Lecture notes for the course

554017 ADVANCED QUANTUM

CHEMISTRY
Pekka Manninen
2012

Contents
1 Many-electron quantum mechanics
1.1 Molecular Schrodinger equation . . . . . . . . . . . . .
1.1.1 Molecular Hamiltonian . . . . . . . . . . . . . .
1.1.2 Time-independent Schrodinger equation . . . .
1.1.3 Limitations . . . . . . . . . . . . . . . . . . . .
1.1.4 Example: Helium-like atom . . . . . . . . . . .
1.1.5 Example: Dihydrogen ion . . . . . . . . . . . .
1.2 Electron spin and wave function antisymmetricity . . .
1.2.1 Spin functions . . . . . . . . . . . . . . . . . . .
1.2.2 Antisymmetry principle . . . . . . . . . . . . .
1.2.3 Example: Spin orbitals of 1s2 state of He . . . .
1.2.4 The Slater determinant . . . . . . . . . . . . . .
1.2.5 Example: Approximate wave functions for He as
1.2.6 Example: Helium-like atom revisited . . . . . .
1.2.7 Slaters rules . . . . . . . . . . . . . . . . . . .
1.3 Concerning interpretation . . . . . . . . . . . . . . . .
1.4 One- and N -electron expansions . . . . . . . . . . . . .
1.5 Further reading . . . . . . . . . . . . . . . . . . . . . .
2 Exact and approximate wave functions
2.1 Characteristics of the exact wave function .
2.2 Electron correlation . . . . . . . . . . . . . .
2.2.1 Electron density functions . . . . . .
2.2.2 Example: Density functions of He . .
2.2.3 Generalized density functions . . . .
2.2.4 The pair function . . . . . . . . . . .
2.2.5 Example: Coulomb and nuclear cusps
2.2.6 Electron correlation summary . . . .
2.3 Electron density . . . . . . . . . . . . . . . .
2.3.1 Electronic energy by density . . . . .
2.3.2 FockDirac density matrix . . . . . .
2.3.3 Example: Closed-shell energy . . . .
2.3.4 Other density functions . . . . . . . .
3

. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
in He
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

7
. . . . . . . . . . .
7
. . . . . . . . . . .
7
. . . . . . . . . . .
8
. . . . . . . . . . .
9
. . . . . . . . . . .
9
. . . . . . . . . . . 11
. . . . . . . . . . . 11
. . . . . . . . . . . 11
. . . . . . . . . . . 12
. . . . . . . . . . . 13
. . . . . . . . . . . 13
Slater determinants 14
. . . . . . . . . . . 14
. . . . . . . . . . . 15
. . . . . . . . . . . 16
. . . . . . . . . . . 17
. . . . . . . . . . . 19
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

21
21
23
23
23
24
25
26
28
29
29
29
31
31

2.4

2.5

On the optimization of the wave function . . .

2.4.1 Variation method . . . . . . . . . . . .
2.4.2 Electronic gradient and Hessian . . . .
2.4.3 HellmannFeynman theorem . . . . . .
2.4.4 The molecular electronic virial theorem
Further reading . . . . . . . . . . . . . . . . .

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

32
32
33
33
34
35

3 The HartreeFock theory

3.1 The HartreeFock approximation . . . . . . . .
3.1.1 The HF wave function . . . . . . . . . .
3.1.2 The HF equations and the Fock operator
3.1.3 Self-consistent field method . . . . . . .
3.1.4 Restricted and unrestricted HartreeFock
3.1.5 Example: HartreeFock treatment of H2
3.1.6 RoothaanHall equations . . . . . . . . .
3.2 Further topics in HF self-consistent field theory
3.2.1 On the convergence of the SCF method .
3.2.2 Integral-direct SCF . . . . . . . . . . . .
3.2.3 Linear scaling aspirations . . . . . . . .
3.3 MllerPlesset perturbation theory . . . . . . .
3.4 Further reading . . . . . . . . . . . . . . . . . .

. . . .
. . . .
. . . .
. . . .
theory
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

37
37
37
37
38
39
40
40
42
42
44
44
45
46

4 Configuration interaction
4.1 Configuration-interaction wave function . . . . .
4.1.1 Single-reference CI wave functions . . . .
4.1.2 Multi-reference CI wave functions . . . .
4.1.3 Optimization of the CI wave function . .
4.1.4 On the disadvantages of the CI approach
4.2 Multi-configurational SCF theory . . . . . . . .
4.2.1 MCSCF wave function . . . . . . . . . .
4.2.2 Example: MCSCF wave function of H2 .
4.2.3 Example: Selection of the active space in
4.3 Further reading . . . . . . . . . . . . . . . . . .

. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
H2 O
. . .

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

49
49
50
50
52
53
53
53
53
55
55

.
.
.
.
.
.
.
.

57
57
57
57
58
59
61
61
62

5 Coupled-cluster theory
5.1 Cluster expansion . . . . . . . . . . . . . . . . . .
5.1.1 Coupled-cluster wave function . . . . . . .
5.1.2 The CC Schrodinger equation . . . . . . .
5.1.3 The coupled-cluster exponential ansatz . .
5.1.4 Coupled-cluster model hierarchy . . . . . .
5.1.5 CI and CC models compared . . . . . . .
5.2 Solving the coupled-cluster equations . . . . . . .
5.2.1 On coupled-cluster optimization techniques
4

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.

63
64
64
65
65
66
67

.
.
.
.
.
.
.
.
.
.
.
.
.
.

69
69
69
70
70
71
72
73
76
76
76
78
78
79
80

.
.
.
.
.
.
.
.
.
.
.
.
.

81
81
81
81
82
83
85
85
85
86
86
86
88
95

A Second quantization
A.1 Basic concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A.1.1 Occupation number vectors . . . . . . . . . . . . . . . . . . . . . .
A.1.2 Creation and annihilation operators . . . . . . . . . . . . . . . . . .

97
97
97
98

5.3

5.4

5.2.2 The closed-shell CCSD model . . . . . . . . . .

Special formulations of coupled-cluster theory . . . . .
5.3.1 The equation-of-motion coupled-cluster method
5.3.2 Orbital-optimized coupled-cluster theory . . . .
5.3.3 Coupled-cluster perturbation theory . . . . . . .
5.3.4 Local correlation methods . . . . . . . . . . . .
Further reading . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

.
.
.
.
.
.
.

6 Basis sets and molecular integrals

6.1 Atomic basis functions . . . . . . . . . . . . . . . . . . . . . .
6.1.1 General considerations . . . . . . . . . . . . . . . . . .
6.1.2 Hydrogenic functions . . . . . . . . . . . . . . . . . . .
6.1.3 The Laguerre functions . . . . . . . . . . . . . . . . . .
6.1.4 Slater-type orbitals . . . . . . . . . . . . . . . . . . . .
6.1.5 Gaussian-type orbitals . . . . . . . . . . . . . . . . . .
6.2 Gaussian basis sets . . . . . . . . . . . . . . . . . . . . . . . .
6.3 Integrals over Gaussian basis sets . . . . . . . . . . . . . . . .
6.3.1 Gaussian overlap distributions . . . . . . . . . . . . . .
6.3.2 Simple one-electron integrals . . . . . . . . . . . . . . .
6.3.3 The Boys function . . . . . . . . . . . . . . . . . . . .
6.3.4 ObaraSaika scheme for one-electron Coulomb integrals
6.3.5 ObaraSaika scheme for two-electron Coulomb integrals
6.4 Further reading . . . . . . . . . . . . . . . . . . . . . . . . . .
7 Accounting for the effects of special relativity
7.1 Relativistic quantum mechanics . . . . . . . . . . .
7.1.1 Theory of special relativity . . . . . . . . . .
7.1.2 KleinGordon equation . . . . . . . . . . . .
7.1.3 Dirac equation . . . . . . . . . . . . . . . .
7.1.4 On the origin of magnetic energy levels . . .
7.2 Relativistic Hamiltonians for many-particle systems
7.2.1 DiracCoulomb Hamiltonian . . . . . . . . .
7.2.2 BetheSalpeter equation . . . . . . . . . . .
7.2.3 DiracCoulombBreit Hamiltonian . . . . .
7.3 Quasi-relativistic Hamiltonians . . . . . . . . . . .
7.3.1 Foldy-Wouthuysen transformation . . . . . .
7.3.2 Breit-Pauli Hamiltonian . . . . . . . . . . .
7.3.3 Other quasi-relativistic Hamiltonians . . . .

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

A.2
A.3
A.4
A.5

A.1.3 Number-conserving operators . . . . . . . . .

The representation of one- and two-electron operators
Commutators and anticommutators . . . . . . . . . .
Orbital rotations . . . . . . . . . . . . . . . . . . . .
Spin in second quantization . . . . . . . . . . . . . .
A.5.1 Notation . . . . . . . . . . . . . . . . . . . . .
A.5.2 Operators in the orbital basis . . . . . . . . .
A.5.3 Spin properties of determinants . . . . . . . .

B Performance of the electronic-structure models

B.1 Total electronic energies . . . . . . . . . . . . . .
B.2 Molecular equilibrium geometries . . . . . . . . .
B.3 Molecular dipole moments . . . . . . . . . . . . .
B.4 Reaction enthalpies . . . . . . . . . . . . . . . . .
References

.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

100
101
103
105
106
106
106
109

.
.
.
.

111
111
112
113
113
119

Chapter 1
Many-electron quantum mechanics
1.1

Molecular Schr
odinger equation

The Schrodinger equation for a quantum system of N electrons and K nuclei a molecule
is1

H(x
(x1 , x2 , . . . , xN ; X1 , X2 , . . . , XK ; t)
1 , x2 , . . . , xN ; X1 , X2 , . . . , XK ; t) = i
t

(1.1)

The wave functions describe all the chemical and physical properties of the system;
and the approximations made in the
subject only to limitations in the Hamiltonian H
solution of this equation. The variables {x} and {X} are used to collectively symbolize all
variables needed to describe the electrons and nuclei; here the position in space and the
internal degree of freedom, spin (see Section 1.2).

1.1.1

Molecular Hamiltonian

is the molecular Hamiltonian, which is for a molecule containing N electrons

In (1.1), H
and M nuclei with charges {ZK }(K = 1, . . . , M ) [13]
= H
e + H
n + H
en
H
N
N
N
1X 2 X X

He =
+
|ri rj |1
2 i=1
i=1 j=i+1

M
M
M
X
X
1X 1 2
ZK ZL

Hn =
K +
2 K=1 mK
|RK RL |
K=1 L=K+1

en =
H
1

M X
N
X

K=1 i=1

ZI
.
|ri RK |

In atomic units used throughout this course, where ~ = me = 40 = e = 1; c = 1/ 137.

(1.2)
(1.3)

(1.4)
(1.5)

The wave function is correspondingly dependent on both the electron and nuclear variables. Usually, because of the large ratio mn /me , it is a good approximation to treat
the nuclei fixed in space. This will decouple the electronic and nuclear degrees of freedom, and we may turn our attention to the electronic problem. This is referred to as the
BornOppenheimer approximation. Consequently, the first term in (1.5) vanishes and its
second will not affect the solution of the electronic Schrodinger equation. The result is the
standard framework in quantum chemistry, molecular electronic Hamiltonian,
=
H

N
X

i +
h

i=1

N X
N
X

gij

(1.6)

i=1 j=i+1

i is the one-electron Hamiltonian of electron i moving in an electrostatic potential

where h
caused by the nuclei, and gij is the interaction between the electrons i and j:
M
X
ZK
hi = 1 2
i
2
|ri RK |
K=1

gij = |ri rj |1 .

(1.7)
(1.8)

In the case of fixed nuclei, the contribution from (1.5) reduces to a simple summation,
Etot

1.1.2

M
M
ZK ZL
1X X
.
= Eelectronic +
2 K=1 L=K+1 |RK RL |

(1.9)

Time-independent Schr
odinger equation

In quantum chemistry, usually the time-independent form of (1.1), with now also invoking
the BornOppenheimer approximation,
n (x1 , x2 , . . . , xN ) = En n (x1 , x2 , . . . , xN ),
H

(1.10)

is under consideration. The corresponding eigenvalues En are in the present context the
quantized energy levels of the electronic cloud of the molecule. The eigenfunctions or wave
functions n describe the electronic structure and properties of the molecule.
Any two eigenfunctions that correspond to different energy values En , Em possess the
orthogonality property
Z
(1.11)
n m dx1 dx2 dxN = 0.
Even if two or more eigenvalues happen to be identical, corresponding to a degenerate
state of the system, the different eigenfunctions may still be assumed orthogonal.
Furthermore, it is necessary that the wave function is square-integrable,
Z
(x1 , x2 , . . . , xN )(x1 , x2 , . . . , xN )dx1 dx2 dxN = a R.
(1.12)

Usually we also assume that the wave function is normalized, i.e., a = 1.

8

1.1.3

Limitations

Eq. (1.10) provides a basis for molecular quantum mechanics and all the static electronic
properties of the molecule may be obtained by expectation values of the appropriate Hermitian operators or using perturbation theory. However, we should recognize the limitations
of this model:
BornOppenheimer approximation. The nuclear-motion Hamiltonian (1.2) plays a
central role in molecular spectroscopy. The motion of nuclei can sometimes affect
even those properties that are considered electronic. This is the case e.g. in the
JahnTeller effect and the Renner effect.
Other limitations in the Hamiltonian (1.6): non-relativistic kinematics, instant interactions, point-like nuclei, and electrostatic interactions only. We will address the
two first of these in Chapter 7.
Eq. (1.1) has analytical solutions only for a few trivial one-electron systems, such as
hydrogen-like atom or the molecule H+
2 . Therefore we are resorted to make approximate solutions, aiming at systematic and controllable approximations, as we will
discuss in later Chapters.
Real experiments are not carried out for an isolated molecule, and observation always
involves interaction with the system, and thus a time-independent description is not
sufficient; but the time-dependent form (1.1) should be applied. There are also means
for treating dynamical effects in the electronic cloud also within the framework of the
time-independent eigenvalue equation (1.10) [4]. Addressing these methods is out of
the scope of this course.

1.1.4

Example: Helium-like atom

For a helium-like atom with a point-like nucleus of charge Z the electronic Hamiltonian,
Eq. (1.6), is
1 + h
2 + g12 = 1 2 Z 1 2 Z + 1 .
=h
H
2 1 r1 2 2 r2 r12
Due to g12 , this is a three-body problem, and thereby no closed form solution exists for
the eigenvalue equation (1.1). We start our consideration by an approximation, where g12
1 + h
2 . We can separate the variables in the eigenvalue equation
0 = h
is set to zero, i.e. H
H0 0 = E0 by substituting2
0 = (r1 , r2 ) = 1 (r1 )2 (r2 ),
2

Considering only the variables r (position vectors), which in this case are most conveniently represented
as the variables of spherical coordinates, r = (r, , )

to the eigenvalue equation and dividing by , obtaining

2 2 (r2 )
1 1 (r1 ) h
h
+
= E,
1 (r1 )
2 (r2 )
meaning that is a solution if (denoting E = 1 + 2 )
1 1 (r1 ) = 1 1 (r1 )
h
2 2 (r2 ) = 2 2 (r2 ),
h
= .
meaning that 1 and 2 are solutions of the one-electron eigenvalue problem h

is a simultaneous eigenstate of the operators h1 and h2 ,

1 (r1 , r2 ) = 1 (r1 , r2 )
h
2 (r1 , r2 ) = 2 (r1 , r2 ).
h
The 1-electron 1-center eigenfunctions that are solutions to the one-electron eigenvalue
problem (this holds only in the absence of g) are the same with the solutions of the
Schrodinger equation for a hydrogen-like atom (note the factor Z),
nlm (r, , ) = Rnl (r)Ylm (, ),

(1.13)

consisting of the radial part

Rnl (r) =

2Z
n

3/2 s

(n l 1)!
2n(n + 1)!

2Zr
n

l

L2l+1
nl1

2Zr
n

Zr
exp
n

(1.14)

and a spherical harmonic function, explicitly expressed as

Ylm (, ) =

2l + 1 (l m)! m
P (cos ) exp(im).
4 (l + m)! l

(1.15)

In the above, L denotes the associated Laguerre polynomials

Ln (x) =
and P the Legendre polynomials
Plm (x) =


1
dn 
exp(x)x n exp(x)xn+
n!
dx

l+m
(1)m
2 m/2 d
(1

x
)
(x2 1)l ;
2l l!
dxl+m

10

Plm = (1)m

(1.16)

(l m)! m
P (x).
(l + m)! l

(1.17)

1.1.5

Example: Dihydrogen ion

When moving from atoms to molecules, the first complication is that even the one-electron
eigenfunctions, even when g = 0, are not obtainable in closed form. Instead of the onecenter nlm functions (1.13) we need polycentric molecular orbitals (MO) that describe the
states of a single electron spread around the whole nuclear framework.

The one-electron eigenvalue equation in the case of hydrogen molecule ion H+

2 is h =
, where


h = 1 2 1 + 1 .
2
rA rB
This may be solved with high accuracy in confocal elliptic coordinates, the solution being
a product of three factors that are in turn solutions of three separate differential equations.
We will, however, go immediately to approximations. It is reasonable to expect that
(r) cA A
100 (r) (densities close to nucleus A)
(r) cB B
100 (r) (densities close to nucleus B)
where cA and cB are some numerical factors. An appropriate trial MO would then be
B
(r) = cA A
100 (r) + cB 100 (r).

Due to the symmetry of the molecule, it is reasonably to assume that there is no higher
density in the other end than in the other; therefore
2
2
B
2
A
B
|(r)|2 = c2A [A
100 (r)] + cB [100 (r)] + 2cA cB 100 (r)100 (r),

implies that cB = cA . Thus, it appears that we can construct two rudimentary MOs from
the 1s AOs on the two centers, denoted by
B
1 (r) = cA [A
100 (r) + 100 (r)]
B
2 (r) = cB [A
100 (r) 100 (r)],

where the remaining numerical factors cA and cB are to be chosen so that the wave functions
will be normalized.
MOs build up this way, in linear combination of atomic orbitals (LCAO), are of great
importance throughout molecular quantum mechanics.

1.2
1.2.1

Electron spin and wave function antisymmetricity

Spin functions

It is well-known that a single electron is not completely characterized by its spatial wave
function (r), but an intrinsic angular momentum, spin, is required to explain e.g. line
emission spectra of atoms in an external magnetic field. Therefore we must introduce a
11

spin variable in addition to r. We further associate operators sx , sy and sz with the

three components of spin angular momentum. We will adopt x as a variable taking into
account both degrees of freedom, and r, and then introduce spin-orbitals
(x) = (r)().

(1.18)

describes the spin state and is a solution of an eigenvalue equation sz = . If the

orbital state is of energy , then

h
=
sz = ,

(1.19)
(1.20)

and is a state in which an electron simultaneously has definite energy and z-component
of spin.
In the case of electrons, the eigenvalue equation (1.20) has only two solutions:
1
1
sz = ,
sz = ;
2
2
with the eigenfunctions and often referred to as spin-up and spin-down, respectively. Therefore an orbital yields two possible spin-orbitals, = and = .
These spin functions may be thought as spikes at the points sz = +1/2 and sz = 1/2,
respectively, and their normalization and orthogonality properties are
Z
Z

d =
d = 1
(1.21)
Z
Z

d =
d = 0.
(1.22)
Also the total spin operators

Sz =
S2 =

N
X

i=1
N
X

sz,i
s2i

(1.23)

i=1

N
X

s2x + s2y + s2z

i=1

(1.24)

with eigenvalues of M and S(S + 1), respectively, are needed in the discussion of manyelectron system. They commute with a spin-free electronic Hamiltonian; and any exact
stationary state is an eigenfunction of them.

1.2.2

Antisymmetry principle

The fact that electrons are indistinguishable, i.e. |(x1 , x2 )|2 = |(x2 , x1 )|2 , leads to (when
generalized for N -electrons) two possibilities for a wave function:
P (x1 , x2 , . . . , xN ) = (x1 , x2 , . . . , xN )
P (x1 , x2 , . . . , xN ) = P (x1 , x2 , . . . , xN )
12

symmetric wave function

antisymmetric wave function

where the operator P permutes the arguments and P is +1 for an even and 1 for an odd
number of permutations.
The choice for an system consisting of N electrons is once and for all determined by
the antisymmetry principle (Pauli exclusion principle):
The wave function (x1 , x2 , . . . , xN ) that describes any state of an N -electron system
is antisymmetric under any permutation of the electrons.

1.2.3

Example: Spin orbitals of 1s2 state of He

When spin is included, the electrons must each be assigned to spin-orbitals 1s, 1s, 2s, . . .
For the 1s2 configuration, both electrons may occupy the 1s orbital with the following
choices of spin variables:
1 (x1 , x2 )
2 (x1 , x2 )
3 (x1 , x2 )
4 (x1 , x2 )

=
=
=
=

100 (r1 )(s1 )100 (r2 )(s2 )

100 (r1 )(s1 )100 (r2 )(s2 )
100 (r1 )(s1 )100 (r2 )(s2 )
100 (r1 )(s1 )100 (r2 )(s2 ).

All of these functions have the same eigenvalue E = 21s of the electronic Hamiltonian.
It is apparent that 1 and 2 are symmetric under interchange of x1 and x2 , and are
thus non-suitable to describe the electrons. However, we may form linear combinations
of s to obtain the desired behavior under the interchange. For instance, the following
appears to be acceptable:

1 (x1 , x2 ) = [3 (x1 , x2 ) 4 (x1 , x2 )]/ 2

1.2.4

The Slater determinant

The previous argumentation may be directly generalized for any N -electron wave function.
For more than two electrons, a large number of differently permuted product functions
must be combined to yield a fully antisymmetric wave function.
Expanded in particular permutations the exact wave function3 is written as
X
(x1 , x2 , . . . , xN ) =
c (x1 , x2 , . . . , xN )
(1.25)

c M

"

X
P

P P 1 (x1 )2 (x2 ) n (xN ) .

M are normalization coefficients and is most

minant

 1 (x1 )

 1 (x2 )
(x1 , x2 , . . . , xN ) = M 
..
.

 (x )
1
N
3

Exact at the limit

13

(1.26)

conveniently expressed as a Slater deter

2 (x1 ) n (x1 ) 

2 (x2 ) n (x2 ) 
.
(1.27)
..
..
...

.
.

2 (xN ) n (xN ) 

Combinations which are symmetric under variable permutations are automatically rejected
by the Slater determinant, and thus the Pauli principle is fulfilled.
The normalization factor M is reasoned in the following way: when expanded, consists
of N ! products, and thus an integrand would consist of (N !)2 products, each of a form
[1 (xi )2 (xj ) n (xk )][1 (xi )2 (xj ) n (xk )].
It is evident that unless i = i , j = j etc. in a particular product, the product will give
no contribution to the result, due to the orthogonality of spin-orbitals. Therefore, there is
N ! non-vanishing contributions,
and as every contribution is equal to 1 with normalized
R
spin-orbitals, we have dx1 dxN = M 2 N !, hence the normalization factor has the
value M = (N !)1/2 .

1.2.5

Example: Approximate wave functions for He as Slater

determinants

The Slater determinant of the configuration (1s)2 of a helium atom in the minimal orbital
basis is


1  100 (r1 )(1 ) 100 (r1 )(1 ) 
1
1 = 
det |1s 1s|.

2 100 (r2 )(2 ) 100 (r2 )(2 )
2

Note the abbreviated notation that will be used quite extensively hereafter: only the
diagonal elements are displayed, and the explicit expression of variables is omitted the
left-right order of the variables in this notation will be 1, 2, 3, .... Also, note that we have
denoted nlm s with their corresponding AO labels.
Respectively, the proper antisymmetric wave functions for the excited 1s2s configuration
are written as
1
2 = det |1s 2s|
2
1
3 =
(det |1s 2s| + det |1s 2s|)
2
1
4 = det |1s 2s|
2
1
(det |1s 2s| det |1s 2s|) .
5 =
2

1.2.6

Example: Helium-like atom revisited

=
Consider the 2-electron Hamiltonian, now including electron-electron interaction, H
h1 + h
2 + g. Let us assume that the (normalized) wave function 1 approximates the (1s)2
configurationDof the
  He Eatom. The corresponding energy is obtained as an expectation
 
value H11 = 1 H
 1 . Now
Z
1
H11 =
{100 (r1 )100 (r2 ) [(1 )(2 ) (2 )(1 )]
2
14

o
100 (r1 )100 (r2 ) [(1 )(2 ) (2 )(1 )] dr1 dr2 d1 d2 ,
H

but since the Hamiltonian does not operate on spins, we may perform the spin-integration
immediately [c.f. Eq. (1.22)],
Z
[(1 )(2 ) (2 )(1 )] [(1 )(2 ) (2 )(1 )] d1 d2 = 2.
Thus we have
Z

100 (r1 )100 (r2 )dr1 dr2

100 (r1 )100 (r2 )H
Z
Z

=
100 (r1 )h1 100 (r1 )dr1 100 (r2 )100 (r2 )dr2
Z
Z

2 100 (r2 )dr2 (r1 )100 (r1 )dr1

+ 100 (r2 )h
100
Z
+ 100 (r1 )100 (r2 )
g 100 (r1 )100 (r2 )dr1 dr2

H11 =

= 21s + J

This result is still an approximation, with only difference to the independent-particle model
being the contribution J from the two-electron integral. The next step in the determination
of He energy levels would be to include more and more Slater determinants corresponding
to excited state configurations and evaluate the respective matrix elements. As anticipated earlier, although the exact solution would require an infinite number of included
determinants, the energy spectrum would in practise converge towards the correct one
rather rapidly. However, the analysis of the many-electron problem using this kind of
direct expansion would obviously be extremely tedious, and a more general approach is
needed. Also, the evaluation of integrals gets far more complicated in the case of many
atomic cores.

1.2.7

Slaters rules

We may indeed devise more systematic rules for the matrix elements. We begin byPnoting

that the expectation value over the 1-electron part of the molecular Hamiltonian, N
i=1 hi ,
will reduce to the sum of N identical terms, since the coordinates of each electron appear
symmetrically in the corresponding integral:
 +
* 
X 
X D   E

i  =
 r .

h
r h
(1.28)


r
i

The expectation value of the two-electron part is a bit more complicated, as contributions may arise from terms that differ by an interchange of two electrons; but essentially a
15

similar argumentation leads to

 +
* 
X 
X




gij  =
(hr s |
g | r s i hr s |
g | s r i) ,


rs

(1.29)

ij

where the former part is referred to as the Coulomb integral and the latter as the exchange
integral. the expectation value of the energy being simply the sum of (1.28) and (1.29).
We also need the off-diagonal elements. Fortunately enough, there are non-vanishing
elements only in two cases:
one spin-orbital is different between and (r 6= r ) when
D   E D   E X
 
 
g | r s i hr s |
g | s r i) ,
(hr s |
H
 s +
 = r h

(1.30)

s6=r

two spin-orbitals are different (r 6= r , s 6= s ) when

D   E
 
H
g | r s i hr s |
g | s r i) ,
 = (hr s |

(1.31)

These results are called Slaters rules. In Slaters rules, the spin-orbitals are required to be
orthonormal, but the rules may be generalized for matrix elements even in a non-orthogonal
spin-orbital basis.

1.3

Concerning interpretation

What is the physical meaning of the wave functions introduced earlier? First of all, the
general formulation of quantum mechanics is concerned only with symbolic statements involving relationships between operators and operands, and is independent of the employed
representation or interpretation.
In a very orthodox interpretation, MOs as well as spin-orbitals are just mathematical
means for the solution of the molecular Schrodinger equation and no physical meaning can
be given for them or even their squared norms.
Perhaps more commonly, a spin-orbital (x) itself is thought to be just a mathematical entity but |(x)|2 dx is taken to be the probability of a point-like electron being in an
element dx, or in volume element dr with spin between and +d. |(r)|2 dr is the probability of finding an electron with any spin in the volume element dr. Also (x1 , x2 , . . . , xN )
cannot be interpreted; but ||2 dx1 dx2 dxN is a probability of the electron 1 in dx1 , electron 2 simultaneously in dx2 , etc. Then, the probability
of electron 1 being in dx1 while
R
other electrons are anywhere would be equal to dx1 dx2 dxN .
However, it might be better to think of an electron as being delocalized, i.e. not having a
more precise position than that indicated by its spatial part of the wave function (r). Then
|(r)|2 dr would be the amount of charge in an element dr. In other words, in this point
16

of view a MO can be thought to approximately describe an electron really spread around

the nuclear framework. However, as we will see later, MOs can be unitarily rotated, i.e.
their form shaped, and they still correspond to the same definite energy. Which MOs are
then closest to the physical ones?

1.4

One- and N -electron expansions

Let us recall: the Slater determinants do not represent an accurate solution to the molecular Schrodinger equation, but we may express the exact wave function as a linear combination of determinants. Furthermore, we have to expand the molecular orbitals in the
determinants as a linear combination of atomic orbitals
X
a =
Ca (r).
(1.32)

As we will see later, atomic orbitals are further approximated with a set of simpler
functions that are also functions of the coordinates of a single electron. This three-step
procedure is often referred to as the standard model of quantum chemistry.4 To arrive
at the exact solution (within the given Hamiltonian), i) orbitals in terms of which the
determinants are constructed must form a complete set in one-electron space, and ii) in
the expansion of N -electron wave function, we must include all determinants that can be
generated from these orbitals.5
All approximations in the solution of the Schrodinger equation should be unambigious
and precisely defined. Ideally, they should be improvable in a systematic fashion; and to
yield more and more elaborate solutions that approach the exact solution. Thus we speak
of different models, levels of theory and hierarchies of approximations.
Computational electronic-structure theory has a few standard models for the construction of approximate electronic wave functions. At the simplest level, the wave function is
represented by a single Slater determinant (the HartreeFock approximation); and at the
most complex level the wave function is represented as a variationally determined superposition of all determinants in the N -electron Fock space (full configuration-interaction, FCI).
Between these extremes, there are a vast number of intermediate models with variable cost
and accuracy. These will be discussed in Chapters
It should be noted that none of the models is applicable to all systems. However, each
model is applicable in a broad range of molecular systems, providing solutions of known
quality and flexibility at a given computational cost.
Furthermore, there are in fact three general features determining the accuracy of quantum chemical methods:6 electron correlation treatment (truncation of the many-particle
4

In passing, we note that this orbital-based approach is not the only way for describing the electron
cloud, but there are numerous alternatives, for instance by optimizing the electron density (wave function)
in real space (see e.g. [5] and references therein.
5
Of course the other and today more popular way for describing the N -electron expansion is the
density-functional theory, see e.g. [6]; a topic that deserves its own lecture course.
6
Assuming a molecule in vacuo, at 0 K

17

Figure 1.1: The two-way hierarchy of approximate molecular wave functions

space), and the description of the one-particle space, i.e. the basis set used as well as
the choice of the molecular Hamiltonian. This is schematically illustrated in Fig. 1.2. A
treatment based on the Schrodinger equation (1.1) with an effective electronic Hamiltonian
describing non-relativistic (NR) kinematics (1.6), is adequate only for systems containing
the lightest elements, located in the first and second row of the Periodic Table. For systems containing heavier elements, a treatment taking into account the effects from special
relativity, will be needed. This will be addressed in Chapter 7.

Figure 1.2: The three features determining the accuracy of a quantum chemical calculation.

18

1.5

Further reading

R. McWeeny, Methods for Molecular Quantum Mechanics (Academic Press 1992)

R. E. Moss, Advanced Molecular Quantum Mechanics (Chapman & Hall 1973)

19

20

Chapter 2
Exact and approximate wave
functions
2.1

Characteristics of the exact wave function

We shall need to approximate the molecular wave function perhaps drastically in order
to reach the applicability to molecules with multiple atomic cores and several electrons.
The approximations should, however, carried out with care. For this reason, we will now list
down the properties (either rationalized earlier or simply taken for granted) that an exact
wave function would possess, and on this basis we should seek an approximate solution
that retains as many of the following properties as possible.
The exact molecular electronic wave function
1. is antisymmetric with respect to the permutation of any pair of the electrons:
P (x1 , . . . , xN ) = P (x1 , . . . , xN )

(2.1)

where P = 1 for even and odd number of permutations, respectively.

2. is square-integrable everywhere in space,
h|i = a R

(2.2)

3. is variational in the sense that for all possible variations , which are orthogonal to
the wave function, the energy remains unchanged:
h|i = 0 h|H|i = 0

(2.3)

4. is size-extensive; for a system containing non-interacting subsystems the total energy

is equal to the sum of the energies of the individual systems.
21

0.8
0.6
0.4
0.2
0
-0.2

1
1*

-0.4
-0.6
-3

-2

-1

Figure 2.1: Nuclear cusps. Wave functions corresponding to the two lowest eigenvalues of
H2 molecule, along the HH bond.
5. Within the non-relativistic theory, the exact stationary states are eigenfunctions of
the total and projected spin operators,
S2 = S(S + 1)
Sz = M

(2.4)
(2.5)

6. The molecular electronic Hamiltonian (1.6) is singular for ri = rj and thus the exact
wave function must possess a characteristic non-differentiable behavior for spatially
coinciding electrons, known as the electronic Coulomb cusp condition,



1
lim
(2.6)
= (rij = 0),
rij 0
rij ave 2
the description of which is a major obstacle in the accurate practical modelling of the
electronic wave function. There exists a condition similar to (2.6) also for electrons
coinciding point-like nuclei, known as the nuclear cusp condition.
7. It may shown that at large distances the electron density decays as

(r) exp(23/2 Ir),

(2.7)

where I is the ionization potential of the molecule.

8. The exact wave function transforms in a characteristic manner under gauge transformations of the potentials associated with electromagnetic fields, ensuring that all
molecular properties described by the wave function are unaffected by the transformations.
22

As we saw in Chapter 1, some of these square integrability and Pauli principle are included straightforwardly, whereas size-extensivity and the cusp condition are more difficult
to impose but still desirable. Others are of interest only in special situations.

2.2

Electron correlation

2.2.1

Electron density functions

Ignoring any concerns about the interpretation, let us define the function
Z
1 (x1 ) = N (x1 , x2 , . . . , xN ) (x1 , x2 , . . . , xN )dx2 dxN ,

(2.8)

and an electron density function that are observable by e.g. X-ray crystallography, by
integrating over spin,
Z
P1 (r1 ) = (x1 )d1 .
(2.9)

It is possible to introduce corresponding density functions for different configurations

of any number of particles; e.g. for two particles as
Z
2 (x1 , x2 ) = N (N 1) (x1 , x2 , . . . , xN ) (x1 , x2 , . . . , xN )dx3 dxN (2.10)
Z
P2 (r1 , r2 ) =
(x1 , x2 )d1 d2 .
(2.11)

These determine the two-electron densities and tell us how the motions of two electrons are
correlated as a result of their interaction. Because electrons interact only in pairs (i.e. there
are no many-body effects), there is no need to consider any higher distribution functions
than the pair functions.

2.2.2

Example: Density functions of He

Let us consider the 1s2s singlet and triplet states of the helium atom,
1
[100 (r1 )200 (r2 ) + 200 (r1 )100 (r2 )] [(1 )(2 ) (1 )(2 )]
2
1
3
(x1 , x2 ) = [100 (r1 )200 (r2 ) 200 (r1 )100 (r2 )]
2
(1 )(2 )
M = +1

[(
)(
)
+
(
)(
)]
M= 0

1
2
1
2
2
(1 )(2 )
M = 1
1

(x1 , x2 ) =

For the singlet state, we have

Z
1
|100 (r1 )200 (r2 ) + 200 (r1 )100 (r2 )|2 |(1 )(2 ) (1 )(2 )|2 dr2 d2
(x1 ) = 2
4
23

=
P (r1 ) =



1
|100 (r1 )|2 + |200 (r1 )|2 |(1 )|2 + |(1 )|2
Z2
(x1 )d1 = |100 (r1 )|2 + |200 (r1 )|2 .

can be rewritten as
(x1 ) = P (r1 ) |(1 )|2 + P (r1 ) |(1 )|2 ,
where the first term vanishes unless 1 = +1/2 and the second unless 1 = 1/2, and thus
1
P (r1 ) density for spin-up electrons
2
1
P (r1 ) =
P (r1 ) density for spin-down electrons
2

P (r1 ) =

These densities are always equal in a singlet state, but in non-singlet states the and
-densities differ, giving a resultant spin density. For the first triplet state we obtain
P (r1 ) = 1, P (r1 ) = 0
and for the third
P (r1 ) = 0, P (r1 ) = 1,
while for the second we get
1
P (r1 ) = P (r1 ) = P (r1 ).
2

2.2.3

Generalized density functions

Consider a 1-electron system, an electron described by a spin-orbital . The expec

tation
R value in this state of any quantity described by operator F is given by hF i =

(x)F (x)dx. We can rewrite this as

Z
Z

hF i =
F (x) (x )dx =
F (x; x )dx
x =x

x =x

by having agreed that F works on functions of x only and setting x = x after operating
with F but before completing the integration. We define the generalized density functions
as
Z

(x1 ; x1 ) = N (x1 , x2 , . . . , xN ) (x1 , x2 , . . . , xN )dx2 dxN

(2.12)
Z

(x1 , x2 ; x1 , x2 ) = N (N 1) (x1 , x2 , . . . , xN ) (x1 , x2 , . . . , xN )dx3 dxN(2.13)

.
24

We may discuss all one-electron properties in terms of (x1 ; x1 ) and all two-electron properties in terms of (x1 , x2 ; x1 , x2 ). The spinless analogues are given by
Z

P (r1 ; r1 ) =
(x1 ; x1 )d1
(2.14)

1 =1
Z
Z

(r1 , r2 ; r1 , r2 ) =
(x1 , x2 ; x1 , x2 )d1 d2 .
(2.15)
1 =1

2.2.4

2 =2

The pair function

Let us write the spin-dependence explicitly also in the case of :

(x1 , x2 ; x1 , x2 ) = , (r1 , r2 ; r1 , r2 )(1 )(2 ) (1 ) (2 )
+, (r1 , r2 ; r1 , r2 )(1 )(2 ) (1 ) (2 ) + . . .

(2.16)

which reduces to a four-component quantity when spin is integrated over. Eq. (2.16) can
be written in our notation for the diagonal components, , (r1 , r2 ; r1 , r2 ) = (r1 , r2 )
etc., as
(r1 , r2 ) = (r1 , r2 ) + (r1 , r2 ) + (r1 , r2 ) + (r1 , r2 ),
(2.17)
where each component resembles a contribution to the pair function from electrons at
r1 and r2 with spins indicated by the subscript as well as the probability of each spinconfiguration. The two other six non-zero terms in (2.16) vanish in the spin-integration.
The pair-function components in the simple case of a system described by one determinant of spin-orbitals [c.f. Eqs. (2.37) and (2.36)] are (with the similar terms obtained
by )
(r1 , r2 ) = P (r1 )P (r2 ) P (r2 ; r1 )P (r1 ; r2 ),
(r1 , r2 ) = P (r1 )P (r2 ).

(2.18)
(2.19)

From these results we can begin to construct a picture on electron correlation, keeping the
one-determinant approximation in mind: From the latter equation we see that there is no
correlation between the positions of electrons of opposite spins, whilst electrons of like spin
are correlated as their pair function vanishes for r2 r1 . This special type of correlation,
referred to as Fermi correlation, prevents electrons of like spin to coincidence in space.
If we leave the one-determinant approximation and suppose that is an exact manyelectron wave function, the following ramifications for electron correlation hold.
2
for r2 r1 , giving zero probability
1. (r1 , r2 ) and (r1 , r2 ) both vanish like r12
of finding two electrons of like spin at the same point in space. This follows from
the antisymmetry of and the resulting antisymmetry of , (r1 , r2 ; r1 , r2 ) and
, (r1 , r2 ; r1 , r2 ) in both pairs of variables; and may be regarded as the most
general statement of the Pauli exclusion principle.

25

2. By writing (r1 , r2 ) = P (r1 )P (r2 )[1 + f (r1 , r2 )] we may define the correlation
hole or the Coulomb hole

P (r1 ) = P (r1 )f (r1 , r2 ),

P (r2 )
i.e., the difference between the probability of finding a spin-up electron in r1 when
one spin-up electron is at r2 and the same probability in the absence of the second
electron. The form of the correlation factor f is known for dilute electron gas, but
unknown in the case of complicated potential present in a molecular system. Clearly,
the Fermi correlation requires that f (r1 , r2 ) 1 for r2 r1 . This hole
integrates to 1:
Z
P (r1 )f (r1 , r2 )dr1 = 1 r2 .

3. Similarly, if we write
(r1 , r2 ) = P (r1 )P (r2 )[1 + f (r1 , r2 )]
then the correlation hole integrates to zero:
Z
P (r1 )f (r1 , r2 )dr1 = 0 r2 .
The probability should be decreased (i.e. f < 0) near the reference electron due to
the Coulomb repulsion but enhanced further away.
4. The hole function is also known to have a discontinuity of slope, or cusp for r1 r2 .
See Section 2.2.5 for further discussion.

2.2.5

Example: Coulomb and nuclear cusps in He

The non-relativistic Hamiltonian of the helium-like atom with the origin at the nucleus,
1
= 1 21 1 22 Z Z +
H
2
2
|r1 | |r2 | |ri r2 |
has singularities for r1 = 0, r2 = 0, and for r1 = r2 . At these points, the exact solution
to the Schrodinger equation must provide contributions that balance the singularities such
that the local energy

H(r
1 , r2 )
(2.20)
(r1 , r2 ) =
(r1 , r2 )
remains constant and equal to the eigenvalue E. The only possible source for this
balancing is the kinetic energy. It is convenient to employ the symmetry of the helium
26

atom and express the kinetic energy operator in terms of three radial coordinates r1 , r2
and r12 , such that the Hamiltonian is written as
  2

2 
1 X 2
2

2Z
1
2

H=

+
+

+
2
2 i=1 ri2 ri ri
ri
r12
r12 r12 r12



r2 r21
r1 r12

r1 r12 r1 r2 r21 r2 r12

(2.21)

The terms that multiply 1/r12 at r12 = 0 must vanish in H,

which imposes a condition

1

(2.22)
= (r12 = 0)

r12 r12 =0 2

on the wave function. Similarly, the singularities at the nucleus are now seen to be balanced
by the kinetic energy terms proportional to 1/ri :


= Z(ri = 0).
(2.23)
ri ri =0

Eq. (2.22) describes the situation when the electrons coincide in space and is referred to as
the Coulomb cusp condition; whereas Eq. (2.23) establishes the behavior of the ground-state
wave function in the vicinity of the nucleus and is known as the nuclear cusp condition.
Expanding the ground-state helium wave function around r2 = r1 and r12 = 0 we obtain




(r1 , r2 , 0) = (r1 , r1 , 0) + (r2 r1 )
+ r12
+ ...,
r2 r2 =r1
r12 r2 =r1
which gives, when the cusp condition (2.22) is applied

1

+ |r2 r1 |(r1 , r1 , 0) + . . .
(r1 , r2 , 0) = (r1 , r1 , 0) + (r2 r1 )

r2 r2 =r1 2

Therefore, the cusp condition leads to a wave function that is continuous but not smooth
(discontinuous first derivative) at r12 .
The nuclear cusp condition for the first electron when the wave function does not
vanish at r1 = 0 (such as the helium ground state) is satisfied if the wave function exhibits
an exponential dependence on r1 close to the nucleus:
(r1 , r2 , r12 ) = exp(Zr1 )(0, r2 , r12 ) (1 Zr1 )(0, r2 , r12 ).
Molecular electronic wave functions are usually expanded in simple analytical functions
centered on the atomic nuclei (AOs), and close to a given nucleus, the behavior of the
wave function is dominated by the analytical form of the AOs. In particular, the Slatertype orbitals, introduced in Section 6.1.4, are compatible with the nuclear cusp condition,
while the Gaussian-type orbitals are not.
27

It should be noted that the cusp conditions in Eqs. (2.22) and (2.23) are written for the
totally symmetric singlet ground state of the helium atom. The cusp conditions in a more
general situation (but still for a wave function that does not vanish at the singularities)
should be written as


1
lim
=
(rij = 0)
rij 0
rij ave
2



= Z(ri = 0),
lim
ri 0
ri ave

(2.24)
(2.25)

where the averaging over all directions is implied.

2.2.6

Electron correlation summary

Let us at this stage review some nomenclature of electron correlation used in molecular
electronic-structure theory.
A commonly-used concept, correlation energy [7], has a pragmatic definition
Ecorr = Eexact EHF ,
where, in a given one-electron basis, EHF is a best one-determinant total energy and
Eexact the exact energy featuring all possible Slater determinants of the system in
the given one-electron basis. This definition is most usable when speaking of molecular ground states and equilibrium geometries, while outside them it is untenable.
Fermi correlation arises from the Pauli antisymmetry of the wave function and is
taken into account already at the single-determinant level.
Static correlation, also known as near-degeneracy or nondynamical correlation, arises
from the near-degeneracy of electronic configurations.
Dynamical correlation is associated with the instantaneous correlation among the
electrons arising from their mutual Coulombic repulsion. It is useful to distinguish
between
long-range dynamical correlation and
short-range dynamical correlation, which is related to the singularities in the
Hamiltonian and correspondingly to the Coulomb cusp in the wave function.
28

2.3

Electron density

2.3.1

Electronic energy by density

The individual expectation values for terms in the electronic Hamiltonian obtain the following expressions using the general density functions:
+
*
Z
X
hi
1 (x1 ; x )dx1
=
h
(2.26)
1
*

x1 =x1

Vi

gij

ij

=
=

x1 =x1

x1 =x1

V1 (x1 ; x1 )dx1 =
Z

x2 =x2

V1 (x1 )dx1

g12 (x1 , x2 ; x1 , x2 )dx1 dx2

(2.27)
=

g12 (x1 , x2 )dx1 dx2(2.28)

.

P
P
The latter expressions for h i Vi i and h ij gij i are obtained by noting that V1 = V (r1 ) is
just a factor in the integrand, as is g12 , and thus the primes may be omitted. Therefore,
the general expression for the N -electron system with the usual molecular Hamiltonian
Eq. (1.6) takes a simple form
1
E=
2

(x1 )dx1 +

K Z
X
I=1

|r1 RI |

1
(x1 )dx1 +
2

|r1 r2 |1 (x1 , x2 )dx1 dx2 ,

(2.29)

or, by carrying out the spin-integration,

1
E =
2

P (r1 )dr1 +

= T + Ven + Vee .

K Z
X
I=1

|r1 RI |

1
P (r1 )dr1 +
2

|r1 r2 |1 (r1 , r2 )dr1 dr2

(2.30)

It should be emphasized that these results are valid for all kinds of states and their approximate wave functions of any system. Furthermore, for evaluation of molecular energies and
properties, we do not even need to construct the N -electron wave function, but everything
is exactly described by the electron densities.

2.3.2

FockDirac density matrix

Let us carry out the same for a single Slater determinant , that we recall to provide an
useful approximation for the ground state wave function, by considering N spin-orbitals
having the orbital factor corresponding to an occupied MO. The energy expression is given
by
D   E
X D   E 1 X
 
 r +
r h
E = H  =
(hr s |
g | r s i hr s |
g | s r i)(2.31)
.
2 rsocc
rocc
29

Assume that we may expand (x1 , x1 ) in spin-orbitals as

X
(x1 , x1 ) =
rs r (x1 )s (x1 ),

(2.32)

rs

where the rs are numerical coefficients; then the expectation value (2.26) can be written
as the trace of a matrix product:
+
*
D   E
X
X
 
i =
rs s h
h
(2.33)
 r = Tr[h],
i

rs

and by comparing to the general expression, we observe that in the one-determinant approximation RS = RS . The one-determinant approximation to (x1 ; x1 ) may now be
written as
X
(x1 ; x1 ) =
r (x1 )r (x1 ).
(2.34)
rocc

The expression for follows in a similar manner, and we find

X
(2.35)
[r (x1 )s (x2 )r (x1 )s (x2 ) r (x2 )s (x1 )r (x2 )s (x1 )]
(x1 , x2 ; x1 , x2 ) =
rs

= (x1 ; x1 )(x2 ; x2 ) (x2 ; x1 )(x1 ; x2 ).

(2.36)

The factorization of the two-electron density matrix in terms of the one-electron is characteristic to the one-determinant approximation: it means that in this approximation
everything is determined by the function (often called as the FockDirac density matrix).
So far, we have included spin implicitly through the use of spin-orbitals. In the onedeterminant case the we may write spin explicitly, for the case x1 = x1 ,
(x1 ) = P (r1 )(1 ) (1 ) + P (r1 )(1 ) (1 ),
where the densities for spin-up and spin-down electrons are
X
X
P (r1 ) =
s (r1 )s (r1 ),
P (r1 ) =
s (r1 )s (r1 ).
s()

(2.37)

(2.38)

s()

By integrating over spin, we obtain the electron density in the form

P (r1 ) = P (r1 ) + P (r1 )

(2.39)

that is, a sum of spin-up and spin-down densities. The excess of spin-up density over
spin-down density,
1
(2.40)
Qz (r1 ) = [P (r1 ) P (r1 )] ,
2
is an useful quantity, which is in fact the density of spin angular momentum around the zaxis. P (r) and Qz (r) are enough to determine all the properties of the electron distribution
in the one-determinant approximation.
30

2.3.3

Example: Closed-shell energy

Let us finally consider an important special case, a closed-shell system, in which all spinorbitals are doubly occupied, once with and once with electron; meaning that P =
P = P/2 and the spin density is everywhere zero. In this case,
X
P (r1 ) = 2
s (r1 )s (r1 ),
(2.41)
socc

and by substituting Eq. (2.37) to Eq. (2.36) we find

1
(r1 , r2 ) = P (r1 )P (r2 ) P (r2 ; r1 )P (r1 ; r2 ).
2

(2.42)

Substituting these expressions to Eq. (2.30) we obtain an often-used expression for the
closed-shell energy:
X D   E X
 r +
[2 hr s |
g | r s i hr s |
g | s r i] .
(2.43)
E=2
r h
r

2.3.4

rs

Other density functions

Other property densities, in addition to the charge density P (r) and the spin-density Qz (r),
could be defined in the same way. For example, referring to Eq. (2.30),

 2
p

P (r; r )
(
p2 = p2x + p2y + p2z )
T (r) =
2

r =r
would be a kinetic-energy density, since T (r)dr is the contribution to hT i associated with
volume element dr at r. However, point contributions to real observables are not necessarily
real, and the proper definition of the kinetic-energy density is


p

T (r) =
P (r; r )
.
(2.44)
2
r =r
where the adjoint operator is understood to work on the primed variable only.
Another often encountered density function is the current density
J (r) =


1
(
p + p )P (r; r ) r =r = Re [
p P (r; r )]r =r ,
2

(2.45)

where denotes the Cartesian x, y, z components. For a system in a stationary state

described by a real wave function J = 0, but for a system in the presence of a magnetic
field, circulating currents are always present. A more general form of J is obtained on
introducing field terms in the Hamiltonian via the gauge-invariant momentum operator,

=p
+ A,
p
31

(2.46)

where A is the magnetic vector potential from the applied field is derived. The corresponding current density is
J (r) = Re [
P (r)] .
(2.47)
The kinetic energy density is also modified by the presence of the field, and (2.44) becomes
"

T (r) =
P (r; r )
2

.
r =r

Finally; it may be shown that for a wave function that satisfies the time-dependent
Schrodinger equation the charge density and the current density satisfy the conservation
equation
P (r)
.
(2.48)
J(r) =
t
When is an exact stationary-state function, P is time-independent and the net flux of
density out of any region is zero; but with A 6= 0 there will be in general a steady-state
distribution of non-zero currents, satisfying the usual continuity equation J = 0.

2.4
2.4.1

On the optimization of the wave function

Variation method

According to the variation principle, the solution of the time-independent Schrodinger

|i = E |i is equivalent to an optimization of the energy functional
equation H
D   E

E[] = D  E ,

(2.49)

 E

is some approximation to the eigenstate |i. It provides a simple and powerful
where 
procedure for generating approximate wave functions: for some proposed model for the
wave function, we express the electronic state |Ci in terms of a finite set of numerical
parameters; the stationary points of energy function
D   E
 
C H
C
E(C) =
(2.50)
hC |Ci
are the approximate electronic states |Ci and the values E(C) at the stationary points
the approximate energies. Due to the variation principle, the expectation value of the
Hamiltonian is correct to second order in the error.
32

2.4.2

Electronic gradient and Hessian

A simple realization of the variation method is to make a linear ansatz for the wave function,
|Ci =

m
X
i=1

Ci |ii ,

i.e. the approximate state is expanded as a linear combination of m Slater determinants.

We further assume here that the wave function is real. The energy function of this state is
given by Eq. (2.50). In order to locate and to characterize the stationary points, we shall
employ the first and second derivatives with respect to the variational parameters:
D   E
 
i H
 C E(C) hi |Ci
E(C)
(1)
Ei (C) =
=2
Ci
hC |Ci
D   E
 
i H
 j E(C) hi |ji
hj |Ci
hi |Ci
2 E(C)
(1)
(1)
(2)
2Ei (C)
2Ej (C)
,
=2
Eij (C) =
Ci Cj
hC |Ci
hC |Ci
hC |Ci
known as the electronic gradient and electronic
respectively.
E
D   Hessian,
 
The condition for stationary points, i H  C = E(C) hi |Ci is in a matrix form equal
D   E
 
to HC = E(C)SC, where the Hamiltonian and overlap matrix are given by Hij = i H
j
and Sij = hi |ji. Assuming that S = 1, we have to solve a standard m-dimensional
eigenvalue problem with m orthonormal solutions CK = (C1K C2K CmK )T , CTK CK =
KL , with the associated real eigenvalues EK = E(CK ), E1 E2 P
Em .
The eigenvectors represent the approximate wave functions |Ki = M
i=1 CiK |ii with a
corresponding approximate energy EK . To characterize
the
stationary
points,
we note that
D   E
 
K (2)
the Hessian is at these points Eij (CK ) = 2( i H  j Ek hi |ji). We may also express
the Hessian in the basis formed of the eigenvectors:
D   E
 
(2)
K (2)
EM N = 2( M H
 N EK hM |N i) = 2(EM EN )M N K EM M = 2(EM EK ).

(2)

EM M is thus singular and K th state has exactly K 1 negative eigenvalues. Therefore,

in the space orthogonal to CK , the first solution is a minimum, the second a first-order
saddle point, the third a second-order saddle point, and so forth.

2.4.3

HellmannFeynman theorem

Many of the theorems for exact wave functions hold also for the approximate ones that
are variationally determined. One of the most important is the HellmannFeynman theorem, which states that the first-order change in the energy due to a perturbation may be
calculated as the expectation value of the perturbation operator V : Let | i be the wave
33

 + V and |i unperturbed wave function at = 0. Then

function associated with H
 E
D 






  D   E

H
+

dE() 
 



= 2 Re
=
H E(0) + V 


d =0

h | i


=0
=0
D   E
 
= V  .
(2.51)

In fact, for the HellmannFeynman theorem to hold, we need to demand from the
approximate wave functions that they are optimized with respect to the changes induced by
the perturbation: |i |i + |/i. Usually in molecular calculations we construct
our wave function from a finite set of analytical functions attached to the atomic nuclei.
For example, when we distort the molecular geometry, we change the basis set in terms
of which our electronic wave function is expanded, and a wave function optimized at one
particular geometry is not accurately represented in terms of a basis set associated with
another geometry. As a result, the unperturbed electronic is not variational with respect
to these changes and the conditions of the HellmannFeynman theorem is not satisfied.
However, this is an artificial difficulty arising from the coordinate representation of
quantum mechanics we presently operate in. The picture presented by the second quantization, the necessary conditions are fulfilled and we are able to employ the theorem for
approximate wave functions in finite bases in exactly the same manner as for exact wave
functions.

2.4.4

The molecular electronic virial theorem

The exact molecular energy is variational with respect to an arbitrary change in the wave
function, particularly of interest are the uniform, normalization-preserving scaling of the
electron coordinates
(ri ) (ri ) = 3N/2 (ri ).

Let us partition the electronic Hamiltonian in a bit different manner than earlier:

H(R)
= T + V (R)
(2.52)
X
1
2i
(2.53)
T =
2 i
X
X
X ZI ZJ
1
ZI
V (R) =

+
,
(2.54)
|ri rj |
|ri RI | I>J |RI RJ |
i>j
iI

and applying this decomposition we may find the stationary points of the energy function
are found at1
 Ei
 E
D 
D 
h 2 D   E




0 =
=
H(R) 
T  + V (R) 

=1
=1
  

 
 E
D   E D 
 dV (R)  

dE(R)
HF


 


 (2.55)


.
2 T  + V (R) = 
=

d =1 
d =1
1

Note that the unperturbed wave function corresponds here to = 1 instead of = 0.

34

This result (2.55) is the quantum-mechanical virial theorem for a field-free non-relativistic
molecular Hamiltonian. The most important observation is that in the molecular equilibrium geometries (R = Re ), and in the case of atoms
 E
D   E
1 D 

 
T  = V (Re ) .
2

Furthermore,

 E
D 


V (Re ) = 2E(Re ).

D   E
 
T |R=Re = E(Re ),

(2.56)

(2.57)

Also these expressions hold for any approximate wave function that is variational with
respect to a uniform scaling of the nuclear as well as electronic coordinates. Note that
according to the last result, no stationary points of positive energy may exist.
The scaling force may be easily related to classical Cartesian forces on the nuclei,
FI (R) = dE(R)/dRI , by invoking the chain rule


X d(RI ) 
X

dE(R) 
dE(R
)
I


=
=

RI FI (R)


d =1
d
d(R
I
=1
=1
I
I

and combining this with virial theorem, we may extract the Cartesian forces experienced
by the nuclei,
 E
D   E D 
X


 
(2.58)
RI FI (R) = 2 T + V (R) .
I

2.5

Further reading

D. P. Tew, W. Klopper, and T. Helgaker, Electron correlation: the many-body problem

at the heart of chemistry, J. Comput. Chem. 28, 1307 (2007)

35

36

Chapter 3
The HartreeFock theory
3.1

The HartreeFock approximation

In the HartreeFock approximation (HF), the electronic wave function is approximated by

a single Slater determinant, and the energy is optimized with respect to variations of these
spin-orbitals.

3.1.1

The HF wave function

In the HF framework, the approximate wave function may be written in a form

|i = exp(
) |0i ,

(3.1)

where |0i is some reference configuration and exp(

) an operator that carries out unitary
transformations among the spin orbitals; these orbital-rotation parameters for the Hartree
Fock ground state are obtained by minimizing the energy:
D   E
 
EHF = min H
(3.2)
 .

The optimization could be carried out using standard numerical analyisis; however, for
the most of purposes more spesific methods are required in order to cope with the
computational cost.

3.1.2

The HF equations and the Fock operator

We recall that the Slater determinant represents a situation where electrons behave as
independent particles, but are subject to the Fermi correlation. Therefore, the optimal
determinant can be found by solving a set of effective one-electron Schrodinger equations
the HartreeFock equations for the spin-orbitals, with the associated Hamiltonian being
the Fock operator
X
f =
fP Q aP aQ ,
(3.3)
PQ

37

where the elements fP Q are called the Fock matrix. A very essential feature of the Fock
operator is that the one-electron part of the true molecular Hamiltonian is retained,
+ V .
but the two-electron part is replaced with the effective Fock operator, so that f = h
The Fock operator incorporates in an average sense the Coulomb interaction among the
electrons, corrected for Fermi correlation:
V =

VP Q aP aQ

(3.4)

(gP QII gP IIQ ) ,

(3.5)

PQ

VP Q =

X
I

where the integrals g are given in Eq. (A.33), with I running over the occupied and P and
Q all spin-orbitals. The first term in VP Q describes the classical Coulomb interaction of the
electron with the charge distribution caused by the others, and the second one, exchange
term, is a correction that arises from the antisymmetry of the wave function.

3.1.3

Self-consistent field method

The HartreeFock equations are solved by diagonalizing the Fock matrix. The resulting
eigenvectors are called the canonical spin-orbitals and the the eigenvalues of the Fock
matrix are the orbital energies, fP Q = P Q P .
Since the Fock matrix is defined in terms of its own eigenvectors, the canonical spinorbitals and the orbital energies can only be obtained using an iterative procedure; where
the Fock matrix is constructed and diagonalized in every iteration until the spin-orbitals
obtained in a particular iteration become identical to those from which the Fock matrix
was constructed. This procedure is referred to as the self-consistent field (SCF) method
and the resulting wave function the SCF wave function.
In the canonical representation, i.e., in the frame where the Fock matrix is diagonal,
the electrons occupy the spin-orbitals in accordance with the Pauli principle and move
independently one another in the electrostatic field caused by the stationary nuclei and by
the charce distribution of the N 1 electrons. The orbital energies are the one-electron
energies associated with the independent electrons and may thus be interpreted as the
energies required to remove a single electron, that is, to ionize the system.
The identification of the orbital energies with the negative ionization potentials is known
as the Koopmans theorem.
The HartreeFock state is invariant to unitary transformations among the occupied
spin-orbitals. Therefore, the spin-orbitals of the HartreeFock state are not uniquely determied by the condition (3.2) and the canonical orbitals are just one possible choice of
spin-orbitals for the optimized N -particle state. Any set of energy-optimized orbitals decomposes the Fock matrix into two non-interacting blocks: one for the occupied, and
another for the unoccupied spin-orbitals. When these subblocks are diagonalized, the
canonical spin-orbitals are obtained.
38

Figure 3.1: The self-consistent field procedure.

The final electronic state is obtained as an antisymmetrized product of the canonical
spin orbitals
!
Y
(3.6)
ai ai |vaci .
|HFi =
i

Finally, let us make a remark that the HartreeFock state is an eigenfunction of the Fock
operator with an eigenvalue equal to the sum of the orbital energies of the canonicial
spin-orbitals
X
f |HFi = 2
i |HFi .
(3.7)
i

3.1.4

Restricted and unrestricted HartreeFock theory

The exact wave function is an eigenfunction of the total and projected spin operators. The
HartreeFock wave function, which is not an eigenfunction of the Hamiltonian (but of the
effective one, the Fock operator), does not possess these symmetries, they must be imposed
on the HartreeFock solution.
In the restricted HartreeFock (RHF) approximation, the energy is optimized subject
to the condition that the wave function is an eigenfunction of the total and projected
spins. In a practical implementation of the RHF scheme, it is advantageous to write
the wave function as a configuration state function rather than a Slater determinant.
In the unrestricted HartreeFock (UHF) approximation, the wave function is not
required to be a spin eigenfunction, and different spatial orbitals are used for different
spins.
39

 In some systems, usually close to the equilibrium geometry, the symmetries of the
exact state are present in the UHF state as well. In such cases the UHF and RHF
states coincide.

3.1.5

Example: HartreeFock treatment of H2

For the hydrogen molecule, the RHF wave function is a singlet spin eigenfunction with
doubly occupied symmetry-adapted 1g orbitals;
|RHFi = a1g a1g |vaci

|UHFi = a1 a2 |vaci .

Let us plot these wave functions as well as the exact wave function for comparison using
the LCAO approximation for the orbital basis:

At short internuclear distances the RHF and UHF wave functions are identical, providing a crude but reasonable representation of the potential energy surface. At larger
separations, the RHF wave function behaves poorly for energy, whereas the UHF wave
function falls towards the correct FCI curve. On the other hand, while the RHF wave
function remains a true singlet at all distances, the UHF wave function behaves incorrectly
with respect to the spin, yielding a spin intermediate of a singlet and a triplet. This effect
is referred to as the spin contamination.

3.1.6

RoothaanHall equations

We will now consider the classical RoothaanHall formulation

P[8, 9] of the HartreeFock
theory, in which the MOs are expanded in a set of AOs, p = Cp , whose expansion
coefficients C are used as the variational parameters.
The HartreeFock energy is the expectation value of the true molecular electronic
Hamiltonian (A.62), and is given in the orbital basis by
E

(0)

 
E X
1X
 
hpq Dpq +
= HF H  HF =
gpqrs dpqrs
2 pqrs
pq
D

40

(3.8)

where
Dpq =

 
E
 
HF Epq  HF

dpqrs = hHF |
epqrs | HFi .

(3.9)
(3.10)

For a closed-shell state, the only nonzero elements of (3.9) and (3.10) are
Dij = 2ij
1
dijkl = Dij Dkl Dil Dkj = 4ij kl 2il kj ,
2
and thus the total HartreeFock energy is
X
X
E (0) = 2
hii +
(2giijj gijji ).
i

(3.11)

ij

The energy must be optimized subject to the orthonormality of the MOs. We introduce
the HartreeFock Lagrangian
X
L(C) = E(C) 2
ij (hi |j i ij )
ij

and the variational conditions are written in terms of it as

0=

X
X
L(C)
= 4hk + 4
(2gkjj gjjk ) 4
Sj jk ,
Ck
j
j

which gives the condition for the optimized HartreeFock state

X
fk =
Sj jk .

(3.12)

Since the multiplier matrix is symmetric (for real orbitals), it can be diagonalized by
an orthogonal transformation among the occupied orbitals, = UUT , the HartreeFock
energy being invariant to such transformations. The off-diagonal Lagrange multipliers may
thus be eliminated by an orthogonal transformation to a set of occupied MOs that satisfy
the canonical conditions. In the canonical basis, the variational conditions become
X
X
AO
f
Ck = k
S Ck ,
(3.13)

where the elements of the AO Fock matrix are

X
AO
f
= h +
(2gii gii ).
i

41

(3.14)

Figure 3.2: RoothaanHall procedure.

Since f AO is symmetric, we can extend the variational conditions to yield a set of orthonormal virtual MOs that satisfy the same canonical conditions as the occupied MOs, i.e.,
fab = ab a . Now, we are able to write the HartreeFock variational conditions as
f AO C = SC,

(3.15)

where is a diagonal matrix that contains the orbital energies. This equation is called the
Roothaan-Hall equations.
The AO Fock matrix can be evaluated entirely in the AO basis,


X
1
AO
AO
f = h +
D g g ,
(3.16)
2

where DAO is the AO representation of the one-electron density,

DAO = CDCT .

3.2
3.2.1

(3.17)

Further topics in HF self-consistent field theory

On the convergence of the SCF method

A straightforward implementation of the RoothaanHall SCF method (given in the figure) may fail to converge or may converge slowly. Several schemes to improve the SCF
42

convergence have been introduced; of which the direct inversion in the iterative subspace
(DIIS) method [10,11] is amongst the simplest and one of the most successful. In the DIIS
method, the information from the preceding iterations is used instead of generating the
next density from the last AO Fock matrix:
f AO =
n

n
X

wi fiAO ,

(3.18)

i=1

where the DIIS

P weights w are obtained by minimizing the norm of the averaged error
n = ni=1 wi ei ,
vector e
n X
n
X
2
kek =
wi hei |ej i wj ,
| {z }
i=1 j=1

subject to the constraint that

Bij

Pn

wi = 1. For this purpose, we construct the Lagrangian

!
n
n
X
X
wi 1 ,
L=
wi Bij wj 2
i=1

i=1

i,j=1

the minimization of which

B11
B21
.
..

B
n1
1

leads to the set of linear equations:

B12 B1n 1
w1
0
B22 B2n 1 w2 0
. .
..
..
..
...
. .
.
.
.
. = .
Bn2 Bnn 1 wn 0
1
1 1
0

from which the weights are solved.

P
The C2-DIIS method [12] is similar to the DIIS, but the constraint is ni=1 wi2 = 1.
This leads to a solution of an eigenvalue problem instead of the set of linear equations as
in DIIS; the scheme may handle singularities better.
In the EDIIS (energy DIIS) scheme [13], the error-vector-norm minimization is replaced by a minimization of an approximation to the true energy function. It improves the
convergence in cases where the start guess has a Hessian structure far from the optimal
one. It also can converge in cases where DIIS diverges. For non-problematic cases EDIIS
has actually a slower convergence rate than DIIS.
These are based on either the diagonalization of the Fock matrix or on a direct optimization of the AO density matrix. Nevertheless, in some cases, even the DIIS method
(and alike) is difficult to converge, when it might be better to use the second-order Newton
method to the optimization of the HartreeFock energy. This has been applied in the
literature to the optimization of the AO density matrix (the basis for the trust-region or
restricted-step method) as well as in a method that carries out rotations among the MOs.
The trust-region SCF (TRSCF) [14] is basically a globally convergent black-box
method, where each SCF iteration consists of a trust-region density subspace minimization
and a trust-region Roothaan-Hall step.
43

100
10

SCF
DIIS

Eh

1
SCF
DIIS

0.1
0.01

0.001
1e-04
1e-05
10
Iteration

20

10
Iteration

20

Figure 3.3: On the left: H2 O in its equilibrium geometry, on the right: H2 O with its
bonds stretched to double (angle remaining constant). Calculations are carried out in the
cc-pVQZ basis.

3.2.2

Integral-direct SCF

In every RoothaanHall SCF iteration a large number of two-electron integrals is needed

to construct the Fock matrix. Since the same AO integrals are used in each iteration, these
integrals can be written to disk and read in when needed.
In large-scale calculations, this approach becomes impractical due to the large number
of needed integrals. In such cases, the AO integrals are instead recomputed in every
iteration as soon as a batch of two-electron integrals has been produced, it is contracted
with the appropriate densities. This procedure is known as the direct SCF method [15].
If the value of an integral is very small, or if it is combined with small density-matrix elements, its contribution to the Fock matrix may be neglected. Therefore, by pre-estimating
the integrals by having some kind of upper limit for them, the number of calculated integrals in the direct SCF scheme can be reduced dramatically. The computational cost of the
SCF method scales formally as O(n4 ), where n is the number of basis set functions, but it
can be shown that only O(n2 ) of the two-electron integrals are significant. Therefore, the
direct SCF scheme combined with the pre-screening of the integrals is much cheaper for
large systems.

3.2.3

Linear scaling aspirations

It is necessary to reduce the formal scaling of the models in order to apply them to larger
molecular systems. This have been under intense study for many years, and significant
progress have been achieved in the cases of SCF (and density-functional theory), which are
based upon
The fast multipole method [16]: the number of significant two-electron integrals scales
44

as O(N 2 ), but only O(N ) have to be evaluated using the standard quantum-chemical
means the rest can be approximated by far simpler formulas. This applies to AOs
whose overlap is close enough to zero.
Linear scaling Fock-matrix construction: the contributions to the Fock matrix in the
AO basis
X
AO
f = h +
D
(2g g )
(3.19)

can be prescreened by the AO density matrix.

Direct optimization of the density matrix [14]: the diagonalization step in the Roothaan
Hall SCF is avoided. These involve plenty of matrix-matrix multiplications, but the
involved matrices are hoped to be sparse enough (in spatially large molecules) to
allow O(N ) evaluation of the multiplications.

3.3

MllerPlesset perturbation theory

Finally, we consider less rigorous but computationally more appealing improvements of the
HartreeFock description, based on perturbation theory. They are rather successful, when
the HartreeFock wave function is reasonably accurate.
In MllerPlesset perturbation theory (MPPT) [17, 18] the electronic Hamiltonian is
partitioned as
= f +

H
(3.20)
the fluctuation potential, which is the difference
where f is the Fock operator (3.3) and
and the effective one-electron pobetween the true two-electron Coulomb potential in H
= g V . The zero-order state is represented by the
tential V of the Fock operator,
P
HartreeFock state in the canonical representation, that is f |HFi = I I |HFi, and by
applying the standard machinery of perturbation theory, we obtain to second order in the
perturbation
 
E X
D
 
(0)
I
(3.21)
EMP = HF f  HF =
(1)
EMP
(2)

 
D
E

= HF 
 HF

EMP =

A>B,I>J

|gAIBJ gAJBI |2
.
A + B I J

(3.22)

(3.23)

Thus, the HartreeFock energy is equal to the sum of the zero- and first-order contributions; and by adding the second-order correction, we obtain the second-order MllerPlesset
energy (MP2)
X |gAIBJ gAJBI |2
EMP2 = EHF
,
(3.24)

A + B I J
A>B,I>J
45

Figure 3.4: The MPn dissociation curves of the C2v water molecule in the cc-pVDZ basis.
Restricted MP2 with full and unrestricted with dashed line.
which is very popular and successful approach to introducing electron correlation on top of
the HartreeFock wave function; providing surprisingly accurate and size-extensive correction at low computational cost. Higher-order corrections are derived in a similar manner,
yielding MP3 and MP4 models, but these become rather expensive and are not as appealing compromise between cost and accuracy as MP2 is. Furthermore, the MPPT series have
been shown to be inherently divergent.

3.4

Further reading

T. Helgaker, P. Jrgensen, J. Olsen, Molecular Electronic-Structure Theory (Wiley

2002), pp. 433522.
J. Almlof, Notes on HartreeFock theory and related topics, in B. O. Roos (ed),
Lecture Notes in Quantum Chemistry II (Springer-Verlag 1994).
46

 J. Almlof, Direct methods in electronic structure theory, in D. R. Yarkony (ed),

Modern Electronic Structure Theory (World Scientific 1995), p. 110.

47

48

Chapter 4
Configuration interaction
4.1

Configuration-interaction wave function

The configuration-interaction (CI) wave function is constructed as a linear combination of

Slater determinants (or configuration-state functions):
X
|Ci =
Ci |ii
(4.1)
i

with the coefficients determined by a variational optimization of the expectation value of

the electronic energy, i.e., by minimization for the ground state
D   E
 
C H
C
.
(4.2)
ECI = min
C
hC |Ci

As discussed earlier, this is equivalent to the solution of an eigenvalue equation HC =

ECI C. The construction of the DCI wave
 Efunction may thus be carried out by diagonalizing
 
the Hamiltonian matrix Hij = i H  j , but more often iterative techniques are used.
In particular, we have to construct the CI wave function in a finite one-electron basis.
The full CI (FCI) wave function consists of all the configurations that can be generated
in a given basis, and it is most convenient to think it as generated from a single reference
configuration, which dominates the wave function and usually is the HartreeFock state,
by the application of a linear combination of spin-orbital excitation operators
!
X
X
1
AB
IA +
IJ
|FCIi = 1 +
X
X
+ |HFi ,
(4.3)
2
AI
ABIJ
where for example
IA |HFi = CIA a aI |HFi
X
A
AB
AB

XIJ |HFi = CIJ aA aB aI aJ |HFi .

49

RHF
CISD
CISDT
CISDTQ
CISDTQ5

R = Rref
E EFCI Weight
0.2178
0.9410
0.0120
0.9980
0.0090
0.9985
0.0003
1.0000
0.0001
1.0000

R = 2Rref
E EFCI Weight
0.3639
0.5896
0.0720
0.9487
0.0560
0.9590
0.0058
0.9988
0.0022
0.9995

Thus, we may characterize the determinants in the FCI expansion as single (S), double (D),
triple (T), quadruple (Q), quintuple (5), and so forth excitations relative to the Hartree
Fock state.
The number of determinants in an FCI wave function is
 2 
2
n!
n
Ndet =
=
,
(4.4)
k
k!(n k)!
for a system with n orbitals containing k alpha and k beta electrons. For example, when
n = 2k, Ndet 16k /k for large k, the number of determinants increases by a factor of
16 for each new pair of electrons and orbitals!

4.1.1

Single-reference CI wave functions

As the FCI wave function is obtainable only for the smallest systems, it becomes necessary
to truncate the expansion, so that only a subset of the determinants are included, based
on the excitation level. This is justified, since in general the lower-order excitations are
more important than those of high orders. In principle, this procedure of hierarchal truncation may be continued until the FCI wave function is recovered. Since the CI model is
variational, the FCI energy is approached from the above.
The electronic energies of truncated CI wave functions for the water molecule (cc-pVDZ
basis) relative to the FCI energy:
Whereas the contributions drop monotonically, the step from an even-order excitation
level to an odd-order is far more pronounced than the step from an odd to an evenorder excitation level; hence the CI expansion is preferably truncated at even orders. For
example, the first useful truncated CI wave function the CI singles-and-doubles wave
function (CISD) [19] recovers 94.5% of the correlation energy, the inclusion of triples on
top of that improves the treatment to 95.9%, while as much as 99.9% of the correlation
energy is recovered at the CISDTQ level.

4.1.2

Multi-reference CI wave functions

In the stretched geometry 2Rref , the HartreeFock determinant is less dominant. Clearly,
the strategy of defining a correlation hierarchy in terms of excitations out of a single
50

Figure 4.1: CI models

51

Figure 4.2: Performance of the MRSDCI model.

reference determinant such as the HartreeFock determinant does not provide the best
possible description of the dissociation process. To overcome this, we may introduce CI
wave functions based on the idea of a reference space comprising more than a single determinant: The multi-reference CI (MRCI) wave functions is generated by including in the
wave function all configurations belonging to this reference space as well as all excitations
up to a given level from each reference configuration, yielding, e.g., the multi-reference
singles-and-doubles CI (MRSDCI) wave function.
The construction of a MRCI wave function begins with the generation of a set of orbitals
and the reference space of configurations. For example, in the water molecule, we would
include in our reference space for instance the configurations obtained by distributing the
eight valence electrons among the six valence MOs in all possible ways (consistent with the
spin- and space-symmetry restrictions).
MRSDCI model is very accurate, if all the important configurations are included in
the reference space.
The number of important configurations becomes, however, very large, making thus
the model applicable only to quite small systems.
The choice of the active space requires a lot of chemical intuition and patience.
Another way of truncating the FCI expansion is to construct the wave function from
individual configurations, the selection of which is based on physical intuition or perturbational estimates.

4.1.3

Optimization of the CI wave function

For large CI expansions it is impossible to set up and diagonalize the Hamiltonian matrix
of the eigenvalue equation. However, usually only a few of the lowest eigenvalues are of
interest, and we may determine them by iterative methods, where the eigenvectors and
eigenvalues are generated by a sequence of linear transformations of the form = HC,
where C is some trial vector. These methods can be devised for both of the CI eigenvalue
problem and the optimization of the CI energy (4.2). The most often used strategies rely
on the second-order Newtons method and its approximations.
52

4.1.4

On the disadvantages of the CI approach

CI wave functions truncated at a given excitation level relative to a reference configuration

do not provide size-extensive energies.
For example, if two fragments are described at the CISD level, then a size-extensive
treatment of the compound system requires that the wave function is augmented
with certain triples and quadruples; more precisely those that represent products of
single and double excitations in the two subsystems.
An approach called quadratic CI (QCI) is a size-extensive revision of the CI model.
Other widely used approach to overcome the lack of size-extensivity is the Davidson
correction applied on top of the truncated CI energy.
The CI description of the electronic system is not at all compact. Even though higher
excitations are less important than those of lower orders, their number is enormous. Therefore, the CI wave function converges slowly with the number of variational parameters.
Both of these problems are due to the linear parameterization in the CI model (4.1).

4.2

Multi-configurational SCF theory

The multi-configurational self-consistent field (MCSCF) theory is a generalization of the

HartreeFock wave function to systems dominated by more than one electronic configuration; that is, cases with pronouced static correlation. This method is particularly useful in
the description of bond breakings and molecular dissociation processes.

4.2.1

MCSCF wave function

In MCSCF theory, the wave function is written as a linear combination of determinants,

whose expansion coefficients Ci are optimized simultaneously with the MOs according to
the variation principle:
X
|, Ci = exp(
)
Ci |ii .
(4.5)
i

The ground-state MCSCF wave function is obtained by minimizing the energy with respect
to the variational parameters,
 
E
D
 
, C H
 , C
EMCSCF = min
.
(4.6)
,C h, C |, Ci

4.2.2

Example: MCSCF wave function of H2

A two-configuration MCSCF wave function of the hydrogen molecule is |MCi = C1 a1g a1g |vaci+
C2 a1u a1u |vaci where the variation principle is invoked to optimize the configuration
53

coefficients as well as the orbitals. The solid line is the potential energy curve in the
cc-pVQZ basis set. The thick and the dashed lines are for FCI and RHF, respectively.
For any larger system than the H2 molecule the selection of configuration space is
the greatest difficulty in the MCSCF calculation. It is very often impossible to generate
a sufficient MCSCF configuration, which is computationally tractable. The selection is
usually carried out by dividing the MO space in inactive, active, and secondary orbitals.
The inactive orbitals are set to be doubly occupied.
The active orbitals have variable occupancies of 0, 1, or 2. The MCSCF expansion
is then obtained by distributing the active electrons in all possible ways among the
active orbitals. In the final optimized state, the active orbitals have non-integer
occupation numbers between 0 and 2.
The secondary orbitals are unoccupied throughout the optimization.
This scheme is called the complete active space SCF method (CASSCF) [20]. We note that
the CASSCF wave function is identical to the FCI wave function, when all the orbitals
of the system are active; and reduces to the HartreeFock wave function when the active
space is empty.
In restricted active space SCF (RASSCF) [21, 22] calculations, the active orbital space
is further divided into three subspaces: RAS1, with an upper limit on the allowed number
of holes in each configuration; RAS2, with no restrictions enforced; and RAS3, with an
upper limit on the allowed number of electrons in each configuration. This allows for larger
active spaces than CASSCF.
In practice, the simultaneous optimization of orbitals and CI coefficients is a difficult
nonlinear optimization problem, which restricts the MCSCF expansions to be significantly
smaller than those encountered in the CI wave functions. By itself, the MCSCF model
is not suited to the treatment of dynamical correlation, that requires large basis sets and
long configuration expansions.
54

4.2.3

Example: Selection of the active space in H2 O

One popular approach for the choice of the MCSCF active spaces is based on the natural
occupation numbers obtained from a preceding second-order MllerPlesset perturbation
theory calculation. For the water molecule in the cc-pVTZ basis in the order (A1 , B1 , B2 ,
A2 ) we obtain
Natural orbital occupation numbers, symmetry 1
1.99953982
0.00648257
0.00047604
0.00008516
0.00002067

1.98393843
0.00552982
0.00036853
0.00007619
0.00001341

1.96476967
0.00113424
0.00035173
0.00005837
0.00000675

0.02261760
0.00083070
0.00019527
0.00004768

0.01156239
0.00060446
0.00013628
0.00003078

0.00068405
0.00005653

0.00064037
0.00004947

0.00104230
0.00013435
0.00001667

0.00086705
0.00011930
0.00001255

0.00012271

0.00006547

Natural orbital occupation numbers, symmetry 2

1.96724668
0.00042732
0.00001646

0.01959000
0.00035178

0.00552930
0.00013127

Natural orbital occupation numbers, symmetry 3

1.96415036
0.00065869
0.00006217
0.00000755

0.02452309
0.00039618
0.00005180
0.00000429

0.00641995
0.00019746
0.00003331

Natural orbital occupation numbers, symmetry 4

0.00626166
0.00006019

0.00079588
0.00001431

0.00038289

We can choose active spaces of different flexibility:

Smallest possible active space seems to be (2,0,2,0), i.e. four electrons in four MOs
3a1 , 4a1 , 1b2 , and 2b2 . This corresponds to a combination of 12 determinants.
For more accurate description, eight electrons distributed on (3,1,2,0) active space,
corresponding to 37 configurations, gives almost similar potential curve.
For further improvement, an eight-electron (6,3,3,1) active space is of high quality
and sufficient for incorporating dynamical correlation effects as well. However, we get
almost the same curve by supplementing the CAS(3,1,2,0) wave function by RAS3
space (3,2,1,0) limiting the number of electrons in RAS3 space to be 2 at maximum.
This reduces the number of determinants from 36240 to 2708.

4.3

Further reading

I. Shavitt, The Method of Configuration Interaction, in H. F. Schaefer III (ed), Methods of Electronic-Structure Theory (Plenum Press 1977)
55

 H.-J. Werner, Matrix-Formulated Direct Multiconfiguration Self-Consistent Field and

Multiconfiguration Reference Configuration-Interaction Methods, in K. P. Lawley
(ed), Ab initio Methods in Quantum Chemistry (Wiley 1987)

56

Chapter 5
Coupled-cluster theory
5.1
5.1.1

Cluster expansion
Coupled-cluster wave function

As noted earlier, the shortcomings of the CI hierarchy the lack of size-extensivity and
the slow convergence towards the FCI limit are due to the linear expansion.
However, we may recast the FCI expansion (4.3) in the form of a product wave function
#
"
Y
(5.1)
(1 + t ) |HFi ,
|CCi =

where we have introduced a generic notation for an excitation operator of unspecified

excitation level (D, T, Q,...) and the associated cluster amplitude t . This (full) coupledcluster (CC) wave function differs from the FCI wave function by the presence of terms
that are nonlinear in the excitation operators. Of course, the CC and FCI wave functions
are completely equivalent provided that all the excitations are included.
Since the s always excite from the set of occupied HartreeFock spin-orbitals to the
virtual ones, [
, ] = 0. Due to the presence of the product excitations in the coupledcluster state (5.1), each determinant can thus be reached in several different ways. For
example, the determinant |i = |HFi = |HFi may be reached with an overall
amplitude equal to the sum of the individual amplitude. With respect to this determinant,
the double-excitation amplitude t is referred to as a connected cluster amplitude and
subsequent singles excitation amplitudes t t as a disconnected cluster amplitude.

5.1.2

The CC Schr
odinger equation

By analogy with the CI theory, we could attempt to determine the CC state by minimizing
the expectation value of the Hamiltonian with respect to the amplitudes. Due to the
nonlinear parameterization of the CC model, the analogous variational conditions would
give rise to an intractable set of nonlinear equations. Hence, in practice coupled-cluster
57

models are not solved using the variation principle, and the CC energies are therefore
non-variational.
|CCi = E |CCi by h| = hHF| we
Instead, by multiplying the eigenvalue equation H
obtain the projected coupled-cluster equations
E
D  
 
(5.2)
H
 CC = E h |CCi ,

where the CC energy is obtained by projection against the HartreeFock state (note that
hHF |CCi = 1)
 
E
D
 
(5.3)
E = HF H  CC .

These equations are nonlinear in amplitudes as well, but unlike the variational conditions,
this expansion will, due to the Slaters rules (1.30-1.31), terminate after few terms, as we
shall see later.

5.1.3

The coupled-cluster exponential ansatz

Because the excitation operators in (5.1) commute, i.e. 2 = 0, the correlating operators
can be written in more convenient form, known as the exponential ansatz for the coupledcluster wave function,
|CCi = exp(T) |HFi ,
(5.4)
where we have introduced the cluster operator
X
T =
t .

(5.5)

In CC theory, a hierarchy of approximations is established by partitioning the cluster

operator in the form
T = T1 + T2 + T3 + + Tn ,
(5.6)
where, for example,
T1 =

tA
IA =
I

AI

T2 =

tA
I aA aI

(5.7)

AI

AB
tAB
IJ
=
IJ

A>B,I>J

1 X AB
tIJ aA aI aB aJ .
4 AIBJ

(5.8)

Each excitation operator in T excites at least one electron from an occupied HartreeFock
spin-orbital to a virtual one. It should be noted that for a system with N electrons,
the expansion terminates after N terms. We further note that [Ti , Tj ] = 0 as a trivial
consequence of the individual excitation operator commutation relations. Moreover,
hHF| = 0
|HFi = 0,
58

because it is impossible to excite an electron from an unoccupied orbital.

To compare the CC model with the CI model, we expand the exponential operator and
collect terms to the same order in excitation level:
exp(T) |HFi =

N
X
i=0

Ci |HFi .

The lowest-order operators Ci are given by

C0 = 1
C1 = T1
1
C2 = T2 + T12
2!
1
C3 = T3 + T1 T2 + T13
3!
1
1
C4 = T4 + T3 T1 + T12 T2 + T14 ,
2!
4!
demonstrating the composition of the excitation processes. The advantages of the cluster
parameterization arise in truncated wave functions and are related to the fact that even at
the truncated level, the coupled-cluster state contains contributions from all determinants
in the FCI wave function, with weights obtained from the different excitation processes.

5.1.4

Coupled-cluster model hierarchy

Similarly with the CI theory, a hierarchy of CC models is introduced by truncating the

cluster operator at different excitation levels. The success of the models discussed below
is illustrated in Figure 5.1.
The simplest successful CC wave function is the coupled-cluster doubles (CCD) model [23],
where only the T2 operator is included to describe the electron-pair interactions.
For better accuracy the singles operator should be included too, yielding the coupledcluster singles-and-doubles (CCSD) model [24]. In this model, the T2 operator describes
the electron-pair interactions and T1 carries out the corresponding orbital relaxations. Its
computational cost scales as O(N 6 ), similarly with CCD.
For higher accuracy, the connected triple excitations have to be taken into account,
which leads to the coupled-cluster singles-doubles-and-triples (CCSDT) model. This model
describes already the dynamical correlation almost perfectly, but is computationally very
demanding, O(N 8 ) and is thus applicable to smallest molecules only.
The triples excitations can be estimated in a perturbational way, giving rise to the
CCSD(T) method [25], which is a good compromise between the computational cost (exhibiting an O(N 7 ) scaling) and accuracy. It is often considered to be the best singlereference treatment of molecular Schrodinger equation that is applicable in practise.

59

Figure 5.1: Success of the CC model hierarcy. On the left: The CC dissociation curves
of the C2v water molecule in the cc-pVDZ basis. On the right: corresponding differences
between the CC and FCI energies. RHF reference state is plotted the full line and UHF
with the dashed.

60

Figure 5.2: The error with respect to FCI of CC and CI wave functions as a function of
the excitation level. Calculation for the water molecule at the equilibrium geometry in the
cc-pVDZ basis.

5.1.5

CI and CC models compared

Let us at this point compare the CI and CC hierarchies; see Figure 5.2. Owing to the
disconnected clusters, CC wave functions (full line) truncated at a given excitation level
also contain contributions from determinants corresponding to higher-order excitations;
whereas CI wave functions (dashed line) truncated at the same level contain contributions
only from determinants up to this level. This is reflected in the convergence towards the
FCI limit, which is typical of small systems.
A further notion outside the figure is that for larger systems, CI starts to behave very
badly, while the CC description is unaffected by the number of electrons.

5.2

Solving the coupled-cluster equations

According to the previous discussion, the projected coupled-cluster equations are written
as


E
D


(5.9)
exp(T) HF = E
HF H


D 
E
D 
E




H
exp(T) HF = E exp(T) HF .
(5.10)

The excited projection manifold h| comprises the full set of all determinants up to
the chosen truncation level.

For the full CC wave function, the number of equations is equal to the number
of determinants and the solution of the projected equations recovers the FCI wave
function.
61

 The nonlinear equations (5.9) must be solved iteratively, substituting in every iteration the CC energy calculated from (5.10).
It is convenient to rewrite the projected CC equations using the non-Hermitian similarity = exp(T)H
exp(T):
transformed Hamiltonian H


E
D

exp(T) HF = E
(5.11)
HF exp(T)H

E
D 

exp(T) HF = 0
(5.12)
exp(T)H

The equations (5.10) and (5.12) are completely equivalent (yet the equivalence is nontrivial
for all the truncated CC models!), meaning that they yield same amplitudes and energy
upon solution. They are referred to as the unlinked and linked coupled-cluster equations,
respectively. Their computational cost is practically the same, but the linked equations
have some advantages: Although both yield the same, size-extensive wave function, the
linked equations are size-extensive term by term, that simplifies some perturbational treatments; in addition, the CC approach excited states in mind is more fruitful to start from
the linked equations.
The main reason is however the following: When the BakerCampbellHausdorff (BCH)
expansion exp(A)B exp(A) = B+[A, B]+ 2!1 [A, [A, B]]+ 3!1 [A, [A, [A, B]]]+ is applied
to the similarity-transformed Hamiltonian, one obtains
T], T] + 1 [[[H,
T], T], T] + 1 [[[[H,
T], T], T], T],
exp(T) = H
+ [H,
T] + 1 [[H,
exp(T)H
2
6
24
(5.13)
the coupled-cluster equations are therefore no higher than quartic in the cluster amplitudes
for any truncation level or even for full CC expansion.
Due to the Brillouin theorem, the one-particle operators contribute only to second
order. Furthermore, cluster operators higher than the doubles do not contribute to the CC
is a two-particle operator. As a result, only singles and doubles amplitudes
energy since H
contribute directly to the CC energy irrespective of the truncation level of the cluster
operator. Of course, the higher-order excitations contribute indirectly, since all amplitudes
are coupled by the projected equations.

5.2.1

On coupled-cluster optimization techniques

The elements of the left-hand side of the CC amplitude equation (5.12) constitute the
coupled-cluster vector function (t), which can be expanded around the set of amplitudes
of the current iteration t(n) :
(t(n) + t) = (0) (t(n) ) + (1) (t(n) t + ,

where (0) is the vector function calculated from the amplitudes t(n) and (1) the CC
Jacobian

E
D 

(n)
(n) 
(1)

(5.14)
= exp(T )[H, ] exp(T ) HF .

Then an iterative scheme is established:

62

1. Solve the set of linear equations

(1) (t(n) )t(n) = (0) (t(n) )

(5.15)

for t(n)
2. Form the improved estimate
t(n+1) = t(n) + t(n)

(5.16)

3. Iterate Equations (5.15) and (5.16) until convergence in the amplitudes.

In more sophisticated implementations, this Newton-method-like approach is replaced by
a scheme, where the solution of the linear equations as well as the construction of the
Jacobian are avoided by the approximation
1 (0) (n)
t(n)
),
= (t

(5.17)

where, e.g. ab
ij = (a + b i j ).
Usually, the convergence is accelerated by the DIIS method (see the discussion in section
3.1).

5.2.2

The closed-shell CCSD model

We consider now in detail the important special case of CC theory: the closed-shell CCSD
model. We recall that the singlet CCSD state is generated by |CCi = exp(T1 + T2 ) |HFi,
hence only those terms in Ti that transform as singlet operators should be retained in the
cluster operator. It is easy to show that
h
i
S , Ti = 0
(5.18)
h
i
Sz , Ti = 0.
(5.19)
These impose important constraints on the cluster amplitudes in the spin-orbital basis.
Due to these, the singles cluster operator becomes
X
T1 =
tai Eai
(5.20)
ai

and the doubles operator

1 X ab
t Eai Ebj .
T2 =
2 aibj ij
ba
The cluster amplitudes are symmetric: tab
ij = tji .

63

(5.21)

The CCSD energy is obtained from Eq. (5.11). By noting that hHF| exp(T) = hHF|,
expanding the cluster operators and after some algebra we obtain the form




E
E D
1D
 
 
ECCSD = EHF +
(5.22)
T2 ] HF ,
T1 ], T1 ] HF + HF [H,
HF [[H,
2
 
D
E
 
where the HartreeFock energy is given by EHF = HF H
HF
. By inserting the CCSD

operators to the energy expression and calculating the commutators explicitly we obtain [26]




E 1X D
E
1X a bD


 
ab

t t HF [[H, Eai ], Ebj ] HF +

t HF [H, Eai Ebj ] HF
ECCSD = EHF +
2 aibj i j
2 aibj ij


E

D
1 X ab


a b

= EHF +
tij + ti tj HF [[H, Eai ], Ebj ] HF
2 aibj
X


a b
tab
= EHF +
(5.23)
ij + ti tj Liajb ,
aibj

where we have denoted Lpqrs = 2gpqrs gpsrq .

5.3
5.3.1

Special formulations of coupled-cluster theory

The equation-of-motion coupled-cluster method

In CC theory, we arrive at a good description of the ground state. The excited states
could be examined by carrying out separate, independent calculation for each state of
interest, using in each case some appropriate zero-order reference determinant. However,
it is usually difficult or even impossible to determine these adequate reference determinants;
it is expensive to carry out individual calculations; and the states obtained in this manner
are not orthogonal, which hinders the identification of the excited states.
Instead, we calculate the excited states in the spirit of CI theory, by a linear expansion
in the space spanned by all states of the form
X
X
|ci =
c |CCi = exp(T)
c |HFi ,
(5.24)

where the summation is carried out over all the excitation operators present in the cluster
operator. This expansion is referred to as the equation-of-motion coupled-cluster (EOMCC) model. The expansion parameters are optimized by the minimization of the energy
 E
D 



c exp(T )H exp(T ) c
.
(5.25)
E(c, c) =
hc |ci

The EOM-CC theory can be thought as conventional CI theory with a similaritytransformed Hamiltonian, which carries the information about electron correlation; while
the configuration expansions carry the information about the excitation structure of the
electronic states.
64

5.3.2

Orbital-optimized coupled-cluster theory

In the standard CC theory, we use the HartreeFock orbitals and then determine a set of
non-zero single-excitation amplitudes together with higher-excitation amplitudes. Alternatively, we could use exp(T1 ) to generate an orbital transformation to a basis in which the
single-excitation amplitudes vanish. This is the idea behind the orbital-optimized (OCC)
theory, in which the orbital-rotation operator exp(
) is used instead of exp(T1 ) (as they
generate the same state to first order),
|OCCi = exp(
) exp(TO ) |0i ,

(5.26)

where TO = T2 + T3 + + TN . The OCC energy is obtained in the usual manner by

projecting the OCC Schrodinger equation against the reference state
 E
D 


exp(
(5.27)
) H
) exp(TO ) 0 ,
EOCC = 0 exp(TO ) exp(

whereas the cluster amplitudes are determined by projection against the manifold |O i
spanned by TO
 E
D 


exp(
(5.28)
) exp(TO ) 0 = 0.
O exp(TO )H

The Brueckner coupled-cluster theory (BCC) is closely related to OCC and differs in some
details of the solution of orbital rotation parameters. Unlike in standard CC theory, the
OCC and BCC orbitals are optimized simultaneously with the optimization of the cluster
amplitudes compare with the MCSCF theory. Surprisingly, the differences between standard and OCC/BCC wave functions are small. Only in the rare cases that are characterized
by the sc. HartreeFock singlet instability the OCC/BCC are clearly more successful than
the standard formulation.

5.3.3

Coupled-cluster perturbation theory

In the coupled-cluster perturbation theory (CCPT) [27] the Hamiltonian in the CC equations (5.11) and (5.12) is partitioned as in Eq. (3.20). The Fock and the cluster operator
obey the following results:
i
h
X

f, T =
t

hh

f, T , T

= [[[f, T], T], T] = = 0.

Thus we may write the non-Hermitian similarity-transformed Fock operator as

f = f +

65

t ,

(5.29)

and we arrive at the working equations of the CCPT:

 
E
D
 
E = E0 + HF 
 HF
E
D  
 
t =  HF .

(5.30)
(5.31)

By expanding the cluster operator in the orders of the fluctuation potential we arrive at
the amplitude equations

E
D 
 (n) 
(n)
(5.32)
t = []  HF

(n) contains the nth order part of the similarity-transformed fluctuation potential.
where []
Then the amplitudes would be determined self-consistently from the projected equations
after truncating the cluster operator at some excitation level. The CCPT is most often
used to establish approximations to contributions from some excitations the most famous
is the non-iterative triples, or (T) correction to the CCSD wave function. Also very successful iterative CCPT schemes have been set up, e.g. the CC2 model approximates the
CCSD model with the computational cost O(N 5 ) and CC3 the CCSDT model with the
computational cost O(N 7 ). The CCPT series provides more systematic improvement to
the wave function than the MllerPlesset perturbation theory (MPPT).
In passing, we note that both MPPT and CCPT are limited to the ground states
of systems dominated by a single configuration. To overcome this, multiconfigurational
perturbation theory has been developed. Most important such theory is CAS perturbation
theory (CASPT) [28], where the zero-order state is taken to be a CASSCF wave function.
At the moment, it is the only generally applicable method for the ab initio calculation
of dynamical correlation effects of open- and closed-shell multiconfigurational electronic
system.

5.3.4

Local correlation methods

Linear scaling approximations for HartreeFock self-consistent field method are carried
out almost routinely at the moment. However, linear scaling black-box correlated wavefunction theory is far away from reality.
Traditional correlated electronic wave function methods, which are built upon the delocalized HartreeFock molecular orbitals (MO) do not exploit the fact that the electron
correlation i.e., the electron-electron Coulomb cusp is spatially a rather local phenomenon.
In the recent literature, localized correlation methods have been pursued to overcome
the scaling wall encountered in the correlated wave function methods. However, despite
recent progress, there is still a need to formulate and implement new approaches that
exploit the locality of the electron correlation problem. Approaches based on introducing
locality in the correlation treatment using localized MOs (LMO) appeared already in 1960s
and in the context of CC methods in 1980s [29, 30].
66

In the approach presented by Head-Gordon et al. CC equations are formulated in a

general non-orthogonal [e.g., the atomic orbital] basis using a general tensor-based formulation [31]. Scuseria and Ayala presented a transformation of the standard MO CC doubles
(CCD) equations into the AO basis [32,33]. Scuseria and Ayala argued that the solution of
the CC equations in the AO basis scales in principle linearly with respect to N , but their
approach seems so far not to have received any widespread use.
A fully operational local CCSD implementation have been presented by Werner et al.
This implementation, referred to as the local CCSD (LCCSD) method [3440], is based on
localizing the occupied space as suggested by Pulay and coworkers [29], using traditional
MO localization schemes, and using for the virtual space a projected atomic orbital (PAO)
basis. LCCSD further relies on a spatial division of the system into domains. Restrictions
on the extent to which the interactions between different domains are accounted for is
important for achieving the low-power scaling of LCCSD making the results deviate from
the standard CCSD results.
The equations of correlated wave function theories may also be derived in a biorthogonal
PAO basis (thus without MOs altogether), aiming at a black-box formulation that could
be applied to large systems [41, 42]. This formulation also constitutes a good platform to
introduce efficient approximation schemes to reduce the number of parameters and thus
the computational scaling. The locality of the PAO basis is reflected in the fact that
the cluster amplitudes, the integrals as well as the correlation energy contributions decay
exponentially with respect to the distance between the centers of the participating PAOs.

5.4

Further reading

R. J. Bartlett, in Modern Electronic Structure Theory Vol II, edited by D. R. Yarkony

(World Scientific, Singapore 1995).
T. Helgaker, P. Jrgensen, and J. Olsen, Molecular Electronic-Structure Theory (Wiley 2002), pp. 648816.
T. J. Lee and G. E. Scuseria, in S. R. Langhoff (ed), Quantum-Mechanical Electronic
Structure Calculations with Chemical Accuracy (Kluwer 1995).
O. Christiansen, Coupled cluster theory with emphasis on selected new developments,
Theor. Chim. Acta 116, 106 (2006).

67

68

Chapter 6
Basis sets and molecular integrals
6.1

Atomic basis functions

Molecular orbitals can be constructed either numerically or algebraically, of which the former provides greater flexibility and accuracy, but is tractable only for atoms and diatomic
molecules; for polyatomic systems we are forced to expand the MOs a (r) in a set of some
simple analytical one-electron functions.

6.1.1

General considerations

Perhaps surprisingly, it turns out that the hydrogenic functions Eq. (1.13) are not ideal
for this purpose; as they do not constitute a complete set by themselves but must be
supplemented by the unbounded continuum states. Secondly, they become very diffuse
and a large number of them is needed for a proper description of both the core and valence
regions of a many-electron atom, especially when located in a molecule.
Therefore, in the molecular electronic-structure theory, s are usually functions of
slightly more unphysical origin. Ideally, they should
1. allow orderly and systematic extension towards completeness with respect to oneelectron square-integrable functions
2. allow rapid convergence to any atomic or molecular electronic state, requiring only a
few terms for a reasonably accurate description of molecular electronic distributions
3. have an analytical form for easy and accurate manipulation especially for molecular
integrals
4. be orthogonal or at least their non-orthogonality should not pose problems related
to numerical instability.
69

6.1.2

Hydrogenic functions

Let us begin by reviewing the properties of the hydrogenic functions. As we noted earlier,
they are the bounded eigenfunctions of the Hamiltonian
Z
1
H = 2 .
2
r
and are written as
nlm (r, , ) =

2Z
n

3/2 s

(n l 1)!
2n(n + 1)!

l




Zr
2Zr
2Zr
2l+1
exp
Lnl1
n
n
n
{z
}
Rnl (r)

2l + 1 (l m)! m
P (cos ) exp(im).
4 (l + m)! l
{z
}
|
Ylm (,)

The energy of the bounded hydrogenic state nlm is given by

En =

Z2
.
2n2

(6.1)

The degeneracy of the hydrogenic states of different angular momenta that is, 2l + 1
degenerate states for every l is peculiar to the spherical symmetry of the Coulomb
potential and is lifted in a many-atom system. The notable features of the radial functions
include
the presence of the exponential
that Laguerre polynomials introduce n l 1 radial nodes in the wave function.
The diffusiveness of nlm for large n is seen from the expectation value of r
hnlm |
r| nlm i =

3n2 l(l + 1)
2Z

(6.2)

In the following we consider basis functions that retain the product form of the hydrogenic
wave functions but have a more compact radial part.

6.1.3

The Laguerre functions

In order to retain the correct exponential behavior but avoid the problems with continuum
states we introduce the following class of functions:
LF
LF
nlm (r, , ) = Rnl (r)Ylm (, )
s
(n l 1)!
LF
(2r)l L2l+2
= (2)3/2
Rnl
nl1 (2r) exp(r),
(n + l + 1)!

70

(6.3)
(6.4)

1s

10

2s

10

2p

10

3s

10

Figure 6.1: Lagurerre functions.

referred to as the Laguerre functions. For fixed l and the radial functions (6.4) with n > l
constitute a complete orthonormal set of functions, for which

 2n + 1
LF
r| LF
,
nlm |
nlm =
2

(6.5)

being thus considerably more compact than the hydrogenic functions for large n and independent of l. The Laguerre functions exhibit the same nodal structure as the hydrogenic
ones but their radial distributions R2 r2 are quite different ( = 1/n, Z = 1).
One should note that the orthogonality of the Laguerre functions is valid only with
fixed exponents. It turns out, however, that the expansion of the orbital of a given l-value
of an atomic system requires a large number of fixed-exponent Laguerre functions with
different n. For non-orthogonal, variable-exponent functions Laguerre functions are not
an optimal choice, but we need to introduce functions specifically tailored to reproduce as
closely as possible the different orbitals of each atom.

6.1.4

Slater-type orbitals

LF
In simplifying the polynomial structure of the Laguerre functions, we note that in Rnl
r
occurs to power l and to powers n l 1 and lower. Thus a nodeless one-electron function
that resembles LF
nlm and the hydrogenic functions closely for large r is obtained by retaining
in the Laguerre function only the term of the highest power of r, yielding the Slater-type
orbitals (STO)
STO
STO
nlm (r, , ) = Rnl (r)Y (, )
(2)3/2
RnSTO (r) = p
(2r)n1 exp(r)
(2n + 1)

(6.6)
(6.7)

For the STOs we use the same 1s, 2p,... notation as for the hydrogenic functions, keeping
in mind that STOs are nodeless and now n only refers to the monomial factor rn1 . The
expectation value of r and the maximum in the radial distribution curve are given by

STO
r| STO
nlm |
nlm

STO
rmax

71

2n + 1
2
n
=

(6.8)
(6.9)

Also STOs would constitute a complete set of functions for a fixed exponent , and we
could choose to work with a single exponent and include in our basis orbitals (1s, 2s, . . .), (2p, 3p, . . .),
(3d, 4d, . . .), all with the same exponent. Alternatively, we may describe the radial space
by functions with variable exponents. For each l, we employ only the functions of the lowest n, yielding a basis of the type ((1s(1s ), 1s(2s ), . . .), (2p(1p , 2p(2p ), . . .), . . .) These
variable-exponent STOs are given by
nl lm (r, , ) = RSTO
(r)Ylm (, )
nl l

(6.10)

3/2

(2nl )
RSTO
= p
(2nl r)l exp(nl r).
nl l
(2l + 2)!

(6.11)

In practice, a combined approach is employed. For example, for the carbon atom, we would
introduce sets of 1s, 2s and 2p functions, all with variable exponents that are chosen to
ensure an accurate representation of the wave function.

6.1.5

Gaussian-type orbitals

The STO basis sets are very successful for atoms and diatomic molecules, but the evaluation
of many-center two-electron integrals becomes very complicated in terms of them. The
basis sets that have established to be the standard practice in molecular electronic-structure
theory are less connected to the hydrogenic functions or actual charge distributions in
molecules. They are called Gaussian-type orbitals (GTO) as they employ the Gaussian
distributions exp(r2 ) instead of the exp(r) as in STOs. A large number of GTOs are
needed to describe an STO (or an AO) properly, but this is more than compensated by a
fast evaluation of the molecular integrals.
The spherical-harmonic GTOs are
GTO
(r)Ylm (, )
GTO
nlm (r, , ) = Rnl
s
3/4

2(2)
22nl2
GTO
Rnl
(r) =
(
2r)2nl2 exp(r2 ).
1/4
(4n 2l 3)!!

(6.12)
(6.13)

They form a complete set of nonorthogonal basis functions. In above, the double factorial
function is

1
n=0

n(n 2)(n 4) 2 even n > 0

n!! =
(6.14)
n(n 2)(n 4) 1 odd n > 0

1
odd n < 0
(n+2)(n+4)1

The spatial extent of the GTOs are


GTO

GTO
r| nlm
nlm |

GTO
rmax

2n l 2
2
r
2n l 1
=
2

72

(6.15)
(6.16)

2.5

5.0

5.0

1s STO (=1)
2s STO (=1)

7.5

1s GTO (=1)
2s GTO (=1)
2p GTO (=1)

1s
STO (=1)
GTO (=1)

1s
GTO (=3.42)
GTO (=0.62)
GTO (=0.16)
5.0

2.5

2.5
2.5

0.0

0.0
0

0.0
0

0.0
0

Figure 6.2: Slater and Gaussian-type orbitals compared.

When we compare these results with the case of STOs, we note that much larger number
of GTOs are needed for a flexible description of outer regions of the electron distribution.
In practice, the use of GTOs with variable exponents differs from the procedure for
STOs. For GTOs, we describe the radial space exclusively by means of variable exponents,
using for this purpose only spherical-harmonic GTOs with n = l + 1, only powers of r
introduced are thus associated with l of the spherical harmonics. The following set of
spherical-harmonic GTOs is used:
GTO
GTO
nl lm (r, , ) = Rnl l (r)Ylm (, )
s
3/4

2(2
)
2l
nl
RGTO
(r)
=
(
2r)l exp(nl r2 ).
l
nl
1/4
(2l + 1)!!

(6.17)
(6.18)

Let us finally consider the Gaussian product rule, that is the one of the main reasons
2
why GTOs are so widely used. The product of two s GTOs A = exp(rA
) and B =
2
exp(rB ) centered on rA and rB is a third Gaussian




2
AB = exp
(6.19)
rAB exp ( + )rC2
+
centered on rC given by

rC =

6.2

rA + rB
.
+

Gaussian basis sets

Of the three general features determining the accuracy of quantum chemical methodsthe
choice of the Hamiltonian, electron correlation treatment or truncation of the many-particle
space, and the description of the one-particle space, i.e., the basis set usedthe choice of the
basis set is the most crucial one, as insufficient basis sets yield erroneous results regardless
of the level of theory, whereas with a proper basis set, at least qualitatively correct results
can be obtained for many problems already at the uncorrelated level of theory and using
the non-relativistic Hamiltonian.
73

The majority of quantum chemical applications employ contracted GTO (CGTO) sets,
i.e. linear combinations of GTO functions with coefficients optimized regarding some criterion (usually SCF energy) that significantly increase efficiency. There are two widely used
contraction schemes: segmented contraction, where each GTO contributes only to a single
CGTO, and general contraction, where such a restriction is not applied. Examples of the
segmented scheme include the Pople-style basis set families (3-21G, 6-31G,...) [43, 44] and
the Karlsruhe basis sets [4547]. The correlation-consistent (cc) basis set families [4853] by
Dunning and co-workers are the most widely used generally contracted basis sets. The basic
idea behind the cc sets is that functions that contribute approximately the same amount to
the correlation energy are added to the basis in groups. The cc basis sets provide smooth,
monotonic convergence for the electronic energy as well as for many molecular properties,
especially those originating in the valence region. The polarized valence (cc-pVXZ, X = D,
T, Q, 5, 6 corresponding to the number of CGTOs used to represent an occupied atomic
orbital) and core-valence (cc-pCVXZ) sets can be augmented with diffuse functions (augcc-pVXZ and aug-cc-pCVXZ; d-aug-cc-pVXZ and t-aug-cc-pVXZ for doubly and triply
augmented sets, respectively). While these features are favorable and the sets are widely
used, the cc-basis sets become very large at large X, and it is not straightforward to extend
the family beyond the published X values or to new elements. The atomic natural orbital
(ANO) basis sets [54, 55] provide another general contraction approach. The contraction
coefficients therein are atomic ANO coefficients that are obtained by optimizing atomic
energies.
The use of increasingly large basis sets that produce results converging to some particular value is usually regarded as a solution to the problem of basis set incompleteness. In
calculations of molecular properties that originate, e.g., in the region close to the nuclei
(examples include indirect spin-spin couplings and hyperfine couplings), approaching the
basis-set limit using the cc or comparable energy-optimized paradigms may lead to excessively large basis sets prohibiting calculations of large molecules. An alternative and often
used approach is to uncontract the basis set and to supplement it by additional steep basis
functions in the l-shells relevant for the property under examination.
The performance of the different basis sets for a certain molecular property can be
qualitatively understood within a concept, which can be measured by the completeness
profiles [56]
X
Y() =
hg()|m i2 .
(6.20)
m

Here {} denotes a set of orthonormalized, contracted or primitive, basis functions, and

g() is a primitive test GTO with the exponent , used to probe the completeness of
{}. The completeness profile becomes unity for all , for all l-values in a CBS. Y()
is typically plotted in the logarithmic scale, against x = log . In a certain exponent
interval [min , max ] this quantity is intuitively connected to the possibility from the
point of view of one-particle space of describing all details of the wave function in the
corresponding distance range from the atomic nuclei. Put simply, atomic properties that

obtain relevant contributions roughly within [1/ max , 1/ min ] from the nucleus can be
74

1.0
Y()

0.75

s
p
d

(a)

0.5

s
p
d
f

(b)

0.25
1.0
Y()

0.75

s
p
d
f
g

(c)

0.5
0.25
1.0

Y()

0.75

s
p
d
f
g
h
i

(e)

0.5
0.25

Y()

1.0
0.75
0.5

s
p
d
f

(f)

s
p
d
f

(g)

s
p
d
f
g
h

(d)

s
p
d
f

(h)

0.25
4

1 0 -1 -2 -3 -4
Log ()

1 0 -1 -2 -3 -4
Log ()

Figure 6.3: Completeness profiles of (a) cc-pVDZ (b) cc-pVTZ (c) cc-pVQZ (d) cc-pV5Z
(e) cc-pV6Z (f) cc-pCVTZ (g) aug-cc-pVTZ and (h) aug-cc-pCVTZ basis sets of fluorine.

reproduced by a basis that has Y() = 1 in this interval. This way of thinking can be
generalized to molecular properties that may be dominated by phenomena occurring close
to the expansion centers of the basis functions, i.e., atomic nuclei (region described by highexponent basis functions) and/or in the valence region, further away from the nuclei. It has
been later demonstrated that this property only can be the guiding factor for successful
basis sets in form of the completeness-optimized basis sets [57].
75

6.3

Integrals over Gaussian basis sets

In the following, we will evaluate the one- and two-electron integrals

D   E Z
 

O = O
 = (r)O(r)
(r)dr
 
D
E
 
O = (1) (1) O
 (2) (2)
Z

=
(r1 ) (r1 )O(r)
(r2 ) (r2 )dr1 dr2

(6.21)

(6.22)

such that the AOs () are taken as fixed linear combinations of real-valued primitive
Cartesian GTOs
2
Gijk (r, a, R) = xiR yRj zRk exp(arR
),
(6.23)
where the Cartesian quantum numbers i, j, k are greater than zero and l = i + j + k for
a given total angular momentum quantum number.
The integrals over primitive Cartesian GTOs may subsequently be transformed to integrals over contracted and/or spherical-harmonic GTOs (6.17) by linear combinations.

6.3.1

Gaussian overlap distributions

An important property of the primitive Cartesian GTOs is that they can be factorized in
the three Cartesian directions,
Gijk (r, a, R) = Gi (x, a, Rx )Gj (y, a, Ry )Gz (z, a, Rz ),

(6.24)

where for example Gi (x, a, Rx ) = xiR exp(ax2R ). We will need also a concept of Gaussian
overlap distribution
ab (r) = Gijk (r, a, RA )Glmn (r, b, RB ),
(6.25)
that may also be factorized in the same way
ab (r) = xij (x, a, b, RA,x , RB,x )ykl (y, a, b, RA,y , RB,y )zmn (z, a, b, RA,z , RB,z ),

(6.26)

where the x component (for example) is given by

xij (x, a, b, RA,x , RB,x ) = Gi (x, a, RA,x )Gj (x, b, RB,x ).

6.3.2

(6.27)

Simple one-electron integrals

We are now ready to consider simple one-electron integrals, by which we mean molecular
integrals that do not involve the singular Coulombic 1/r interaction. All such integrals can
be factorized in the three directions as
Sab = Sij Skl Smn .
76

(6.28)

According to the Gaussian product rule (6.19), any Cartesian component of the overlap
distribution can be written as a single Gaussian at the centre of charge Px , and we may
thus evaluate the integral
r
Z
Z

2
x
2
2
exp(XAB
),
(6.29)
00 dx = exp(XAB )
exp(pxp )dx =
p

where denotes the reduced exponent ab/(a + b), p = a + b, and

aRA,x + bRB,x
p
= RA,x RB,x .

Px =
XAB

(6.30)
(6.31)

This result provides a basis for a set of recurrence relations by which we may evaluate
simple integrals overlap or more complicated ones over GTOs of arbitrary quantum
numbers. This procedure is known as ObaraSaika scheme [58, 59]. There exist other
methods for molecular integral evaluation but we will not discuss them here. The relations
are obtained by considering the behaviour of the integral under coordinate transformation
and their detailed derivation is omitted here but recommended further reading.
The simplest case is the overlap integral
Sab = hGa |Gb i ,

(6.32)

for an Cartesian component of which the ObaraSaika relations are written as

1
(iSi1,j + jSi,j1 )
2p
1
= XP B Sij + (iSi1,j + jSi,j1 ).
2p

Si+1,j = XP A Sij +

(6.33)

Si,j+1

(6.34)

For the linear and angular momentum as well as for kinetic energy integrals we need
the integrals over differential operators
  e f g  
 
ef g
(6.35)
Dab = Ga  e f g  Gb .
x y z

In one direction the relations are

1
e
e
(iDi1,j
+ jDi,j1
2beDije1 )
2p
1
e
e
+ jDi,j1
+ 2aeDije1 ),
= XP B Dije + (iDi1,j
2p

e
Di+1,j
= XP A Dije +

(6.36)

e
Di+1,j

(6.37)

with the e = 0 case providing the starting point and being equivalent with the overlap
integral.
77

6.3.3

The Boys function

We will now introduce a special function that has a key role in the evaluation of one- and
two-electron Coulomb integrals, the Boys function of order n, defined as
Z 1
Fn (x) =
exp(xt2 )t2n dt.
(6.38)
0

It is a strictly positive and decreasing function. We observe that the values of Boys function
of order n at x = 0 have a closed-form expression
Z 1
1
Fn (0) =
t2n dt =
.
(6.39)
2n + 1
0
Different methods for evaluating the Boys function has been introduced. One way is
through recursion. By partial integration of the function we see that the different orders
are related by
Fn+1 (x) = (2n + 1)Fn (x) exp(x)2x
(6.40)
or
Fn (x) = 2xFn+1 (x) exp(x)2n + 1.

(6.41)

We would therefore need to calculate the Boys function for the highest or the lowest order
needed, and the others are obtained through the downward or upward recursion, of which
the former is numerically more robust.

6.3.4

ObaraSaika scheme for one-electron Coulomb integrals

The Cartesian one-electron Coulomb integrals,

  
Vab = Ga rC1  Gb =

ab (r)
dr,
rC

(6.42)

needed for the one-electron part of the molecular Hamiltonian, may also be evaluated by
using the ObaraSaika recursion relations.
We again factorize the distribution according to (6.26), and are about to obtain the
integrals through the recursion relations (c.f., below) for auxillary integrals N
ijklmn :
1
N
(iN
i1 jklmn + ji j1 klmn )
2p
1
N +1
+1
+1
(iN
XP C N
i1 jklmn + ji j1 klmn )
ijklmn
2p
1
N
= XP B N
(iN
ijklmn +
i1 jklmn + ji j1 klmn )
2p
1
+1
N +1
+1
(iN
XP C N
i1 jklmn + ji j1 klmn )
ijklmn
2p

N
N
i+1 jklmn = XP A ijklmn +

N
i j+1 klmn

78

(6.43)

(6.44)

with the special cases (that also serve the starting points of the recursion)
0ijklmn = Vijklmn = Vab
2 xyz
N
Kab FN (pRP2 C ),
000000 =
p

(6.45)
(6.46)

where FN is the Boys function, and we have denoted

x
2
Kab
= exp(XAB
).

6.3.5

(6.47)

ObaraSaika scheme for two-electron Coulomb integrals

The evaluation of electron-electron repulsion integrals is a highly non-trivial task and due
to their bottleneck status, under intense study since the early days of quantum chemistry.
We will discuss here only one of the numerous schemes that is in accordance with the
earlier discussion, namely the ObaraSaika scheme for Cartesian two-electron integrals,
Z Z
 1 


ab (r1 )cd (r2 )


gabcd = Ga (r1 )Gb (r1 ) r12 Gc (r2 )Gd (r2 ) =
dr1 dr2 .
(6.48)
r12

The relations feature similarly to the one-electron corresponds a set of auxillary integrals
. Again, two special cases of them are required:
2 5/2
xyz xyz

Kab
Kcd FN (RP2 Q )
pq p + q
= gix jx kx lx iy jy ky ly iz jz kz lz = gabcd .

N
0000 0000 0000 =
0ix jx kx lx iy jy ky ly iz jz kz lz

(6.49)
(6.50)

xyz
y
z
x
In the above, we have denoted Kab
= Kab
Kab
Kab
[c.f. Eq. (6.47)] and is the reduced
exponent pq/(p + q). FN is again the Boys function. In the following, the indices i, j,
k and l are the Cartesian quantum numbers for orbitals a, b, c and d in the x-direction.
Starting from these, first a set of integrals with j = k = l = 0 is generated by


i
N +1

N +1
N
N
(6.51)
i1 000 i1 000 .
i+1 000 = XPA i000 XPQ i000 +
p
2p
p

Then, the second electron is treated using the integrals generated in (6.51):
i0 k+1 0 =

bXAB + dXCD N
i
k N
p
i0k0 N
i0 k1 0 N
.
i1 0k0 +
q
2q
2q
q i+1 0k0

(6.52)

In the final step, the Cartesian powers are transferred between the orbitals of the same
electron:
i j+1 kl = i+1 jkl + XAB N
ijkl

(6.53)

ijkl+1 = ij k+1 l + XCD N

ijkl .

(6.54)

In this way, we may construct the full set of Cartesian two-electron Coulomb integrals.
79

6.4

Further reading

T. Helgaker and P. Taylor, Gaussian basis sets and molecular integrals, in Modern
Electronic Structure Theory, Part II, D. R. Yarkony (ed.), (World Scientific 1995),
pp. 725856
R. Lindh, Integrals of electron repulsion, in P. v. R. Schleyer et al. (eds.), Encyclopedia of Computational Chemistry Vol. 2, (Wiley 1998) p. 1337.

80

Chapter 7
Accounting for the effects of special
relativity
7.1
7.1.1

Relativistic quantum mechanics

Theory of special relativity

In NR classical mechanics, two systems O and O moving relative to one another with
speed v are related with the Galileo transformation x = x + vt, with time t treated as
an independent parameter. However, the famous Maxwell equations describing elecromagnetism are not invariant under Galileo transformation. This problem was culminated in
the Michelson-Morley experiment, in which the speed of light was measured to be constant
in all the directions of movement of the Earth. This was hard to explain with the classical
theory, and the problem was solved in the theory of special relativity by Albert Einstein
(1905), which is based on two postulates: 1. The laws of physics are the same in all inertial
frames of reference, and 2. The speed of light in free space has the same value in all inertial
frames of reference. These are satisfied in Lorentz transformation (1904) 1
x = (x vt)

v x
t = t 2
c

(7.1)
(7.2)

p
where = 1/ 1 (v/c)2 . The direct consequence of the theory of special relativity is that
systems moving relative to each other with constant speed are equivalent, so no absolute
reference system can exist.

7.1.2

KleinGordon equation

The straightforward relativistic analogue for the non-relativistic (NR) Schrodinger equation, H = i/t, using a relativistic energy-momentum relation, E 2 = p2 c2 + m2 c4 , is
1

Atomic unit system. In the four-algebra, the metric tensor g = diag(1, 1, 1, 1). Einsteins summation convention is used.

81

the Klein-Gordon equation


1 2
2
+ = m2 c2 .
2
2
c t

(7.3)

The problem with the Klein-Gordon equation lies with the negative energy solutions that
involve a negative probability density.
However, the equation is usable with re-interpreatation of the solutions Pauli and
Weisskopf interpreted the solution as a current density and Feynman and St
uckelberg
interpreted the negative energy solutions to describe either a particle moving backwards
in time or an antiparticle moving forward in time. The Klein-Gordon equation is sufficient
relativistic wave-equation for spinless paricles.

7.1.3

Dirac equation

To avoid these, Paul Dirac suggested a form of an equation [60, 61] that is linear in both
/t and as
H Dirac = (c p + mc2 ),
(7.4)
where the factors i (i = 1, 2, 3) and are defined by demanding that the relativistic
energy-momentum relation be satisfied.
A comparison of Eq. (7.4) with the energy-momentum relation shows that i and
anticommute and that 2i = 2 = 1. i and are N N matrices, with N = 4 being
the lowest dimension that satisfies the equation. The equation can be written also in
terms of larger N . The representation of and is not unambiguous. The Dirac-Pauli
representation




1 0
0 i
,
(7.5)
,
=
i =
0 1
i 0
where i are the Pauli spin matrices






1 0
0 i
0 1
,
, 3 =
, 2 =
1 =
0 1
i 0
1 0

(7.6)

is the most commonly used.

By multiplying (7.4) with from left and denoting
= (ik , ), k = 1, 2, 3,

(7.7)

the resultant equation can be written into compact form as

( + mc) = 0,

(7.8)

where denotes differentation with respect to x = (x, ict). The gamma-matrices (7.7)
can in fact be defined with condition [ , ]+ = 2g , and every four-matrices fulfilling
this condition can be used as their representation. The representation corresponding to
82

Dirac-Pauli representation is the most common. This form of the the Dirac equation for
a free electron can be shown to be Lorentz covariant and describe spin-half particles. The
solution of the Dirac equation is a four-component Dirac spinor.
The solution also contains negative energy solutions, but with a positive probability
density. They were interpreted as positrons, electrons with opposite charge. The positrons
were experimentally observed a few years later [62]. The Dirac spinor can be divided into


+
=
,
(7.9)

where + and are the large- and small-component solutions. The solutions with large
+ correspond mainly to the positive energy i.e. electronic solutions, whereas solutions
with large correspond to the positronic states.2

7.1.4

On the origin of magnetic energy levels

The substitution p = p qA /c, where A is the 4-vector potential (A, i), yields
the Dirac equation for a particle with charge q in the presence of an electromagnetic field,


(7.10)
c (p qA/c) + q + mc2 = i4 .

For a constant magnetic field, the scalar potential = 0 and the stationary state of the
Eq. (7.10) this equation in terms of large and small components becomes





+
mc2 c~
+
=E
.
(7.11)
c~ mc2

One obtains an equation involving only the upper components by solving for one pair in
terms of the other,

( )2
+ = +
2m 

2
q
+
B + = + .
2m 2mc

(7.12)

Here = 12 (mc2 + E 2 /mc2 ) is approximately the energy of the particle minus the energy
related to its mass. The first term is analogous to NR treatment of a charged particle in
a magnetic field, which solutions |n, k, ml i that are eigenfunctions of Hnr , pz , and Lz (L
is the orbital angular momentum operator) with eigenvalues p2 /2m + (2n + 1)eB/2m, p,
and ml , respectively. These are consistent with the classical picture of the charged particle
moving in a helix with axis parallel to the field. The second term describes the interaction
2

The non-relativistic limit (c ) of the Dirac equation, often referred to as the Levy-Leblond
equation [63], is a Galilei-invariant four-component equation that describes spin one-half particles. It
describes correctly also the interaction between the electron spin and an external magnetic field, with the
gyromagnetic ratio equalling to 2.

83

of the spin magnetic moment of the particle with the field. If the uniform field is chosen
to be in x3 -direction in (7.12),


2
2m

q
+ 2mc
B
0

2
2m

0
q
2mc
B



1
2

1
2

and hence either 1 |n, k, ml i and 2 0 with = (2n + 2)qB/2mc + p2 /2m or 2

|n, k, ml i and 1 0 with = neB/mc + p2 /2m. + is an eigenfunction of perpendicular
part of the Hamiltonian with the field, p3 , L3 , and 3 . The eigenvalue of the last is labeled
as 2ms , which can be 1. The lower components are obtained from (7.11). The operator
Lz does not commute with the Hamiltonian, and different components of the wavefunction
are associated with different ml values. However, it can be shown with a direct calculation
that the operator
1
Jz = Lz + z
2

(7.13)

does commute with both the Hamiltonian and pz , and hence the relativistic states are
characterised by quantum numbers n, k, and mj . The complete four-component
functions
q
are eigenfunctions of Dirac Hamiltonian with eigenvalues E = mc2

+
n,k,m
j

|n 1, k, mj 12 i

0
= C+
kc
1 ,
En +mc2 |n 1, k, mj 2 i

2
,
mc2

(7.14)

2c(n)1/2

n,k,m
=
C

|n, k, mj + 21 i

0
|n 1, k, mj + 12 i

2c(n)1/2
1 .
i
2 |n, k, mj +
En +mc2

En +mc
kc
En +mc
2 |n, k, mj

(7.15)

2
+ 21 i

Here n = 1, 2, 3, . . ., k may obtain any value and mj any half-integer value less or equal to
n 21 . When n = 0, only the second function is defined. The ml and ms are not defined
in the fully relativistic picture, but they are for the large components. By introducing a
quantum number n = n 12 (|ml | + ml ), the energy can be expressed as
=

k2
q
+
((2n + |ml |) + ml + 2ms )B
2m 2mc

(7.16)

which expresses the contributions of the dipole-field interaction explicitly. Therefore, the
quantised magnetic energy levels (Zeeman levels) arise naturally from Dirac theory.
84

7.2
7.2.1

Relativistic Hamiltonians for many-particle systems

DiracCoulomb Hamiltonian

As the essence of chemistry is a many-body quantum mechanical problem, it would be

tempting to introduce a way of treating interacting electrons within the Dirac picture
and hence obtain a proper description for the problem. The easiest way to construct a
relativistic Hamiltonian for a many-electron system would be to include the potential terms
that describe nucleus-electron and electron-electron Coulombic interactions,
!
X
X ZK
1X 1
,
(7.17)
H DC =
HiDirac +
+
r
2
r
iK
ij
i
K
i6=j
The potentials of this Dirac-Coulomb Hamiltonian affect instantly, and hence H DC is not
Lorentz covariant.

7.2.2

BetheSalpeter equation

The BetheSalpeter equation [64, 65], which is essentially a generalisation of the Dirac
equation for two particles, transforms properly under the Lorentz transformation. It is
an integro-differential equation, which at least in principle can be made accurate to an
arbitrarily high order. The derivation of the equation can be found from text books of
relativistic quantum mechanics, and the equation in the most compact form is written
as [66]




(1) + m(1) c (2) (2) + m(2) c (x(1) , x(2) )
Z
=
K(x(1) , x(2) , x(3) , x(4) )(x(3) , x(4) )dx(3) dx(4) ,
(7.18)
where superscripts denote particles. The operator K includes Dirac matrix operators and
x-dependent operators. The equation (7.18) can be written into different form separating
space and time variables for each particle as



(2)
(1)
(7.19)
D(1) i4 = K,
D(1) i4

where D(i) is the Dirac Hamiltonian for the free ith particle [Eq. (7.4)], and K is c2 (1) (2)
times the integral operator in (7.18). K can be derived from the quantum electrodynamics
as a series valid to any order. The evaluation of K and t The solution of the resulting
equation is problematic, in the use of the Bethe-Salpeter equation. In addition, and it
features different time variables for the two particles. This is essential for the covariance
but makes physical interpretation more difficult.
85

7.2.3

DiracCoulombBreit Hamiltonian

An approximate relativistic many-electron Hamiltonian can be constructed from the Dirac

Hamiltonian by adding the Breit term [67]


(i rij )(j rij )
1
Breit
i j
(7.20)
Hij =
rij
2rij2
that consists of the leading terms in the series obtained from the BetheSalpeter equation.
It describes other pairwise additive interactions than the standard Coulomb interaction of
electrons i and j, the first being the magnetic Gaunt term [68], and the second being a
term that describes retardation, i.e. the finite speed of interaction. With this term, the
DiracCoulombBreit (DCB) Hamiltonian is written as
H DCB =

X
i

HiDirac +

X 1
1 X 1
1 X Breit
+
+
Hij .
r
2
r
2
iK
ij
i,j
i,j
i,K

(7.21)

It implicitly includes all relativistic effects on the kinetic energy as well as spin-orbit (SO)
interaction. This Hamiltonian is, however, not Lorentz covariant, and it treats the effects
of quantum electrodynamics perturbationally up to O(2 ). Practical quantum chemistry
with H DCB is computationally too expensive for many purposes.
In addition, there are some theoretical defects in the DCB Hamiltonian, such as the
BrownRavenhall disease [69], or the continuum dissolution, in which any bound multiparticle state is degenerate with a state containing negative energy continuum solutions.
Another problem is related to the self-consistent field solution of the four-component equations [the DiracFock(Breit) method], in which solutions are obtained that not always are
true upper bounds to the exact solution [70]. This is called the finite basis set disease. It
can be at least partially treated using kinetic balance [71], which, however, leads to very
large basis sets. These problems have kept applications of true chemical interest largely,
but not completely, beyond the reach of rigorous four-component methods.

7.3

Quasi-relativistic Hamiltonians

Since chemistry is only concerned with the positive energy solutions of the Dirac equation,
it would be useful to decouple the positive and negative energy components of (7.9) to give
a two-component equation for the positive energy solutions, which could be used in the
same manner as NR equations. There are several ways to reduce H DCB into two-component
form. Such approximate Hamiltonians are usually called quasi-relativistic Hamiltonians.

7.3.1

Foldy-Wouthuysen transformation

One way to decouple the large and small component solutions is to apply the FoldyWouthuysen transformation [72] (FW) on the Dirac Hamiltonian.
86

For an electron in potential V , when the zero-level of energy is chosen to be +me c2 ,

Eq. (7.11) is written as





+
V
c~
+
=E
.
c~ V 2mc2

Elimination of the small component (ESC) gives


1
EV
1
c~
+,
1+
+ =
~ + X
= 2
2c + E V
2c
2c2

and therefore

+ = E+ .
H ESC + V + + c~ X

(7.22)

The Hamiltonian H ESC is dependent on the energy E and operates on the unnormalised
large component. In the normalised ESC approach [73], a normalised two-component wave
function is generated, which affects the sc. picture change [74].
Foldy and Wouthuysen [72] introduced a systematic procedure, the Foldy-Wouthuysen
(FW) transformation, for decoupling the large and small components to a successively
higher order of c1 by finding a unitary transformation that block-diagonalises the fourcomponent Hamiltonian.
For a positive-energy two-component effective Hamiltonian, the transformation matrices
!
1
1
X

1+X X
1+X X
U =
,
(7.23)
1
1
X
1+X X

U 1 = U =

1+X X

1+X X
1

1+X X

1
1+X X

1+X X

(7.24)

generate the desired two-component FW wave function


 

+

=U
.
(7.25)
0

in Eq. (7.22) is expanded with respect to (E V )/2c2

When the energy-dependent X
to the lowest order, X (1/2c2 )c~ p is obtained, and to order c2 in the absence of
electromagnetic fields

c~
p 
p2
1 8c
2
2
2c
U=
.
(7.26)
p
p2
c~2c
1

2
8c2
This, used with the FW transformation, gives rise to the Pauli Hamiltonian [75] (writing
instead of c1 ),
1
p2 1 2 4
(p 2 V ) + 2~ (V p).
(7.27)
2
8
4
Here the first two terms correspond to the NR Hamiltonian. The third and the fourth
term represent, respectively, the mass-velocity (mv) and Darwin (Dar) corrections that are
often called the scalar relativistic (SR) corrections. The last term is the SO coupling term.
H Pauli = V +

87

7.3.2

Breit-Pauli Hamiltonian

A two-component quasi-relativistic Hamiltonian for many-electron systems with the inclusion of the Breit term, i.e. the Breit-Pauli Hamiltonian, is gained from the Dirac-CoulombBreit Hamiltonian (7.21) for two-electron systems through the FW transformation and a
generalisation to N electrons [1, 3, 66]. Omitting terms that describe the motion of the
nuclei, the electronic part is written as
H BP =

hi +

1 X
hij ,
2 i,j

(7.28)

where the one-electron part is in the presence of electromagnetic fields

hi =
+

+
+

X ZK
K

riK

1 2
( + i~i i i )
2 i
1 2 2
(i + i~i i i )2
8
1 2 X ZK
~i (riK i )

3
4
r
iK
K
2X
ZK (riK ).

2
K

(7.29)
(7.30)
(7.31)
(7.32)
(7.33)

Eq. (7.29) is the non-relativistic nuclear attraction potential, (7.30) describes non-relativistic
kinematics and spin Zeeman interaction, term (7.31) is a relativistic correction to (7.30),
(7.32) is the one-electron relativistic spin-orbit term, and (7.33) represents the one-electron
Darwin effect. The two-electron part of the Breit-Pauli Hamiltonian consists of
1
rij
1 2 ~i (rij i ) ~j (rij j )

4
rij3

hij =

2 (rij )


1 2
j ( i rij )(rij j )

i
+
2
rij
rij3
1 2 ~i (rij j ) ~j (rij i )

2
rij3

1 2 8

(rij )(~i ~j )

4
3

(~i ~j ) 3(~i rij )(~j rij )

+
rij3
rij5
88

(7.34)
(7.35)
(7.36)
(7.37)
(7.38)

(7.39)

Here (7.34) is the Coulomb repulsion between the electrons, (7.35) the two-electron spinsame orbit term, (7.36) the two-electron Darwin term, (7.37) the orbit-orbit interaction,
(7.38) the spin-other orbit interaction, and (7.39) the electronic spin-spin interaction.
We will in the following form an approximate electronic Hamiltonian for a molecule
~ K = K IK , where K is the nuclear gyrocarrying a point-like nuclear magnetic dipole
magnetic ratio and IK the nuclear spin, in external magnetic field B0 , obtained by making
a substitution p + A into H BP . This substitution and the corresponding expansion
is presented in detail in Paper I. In this treatment, the vector potential A is expanded as
a sum A = A0 + AK + AL , where A0 is the vector potential corresponding to B0 in the
Coulomb gauge A = 0, and
1
B0 riO
2
IK riK
AK (ri ) = 2 K
3
riK
A0 (ri ) =

(7.40)
(7.41)

are the fields caused by the point-like magnetic dipole moments of nuclei K and L. In
Eq. (7.40), riO = ri RO , where O refers to the gauge origin. The terms containing a
third or higher power in B0 or a second or higher power in IK or IL are omitted. Explicit
forms and physical interpretations of the obtained terms are presented in Tables 7.1, 7.3.2
and 7.3.2.
From the one-electron part, (7.30) is expanded as
1X 2
PSO
DS
DSO
hKE + hOZ
+ hPSO
+ hSUSC
+ hDS
B0 + h K
L
KB0 + hLB0 + hKL
B02
2 i i
for the NR kinematics. The NR spin Zeeman term

(7.42)

becomes

1 X
1X
FC
FC
SD
SD
i
~i ( i i ) =
~i Bi = hSZ
B0 + h K + h L + h K + h L .
2 i
2 i

(7.43)

In (7.31), i2 and i~i i i do not commute when spatially non-uniform nuclear dipole
fields are present, and thus the term is first expanded as

1 X 4 1 2 X 2
R
i , ~i i i +
i i
Hkin
= 2
8
8
i
i
X
1 2
(~i i i )2 ,
+
8
i
3

(7.44)

Some text books, e.g. [66, 76], consider these terms, especially the Fermi contact interaction hFC , as
relativistic effects similarly with other terms that arise from the reduction from the Dirac equation to
(Breit-)Pauli form, such as the Darwin term. However, according to analysis by Kutzelnigg [77], these
hyperfine interactions are not relativistic effects, but may be derived also from the NR Levy-Leblond
equation.

89

where the anticommutator of the operators A and B is denoted as [A, B]+ = AB + BA.
Expanding the first term gives rise to
1 X 4
OZKE
PSOKE
DSKE
DSKE
hmv + hB
+ hK
+ hLPSOKE + hKB
+ hLB
2
0
0
0
8
i

PSOPSO
SUSCKE
PSOOZ
DSOKE
PSOOZ
+ hKL
+ hB
+ hLB
, (7.45)
+ hKL
+ hKB
2
0
0
0

and the electron-spin dependent second term in Eq. (7.44) becomes

X


1
1 X 2
2 i
i , ~i Bi +
i2 , ~i i i += 2
8
8
i
i

FCKE
SDKE
hSZKE
+ hK
+ hLFCKE + hK
+ hLSDKE .(7.46)
B0

R
The last term to arise from Hkin
is

1 2X
1 X
(~i i i )2 = 2
(~i Bi )2

8
8
i
i

dip
dip
condip
dipdip
con
hcon
+ hKL
, (7.47)
KB0 + hLB0 + hKB0 + hLB0 + hKL

in which the dependence on electron spin vanishes, i.e. (~i Bi )2 = Bi2 , as can be seen
when applying (~ A) (~ B) = A B + i~ A B. The corresponding con-con cross
term, which is proportional to the product of the contact fields of two nuclei that do
not coincide, vanishes everywhere in space. The one-electron SO term (7.32) divides into
field-independent and -dependent SO interactions
1 2 X ZK
SO(1)
SO(1)
SO(1)
~i (riK ) hSO(1) + hB0 + hK
+ hL .

3
4
r
iK
K

(7.48)

Combining Eqs. (7.35) and (7.38) of the two-electron part gives rise to field-dependent
and -independent two-electron spin-orbit interactions
1 ~i (rij i ) ~j (rij j )
SSO(2)
SOO(2)
2
hSSO(2) + hSOO(2) + hB0
+ h B0
4
rij3
SSO(2)

+hK

SOO(2)

+ hK

SSO(2)

+ hL

SOO(2)

+ hL

(.7.49)

From the first part of Eq. (7.37), two-electron counterparts of the operators in (7.42)
appear, but they are neglected here. This term gives rise to the field-independent electronelectron orbital interaction term, listed e.g. in Ref. [66], and it is included in Table 7.3.2.
The terms in Tables 7.17.3.2 are classified according to their appearance in the orders
of , and further in accordance with their dependence on i) the electron spin, ii) the nuclear
spin, and iii) the external magnetic field. Indices i, j are used to label electrons and K, L
to label nuclei. The operators with dependence on the index K will appear in a similar
form for nucleus L.
90

Table 7.1: The O(0 ) terms of the approximate Hamiltonian, H BP , in the presence of
point-like nuclear magnetic dipoles and an external magnetic field.
Label
hKE
hNE
hEE
hNQCC
K
hOZ
B0
hSUSC
B2
0

hSZ
B0

Description
Kinetic energy
Electron-nuclear Coulomb interaction
Electron-electron Coulomb interaction
Spin-dependence, no field-dependence
Nuclear quadrupole interaction
Field-dependence, no spin-dependence
Orbital Zeeman interaction
Diamagnetic susceptibility
Spin- and field-dependence
Electron spin Zeeman

Operator
P
21P i 2i
K
i,K rZiK
P 1
1
i,j rij

QK
21 IK (2I
K 1)
1
2
1
8

2 3r
1riK
iK riK
3
riK

P
Pi liO B0 2
i B0 (1riO riO riO ) B0

1
g
2 e

i si

B0

The nuclear quadrupole interaction (NQCC) listed in Table 7.1 does not arise naturally
from the present treatment, in which a point-like nucleus is assumed. It arises as the leading
non-trivial term from the treatment of the Coulombic Z/riK term in presence of a finite
i.e. physically correct nuclear model. However, its inclusion is necessary due to its major
importance in the consideration of the spectral parameters of nuclear magnetic resonance.
The operators of H BP are intuitively interpretable in terms of familiar NR concepts,
and provide an interpolation step between the NR and Dirac regimes. There are inherent
difficulties in Hamiltonians arising from the FW-transformation, as investigated e.g. by
Moss and co-workers [7881]. Farazdel and Smith criticised the mass-velocity term [82],
pointing out that for large momenta (p > 1 ), the mass-velocity term is not obtained
unless the expansion of the square-root operator outside its radius of convergence is used.
Most of the difficulties originate from the fact that the expansions implicitly or explicitly
rely on expansion (E V )/2c2 , which is invalid for particles in a Coulomb potential, where
there will always be a region of space close to the nucleus in which (E V )/2c2 > 1.
Furthermore, several terms pose a -function singularity at the origin, which is absent in
the Dirac Hamiltonian. These features are ostensibly problematic, as it has been demonstrated that relativistic effects originate mostly from the region close to nucleus. Numerous
quantitative studies verify, however, the practical feasibility of H BP for many atomic and
molecular properties.
Other difficulties arise also with regard to the self-consistent treatment of FW transformed Hamiltonians, as the eigenvalue equation H Pauli = E yields a continuous eigenvalue spectrum without lower bound due to the mass-velocity and SO operators, unless
the operator is not restricted to the solution space of the NR Hamiltonian [83].
Due to difficulties in the iterative solution of FW transformed Hamiltonians, perturbation theory has to be applied. In some cases, e.g. with the heaviest nuclei, the relativistic
91

hmv
hDar(1)
hDar(2)
hOO

Description
Scalar relativistic corrections
Mass-velocity term
Electron-nuclear Darwin term
Electron-electron Darwin term
Electron-electron orbital interaction

hSO(1)

Spin-dependence, no field-dependence
Electron spin
Spin-orbit interaction

hSSO(2)

Electron-electron spin-orbit interaction

hSOO(2)

Electron-electron spin-other-orbit interaction

hSSD(2)

Spin-spin dipolar interaction

hSSC(2)

Spin-spin contact interaction

Nuclear spin
Orbital hyperfine interaction
Electron and nuclear spins
Dipolar hyperfine interaction
Fermi contact hyperfine interaction

hPSO
K
hSD
K
FC
hK

Operator
P 4
2
81 P
i i
2

i,K ZK (riK )
2
P
2

2
i,j (rij )
P
(i rij )(rij j )
1 2
i j

i,j rij
4
r3
ij

P
1 2
ge i,K rZ3K si liK
4
P iK l
41 2 ge i,j si rij3
ij
P
lij
1 2
2 ge i,j sj r3
ij
P rij2 (si sj )(si rij )(rij sj )
1 2 2
ge i,j
5
4
rij
ge2 2 P
3
i,j si sj (rij )
2 K

i IK

liK
3
riK

2
P
3riK riK 1riK
IK
5
i si
r
iK
P
i (riK )si IK

1 2
ge K
2
4 2
ge K
3

Table 7.2: Partial listing of O(2 ) terms of the approximate BreitPauli Hamiltonian.

92

Label

(Table 2.2 continued)

OZKE
hB
0
SUSCKE
hB
2
0

93

hSZKE
B0
SO(1)
h B0
SSO(2)
h B0
SOO(2)
h B0

hDS
KB0
PSOOZ
hKB
0

Field-dependence, no spin-dependence
Orbital Zeeman kinetic energy correction
Diamagnetic susceptibility kinetic energy correction
Spin- and field-dependence
Electron spin
Electron spin Zeeman kinetic energy correction
Spin-orbit Zeeman gauge correction
Electron-electron spin-orbit Zeeman gauge correction
Spin-other-orbit Zeeman gauge correction
Nuclear spin
Electron nuclear Zeeman modification
Orbital hyperfine-orbital Zeeman interaction

P
2
i
Pi liO B0
2
2
B

[
,
0
i (1riO riO riO )]+ B0
i

1 2

4
1 2

32

1 2
ge i si B0 2i
4
P
P
1 2
ge N ZN i si 1(riO riNr3)riO riK
8
iK
P
1(r r )r r
18 2 ge i,j si iO ijr3 iO ij B0
ij
P
1(r r )r r
41 2 ge i,j sj iO ijr3 iO ij B0
ij

1 2
K
2
1 2
K
4

P
P

i IK

i IK

1(riO riK )riO riK

3
riK
i
liK
B0
3 , liO
riK
+

B0

B0

Label

Description
Spin-dependence, no field-dependence
Nuclear spin

Operator

PSOKE
hK

Kinetic energy correction of the orbital hyperfine interaction

Electron coupled nuclear spin-spin interaction

1 4
K
4

hDSO
KL

1 4
K L
2

DSOKE
hKL

Kinetic energy correction of the electron coupled nuclear spin-spin interaction

1 6
K L
8

PSOPSO
hKL

Orbital hyperfine coupled nuclear spin-spin interaction

1 6
K L
2

dipdip
hKL

SDKE
hK
FCKE
hK
SO(1)

hK

SSO(2)

hK

SOO(2)

hK

P
P
P

hcon
KB0
hdip
KB0

i IK

i IK

i IK

1(riK riL )riK riL

IL
3 r3
riK
iL
i
h
iK riL
2i , 1(riK rr3iL )r
3
iK riL
+

liK liL
3 , r3
riK
iL +

IL

IL

i
h
2
2
P
3riK riK 1riK
3riL riL 1riL
6
IL
+
(r
)

(r
)
Contact hyperfine coupled nuclear spin-spin in- 2
5
5
iL
K L
iK
i K
3
riL
riK
teraction
P
2
2
Dipolar hyperfine coupled nuclear spin-spin in- 14 6 K L i IK r5 1r5 [9riK riL (riK riL ) 3riK riL (riK
+ riL
)
iK iK
teraction
2
2
+1(riK
riL
)] IL
Electron and nuclear spins
i
h
P
3r r 1r 2
Kinetic energy correction of the dipolar hyper- 81 4 ge K i si 2i , iK riK5 iK IK
iK
+
fine interaction
P
Kinetic energy correction of the Fermi contact 3 4 ge K i si [2i , (riK )]+ IK
interaction
P
P
1 4
iK riN
Spin-orbit hyperfine correction
IK
ge K N ZN i si 1(riK rr3iN r)r
3
4
iK iN
P
1(r
r
)r
r
Spin-same-orbit hyperfine correction
14 4 ge K i,j si iK r3ij r3 iK ij IK
iK ij
P
1(r r )r r
Spin-other-orbit hyperfine correction
12 4 ge K i,j sj iK r3ij r3 iK ij IK
iK ij

Spin- and field-dependence

Nuclear spin
DSKE
hKB
0

iK

94

condip
hKL

i
h
2 liK
I

,
i r3
i K

1 4
K
8

)riO riK
2i , 1(riO riK
r3

Kinetic energy correction of the electron nuclear

i IK
iK
Zeeman modification
P
Nuclear spin Zeeman contact hyperfine interac- 2
4 K i (riK )IK B0
3
tion
P
3r r 1r 2
Nuclear spin Zeeman dipolar hyperfine interac- 14 4 K i IK iK riK5 iK B0
iK
tion

B0

Table 7.3: Partial listing of O(4 ) and O(6 ) terms of the approximate Hamiltonian.

Figure 7.1: A schematic illustration of the behaviour of ZORA and Pauli approximations
with respect to the Dirac equation. Drawn after Ref. [74].
effects may be so large that a perturbation theoretical approach is inadequate.

7.3.3

Other quasi-relativistic Hamiltonians

As discussed above, there are difficulties in two-component transformations of the Dirac

Hamiltonian, which arise in the form of divergent terms and singularities when r 0.
An alternative approach for decoupling the large and small components of the Dirac
equation is the zeroth-order regular approximation (ZORA) Hamiltonian [84]. The derivation of the ZORA Hamiltonian proceeds almost in complete analogy with that of the Pauli
Hamiltonian, except here the expansion is carried out with respect to E/(2c2 V ). In
the derivation of the ZORA Hamiltonian, one proceeds almost in complete analogy with
Pauli Hamiltonian, but the expansion is carried out with respect to E/(2c2 V ). The
Hamiltonian becomes
1
i
H ZORA = V + pKp + ((pK) p),
2
2

(7.50)

where K = (1 21 2 V )1 . It has similarities with the earlier Chang-Pelisser-Durand zerothorder effective Hamiltonian [85]. It has been shown that ZORA behaves well near to nucleus, and that it is variationally stable at zeroth-order. ZORA contains similar relativistic
corrections as are present in the Pauli Hamiltonian but in a regularised form. The inclusion
of the first-order in the expansion yields FORA (first-order regular approximation). ZORA
is bounded from below, up to Z = 137. It has the flaws of not being gauge invariant, and
the ignoring the change in the metrics is problematic. The features and differences between
the Pauli and ZORA approximations are outlined in Figure 7.1, in which the x-axis is the
distance r from the nucleus.
The Pauli approximation breaks down near the nucleus but approaches the correct limit
at large r. ZORA is bounded from below, up to Z = 137, and it is a good approximation
for |V | |E| > 0, i.e. where the potential dominates, but it does behave erroneously when
95

energies and distances are large [74,86]. for large energies and large distances from nucleus
ZORA possesses a dependence on the choice of the zero of the potential, nd in it only
the eigenvalue spectrum, not the eigenfunctions, reflect the actual physics. This is often
referred to as the picture change effect. This difference is due to the fact that the change
of picture [87] is neglected and that the ZORA Hamiltonian does not contain all terms of
the 2 order.
The scaled ZORA includes the change in metric, but the picture change is still ignored.
There are other ideas for the regular approximation scheme. In exponential regular approximation (ERA) [88], an exponential function is added to the definition of the small
component in order to correct the metrics. The idea is to remove the singularity as with
ZORA, without introducing the long-range dependence into the ansatz. In modified ERA
(MERA) [88], part of the relativistic contribution is omitted to avoid spin-orbit terms in
the metric. In infinite-order regular approximation (IORA) [89, 90] and its MERA-type
modification (MIORA) [88], the expansion may be taken to infinite order by using an unnormalised FW transformation, which results in the ZORA Hamiltonian and a non-unit
metric. Filatov and Cremer [91] presented a scaled IORA (SIORA) method, in which
an effective energy-independent relativistic Hamiltonian was obtained by a regular expansion of the exact Hamiltonian for electronic states in terms of linear energy-independent
operators.
Another alternative is an approach with external field projectors. This method has
been developed by Hess [92,93] on the basis of theoretical work by Sucher [94] and Douglas
and Kroll [95]. This scheme is often called the Douglas-Kroll-Hess (DKH) method. In the
method, the FW transformation is carried out up to the first order in the momentum space,
and the result is then split into odd and even parts. After that,
p the FW transformation
is repeated to order n, with antihermitian matrices Wn , Un = 1 + Wn2 + Wn . The form
of the W s is chosen to cancel certain terms. The resulting Hamiltonian is decoupled, i.e.
it operates on the large and small components independently, to order n + 1. Already U1
includes several relativistic phenomena. DKH is widely used in molecular calculations. It
has the benefit of being bounded from below, with no unphysical states occuring. It also
approaches the correct NR limit.
Several quasi-relativistic schemes based on the elimination of the small component or
other approach have been presented in Refs. [96102].
Yet another alternative is to apply perturbation theory straight to the Dirac equation,
and to change the metrics between the large and small component, yielding direct (or
Dirac) perturbation theory (DPT) [103105]. This approach gives non-singular forms for
energy and wave function to arbitrary order. However, it is computationally almost as
expensive as using the four-component framework, and its implementation is considered
to be difficult.
Besides in the Hamiltonian, the relativistic effects can partly be included as artifacts
in the formally non-relativistic calculations themselves by using the relativistic effective
core potential (ECP) approach (see e.g. Refs. [106] or [107] and references therein).

96

Appendix A
Second quantization
A.1

Basic concepts

In the standard formulation of quantum mechanics (such as that employed in the previous
chapters) observables are represented by operators and states as functions. In second
quantization (SQ) formalism [108], also the wave functions are represented in terms of
operators.

A.1.1

Occupation number vectors

In the SQ formalism, each Slater determinant (1.27) of M orthonormal spin-orbitals is

represented by an occupation number vector (ON) k,

1
P occupied
|ki = |k1 , k2 , . . . , kM i ,
kP =
(A.1)
0 P unoccupied
in an abstract vector space called the Fock space.
For an orthonormal set of spin-orbitals, we define the inner product between two ON
vectors as
M
Y
hm |ki =
m P kP .
(A.2)
P =1

This definition is consistent with the overlap between two Slater determinants, but has a
well-defined but zero overlap between states with different electron numbers is a special
feature of the Fock-space formulation. It allows for a unified description of systems with
variable numbers of electrons.

In a given spin-orbital basis, there is a one-to-one mapping between the Slater determinants with spin-orbitals in a canonical order and the ON vectors in the Fock
space.
However, ON vectors are not Slater determinants: ON vectors have no spatial structure but are just basis vectors in an abstract vector space.
97

 The Fock space can be manipulated as an ordinary inner-product vector space.

The ON vectors constitute an orthonormal basis in the 2M dimensional Fock space
F (M ), that can be decomposed as a direct sum of subspaces,
F (M ) = F (M, 0) F (M, 1) F (M, N ),

(A.3)

where F (M, N ) contains all ON vectors obtained by distributing N electrons among the
M spin-orbitals, in other words, all ON vectors for which the sum of occupation number
is N . The subspace F (M, 0) is the vacuum state,
|vaci = |01 , 02 , . . . , 0M i .

(A.4)

Approximations to an exact N -electron wave function

 expressed in terms of vectors in
 are
M
.
the Fock subspace F (M, N ) of dimension equal to
N

A.1.2

Creation and annihilation operators

In second quantization, all operators and states can be constructed from a set of elementary
creation and annihilation operators. The M creation operators are defined by
aP |k1 , k2 , . . . , 0P , . . . , kM i = kP |k1 , k2 , . . . , 1P , . . . , kM i
aP |k1 , k2 , . . . , 1P , . . . , kM i = 0
where
kP

P
1
Y

(1)kQ .

(A.5)
(A.6)

(A.7)

Q=1

The definition can be also combined a single equation:

aP |k1 , k2 , . . . , 0P , . . . , kM i = kP 0 kP |k1 , k2 , . . . , 1P , . . . , kM i

(A.8)

Operating twice with aP on an ON vector gives

aP aP |k1 , k2 , . . . , 0P , . . . , kM i = aP kP 0 kP |k1 , k2 , . . . , 1P , . . . , kM i = 0,
therefore
aP aP = 0.
For Q > P ,
aP aQ |. . . , kP , . . . , kQ , . . .i = aP kQ 0 kQ |. . . , kP , . . . , 1Q , . . .i

= kP 0 kQ 0 kP kQ |. . . , 1P , . . . , 1Q , . . .i .
98

(A.9)

Reversing the order of the operators,

aQ aP |. . . , kP , . . . , kQ , . . .i = aQ kP 0 kP |. . . , 1P , . . . , kQ , . . .i

= kP 0 kQ 0 kP (kQ ) |. . . , 1P , . . . , 1Q , . . .i .

and combining the results, we obtain

aP aQ + aQ aP |ki = 0.
Substitution of dummy indices shows that this holds for Q < P as well, and is true also
for Q = P . Since |ki is an arbitrary ON vector, we conclude the anticommutation relation
aP aQ + aQ aP = [aP , aQ ]+ = 0.

(A.10)

The properties of the adjoint or conjugate operators aP can be reasoned from those of
the creation operators. Thus, the adjoint operators satisfy the anticommutation relation
[aP , aQ ]+ = 0.

(A.11)

Let us invoke the resolution of the identity:

X
aP |ki =
|mi hm |aP | ki ,
m

where, using Eq. (A.8),

hm |aP | ki = hk |aP | mi =

mP 0 m
P
0

if kQ = mQ + QP
.
otherwise

K
From the definition of and from kQ = mQ + QP we see that m
P = P . We may therefore
write the equation as

kP 1 kP if mQ = kQ QP
hm |aP | ki =
.
0
otherwise

Hence, only one term in aP |ki survives:

aP |ki = kP 1 kP |k1 , . . . , 0P , . . . , kM i .

(A.12)

The operator aP is called the annihilation operator.

Combining Eqs. (A.8) and (A.12), we get
aP aP |ki = kP 1 |ki
aP aP |ki = kP 0 |ki

(A.13)
(A.14)

aP aP + aP aP = 1.

(A.15)

that leads to

99

For P > Q, we have

aP aQ |ki = kP 0 kQ 1 kP kQ |k1 , . . . , 0Q , . . . , 1P , . . . , kM i

aQ aP |ki = kP 0 kQ 1 kP kQ |k1 , . . . , 0Q , . . . , 1P , . . . , kM i ,
thus we have the operator identity
aP aQ + aQ aP = 0

P > Q.

(A.16)

Hence,
aP aQ + aQ aP = [aP , aQ ]+ = P Q .

A.1.3

(A.17)

Number-conserving operators

The occupation-number operator that counts the number of electrons in spin-orbital P is

introduced as
P = a aP ,
P |ki = kP |ki
N
N
(A.18)
P
= N
, as well as idempotent, N
2 =
The occupation-number operators are Hermitian, N
P
P . Using the basic anticommutation relations, we obtain
N
P , a ] = P Q a
[N
Q
Q
h
i
P , aQ = P Q aQ
N

(A.19)

P , X]
= NPX X,

[N

(A.21)

(A.20)

consisting of creation and annihilation operators, e.g. X

=
So for an arbitrary string X

become
aP aQ aP aR aS , the commutators with N

minus the number of times aP occurs.

where NPX is the number of times aP occurs in X
All occupation-number operators added together in the Fock space gives the Hermitian
operator
M
X

(A.22)
N=
aP aP ,
P =1

|ki = N |ki, and is therefore

which returns the number of electrons in an ON vector, N
known as the particle-number operator. We find that the number operator commutes with
any string T that contains an equal number of creation and annihilation operators. Such
to a
strings are called number-conserving. In general, the application of the string X
Fock-space vector increases the number of electrons by N X .
The simplest number-conserving operators are the elementary excitation operators
EQP = aP aQ ,
100

(A.23)

which give as applied to an ON vector

aP aQ |ki = aP kQ 1 kQ |k1 , . . . , 0Q , . . . , kP , . . . , kM i

where

= kP 0 kQ 1 kQ kP P Q |k1 , . . . , 0Q , . . . , 1P , . . . , kM i


 kP 1
k k

= P Q Q P (1 kP )kQ 
kQ 0
P Q =

(A.24)

1 P Q
.
1 P > Q

The case P < Q differs from this result only in the interpretation of P Q , whereas the case
P = Q is covered by Eq. (A.18). Therefore we may write


 kP 1

k k
aP aQ |ki = P Q P Q (1 kP + P Q )kQ 
.
(A.25)
kQ P Q
The application of a single such operator gives a single excitation, two a double excitation
and so forth.

A.2

The representation of one- and two-electron operators

Expectation values of Hermitian operators correspond to physical observables and should

thus be independent of the representation of operators and states. Therefore, we require
the matrix element of a second-quantization operator between two ON vectors to be equal
to its first-quantization counterpart.
To determine the numerical parameters fP Q in the second-quantization representation
of one-electron operators, we evaluate the matrix elements of f between two ON vectors
and compare with the usual Slater rules for matrix elements between determinants. We
are lead to the identification
D   E
 
fP Q = P fc  Q
(A.26)
gP QRS = hP R |
g c | Q S i .

(A.27)

The parameters gP QRS possess the particle-interchange symmetry,

gP QRS = gRSP Q .

(A.28)

In the case of real spin-orbitals, the integrals exhibit the following permutational symmetries:
fP Q = fQP
gP QRS = gQP RS = gP QSR = gQP SR ,
101

(A.29)
(A.30)

of which the latter may be further combined with Eq. (A.28) to yield an eight-fold symmetry.
Using the introduced second-quantization representation, we may now construct the
(non-relativistic, spin-free) molecular electronic Hamiltonian in the BornOppenheimer
approximation,
X
1
=
H
hP Q aP aQ + gP QRS aP aR aS aQ + hnuc ,
(A.31)
2
PQ
where
hP Q =

X ZI
1
P (x) 2
2
riI
I

Q (x)dx

P (x)R (x)Q (x)S (x)

dx1 dx2
r12
1 X ZI ZJ
=
.
2 I6=J RIJ

gP QRS =
hnuc

(A.32)
(A.33)
(A.34)

The form of the second-quantization Hamiltonian can be interpreted in the following way:
Applied to an electronic state, the Hamiltonian produces a linear combination of the original state with states generated by single and double electron excitations from this state.
With each such excitation, there is an associated amplitude hP Q or gP QRS , which represents
the probability of this event happening.
c and C c be one-electron operators in first quantization and A,
B
and C be
Let Ac , B
the corresponding second-quantization representations.
c , where a and b are numbers, is represented
The first-quantization operator aAc + bB

by aA + bB.
The standard relations
B
C)
= (AB)
C
A(
= B
A
(AB)

(A.35)
(A.36)

are valid.
For a complete one-electron basis
c C = AB

C c = Ac B
but for finite basis sets this expression does not hold. The second-quantization operators are projections of the exact operators onto a basis of spin-orbitals. For an
incomplete basis, the second-quantization representation of an operator product depends on when the projection is made.
102

First quantization
One-electron operator:
fc =

N
X

Second quantization
One-electron operator:
X
f =
fP Q aP aQ

f (xi )

PQ

i=1

Two-electron operator:
c

g =

N X
N
X

Two-electron operator:
g =

g(xi , xj )

i=1 j=i+1

Operators are independent of the spinorbital basis

Operators depend on the number of electrons
Operators are exact

1 X
gP QRS aP aR aS aQ
2 P QRS

Operators depend on the spin-orbital basis

Operators are independent of electrons
Projected operators

Table A.1: Comparison between first and second quantization representations.

Second quantization treats operators and wave functions in a unified way they are all
expressed in terms of the elementary creation and annihilation operators: any ON vector
may be written as
"M
#
Y
k |vaci =
|ki = X
(aP )kP |vaci
(A.37)
P =1

and therefore the matrix elements can be viewed as the vacuum expectation value of an
operator


E
D   E D
 
 
vac
,
(A.38)
X
O
X
m
=
vac
k O
 k

m

and expectation values become linear combinations of vacuum expectation values.

A.3

Commutators and anticommutators

In the manipulation of operators and matrix elements in second quantization, the commu B]
= AB
B
A and the anticommutator [A,
B]
+ = AB
+B
A of two operators are
tator [A,
often encountered. Referring to the basic relations of the elementary creation and annihilation operators, it is usually possible to simplify the commutators and anticommutators
between strings of elementary operators considerably.
We need to introduce the concepts of operator rank and rank reduction.
The (particle) rank of a string of creation and annihilation operators is simply the
number of elementary operators divided by 2.
103

 Rank reduction is said to occur when the rank of a commutator or anticommutator

is lower than the combined rank of the operators commuted or anticommuted.
A simple rule
Rank reduction follows upon anticommutation of two strings of half-integral rank
and upon commutation of all other strings
is useful for simplifying the expressions.
One useful strategy for the evaluation of commutators and anticommutators is based
on their linear expansion in simpler commutators or anticommutators according to e.g. the
following operator identities, and aim at greatest possible rank reduction:

h
h
h

B
1 B
2
A,

B
1 B
n
A,
h

B
1 B
2
A,

B
1 B
n
A,
h

B
1 B
2
A,

B
1 B
n
A,

B
1 ]B
2 + B
1 [A,
B
2 ]
= [A,
=

n
X
k=1

1 B
k1 [A,
B
k ]B
k+1 B
n
B

B
1 ]+ B
2 B
1 [A,
B
2 ]+
= [A,
n
X

(A.39)
(A.40)
(A.41)

1 [A,
B
k ]+ B
n (n even)
(1)k1 B

(A.42)

B
1 ]B
2 + B
1 [A,
B
2 ]+ = [A,
B
1 ]+ B
2 B
1 [A,
B
2 ]
= [A,

(A.43)

k=1

n
X
k=1

1 [A,
B
k ]+ B
n (n odd).
(1)k1 B

(A.44)

Let us consider the simplest nontrivial commutator, [aP , aQ aR ]. Moving one of the
operators after the comma out of the commutators by using (A.41),
[aP , aQ aR ] = [aP , aQ ]+ aR aQ [aP , aR ]+

(A.45)

and invoking the basic anticommutation relation we have

[aP , aQ aR ] = P R aQ .

(A.46)

[aP , aQ aR ] = P Q aR .

(A.47)

Similarly,
Then, let us evaluate the commutator between two excitation operators, using the
results from the previous example,
[aP aQ , aR aS ] = [aP , aR aS ]aR + aP [aQ , aR aS ] = QR aP aS P S aR aQ .
104

(A.48)

A slightly more complicated commutator is evaluated by applying Eq. (A.39) and the
previous result:
[aP aQ , aR aS aM aN ] = [aP aQ , aR aS ]aM aN + aR aS [aP aQ , aM aN ]
= QR aP aS aM aN P S aR aQ aM aN

+QM aR aS aP aN P N aR aS aM aQ .

(A.49)

The following double commutator can be evaluated by invoking the result (A.48) twice
[aP aQ , [aR aS , aM aN ]] = SM [aP aQ , aR aN ] RN [aP aQ , aM aS ]

= SM QR aP aN SM P M aR aQ RN QM aP aS
+RN P S aM aQ .

(A.50)

This gives rise to a rank reduction by 2.

A.4

Orbital rotations

In many situations, e.g., during the optimization of an approximate electronic state or in

the calculation of the response of an electronic state to an external perturbation, it becomes
necessary to carry out transformations between different sets of orthonormal spin-orbitals.
We shall encounter especially cases where the occupation numbers refer to a set of
orthonormal spin-orbitals P that is obtained from another set P by a unitary transformation
X
P =
Q UQP .
(A.51)
Q

Here, a unitary matrix U is a matrix that fulfills the relation

U U = UU = 1.

(A.52)

The spectral theorem states that any unitary matrix can be diagonalized as
U = VV ,
where V is unitary and a complex diagonal matrix, kk = exp(ik ). Therefore any unitary
matrix can be written as the matrix exponential1 of an anti-Hermitian matrix:
U = V exp(i)V = exp(iVV )
exp(X),
X = X .
By an introduction of the anti-Hermitian operator
X

=
P Q aP aQ ,

=
,
PQ

The matrix exponential is defined as exp(A) =

n=0

105

An /n!.

(A.53)

(A.54)

where the parameters P Q are the elements of an anti-Hermitian

matrix , for which
 

U = exp(), the elementary operators a

P , a
P and state 0 generated by the unitary
transformation (A.51) can be expressed in terms of the untransformed operators and states
as

A.5

a
P = exp(
)aP exp(
)
a
= exp(
)aP exp(
)
 P
0 = exp(
) |0i .

(A.55)
(A.56)
(A.57)

Spin in second quantization

In the formalism of second quantization, there is no reference to electron spin. However,

in non-relativistic theory, many important simplifications follow by taking spin explicitly
into account.

A.5.1

Notation

We shall separate the orbital space and the spin space,2 that is, a spin orbital will be
denoted by P (x) = p (r) () = p . The theory of second quantization holds unchanged;
for example the basic anticommutator now becomes
[ap , aq ]+ = p,q = pq ,

(A.58)

where, for example, ap is the creation operator associated with the spin-orbital p .

A.5.2

Operators in the orbital basis

We may classify the quantum-mechanical operators according to how they affect the orbital
and spin parts of wave functions: Spin-free operators, spin operators and mixed operators.
One-electron are written in the spin-orbital basis as
XX
f =
fp,q ap aq .
p

The integrals vanish for opposite spins in the case of spin-free (or spinless) operators:
Z
fp,q =
p (r) ()f c (r)q (r)()drd
Z
X
= p (r)f c (r)q (r)dr = fpq =
fpq Epq ,
pq

Hereafter, generic spin functions , , and , which may have a value of either or , will be
encountered. Note the slight difference in the notation with the course text book [26].

106

where we have introduced the singlet excitation operator

Epq = ap aq + ap aq .

(A.59)

The two-electron operators are given by

1 X
gp,q,r,s ap ar as aq .
g =
2 pqrs,
Here, the orthogonality of the spin functions reduce the integrals to
gp,q,r,s = gpqrs ,

(A.60)

where gpqrs are the two-electron integrals in ordinary space, that are always symmetric
with respect to the particle-interchange, gpqrs = grspq , and have the Hermitian symmetry

gpqrs = gqpsr
in the case of complex orbitals; and in the case of real orbitals, we have a
permutational symmetry
gpqrs = gqprs = gpqsr = gqpsr .
The second-quantization representation of a spin-free two-electron operator is then
X
1X
1X
1X
g =
gpqrs
gpqrs (Epq Ers qr Eps ) =
gpqrs epqrs . (A.61)
ap ar as aq =
2 pqrs
2 pqrs
2 pqrs

We have introduced the two-electron singlet excitation operator in the last equality.
We can now express the spin-free, non-relativistic molecular electronic Hamiltonian in
the orbital basis:
X
1X
=
H
hpq Epq +
gpqrs epqrs .
(A.62)
2 pqrs
pq

The one- and two-electron integrals are the same as in (A.31), except that the integrations
are over the spatial coordinates only.
The one-electron second-quantization operators associated to operators that work in
spin space only may be written in the general form
XZ
f =
p (r) ()f c ()q (r)()drdap aq
p q

XZ

()f c ()()d

ap ap .

(A.63)

Three important examples of pure spin operators are the raising and lowering operators
S+c and Sc , and Szc . The effect of these operators on the (one-particle state) spin functions
is
S+c =
Sc = 0
1
Szc =
2

S+c = 0
Sc =

1
Szc = + .
2

107

From these we arrive at the following expressions for the basic spin operators:
X
S+ =
ap ap

(A.64)

S =

ap ap

(A.65)


1 X

ap ap ap ap .
Sz =
2 p

(A.66)

The lowering operator is seen to be the Hermitian adjoint of the raising operator:
X
X
S+ =
(ap ap ) =
ap ap = S .
p

The operators for the x and y components of the spin angular momentum can be written
in terms of the raising and lowering operators as


1 c
Sxc =
S+ + Sc
2


1
S+c Sc ,
Syc =
2i
from which we obtain the second-quantization counterparts,

1 X
Sx =
ap ap + ap ap
(A.67)
2 p

1 X
Sy =
ap ap ap ap ,
(A.68)
2i p
which may be combined with (A.66) to give the operator for the total spin, S2 = Sx2 + Sy2 +
Sz2 . As it contains products of one-electron operators, it is a two-electron operator and
therefore somewhat tedious to manipulate. However, the explicit form is seldom needed,
because we are able to employ the standard operator identities:
In contrast to the orbital basis, the spin basis is complete, hence the usual firstquantization commutation relations hold also for the second-quantization spin operators.
One such example is
"
#

X
X
X

[ S+ , S ] =
ap ap ,
a , aq =
ap ap a ap = 2Sz .
(A.69)
q

For example, the fine-structure and hyperfine-structure operators of first quantization

affect both the spatial and spin parts of the wave function. An example would be the
spin-orbit interaction operator (see Section 7.3.2)
c
VSO

N
X
i=1

(riK )ci Sci .

108

The second-quantization representation is obtained in the usual manner as

XZ
c
VSO =
p (r) ()VSO
(r, )q (r)()drdap aq
p q


X
x
y
z
,
Vpqx Tpq
+ Vpqy Tpq
+ Vpqz Tpq

(A.70)

pq

where ( = x, y, or z)
Vpq

p (r)(r)c q (r)dr

and where the Cartesian components of the triplet excitation operators are given by

1
x
ap aq + ap aq
Tpq
=
2

1 
y
Tpq
=
ap aq ap aq
2i

1
z
ap aq ap aq .
Tpq
=
2

(A.71)
(A.72)
(A.73)

The second-quantization representation of

spin-free operators depend on the orbitals but have the same amplitudes (=integrals)
for alpha and beta spins
pure spin operators are independent of the functional form of the orbitals
mixed operators depend on both the spin of the electron and the functional form of
the orbitals.

A.5.3

Spin properties of determinants

Slater determinants are in general not eigenfunctions of the non-relativistic Hamiltonian

(A.62), but are instead non-degenerate eigenfunctions of the spin-orbital occupation-number
operators:
p |ki = kp |ki .
N
(A.74)
We also note that the spin-orbital ON operators commute with the spin-projection operap ] = 0, because the ON operators commute among themselves, and Sz is a
tor, i.e. [Sz , N
linear combination of spin-orbital ON operators. Added by the observation that there is
no degeneracies in (A.74), the Slater determinants are eigenfunctions of the projected spin:
Sz |ki = M |ki .

(A.75)

By contrast, the Slater determinants are not eigenfunctions of the total spin. However, it
is possible to determine spin eigenfunctions as simple linear combinations of determinants.
109

A hint for the procedure is given by an observation that both the total and projected-spin
operators commute with the sum of the ON operators for alpha and beta spins:
h
i
p + N
p = 0
Sz , N
h
i
2

S , Np + Np = 0,

p + N
p are singlet operators. Such spin-adapted functions are known as conbecause N
figuration state functions.

110

Appendix B
Performance of the
electronic-structure models
Having now considered the most important approximations to the molecular electronic
wave function we will now address their performance in terms of a few numerical examples [26]. It should be remembered that none of these models (besides FCI, which is
inapplicable for most cases) is an all-around-good one: good performance in some problem
and/or in some system does not guarantee good performance in another situation.

B.1

Total electronic energies

For the total molecular energy, i.e., E in the molecular Schrodinger equation, there is
no direct experimental counterpart. We examine it in order to establish a feeling on the
severity of the approximations involved in the calculation. We should recall that there were
a third class of approximations in addition to the truncation of one- and N -electron spaces:
To investigate the validity of the use
approximations in the molecular Hamiltonian H.
of the non-relativistic Hamiltonian, we include the leading-order one-electron relativistic
corrections that include the spin-orbit interaction (SO) (7.32), mass-velocity (MV) , and the
Darwin (Dar) corrections, see Table 7.3.2. The leading-order two-electron contributions,
such as the two-electron Darwin contribution and the spin-spin contact interaction, are
smaller by at least one order of magnitude. The MV and Dar corrections are always
of opposite sign. The calculation is carried out using the CCSDT model for the water
molecule in the cc-pCVXZ bases at a CCSD(T)/cc-pCVQZ geometry and presented in
Table B.1.
Some observations:
It is remarkable how a tiny, less than one per cent fraction of the total energy, plays
such an important role in the determination of molecular properties.
The leading-order relativistic correction, -52.18 mEh , is larger than the triples contribution already in water, and gets far more pronounced in systems with heavier
111

elements present!
(T) is a very good approximation to full triples.

DZ
TZ
QZ
5Z

Total energy [Eh ]

CCSDT
HF
-76.24121 -76.02680
-76.33228 -76.05716
-76.35981 -76.06482
-76.36899 -76.06708

HFSD
-211.21
-267.40
-285.98
-292.44

Energy corrections [mEh ]

SD(T) (T)T SO
MV
-3.04
-0.16
-7.65
-0.08
-8.99
-0.01
0.00 -255.87
-9.52
0.04

Dar

203.69

Table B.1: Breakdown of the total electronic energy of water molecule.

B.2

Molecular equilibrium geometries

One of the most common applications of molecular electronic-structure calculations is

the determination of the molecular equilibrium structures. Good geometries are also of
fundamental importance in study of molecular properties.
We consider the mean errors and standard deviations of calculated bond distances and
bond angles as compared to experimental results using a hierarchy of N -electron models
and a hierarchy of Gaussian basis sets. The error is given by = Picalc Piexp , and the
mean error and standard deviation by
n

X
= 1

i
n i=1
v
u
n
u 1 X
t
2.
std =
(i )
n 1 i=1

(B.1)

(B.2)

They can be conveniently represented graphically in terms of the normal distribution

#
"

2
1 P
.
(B.3)
(P ) = Nc exp
2
std
The calculations were carried out for sets of small molecules containing main-group elements. 29 bond distances and 10 bond angles were calculated and presented in Figures B.1
and B.2, respectively. Figure B.3 illustrate the convergence of the hierarchy of correlation
models.
Some observations:
The HartreeFock model is suitable for only preliminary calculations.
112

 The CCSD(T) model is capable of providing structures both bond distances and
angles of high quality.
The MP2 model is less accurate but more generally applicable as it is considerably less
expensive. Higher-order MPPT is significantly less accurate but hideously expensive,
i.e., useless.
The performance of the CCSD model is disappointing. It however performs more
robust than MP2 in some selected cases with complicated electronic structure.
The CISD model is also rather useless, showing no improvement when the basis set
is increased.
The basis-set requirements are different for different models. Whereas there is no
much use to use QZ-level basis sets with the HartreeFock model (as the results
are not improved), the correlated models require at least a TZ-level basis set, but
preferably the cc-pVQZ basis.
The results suggest the following recipe for the determination of molecular structures:
obtain an initial structure with HF using a DZ-level basis and improve it by using MP2
and a TZ basis. If ever possible, calculate the final geometry in the cc-pVQZ basis using
the CCSD(T) model.

B.3

Molecular dipole moments

The comparison between the experimental and calculated molecular dipole moments is
difficult, as the experiments are measuring the dipole moment in the vibrational ground
state 0 , whereas the calculations are carried out for the equilibrium dipole moment e ,
and thus we would have to carry out a vibrational averaging in order to speak of the
same quantity. However, there are a few experimental values for e , and we shall compare
the performance of our standard models with those, reported in Table B.3. Note, that
due to the outer-valence origin of this property, we have to augment our basis sets with
extra diffuse primitive Gaussians; hence we are using the aug-cc-pVXZ basis sets. The
geometries are optimized using the same wave function as for the calculation of the dipole
moment.
The quality of the calculation is more dependent on the correlation treatment than the
cardinal number of the basis set; yet it is necessary to augment the basis set. The further
augmentation (d-aug-cc-pVXZ basis sets) of the basis set does not affect the calculated
dipole moments significantly.

B.4

Reaction enthalpies

Last, we study the reaction enthalpies of 13 chemical reactions among closed-shell molecules.
The calculations are carried out in the HFMP2CCSDCCSD(T) hierarchy, using the
113

Exp.
NH3
HF
H2 O

1.561(5)
1.803(2)
1.8473(10)

HF
QZ
1.544
1.884
1.936

MP2
QZ
1.517
1.811
1.864

CCSD
QZ
1.530
1.813
1.870

DZ
1.541
1.799
1.859

CCSD(T)
TZ
QZ
1.513 1.521
1.797 1.800
1.845 1.853

Table B.2: Molecular dipole moments by experiment and theory.

correlation-consistent basis set hierarchy supplemented with steep functions to provide

core-polarization, which somewhat surprisingly must be included in the high-precision
calculation of reaction enthalpies.

114

Figure B.1: Errors in the calculated bond lengths of 29 molecules as a function of the level
of theory.
115

Figure B.2: Errors in the calculated bond angles of 13 molecules as a function of the level
of theory.

116

 of the cc-pVQZ basis.

Figure B.3: Changes in bond lengths calculated from

Figure B.4: Performance of different models in computation of reaction enthalpies.

117

118

References
[1] R. McWeeny, Methods of Molecular Quantum Mechanics, 2nd edition. Academic
Press, London, 1992.
[2] P. W. Atkins and R. S. Friedman, Molecular Quantum Mechanics, 3rd edition. Oxford
University Press, Oxford, 1997.
[3] R. E. Moss, Advanced Molecular Quantum Mechanics. Chapman & Hall, 1973.
[4] J. Olsen and P. Jrgensen in Modern Electronic Structure Theory, Part II (D. R.
Yarkony, ed.), World Scientific, Singapore, 1995.
[5] T. L. Beck Rev. Mod. Phys., vol. 72, p. 1041, 2000.
[6] R. G. Parr and W. Yang, Density-Functional Theory of Atoms and Molecules. Oxford
University Press, Oxford, 1989.
[7] P.-O. Lowdin Rev. Mod. Phys., vol. 34, p. 80, 1962.
[8] C. C. J. Roothaan Rev. Mod. Phys., vol. 23, p. 69, 1951.
[9] G. G. Hall Proc. Roy. Soc. A, vol. 208, p. 328, 1951.
[10] P. Pulay J. Chem. Phys., vol. 73, p. 393, 1980.
[11] P. Pulay J. Comp. Chem., vol. 4, p. 556, 1982.
[12] H. Sellers Int. J. Quantum Chem., vol. 45, p. 31, 1993.
[13] K. N. Kudin, G. E. Scuseria, and E. Cances J. Chem. Phys., vol. 116, p. 8255, 2002.
[14] P. Salek, S. Reine, F. Pawlowski, T. Helgaker, S. Hst, L. Thgersen, P. M.
P. Jrgensen, J. Olsen, B. Jansk, and S. Coriani J. Chem. Phys., vol. 126, p. 114110,
2007.
[15] J. Almlof, K. Fgri, and K. Korsell J. Comp. Chem., vol. 3, p. 385, 1982.
[16] M. C. Strain, G. E. Scuseria, and M. J. Frisch Science, vol. 271, p. 51, 1996.
[17] C. Mller and M. S. Plesset Phys. Rev., vol. 46, p. 618, 1934.
119

[18] J. A. Pople, J. S. Binkley, and R. Seeger Int. J. Quantum Chem. Symp., vol. 10, p. 1,
1976.
[19] I. Shavitt in Methods of Electronic Structure Theory (H. F. S. III, ed.), Plenum Press,
New York, 1977.
[20] B. O. Roos, P. R. Taylor, and P. E. M. Siegbahn Chem. Phys., vol. 48, p. 157, 1980.
[21] J. Olsen, B. O. Roos, P. Jrgensen, and H. J. A. Jensen J. Chem. Phys., vol. 89,
p. 2185, 1988.
[22] P.-
A. Malmqvist, A. Rendell, and B. O. Roos J. Chem. Phys., vol. 94, p. 5477, 1990.
[23] A. C. Hurley, Electron correlation in small molecules. Academic Press, London, 1976.
[24] C. D. Purvis and R. J. Bartlett J. Chem. Phys., vol. 76, p. 1910, 1982.
[25] K. Raghavachari, G. W. Trucks, J. A. Pople, and M. Head-Gordon Chem. Phys.
Lett., vol. 157, p. 479, 1989.
[26] T. Helgaker, P. Jrgensen, and J. Olsen, Molecular Electronic-Structure Theory. John
Wiley & Sons, Chichester, 2000.
[27] J. Noga and R. J. Bartlett J. Chem. Phys., vol. 86, p. 7041, 1987.
A. Malmqvist, B. O. Roos, A. J. Sadlej, and K. Wolinski J. Phys.
[28] K. Andersson, P.-
Chem., vol. 94, p. 5483, 1990.
[29] P. Pulay Chem. Phys. Lett., vol. 100, p. 151, 1983.
[30] S. Sb and P. Pulay Annu. Rev. Phys. Chem., vol. 44, p. 213, 1993.
[31] P. E. Maslen, C. Ochsenfeld, C. A. White, M. S. Lee, and M. Head-Gordon J. Phys.
Chem. A, vol. 102, p. 2215, 1998.
[32] G. E. Scuseria and P. Y. Ayala J. Chem. Phys., vol. 111, p. 8330, 1999.
[33] P. Y. Ayala and G. E. Scuseria J. Chem. Phys., vol. 110, p. 3660, 1999.
[34] C. Hampel and H. J. Werner J. Chem. Phys., vol. 104, p. 6286, 1996.
[35] M. Sch
utz, G. Hetzer, and H.-J. Werner J. Chem. Phys., vol. 111, p. 5691, 1999.
[36] M. Sch
utz and H.-J. Werner J. Chem. Phys., vol. 114, p. 661, 2001.
[37] M. Schutz Chem. Phys. Phys. Chem., vol. 4, p. 3941, 2002.
[38] H. J. Werner, F. R. Manby, and P. J. Knowles J. Chem. Phys., vol. 118, p. 8149,
2003.
120

[39] J. E. Subotnik and M. Head-Gordon J. Chem. Phys., vol. 122, p. 034109, 2005.
[40] P. E. Maslen, A. D. Dutoi, M. S. Lee, Y. H. Shao, and M. Head-Gordon Mol. Phys.,
vol. 103, p. 425, 2005.
[41] O. Christiansen, P. Manninen, P. Jrgensen, and J. Olsen J. Chem. Phys., vol. 124,
p. 084103, 2006.
[42] V. Weijo, P. Manninen, P. Jrgensen, O. Christiansen, and J. Olsen J. Chem. Phys.
[43] R. Ditchfield, W. J. Hehre, and J. A. Pople J. Chem. Phys., vol. 54, p. 724, 1971.
[44] R. Krishnan, J. S. Binkley, R. Seeger, and J. A. Pople J. Chem. Phys., vol. 72, p. 650,
1980.
[45] A. Schafer, H. Horn, and R. Ahlrichs J Chem Phys, vol. 97, p. 2571, (1992).
[46] A. Schafer, C. Huber, and R. Ahlrichs J. Chem. Phys., vol. 100, p. 5829, 1994.
[47] F. Weigend, F. Furche, and R. Ahlrichs J. Chem. Phys., vol. 119, p. 12753, 2003.
[48] T. H. Dunning J. Chem. Phys., vol. 90, p. 1007, 1989.
[49] R. A. Kendall, T. H. Dunning, and R. J. Harrison J. Chem. Phys., vol. 96, p. 6796,
1992.
[50] D. E. Woon and T. H. Dunning J. Chem. Phys., vol. 98, p. 1358, 1993.
[51] K. A. Peterson, R. A. Kendall, and T. H. Dunning J. Chem. Phys., vol. 99, p. 9790,
1993.
[52] K. A. Peterson, D. E. Woon, and T. H. Dunning J. Chem. Phys., vol. 100, p. 7410,
1994.
[53] A. Wilson, T. van Mourik, and T. H. Dunning J. Mol. Struct. (THEOCHEM),
vol. 388, p. 339, 1997.
[54] P.-O. Widmark, P.-
A. Malmqvist, and B. O. Roos Theor. Chim. Acta, vol. 77, p. 291,
1990.
[55] P.-O. Widmark, B. J. Persson, and B. O. Roos Theor. Chim. Acta, vol. 79, p. 419,
1990.
[56] D. P. Chong Can. J. Chem., vol. 73, p. 79, 1995.
[57] P. Manninen and J. Vaara J. Comp. Chem., vol. 27, p. 434, 2006.
[58] S. Obara and M. Saika J. Chem. Phys., vol. 84, p. 3963, 1986.
121

[59] S. Obara and M. Saika J. Chem. Phys., vol. 89, p. 1540, 1988.
[60] P. A. M. Dirac Proc. Roy. Soc. A, vol. 117, p. 610, 1928.
[61] P. A. M. Dirac Proc. Roy. Soc. A, vol. 118, p. 351, 1928.
[62] C. D. Anderson Phys. Rev., vol. 43, p. 491, 1933.
[63] J. M. Levy-Leblond Commun. Math. Phys., vol. 6, p. 216, 1967.
[64] E. E. Salpeter and H. A. Bethe Phys. Rev., vol. 84, p. 1232, 1951.
[65] E. E. Salpeter Phys. Rev., vol. 87, p. 328, 1952.
[66] J. E. Harriman, Theoretical Foundations of Electron Spin Resonance. Academic
Press, New York, 1978.
[67] G. Breit Phys. Rev., vol. 34, p. 553, 1929.
[68] J. A. Gaunt Proc. Roy. Soc. (London) A, vol. 122, p. 513, 1929.
[69] G. E. Brown and D. G. Ravenhall Proc. Roy. Soc. (London) A, vol. 208, p. 55, 1951.
[70] K. G. Dyall and K. Fgri Chem. Phys. Lett., vol. 174, p. 25, 1990.
[71] R. E. Stanton and S. Havriliak J. Chem. Phys., vol. 81, p. 1910, 1984.
[72] L. L. Foldy and S. A. Wouthuysen Phys. Rev., vol. 78, p. 29, 1950.
[73] K. G. Dyall J. Chem. Phys., vol. 106, p. 9618, 1997.
[74] J. Autschbach and W. H. E. Schwarz Theor. Chem. Acc., vol. 104, p. 82, 2000.
[75] W. Pauli in Handbuch der Physik, vol. 5, Springer, Berlin, 1958.
[76] D. W. Davies, The Theory of the Electronic and Magnetic Properties of Molecules.
Wiley, London, 1969.
[77] W. Kutzelnigg Theor. Chim. Acta, vol. 73, p. 173, 1988.
[78] J. D. Morrison and R. E. Moss Mol. Phys., vol. 41, p. 491, 1980.
[79] I. J. Ketley and R. E. Moss Mol. Phys., vol. 48, p. 1131, 1983.
[80] I. J. Ketley and R. E. Moss Mol. Phys., vol. 49, p. 1289, 1983.
[81] R. E. Moss Mol. Phys., vol. 53, p. 269, 1984.
[82] A. Farazdel and V. H. Smith Int. J. Quantum Chem., vol. 29, p. 311, 1986.
122

[83] T. Ziegler, V. Tschinke, E. J. Baerends, J. G. Snijders, and W. Ravenek J. Phys.

Chem., vol. 93, p. 3050, 1989.
[84] E. van Lenthe, E. J. Baerends, and J. G. Snijders J. Chem. Phys., vol. 99, p. 4597,
1993.
[85] C. Chang, M. Pelissier, and P. Durand Phys. Scr., vol. 34, p. 394, 1986.
[86] R. van Leeuwen, E. van Lenthe, A. Ehlers, E. J. Baerends, and J. G. Snijders J.
Chem. Phys., vol. 101, p. 1272, 1994.
[87] M. Barysz and A. J. Sadlej Theor. Chim. Acta, vol. 97, p. 260, 1997.
[88] D. Sundholm Theor. Chem. Acc., vol. 110, p. 144, 2002.
[89] A. J. Sadlej and J. G. Snijders Chem. Phys. Lett., vol. 229, p. 435, 1994.
[90] K. G. Dyall and E. van Lenthe J. Chem. Phys., vol. 111, p. 1366, 1999.
[91] M. Filatov and D. Cremer J. Chem. Phys., vol. 119, p. 11526, 2003.
[92] B. A. Hess Phys. Rev. A, vol. 32, p. 756, 1985.
[93] B. A. Hess Phys. Rev. A, vol. 33, p. 3742, 1986.
[94] J. Sucher Phys. Rev. A, vol. 22, p. 348, 1980.
[95] M. Douglas and N. M. Kroll Ann. Phys. (N.Y.), vol. 82, p. 89, 1974.
[96] J. G. Snijders and A. J. Sadlej Chem. Phys. Lett., vol. 252, p. 51, 1996.
[97] M. Baryz, A. J. Sadlej, and J. G. Snijders Int. J. Quantum Chem., vol. 65, p. 225,
1997.
[98] T. Nakajima and K. Hirao Chem. Phys. Lett., vol. 302, p. 383, 1999.
[99] L. Visscher and T. Saue J. Chem. Phys., vol. 113, p. 3996, 2000.
[100] F. Wang, G. Hong, and L. Li Chem. Phys. Lett., vol. 316, p. 318, 2000.
[101] D. G. Fedorov, T. Nakajima, and K. Hirao Chem. Phys. Lett., vol. 335, p. 183, 2001.
[102] M. Filatov and D. Cremer Chem. Phys. Lett., vol. 351, p. 259, 2002.
[103] G. L. Sewell Proc. Cambridge Phil. Soc., vol. 45, p. 631, 1949.
[104] A. Rutkowski J. Phys. B, vol. 19, pp. 149, Ibid. 3431, Ibid. 3443, 1986.
[105] W. Kutzelnigg Z. Phys. D, vol. 11, p. 15, 1989.
123

[106] M. Dolg in Modern Methods and Algorithms of Quantum Chemistry (J. Grotendorst,
ed.), vol. 1, John von Neumann Institute for Computing, J
ulich, 2002.
[107] P. Pyykko and H. Stoll in Chemical Modelling: Applications and Theory (A. Hinchliffe, ed.), vol. 1, R. S. C., Cambridge, 2000.
[108] P. Jrgensen and J. Simons, Second Quantization Based Methods in Quantum Chemistry. Academic Press, New York, 1981.

124