
21253 F20101211_AAABHN bokhan_d_Page_088.QC.jpg
5a9158517eb29107ebd1ee7e3efb5b67
49aeacc8594838a319c693bac80add30cc3ecfa2
6658 F20101211_AAABGZ bokhan_d_Page_079.QC.jpg
59fb37a5d38c3416372ebdb4cfdc475a
85b3244a7faaa6e36933091e98e44913d856acb0
26419 F20101211_AAAAEL bokhan_d_Page_082.QC.jpg
80155861e9cee5d961078a14f0d26a0b
b8c59107083e91385df707bb81af5aa49daa3c1c
1051974 F20101211_AAAADX bokhan_d_Page_073.jp2
c8b54329920fb7f882757984b920406b
c6cb209cf5809e7b909eb3eea8b490c625da8c73
18583 F20101211_AAAAFA bokhan_d_Page_003.jpg
ebffab29635cdfc30bfe6354c92a069b
7b45463e03d8780a543f884b5e584dca41373a1b
3673 F20101211_AAABID bokhan_d_Page_096thm.jpg
14103ab0d93b12dfe3f52f38db84c702
59af7590fa264fd8331e81b1320f8f7066eb7dbb
5552 F20101211_AAABHO bokhan_d_Page_088thm.jpg
74e3f51b60e933abe0caab9542489358
4455eb8e19994d2894a2d6f1a96f2a663cc5d6c4
62421 F20101211_AAAAEM bokhan_d_Page_113.jp2
acd35a7a4689c2006a5d96bf5b11bbbc
ccf145a2c81d2de5e3c76d9bd877fc300239a8cb
F20101211_AAAADY bokhan_d_Page_104.tif
cc12db6fc8c50648abf90bf981e5fe68
228b74238456632b93dc9c4649d2ed38bc365fa9
87894 F20101211_AAAAFB bokhan_d_Page_004.jpg
a706b8ae883f4453c2e0c15d5dbec8d1
1097002797a309ded8f331b4c8a0c96c894cd003
27981 F20101211_AAABIE bokhan_d_Page_097.QC.jpg
487308406a4751de5a10bae22efe3144
7361873e0789a3d1c0be61749034bed74c8a0bc6
26582 F20101211_AAABHP bokhan_d_Page_089.QC.jpg
a069933955117f529a6499dbc7b20e1c
748b302d90ab779a3e027cb64612fb79b14a47a1
F20101211_AAAAEN bokhan_d_Page_007.tif
cc9a26c98685f705dc74588c134b478d
7a1980182d7ad42f79332cfa5b731505fe0429de
4697 F20101211_AAAADZ bokhan_d_Page_044thm.jpg
b65146886f76cfc152aeb4480fc54474
2217714476e66889da9fd72bbecc5736c10220d3
11327 F20101211_AAAAFC bokhan_d_Page_006.jpg
cea5b93e817b4e2163fd12acd67a4692
46da0f2e83f95ca2acf213f6d369e6cbee386d9c
6944 F20101211_AAABIF bokhan_d_Page_097thm.jpg
f4b2a8abc028cf4c26ac2b0f36f8a021
c6d821cdd09e4aba498af0dac35211dae1fa25db
6674 F20101211_AAABHQ bokhan_d_Page_089thm.jpg
ab89e14e753884ab57ce3a5f69993721
53a4045c38a8352d6abe35ced2425ace22701b2b
69079 F20101211_AAAAEO bokhan_d_Page_135.jp2
7663e11ff47f778b7ebe9f4e2e4fce98
bf0933b7fcca623a49c53e1e35601106498ccfea
37886 F20101211_AAAAFD bokhan_d_Page_008.jpg
655f9e93dc163b527f6caa91f1f5e0d1
245ffaaecbcdec6b5e78a86b43262f273be0d48e
16752 F20101211_AAABIG bokhan_d_Page_098.QC.jpg
1d7dc038870b4e4de7d9c85b948e3e6d
b0690e7726329641328c14bd212f729231133194
23012 F20101211_AAABHR bokhan_d_Page_090.QC.jpg
fb64f51f4fc2a18531844ce6f7d12ddb
15660eb6f85b332c512de9953947af995eb94fc5
80794 F20101211_AAAAEP bokhan_d_Page_023.jp2
9a3bd2b65190b1df36ea6a4699d7753d
540e2f3d2432fdcb0ea3c0269616dac7796cc2aa
65684 F20101211_AAAAFE bokhan_d_Page_009.jpg
49449c261c66a28cc0e4c62b8351a773
c666aa563ab36819e21a54dffd34c9f6090f3888
4373 F20101211_AAABIH bokhan_d_Page_098thm.jpg
9a2f6a64b9b655ff63fb2f1fd5eea5f4
9c007148596a68ba482fbd18394dc1844f4038ee
5829 F20101211_AAABHS bokhan_d_Page_090thm.jpg
23f100c22b3549032f65066ca46220ce
b1aba86f2bcf6611795f6d02ab725384dcfeb23e
1025336 F20101211_AAAAEQ bokhan_d_Page_042.jp2
62d36285685641dec1d61c3cd864a762
80780c362c3129a38ca6cd945b9228b60c0a72dd
74682 F20101211_AAAAFF bokhan_d_Page_010.jpg
b9e77109ee83bf265db383b22a32e50c
d4f576f984a1ba0fd8712a1ac8ea0b549afc485c
9795 F20101211_AAABII bokhan_d_Page_099.QC.jpg
84ce4c88f9f71aa3bb42ba2beab617e5
0a2bc298e5b91ccfa58b2244516ec30b3ffa23f2
16036 F20101211_AAABHT bokhan_d_Page_091.QC.jpg
b8ffd7186b0eb5f1c2802694d3e2b299
e7693dd2efeba488e5e8364311f1aa1a8c0a5a05
1997 F20101211_AAAAER bokhan_d_Page_003thm.jpg
0f40337882913c49869b930b5db66449
7b11009c8a8a0c00b2f9a78451e678231253fb0f
11432 F20101211_AAAAFG bokhan_d_Page_011.jpg
016887b38d2cb9576e2e0eea68de6b55
f8c8f716296bb7dcd4bc9ee87447e77621476240
2946 F20101211_AAABIJ bokhan_d_Page_099thm.jpg
8494fe1d0ffe89be95612ca582c89d63
f482799aba5cd9a69c0d9cd9671e365a579df74a
4612 F20101211_AAABHU bokhan_d_Page_091thm.jpg
516648b01f75a49da5aa2f18e4c22eaf
36b0d8021715b356f28150a059c4d9464b45652b
15834 F20101211_AAAAES bokhan_d_Page_102.QC.jpg
8947fe9a6e0efd6a16540208a1c1ec08
f56d91e1a8442720ebe9c82d480a9924a2eb232c
61573 F20101211_AAAAFH bokhan_d_Page_012.jpg
bcc968befb5b67bf7f55694a3b5bcedf
cca14d737e2cf9007d65e847adc56e4c2b19fcdb
22841 F20101211_AAABIK bokhan_d_Page_100.QC.jpg
58dab3c2982fadbd57bbc031de29f6b4
a80a5f77f1a5bd8b2ca238492bb82767fd1fb684
18671 F20101211_AAABHV bokhan_d_Page_092.QC.jpg
fbff792ca28a274fb65cc303e393cce4
ed90a87cf515bc0fdb5324ddd7ee31408b2bd365
13133 F20101211_AAAAET bokhan_d_Page_076.QC.jpg
0c947fffdd2c2186c10a81bd5cf6bbf6
fbe983f01abfbe5d6e293abab10743f7f67914bf
39587 F20101211_AAAAFI bokhan_d_Page_014.jpg
4543c67377039fe472ecccb268388277
6d5602e50e71167b87c6c4001fc67274ab731128
21891 F20101211_AAABIL bokhan_d_Page_101.QC.jpg
9c86d03b61fd34b9e4759f005bcc9497
08e986c2138b48bbfdc7b0fbb2e4ceee95f75408
5593 F20101211_AAABHW bokhan_d_Page_092thm.jpg
d9127a16af54a3ae4570c7d78770975e
e6e0b345a7e043bb28a185b08e1608d33b904c53
825169 F20101211_AAAAEU bokhan_d_Page_051.jp2
5b86ab345f28e1b366cf449a9ce0a653
b58ff8f8bb7a1970d50bd54897e2665fb809cc28
67617 F20101211_AAAAFJ bokhan_d_Page_015.jpg
d4666ae9dbeca0fa5e1a9c9b48426555
e8d38bcd32623c34621f0508552084da07e46062
13229 F20101211_AAABJA bokhan_d_Page_110.QC.jpg
b4a3173f51970c2d2796c32f0c6f67ba
97844b241086e5fecb09769c67907534bfd33aaa
24429 F20101211_AAABHX bokhan_d_Page_093.QC.jpg
b860b5708e343a26e743038290bb8c67
a1761c5dfd74ac58e2c1839bb22fc45ee644ac72
202269 F20101211_AAAAEV UFE0021221_00001.xml FULL
012da8abd65a22d5e67354fb4e7fbf30
981fbba9ee2fdf0cdeb31d15f6da248912d96d77
4164 F20101211_AAABJB bokhan_d_Page_110thm.jpg
71db507377903b70bc7acee8615c99cd
28cc9587dab8243a41c4874e5f3cf84f5c5aef5d
5994 F20101211_AAABIM bokhan_d_Page_101thm.jpg
3a2f27fcd93b19824df5d8d9411ca0dc
8cf285f38bdec30540e03febd0872bb9f276052d
6538 F20101211_AAABHY bokhan_d_Page_093thm.jpg
8a29f8656b373ac107650d0b88eb7b6e
40845a5d4522ed3b34ed4dd5414e62086b35d9a0
61507 F20101211_AAAAFK bokhan_d_Page_016.jpg
cec11930157a7137b9091b9c32dfc859
c3ed743a8f202042866823c8e2505136f2e4fd6a
12755 F20101211_AAABJC bokhan_d_Page_111.QC.jpg
33e9bc56c2e9e4d6a4a12b0037a9eae9
21648added73ec42b8444f8b517f2fd96ca4c777
4809 F20101211_AAABIN bokhan_d_Page_102thm.jpg
ba8da22aac6777f49d5da9f0d10d5336
6051c0f1aa0607f990ebdc528208bab45401d060
6325 F20101211_AAABHZ bokhan_d_Page_094thm.jpg
d7e68e0c4b46f91dc93ff471d6fda01a
4201dbc6fb12a9e60236aea217c757f0db2594b8
60761 F20101211_AAAAGA bokhan_d_Page_034.jpg
59809120dfef0a4001d42b3761d49584
eac08d74664b6f236cc8d117e6077e30f0c07207
69138 F20101211_AAAAFL bokhan_d_Page_017.jpg
372bc22651d4f5646841f06f93e529e2
17471cc62ed159e85ec669442af1e0bed71159af
4063 F20101211_AAABJD bokhan_d_Page_111thm.jpg
b1fb6bf917c88f8b78150a57740c0119
17b1586def3a04a5821c0849964ba5ffc0b3a726
13929 F20101211_AAABIO bokhan_d_Page_103.QC.jpg
49ebe3c5342fd07c6c1823d6d96a9a84
669e990b78355f836dc76a4f002111e4521f4ca4
24487 F20101211_AAAAEY bokhan_d_Page_001.jpg
3b2b47fd1a69d5fc4c1be35a219c6d9b
a450ece7deab630d3dc147075978d3a8a823edd8
53215 F20101211_AAAAGB bokhan_d_Page_035.jpg
2fe1e1a5d1fd9ced97709c39fb1aebeb
d8831e7609d30e5444caef9462cbd827cc067df3
55075 F20101211_AAAAFM bokhan_d_Page_018.jpg
71bc3abf3092d0a195047b2c433fe34f
e79fa82b1df0149b5bc73d960bd7f530942f9ab9
12462 F20101211_AAABJE bokhan_d_Page_112.QC.jpg
2b17ae3ae8169212bcfb4694697d9974
00a67b64b813fa7ddde8a5c01966fba8f585f1d9
4532 F20101211_AAABIP bokhan_d_Page_103thm.jpg
11291270e0f234958fbaf5158fb63346
81577191f91e207e2e27184eb592ee9ede313c80
9760 F20101211_AAAAEZ bokhan_d_Page_002.jpg
a681c38ef07e93def6f716b94085a3c5
420d2ad057ca36ee2931f814e60655c3ca6179ca
41778 F20101211_AAAAGC bokhan_d_Page_036.jpg
0ec80c3b8a7b7668924b335fece36609
2661719bdc448aeb19b546e7832adee2d338ba5c
68467 F20101211_AAAAFN bokhan_d_Page_019.jpg
883c4d61b14b35ef459da66fc20c6d25
91d9f847e770c6347c083142b6ce760330eb0020
3807 F20101211_AAABJF bokhan_d_Page_112thm.jpg
c851d223a8d511c8149be7f23657a596
d907e351dfeeed6a35a523ce662319a3d73df2a8
17611 F20101211_AAABIQ bokhan_d_Page_104.QC.jpg
d4c4daa9e0572ad5e56dca921e09efb0
932771dad2523a1db7b096244100c44e9160c23e
67226 F20101211_AAAAGD bokhan_d_Page_037.jpg
978d264ec92ae151510ff6507266754e
0be307173c3bc9941814eb8c851ee8288404ada7
69747 F20101211_AAAAFO bokhan_d_Page_020.jpg
2eda17caaebee60e4a87ae2b8444e16a
ca1f3a14ce17e5213159234b2f45745194464afa
13350 F20101211_AAABJG bokhan_d_Page_113.QC.jpg
2f74ff4424f9c7699baa395e086900af
0b93fd39dd45134aed0058ace3d62d33da9bdd82
5261 F20101211_AAABIR bokhan_d_Page_104thm.jpg
216e8e99ec392528097aaeb2bc86f9fa
ee8285fe318dd55927efa56f230577944e8efc34
57716 F20101211_AAAAGE bokhan_d_Page_039.jpg
c11e69d1c545c23a9c967f75740282f7
348254be90edec6636519716dd9ca6e727ae1e39
60923 F20101211_AAAAFP bokhan_d_Page_021.jpg
a7175ebc30cd477081419048bfeb4058
369892856c8ddc4d4427011e0ca07e16cb0aa91c
4149 F20101211_AAABJH bokhan_d_Page_113thm.jpg
49bf7916c54c2c3be93b023ca958651a
7a81c7c59e0aaa1a13ec5a3eaa27db9652bcb11b
5285 F20101211_AAABIS bokhan_d_Page_105thm.jpg
af3e1c1f09659cecc0a6e1559581731a
5f5e1b9711eec562cb45cbfa75cd305add780f42
62444 F20101211_AAAAGF bokhan_d_Page_040.jpg
e8cb1372f2b62a823df649ce4ef2ff28
13598d84bf364b14d9bab69c19832b641b3b8a98
54432 F20101211_AAAAFQ bokhan_d_Page_022.jpg
387fb6ea3e0af2d789b9633a8ac5ca0e
d7fd9bfbc5c012b4a1dbf190429d8f0c6861d60f
13648 F20101211_AAABJI bokhan_d_Page_114.QC.jpg
a9400ddbd5605ca85717603b9f2f1c12
2980142da5ca1c64eee57127a02d1778939d8ce6
22717 F20101211_AAABIT bokhan_d_Page_106.QC.jpg
6cc334ebafe6b9ab29840893a3d3b152
c97ede8b7109b8bb3e13f3d21f0a0ae182c25691
73775 F20101211_AAAAGG bokhan_d_Page_041.jpg
0dbd8ab98a9f0d12f0455a51d471d259
7177dff20df2c5772bd84fb89089b7a7c95e11c0
54091 F20101211_AAAAFR bokhan_d_Page_023.jpg
64e9bd1805517631729aa6834d213ba0
6d043ce1644c53d4149a9f101b7dc36323b31417
4250 F20101211_AAABJJ bokhan_d_Page_114thm.jpg
cdbb833757ad00d94b68936a3168c545
ab8dca6e0e290bf531f2246b14bc2df5cc827480
6144 F20101211_AAABIU bokhan_d_Page_106thm.jpg
df3a2a154b076d4a683098451e4f92d8
9b3afbb184e2bd9b1ca64d3454bcc82283ce973c
74840 F20101211_AAAAGH bokhan_d_Page_042.jpg
b66473b0e75729636a6b58b3b058b88d
74f5af811d0cad73104d98d89edd6665eb988ced
58617 F20101211_AAAAFS bokhan_d_Page_024.jpg
59cf8f8395457cb5455170811a2f8982
2e41c83b41f93a35b0da5418181b76b964c50e68
12256 F20101211_AAABJK bokhan_d_Page_115.QC.jpg
11032677a99b48a0eb2067ef4343e6fb
2c7cffef9a2748541c1e31655f82a53bc551f283
F20101211_AAABIV bokhan_d_Page_107.QC.jpg
2ea7d8d9bee66d83da38c5234ee7273d
04d1b5e03ddf6a569eec45742cf42a29f3c4e2a1
62771 F20101211_AAAAGI bokhan_d_Page_043.jpg
6fcd8a90cb2c39963d910239f3161e53
6b41e54da9e9f9d7a50283981a7e61a6065dfef8
54219 F20101211_AAAAFT bokhan_d_Page_025.jpg
bd190e9926bf259916854aaabd866a65
0d6823b392ef6f8fc1c7b4ea0f6eaba6a9284495
4019 F20101211_AAABJL bokhan_d_Page_115thm.jpg
b0011cf6baf5f0c2810fb1d30ccdbf35
69ed175f45c425e4a21768f1af7fec6749da5bdf
6091 F20101211_AAABIW bokhan_d_Page_107thm.jpg
d736a70db2489a1535632cca8c9b342d
e85c5b462ccf6fa6a9e3f1e7756c59b84287463b
55358 F20101211_AAAAGJ bokhan_d_Page_044.jpg
02b5453184ab99c0ae8cf3799efc2ec4
a10c5b78443d3aeb2d56e753b7ce17190ad93fc9
81119 F20101211_AAAAFU bokhan_d_Page_027.jpg
7788becbd7d872537aa1aa6ada120a15
350966a814eabc92cf276fad9fa080174e2041a4
3691 F20101211_AAABKA bokhan_d_Page_125thm.jpg
11abf7ae5b93e9b3d1b2289bddfad417
c2f9422bdfb658f5e70693e4276494161b9a1740
4064 F20101211_AAABJM bokhan_d_Page_116thm.jpg
ff03958cd37e7ee28b3e1345afaf2c4b
28da8b303b0bc051aaaf9c0f384cef3b38ba3d95
26038 F20101211_AAABIX bokhan_d_Page_108.QC.jpg
8add4a88db0875b1d8a1f39d9d9ae866
f9c276058c9c1c700ac43d074a30a7831c764d09
61905 F20101211_AAAAGK bokhan_d_Page_045.jpg
5f690e54b80223b48a1b773a1359005b
e87579bd1900483351cc1cc0e68fea1c4d4ae7a2
73943 F20101211_AAAAFV bokhan_d_Page_028.jpg
fa0299bb9be65e530741c870a4cdb766
bd0fb674c04866f4fc4ff12375d72556eff9965f
5348 F20101211_AAABKB bokhan_d_Page_126.QC.jpg
f6f0cc649a4f5e222050c2819787bbb1
b15a415d7f152c8e5b09c445a01c029d20054975
6451 F20101211_AAABIY bokhan_d_Page_108thm.jpg
7913d63466e92400c4498c306ee4119f
f6acdc0aeb62f0225f768079ebb2fa1f3a2cfe3d
69096 F20101211_AAAAFW bokhan_d_Page_029.jpg
7247166d9f1e957292a39b2a30e7102d
8fb052503a0ab9faf82f06f584668c3d2b6957e4
1978 F20101211_AAABKC bokhan_d_Page_126thm.jpg
651eec3d6947c17732bada01a9931044
044bcaa34091d6f951dea6c16534d1f1b12d6e39
12616 F20101211_AAABJN bokhan_d_Page_117.QC.jpg
c957cedc13f3ecd8b5c84b3af6e8d501
215e1ca3b6eb1be798944559a3d5df68f5a83345
3874 F20101211_AAABIZ bokhan_d_Page_109thm.jpg
83a77b89b6ae4f99163bce6bfcb9ff13
b5849107058ee8c62b1fa5c001a84cdfe51b454d
91760 F20101211_AAAAHA bokhan_d_Page_063.jpg
2bf695fd3d48130320a52fbe5365936b
b2e6bc2bbd7c2e990abab2bafe51e176ef4671d3
57555 F20101211_AAAAGL bokhan_d_Page_046.jpg
acaca0cfc3b5d2501b87748a2c6b01a3
e301d7fc453b5a67274b2b1c485f15f4ab7c31ef
51199 F20101211_AAAAFX bokhan_d_Page_030.jpg
317dd7e1a9eb37b3b54b2a9398f0d408
7ec320452e4739e1c7618768f8b878ba5bd77d09
17169 F20101211_AAABKD bokhan_d_Page_127.QC.jpg
71cc4bcecbef4d77c7e6b2cc0cc87d44
9fdc7d746165d6816c05b30464445b6965267544
3986 F20101211_AAABJO bokhan_d_Page_117thm.jpg
2685f5f501e6a777b10a9f9d888b131a
d0452d5c078314661f07a4dd5f606ff96ddea3e0
74599 F20101211_AAAAHB bokhan_d_Page_064.jpg
8d253ca25b02cd5289e6e13b15c1e342
2c7ab5b96ec20d576d740a2ed07c62b876321ee2
86025 F20101211_AAAAGM bokhan_d_Page_047.jpg
5102d94e7f4ac267805b708eb42a1894
9916318f54d72305c531051f9298f84f44fb1b9b
52468 F20101211_AAAAFY bokhan_d_Page_032.jpg
745845e899efd785626f07e675532795
d6d04e6aac5516c188e5f8b6376a1f3d629d02a1
4795 F20101211_AAABKE bokhan_d_Page_127thm.jpg
3e23f3c0c3efea9e868ccad56fbb764d
6e93dbda2916a599048252bf762bd01995900097
13779 F20101211_AAABJP bokhan_d_Page_118.QC.jpg
9226c055dfa6b38537be8d264f997a9a
6e447230834d6c71b2e226b63319c43012ea7e6e
88090 F20101211_AAAAHC bokhan_d_Page_065.jpg
8f06d6f9fa24fa855d4d4304527fb958
ba1103bd3cf1761ff38e6b9f4e3fde1edfe134bf
70193 F20101211_AAAAGN bokhan_d_Page_048.jpg
e1331f6857c0e95b439fe2a56627806f
3134b73e272aff6c7f96f422981f270babf3864b
51246 F20101211_AAAAFZ bokhan_d_Page_033.jpg
ca30c2cbd4b9f51b643c200a5ca31c13
651cf6d43ec2335477cdf5843115d0c9dfc638c0
15735 F20101211_AAABKF bokhan_d_Page_128.QC.jpg
f339745e8a0eb3cd4dd4ec7d71a0d062
66127d80d7b19ded19299b2457c435b1eb3460b2
4199 F20101211_AAABJQ bokhan_d_Page_118thm.jpg
a425f9a2e57336f357e177f6648ef5a3
1001595b71fcaad582519b2fb2998f97290a2b99
76977 F20101211_AAAAHD bokhan_d_Page_066.jpg
2341f70e5d9f80e58af31e5b01e2d9af
240ed33b4c0d8d268c25e21f492451df84186e26
54368 F20101211_AAAAGO bokhan_d_Page_049.jpg
d76d5b6d23cd4d72973a92da27835823
2e8c02e3060f23242a25d82d6807abfc40c4aa21
4381 F20101211_AAABKG bokhan_d_Page_128thm.jpg
fba5f6860be200d3db460f27fa185941
612d7ff013b2ce45a022a0539060adeef3eb62ba
11439 F20101211_AAABJR bokhan_d_Page_119.QC.jpg
2f48848461a2a4a683f4d1ad1bd8a248
eb104adb2289806c69a1eeeabd0804785513bf29
85220 F20101211_AAAAHE bokhan_d_Page_067.jpg
5d1685f1768abeb900cbc0ea5ea8c782
06ae1521621a4fae6b058df023dc7f8a5e72de54
78318 F20101211_AAAAGP bokhan_d_Page_050.jpg
087356c4eb82a989a77718faeae96b4f
7e0bc2fc880ea5747475c9725967d8debf4b3198
15232 F20101211_AAABKH bokhan_d_Page_129.QC.jpg
99b92999b34c86603bfad2da8575c08d
1ccae35525c93600b066d79b87002b218feb1e8d
3658 F20101211_AAABJS bokhan_d_Page_119thm.jpg
41bfcb73271e50a55c60faad116450c6
49ffe342d51493c5f7d9e611034e82fcb5f4e4e3
29334 F20101211_AAAAHF bokhan_d_Page_068.jpg
18d976db30618eb396a9fb8ad0525861
b743455be0744599659243901b9c2cd444e3e718
68655 F20101211_AAAAGQ bokhan_d_Page_052.jpg
65cac8889f62d0e485aaf258596aa901
52f8dba278b9ec0284b56fc8d95a406d926a5c6f
4305 F20101211_AAABKI bokhan_d_Page_129thm.jpg
cb228d9d004e18e080f79cd550053758
ecfeac73ce737a89c92e026ee7b115bf3d59fbcf
11873 F20101211_AAABJT bokhan_d_Page_120.QC.jpg
7f24ae8d422a580dc84aaf3a566fc564
3fe550056f2af39f19ac5bb8c11f392369230804
58501 F20101211_AAAAHG bokhan_d_Page_069.jpg
1a0efd4547495bd21490b7c7a9d2d050
639117c8010954d35c9979dfe7bd4d946584e510
67526 F20101211_AAAAGR bokhan_d_Page_053.jpg
414249a7f77f39493a3711ec2e32edb1
adb3a309958f9a2b1354febfcf811b5586c6caf9
14135 F20101211_AAABKJ bokhan_d_Page_130.QC.jpg
551958e60fe26260def3b173fc7491ee
5e88e2e3048c96e30f7572ff3f6d12322032f337
3858 F20101211_AAABJU bokhan_d_Page_120thm.jpg
7aef9abb4c1639c23ddbed946f759791
550c5e195024b86550a7f7875848f50529dfc6af
68172 F20101211_AAAAHH bokhan_d_Page_070.jpg
0129e66d5012e735e573fe0cbfba2150
bb8f9c0cf98656b519b8c083c57444c662343f41
69124 F20101211_AAAAGS bokhan_d_Page_054.jpg
00696891376e7c7715ae7c2f145a92a1
6f39436601352e6f11501e8ba91d227b51d73340
4227 F20101211_AAABKK bokhan_d_Page_130thm.jpg
a08cd09cd13c88ea0c7da2b1c9089726
4475be6f023ac54de050247eb7b701c2e414871c
3719 F20101211_AAABJV bokhan_d_Page_121thm.jpg
bad24133ead957da64d47d1cf8bb4c4c
359133ea5d8de38606b4c0a69f2d09a6030d56cd
80173 F20101211_AAAAHI bokhan_d_Page_071.jpg
6003832c47e98d0186ff59e72a8efd69
9b013ac9115272aadb149d524f0c46dcbe57cb7b
64912 F20101211_AAAAGT bokhan_d_Page_055.jpg
6183ccb7cd2ff8cc0387d583048bb91a
7c4ec056a99d5b1cb26615f52f6f8c41d24ba9ba
21109 F20101211_AAABKL bokhan_d_Page_131.QC.jpg
d8435889e50b6db7113c342a47b4ddfa
e7c797edf898d39f7d65fe48752f6f6a481119d8
11904 F20101211_AAABJW bokhan_d_Page_122.QC.jpg
f3ab57d85491265cf9eddb16e4e274f6
f253ad8be21b23022e5d46680a5ed399da648469
97991 F20101211_AAAAHJ bokhan_d_Page_072.jpg
4886e0ad957844ebce1eb9625941b322
787bdaec2a1652606a3e308bc16b7a5493512a5f
78456 F20101211_AAAAGU bokhan_d_Page_056.jpg
68fa395a10fae5fd12b075b82b6d9180
d3c63507445c6e49ce230a9da45cf2469b25e167
5792 F20101211_AAABKM bokhan_d_Page_131thm.jpg
8d7432d42b3bdf692afb575166cd3808
e96444da461cad10158957a303691c6bb366aaea
3822 F20101211_AAABJX bokhan_d_Page_122thm.jpg
8589019aae32de739b346bd0de16bebf
e6d15454ef1edc605305432e62156d0bc6630b1f
63370 F20101211_AAAAHK bokhan_d_Page_073.jpg
26504b2517fe4ae33ad39a369b346831
a1d3a78f260522c486bcca3b42f2153071cbf55a
46005 F20101211_AAAAGV bokhan_d_Page_058.jpg
63b4c14d64b1fd9472b41772ae7fb78b
458b3c60aaf2090e87fddc67a7b869c41379fc49
26163 F20101211_AAABKN bokhan_d_Page_132.QC.jpg
956d16378e41d49fbdf381ee17bfcf56
47dc09c44668b0061b0848ce9a5c676955f81c9f
3966 F20101211_AAABJY bokhan_d_Page_123thm.jpg
65358ccc7bcdf394c86c79ae2be66499
9c64ebda9c543d68d4b0a4cf6da4daf8367ffcf0
78420 F20101211_AAAAHL bokhan_d_Page_074.jpg
6c94de3070f281d3f904c98a4214f64f
aa8ee237848b51904387942a674092d43f54eb96
62977 F20101211_AAAAGW bokhan_d_Page_059.jpg
25059d7e95aa382bdaeda364e35165b8
00bf7dffec94cfd176825f61977ae3b20bc82efa
11232 F20101211_AAABJZ bokhan_d_Page_125.QC.jpg
a0c2fae490b409ac3b7989df094cc4c0
e3f7f6a75724cf48871fdff7636a7c4b2c536e89
60587 F20101211_AAAAGX bokhan_d_Page_060.jpg
cea40d35fb1bb532c1afd96bb292ffef
dfb255e30940243c06c00d029d760ee641c63ff7
55685 F20101211_AAAAIA bokhan_d_Page_092.jpg
f3c0a8cc8faac33ff34c6cba42be6c5c
dbdb6055b4adf41b338fba4ccb8313cf6e92062f
6542 F20101211_AAABKO bokhan_d_Page_132thm.jpg
57cc76409074126d010f64b96011c791
b6e3a2c8f1c4841c501ee4af760151eada088e30
76370 F20101211_AAAAHM bokhan_d_Page_075.jpg
99f6f2d94e8e22735c360245c6c5bf9b
166bb81311660a0e0bb3c82a7391ce1d63866ce1
80990 F20101211_AAAAGY bokhan_d_Page_061.jpg
73dd1673e4a1d72ef359cc71b966f386
0a91e40225507f8cd6247ed775ab38f49bed469c
88028 F20101211_AAAAIB bokhan_d_Page_097.jpg
367476418d3dd81bd8e6b47b7ca820c2
5468c2aecff34661511f9b14ff9485b9c7340f0f
22508 F20101211_AAABKP bokhan_d_Page_133.QC.jpg
f9fd1c8d5681f7dfd99e68a345d51d31
126b8055091b5d68342f07140a1cc8a777a170ba
42682 F20101211_AAAAHN bokhan_d_Page_076.jpg
5388090acea9354aad255b4dc75f9688
9597336ea71b7b4ce01f2d5aee7b45dc53b565a3
75861 F20101211_AAAAGZ bokhan_d_Page_062.jpg
f7b410f949764a9768cd6bc02ed49a8e
9ab0eecbe1cb4677c1db3144912a3a3c92906e5d
31234 F20101211_AAAAIC bokhan_d_Page_099.jpg
bac4b3e05b38911fb18db491fd9abd81
8e9ddf26f1d12c29eea8c96cf83d38452ab491d7
5955 F20101211_AAABKQ bokhan_d_Page_133thm.jpg
94532b01f03aaa289ff50f5ed38f590b
bd9d8f9d13e9eb45d14b0bd97e4544a407e57a32
71213 F20101211_AAAAID bokhan_d_Page_100.jpg
bd1792c1d4373597980387ef04238515
8319d852af2d5431720010401e0d5b2c41dae484
19253 F20101211_AAAAHO bokhan_d_Page_077.jpg
02c2f39d5024c1ef84492eaa7c1f137a
8063d5ed5de0dcac5ec69e0f2488c25d9c46d49c
11627 F20101211_AAABKR bokhan_d_Page_134.QC.jpg
a2065358c83d026634e9c21f2c6d2b6b
f7dd3844478cc6eb3d56ac661b50ce4ab21405f3
70870 F20101211_AAAAIE bokhan_d_Page_101.jpg
4fc57eb273fe323bf67036ec0247d93d
4ad5378d42760fc4c0c8e8f06510ee4b1c47eb29
20734 F20101211_AAAAHP bokhan_d_Page_078.jpg
4837ebae276d3fbade879514f9828857
f201ad8f3981178b564406041feff92034adf9c4
3517 F20101211_AAABKS bokhan_d_Page_134thm.jpg
8eb628cf0b41ae3ec24dcb4c1bbd5a44
57ca7f12b96bbdab92ee2e5dead866c03bd08a4b
50762 F20101211_AAAAIF bokhan_d_Page_102.jpg
bba8dae8d908aeaf12957667cf067935
3af276b4b17ea64b2183ba9b10122c39223c6b7e
20741 F20101211_AAAAHQ bokhan_d_Page_079.jpg
be159b72b647bf285225aa14970b5704
1b39bd4badc1259d3505fea59ca30da37a532094
15146 F20101211_AAABKT bokhan_d_Page_135.QC.jpg
dcf7ec29a6058e8d6aeba4007f384bd5
179d7a095cb32f24260bcd74e27e7cd310ff0e5d
46100 F20101211_AAAAIG bokhan_d_Page_103.jpg
862d852efbdb74581ad10d5187231b93
91a0636b2c5e5365a3c2c58e4147ba36b7198003
79903 F20101211_AAAAHR bokhan_d_Page_080.jpg
51876221c194e907a1e7f543a82e5129
fc6343151d4abc23128340d5b5bb0b7018b8110a
3928 F20101211_AAABKU bokhan_d_Page_135thm.jpg
a56f0751b75ee70441104cfe51a8876e
a1fe7e24271398024bf7711de1c9ee9ed330ff40
59103 F20101211_AAAAIH bokhan_d_Page_104.jpg
0c882d3bd1738bd3440ebe88ae9962f3
1f4847c07f7bde4dd4a569f5632243531f1f9bff
89101 F20101211_AAAAHS bokhan_d_Page_081.jpg
a2f94c001522e7f2ffea52d35c949eda
5656ca947d7af5d8a9ac6cbb8900ed983ed29e83
156084 F20101211_AAABKV UFE0021221_00001.mets
e238ab1c743f856c61353240fed28540
4388b924fed98b763daacd9bd05cb0e8880dd015
61446 F20101211_AAAAII bokhan_d_Page_105.jpg
5f477d0e1ee4742e36104476196b8e38
3fbb633cd83028c2b51355bc84f64b98631d76a1
71993 F20101211_AAAAHT bokhan_d_Page_084.jpg
092b20bb65909a171704dba440b57d2a
bfdfd9d4a9ab5556c0d5883ff9dc58719391c8e3
73712 F20101211_AAAAIJ bokhan_d_Page_106.jpg
6b3dba98b6b360ba4cffe23355b2614c
4815b1917b1e2616c301a595be49f41dc1aaf42e
46514 F20101211_AAAAHU bokhan_d_Page_085.jpg
6d1c29bd5e88f418e91dc563b0459131
103be4c11b3e4d143dc6f5e50a2171f8831563ae
79580 F20101211_AAAAIK bokhan_d_Page_107.jpg
c6ab2841fe20e3761cba5a848e5233bf
0bb483828e6d6fe3ccd732ddad4998764c61a4ea
47771 F20101211_AAAAHV bokhan_d_Page_086.jpg
c854a2bda3da2bd9d5228dd377aed3cc
661ebdd78870be71818a1152456ddd713b8dac46
84032 F20101211_AAAAIL bokhan_d_Page_108.jpg
a4c38504b675137895a9cfae4bc6351f
fc88ef13632d8cfda73fe677540832054cf03f2e
89085 F20101211_AAAAHW bokhan_d_Page_087.jpg
2447bc9ac1ba8f64af8a1daa8b76ab89
3a29d9c8ababef20115f5aa929c086a2eae93113
40475 F20101211_AAAAJA bokhan_d_Page_124.jpg
f93c924ebc970069f4631745e5b3d372
555de63f2b3ec1bf220a2a921195794f9eeac1d4
46383 F20101211_AAAAIM bokhan_d_Page_109.jpg
c53b99da58e168c12ce57eeea2d2601d
f590eb4e7e81472e007c03b4f5ebb35d64fadf4d
71175 F20101211_AAAAHX bokhan_d_Page_088.jpg
4a7f5ec3ec368b498e4905fb78a2ccf9
920a6f62884bb59c5e5b43cec8bb9673f1a0f0cd
36624 F20101211_AAAAJB bokhan_d_Page_125.jpg
39c98de960fb6d068bade1259b9364b5
5cf4433c157987e629f91fc587f30ba219dfc3a6
73869 F20101211_AAAAHY bokhan_d_Page_090.jpg
2ba93cc45b5275e96b4de22c2235befb
e06a3b0928c6ab08b270306d8f49801f7fec8678
17200 F20101211_AAAAJC bokhan_d_Page_126.jpg
29375cc98fa3aedc181c0317f5b55cfd
39259180c90fc9bd189d929dcce07891f9e8d312
44490 F20101211_AAAAIN bokhan_d_Page_110.jpg
d8a2ae7ad8747b92d0f5b32081307274
0339d9f4c64c07b005741b4cd99da48e7e4e669f
50501 F20101211_AAAAHZ bokhan_d_Page_091.jpg
e72f8a621088c656770185b854b36580
ba7893dc554ee0594dfe2d057a20f75b38b4129e
50111 F20101211_AAAAJD bokhan_d_Page_128.jpg
ba2b3f3b091c3ddc9e3ec6a55daee02d
00cd41f61256d33829453e24117e6eaba42c0820
42088 F20101211_AAAAIO bokhan_d_Page_111.jpg
93cd8ed75493c4eb057f43516eb093a4
b748df65be2ceddbe7a46a5437a00ed75191e735
48055 F20101211_AAAAJE bokhan_d_Page_129.jpg
b84990e16421cf8574a822831b335763
c27183d8c0f60fff0b9eb73222606bead07f9511
39947 F20101211_AAAAIP bokhan_d_Page_112.jpg
decc2f6ce0b2dd08daa0568df15c9a6f
c4877e84c6d6243c9699e1635fe2d1bac847e68a
45668 F20101211_AAAAJF bokhan_d_Page_130.jpg
340284628f3f05d8deeef0289c07c814
338b57c8b9ce38350d7e100da5fb54b5dc9aa92e
44396 F20101211_AAAAIQ bokhan_d_Page_113.jpg
b1619d70edb57bc5de0fb649f82841a2
c8477f5a47ab020a041957ef28aa197772ff7f54
64787 F20101211_AAAAJG bokhan_d_Page_131.jpg
eb2c94dfb4ff2d85c7f898f05fa7af91
965ed00f63ace5ad7f4e48871010f1281f4d6af4
45487 F20101211_AAAAIR bokhan_d_Page_114.jpg
a7186153b4db51f4d15c72c3d4f00bc9
2e3183cd4635b72b7f77c32759af267ae786adac
88481 F20101211_AAAAJH bokhan_d_Page_132.jpg
34e6163c65b0951c1867f65fac8d8b0c
6aa0122838e90ea59ef725084a9eda7df2cd5182
40985 F20101211_AAAAIS bokhan_d_Page_115.jpg
68490bbcdc64efe577e64850cbd95712
54d10939a1fe7c59b8f448b53e06a2180dd85c60
35756 F20101211_AAAAJI bokhan_d_Page_134.jpg
4e71e6c1e91e073918370df9f30b9b1d
9fab491f303a02ed4f82e01654e8f8b7a5b0fd7e
42225 F20101211_AAAAIT bokhan_d_Page_116.jpg
1ad6d92e661775633d2175516d84fe04
2ac74aa9eb99282585aadcc6641854d9a6c8c9c7
25610 F20101211_AAAAJJ bokhan_d_Page_001.jp2
30fd27f0f6b0b2c4879480e7d7517d80
4e265f4b576aeec1adba542c1ea2b15ad705d650
46186 F20101211_AAAAIU bokhan_d_Page_118.jpg
79ddc9338fcc2d2417d8dffbf33c01df
fefa8822fd8dcc2b031db3ab8c7dc6ac625a1efa
4852 F20101211_AAAAJK bokhan_d_Page_002.jp2
d2851b7a5b1db2da9cbf8339ce06d726
f30d972e3ea44c5cb26fee5bed1d42bb4059e1c5
37650 F20101211_AAAAIV bokhan_d_Page_119.jpg
e7556eb187cfbb1e700c3baa6c425bfb
c23e54e130411cbb87758b0068a3f60a397f6471
1051955 F20101211_AAAAJL bokhan_d_Page_004.jp2
240590d3ec57cee04e0d33a9c3204dda
c062a3ec301e1ac20c2568a10961e8b9683f0f29
38913 F20101211_AAAAIW bokhan_d_Page_120.jpg
90ad6a3c30ff53b3c6cf104d280520ad
02ffc293c2b20c16062544a1a4f5584637529462
F20101211_AAAAJM bokhan_d_Page_005.jp2
e686329251988deacbf696b963ad1221
535a4c93dcc5b3c6b83db2961d84b91efabbd5f5
37490 F20101211_AAAAIX bokhan_d_Page_121.jpg
e4653ccc430796ec6a73073c52d97a8c
c24d09e8faddeb1b27b59e5f8401204f5daf9b81
95555 F20101211_AAAAKA bokhan_d_Page_021.jp2
6384ba14ade1a4d76fcab065cff0f5c0
3bd2c16c46d0e246bdf8e97ae6769aa004774217
64200 F20101211_AAAAJN bokhan_d_Page_006.jp2
8d61391b5ca4c12514916e8e25844aba
9467949b81b88657c6a408af7b8edc6b524eecf1
38843 F20101211_AAAAIY bokhan_d_Page_122.jpg
c40d5f18a8950f90c4bacac1661888f6
186875ed7091e541e31357df1f5c6c2874fa6778
779162 F20101211_AAAAKB bokhan_d_Page_022.jp2
b3b861d94b1f5eb46cbbc4a3e6df7c6c
891725e59e9f8be2fdf6a19a3ecc3ca5ceaa2334
39532 F20101211_AAAAIZ bokhan_d_Page_123.jpg
249a0739675d456b5cabc127d9f4c417
59f1ffe8b387d42b82bb750def4d25eaa179c7a6
766738 F20101211_AAAAKC bokhan_d_Page_024.jp2
cb380b898dfb38c8a85b9ccd0d5d447a
5e4e08be385db8c2e8c6b5a2c0bb5dd9d8d25a7c
1051967 F20101211_AAAAJO bokhan_d_Page_007.jp2
0452ccef4fe36eea29ddf3d5157291b7
2f4f06a01ff199958dce6516215e57d95577173b
744264 F20101211_AAAAKD bokhan_d_Page_025.jp2
660bf95f695bf2e52e0c2d42ff467625
47bbc3f3ac0e67a6836588783358fdd843fc91c8
762596 F20101211_AAAAJP bokhan_d_Page_008.jp2
e740fd60de7e8eb1a8351403b07f0fe0
7928b4662faac325a3dd43cf06ad17d49624efcc
1051976 F20101211_AAAAKE bokhan_d_Page_026.jp2
bcbcd3d0864dc0a5f428a846908db8f7
d8b75162296a384300ed70b2219625290f953da5
102029 F20101211_AAAAJQ bokhan_d_Page_009.jp2
e2afc463f14d3434138666da4b3f1f32
a130ce86491e0c9e031427a8112c92ec518deca5
128997 F20101211_AAAAKF bokhan_d_Page_027.jp2
f1ffceb3082471d8d1eb291e56d01ab2
c316dc9bb797ae9c3fc431b4b314a4f6fdc7b90d
116343 F20101211_AAAAJR bokhan_d_Page_010.jp2
91d1b02f905f1e3589a236df13a54420
272429cdedf8ff0730da61006fa0f5d0fc8d81cc
115681 F20101211_AAAAKG bokhan_d_Page_028.jp2
8e1debfcc585329bac3cf6923a107157
2b853de89ec0628905653aff19f67b8d21b615a6
9170 F20101211_AAAAJS bokhan_d_Page_011.jp2
ecca7f0e287b77023fc9782cf937cd18
b7bc1058ad22133955cc0f43f83a00acf19bdcde
107833 F20101211_AAAAKH bokhan_d_Page_029.jp2
2ef53fe297754be2388096b81c77de37
37f2e00fec87162552e6d6089a68b920dd561811
98799 F20101211_AAAAJT bokhan_d_Page_012.jp2
870ac6cd52d518cb609655d458c0ad40
23fe13029563e4302e3cbab6b18f61541d58abc9
79347 F20101211_AAAAKI bokhan_d_Page_030.jp2
a9d090a36fcc8f316d13595b2901ece5
ac38d324657b525a1f3df780e26887972ce70432
530544 F20101211_AAAAJU bokhan_d_Page_014.jp2
7ee78af1cdf8f503c2f7fa037043e511
55525aab9d668366b090278d34282bb343b86ef6
F20101211_AAAAKJ bokhan_d_Page_031.jp2
4e693404a949edcbeef0f31c50f45078
7cb55275b89c5c0e712c25802ea51dd09edf279c
924880 F20101211_AAAAJV bokhan_d_Page_015.jp2
ede81deb4d5b9ab803d1599abdff416d
9862e99b352de6c7e0696ac565a26e52ba23ee1a
723623 F20101211_AAAAKK bokhan_d_Page_032.jp2
192223e798d8c82364c2e3fef923469a
b8cc9f92a857f5e05229eaed8f2a81530f0ce4af
98663 F20101211_AAAAJW bokhan_d_Page_016.jp2
10a439898c6e8b4aa6e33c3c8bbcd147
19c174f0275116585428e7ec4b48ac413f4c4c34
694282 F20101211_AAAAKL bokhan_d_Page_033.jp2
6358d2cc11e7cc5b0911c26d5a096d51
c5be385b26e6362d9b5ac0f767ea5778ff6609d8
955494 F20101211_AAAAJX bokhan_d_Page_017.jp2
efabaa17d50ed8b591f1c203cdee0266
b5dc5c6ea426bdd29cf2c8e74981b521af5551e0
965550 F20101211_AAAALA bokhan_d_Page_052.jp2
ccd3a0b1c307bf22c1d70f5b40fda7f5
334b551983e6fdb44cb1ec3b930b22a8cfb115aa
818568 F20101211_AAAAKM bokhan_d_Page_034.jp2
53c368c44f1d36631245680716f5cdaa
45ed33c68976642fa6f1f6386bd6d9e7c97e9bba
742131 F20101211_AAAAJY bokhan_d_Page_018.jp2
e2b03571cb6603d77e0aaaf78fcb6696
e2fd0f9d8a991d881afa2915554b2c3e191e5efe
943083 F20101211_AAAALB bokhan_d_Page_053.jp2
6ea96c42d0c7bac7a09176bb919d029c
cc9d249fb420d12950ffc3372328c7adfcdfd3a6
707927 F20101211_AAAAKN bokhan_d_Page_035.jp2
73ed854f9295b1648e019f46adf6ee14
1f2f346c8add73252098c9f0856903253a922f49
932294 F20101211_AAAAJZ bokhan_d_Page_019.jp2
792aee8405afa41ee8cddba1fb3e7b83
e9ee4d3df1e7f60b551dcd8ed0a41c0e1898e4d3
954245 F20101211_AAAALC bokhan_d_Page_054.jp2
7bbc687dc2e9b6d03529c0397bd9c262
63597ef3fd39f7d14947f9499bb02d6f4a902197
515803 F20101211_AAAAKO bokhan_d_Page_036.jp2
2e267f1944810eaf40761afafa97a6cc
36ef8b06418d0bdba80f2e7e17b6accb52ac3dc9
867291 F20101211_AAAALD bokhan_d_Page_055.jp2
49d7f5177860a39221a2a6ce37a14b98
a8d4a5e33f9d702733dc25f7e2f789d9f83e9513
1051970 F20101211_AAAALE bokhan_d_Page_056.jp2
73ef5e2214b36927bac18a576ca4ca60
4f7c890a62f940a8bc05e7233ff3a48791d3b11b
957226 F20101211_AAAAKP bokhan_d_Page_037.jp2
2d3d258cf208d4cda3bf1844b0d6bb5d
1241359a905a03f96303cf6c627e4f70828717fb
985943 F20101211_AAAALF bokhan_d_Page_057.jp2
273a22d4517d120683d25990be0697d8
eb38b62f4ad25cb12eec1ab0f3309bd2dcba4324
907308 F20101211_AAAAKQ bokhan_d_Page_038.jp2
1b1a5606045d553b722685efd4e989f4
902f3c2f97723082a73515ccb72defd973e63164
610984 F20101211_AAAALG bokhan_d_Page_058.jp2
05f9b291c78785097205cadae3e5ff30
f1eea96921552f214397dcbd5319189571b288ad
834299 F20101211_AAAAKR bokhan_d_Page_039.jp2
e8a607b19f511ed86d05547808652420
b318f94ddba75644d1fe82220ebcb94c926b1f17
803271 F20101211_AAAALH bokhan_d_Page_059.jp2
57be4a3333a4483dacf16dd356e1da7c
a09d721270f7fa968a7551123a1a73ddced7da12
868554 F20101211_AAAAKS bokhan_d_Page_040.jp2
5482910435624be3a480c01aaaa89439
c5b56d36eba470a8c865817e5de53bdda01ece38
814845 F20101211_AAAALI bokhan_d_Page_060.jp2
2f064ee1dd40cc8918cb89d52f749999
bcb919ccb19b2469d2f04a538badcab3d8970a4e
1017558 F20101211_AAAAKT bokhan_d_Page_041.jp2
97eaffde8c71cde893704ab51c1e39bb
bad9257608eb688c210efb20b61477842f36a0d7
1051929 F20101211_AAAALJ bokhan_d_Page_061.jp2
02ea606456bfdd0585d0acb53af4cb07
6f7f7712cad6da716980c1ed0c4caaeb40b85b61
808000 F20101211_AAAAKU bokhan_d_Page_043.jp2
b41324ef08ab9ebafbc7fffe022daa53
819ca595a3cf6e807534240569672613f73cfa9d
F20101211_AAAALK bokhan_d_Page_062.jp2
b5707ead5c5721bcba931675643a1c86
66b8626ee6f54d3cdb72194a6db4662507da1cf8
885715 F20101211_AAAAKV bokhan_d_Page_045.jp2
8f734b8c18c645235fcabd35423410c8
75254292a9c2c01906b11348c1ec30579935caad
1051874 F20101211_AAAALL bokhan_d_Page_063.jp2
2b39339c3d8331c4746f9db5cf2758f1
a920fe749b210135d06f5c6e1848d4aa27c5d12c
810258 F20101211_AAAAKW bokhan_d_Page_046.jp2
47c08efac1958f6ca35661ed95aa2cd2
5b70986eeeb4c9af92fbbf7dd8c420893547da02
F20101211_AAAAMA bokhan_d_Page_081.jp2
2a41c4ce7d860bcd4e8c19be192969e7
a0de84acafd80c1aa8793b61c7332b90bf7a24e6
1051917 F20101211_AAAALM bokhan_d_Page_065.jp2
eb1453a46afc0c0ea615bcb77a27eb6d
0631f5612b08f9fd101ae97b60f06d14b81013b7
1051964 F20101211_AAAAKX bokhan_d_Page_047.jp2
a4c4ac3b249ffa5f75a6715cf6430bd6
844920e5d9cff0ed8f930ded460411a9a54b684b
F20101211_AAAAMB bokhan_d_Page_082.jp2
67317394d7e0bb097f5e4e46cfd2ba8c
4b4dbe4dbd225ed6fae933cd71d6543778aee2ee
1051973 F20101211_AAAALN bokhan_d_Page_066.jp2
6ee8ec02ceb4f7e87caa0be967f670de
ed75f90e1f5e2ab91259c98768fa7d20cef7311c
986304 F20101211_AAAAKY bokhan_d_Page_048.jp2
178323a012b48f12ca3e4b7e289f5140
a486c1c5ab2ffe9afa02220a39560088843b4ba7
728578 F20101211_AAAAMC bokhan_d_Page_083.jp2
316d98c40041c24998674a3879f9a25f
6d21f3d1d48e9035a428b9f245782dfbe8376bd6
F20101211_AAAALO bokhan_d_Page_067.jp2
359aad1d5358d51abbf796113c250593
4ae97ddbf284dc2470197f74ef5ec3541b9553e5
758301 F20101211_AAAAKZ bokhan_d_Page_049.jp2
f86ea5b26dcf33d4162853c79b2320d8
d83226c4ec94a1af6fe2cca9f416a9a180bd1bff
998761 F20101211_AAAAMD bokhan_d_Page_084.jp2
152561faa0ed0f66f92a4d0edad235d2
42b519cc198456f16de06ee83391f97efc3ca98d
342536 F20101211_AAAALP bokhan_d_Page_068.jp2
379f0e038fe3367c9534fd5eabe343cb
a2dfc3d821723359831cf77773d47e7e27499afa
673802 F20101211_AAAAME bokhan_d_Page_085.jp2
1a6a1bb531329bc11b655b617f34d5a2
72cee6d6739cdbdd9e7fa5d7ff6efa4d9899f0a4
652254 F20101211_AAAAMF bokhan_d_Page_086.jp2
ca04e78d6f65bdc406bc9116d4e88ab4
153b8b3baa465f540a7e1be0c6664fb18cd6c3e4
805210 F20101211_AAAALQ bokhan_d_Page_069.jp2
a2beb9ce6a502eaa544abbb9490181f0
d21de4605aededa66cd0b113a6c04a08d7aec309
951732 F20101211_AAAAMG bokhan_d_Page_088.jp2
ab6587d7a0fec77c7ceb8a66abe2ed41
176383490f8114ae610177b661a2389516f196d5
929055 F20101211_AAAALR bokhan_d_Page_070.jp2
8b3d93d1e67c693b5b0c803beb4f1046
28c33e2e26d72e0d80a494ceedeef57a66cdb9d1
1044812 F20101211_AAAAMH bokhan_d_Page_090.jp2
8785f6b68cfc0406b4d41c40a91d3abc
5f74f91d00e067ed780ed85394129fe4389d6fb6
F20101211_AAAALS bokhan_d_Page_071.jp2
a462590f8f3a403cbe882aa45a315877
e192678ce4213be5601b41535513553b3c1a8ab8
640637 F20101211_AAAAMI bokhan_d_Page_091.jp2
40e23ac34af5a13ff2fe66794946de6d
70c883fe10a2c19a18fc21b48dfb54923aba30ec
F20101211_AAAALT bokhan_d_Page_072.jp2
85ed55c28ccc7a808064acebb6af3b3a
3f9fa33ea667aa348b26009cc0df283ee6eac43a
807770 F20101211_AAAAMJ bokhan_d_Page_092.jp2
3ce94ee658e1f6591aff8eea6dbd2d06
2e279c5a583e93187f2516a91b11220d757323df
F20101211_AAAALU bokhan_d_Page_074.jp2
72f8a3dd4d62377a89fa9116192bab80
a34a986fc03eacebb691ffda1f4cb14fd81a5184
1051984 F20101211_AAAAMK bokhan_d_Page_093.jp2
a13c7263396f5daf5a9ff597cb282953
54506871237e86da5d768445b4c52cb3a2d90286
1051962 F20101211_AAAALV bokhan_d_Page_075.jp2
8c8c97e597ca081ddad37831bf4be7fc
b4477e88c1e8f5ad1203050ee34fabb55789e82d