Lecture outlines

Coupled Cluster (CC)

Questions:
What is Coupled Cluster (CC) calculations? How it is
different from Configuration Interaction (CI) methods?
 Recovering of correlation energy

In quantum chemistry, there are many approaches to “recovering”

the missing correlation energy. Typically, they fall into two classes:

1. Perturbation theory
2. Wavefunction expansion

• The most common perturbation approach is Moller-Plesset

“MP2” or “MP4”.

• The most common wavefunction expansion approaches are

Configuration Interaction “CI” and Coupled Cluster “CC” or
“CCSD”.
 Coupled Cluster (CC)
• Coupled cluster is one of the post HF methods to recover the missing
correlation energy of HF calculations.

• Coupled cluster calculations are similar to configuration interaction

calculations in that the wavefunction is a linear combination of many
determinants. However, the means/methods for choosing the
determinants in a coupled cluster calculation is more complex than
the choice of determinants in a CI.

• It also takes the basic HF molecular orbital method and construct

multi-electron wavefunctions using the exponential cluster operator to
account for electron correlation.

• This is one of the best methods for small to medium sized molecules
for highly accurate calculations. There are various orders of the CC
expansion, called CCSD, CCSDT, and so on, Like CI. .
• Coupled cluster calculations give variational energies as long as the
excitations are included successively. Thus, CCSD is variational, but
CCD is not. CCD still tends to be a bit more accurate than CID.

• A calculation denoted CCSD(T) is one in which the triple excitations are

included perturbatively rather than exactly. (this is indicated by the T in
parentheses).

• The more accurate (and more expensive) CCSD(T) is often called

“the gold standard of quantum chemistry" for its excellent compromise
between the accuracy and the cost for the molecules near equilibrium
geometries.

• The accuracy of these two methods is very similar. The advantage of

doing coupled cluster calculations is that it is a size extensive method.
Often, coupled-cluster results are a bit more accurate than the equivalent
size configuration interaction calculation results.

• When all possible configurations are included, a full coupled-cluster

calculation is equivalent to a full CI calculation.
• Quadratic configuration interaction calculations (QCI) use an
algorithm that is a combination of the CI and CC algorithms.
Thus, a QCISD calculation is an approximation to a CCSD
calculation.

• These (QCI) calculations are popular since they often give an

optimal amount of correlation for high-accuracy calculations on
organic molecules while using less CPU time than coupled
cluster calculations. Most popular is the single- and double-
excitation calculation, QCISD.

• There is a variation on the coupled cluster method known as the

symmetry adapted cluster (SAC) method. This is a size
consistent method. For excited states, a CI out of this space,
called a SAC-CI, is done. This improves the accuracy of
electronic excited-state energies.

• One of the drawbacks of the method is that it is not variational

The wavefunction of the coupled-cluster Tˆ
theory is written as an exponential ansatz: CC  e  0
 Φ0 is a Slater determinant usually constructed from HF molecular orbitals.
 T is an excitation operator which, when acting on Φ0 produces a linear
combination of excited Slater determinants


1 ˆ 2 1 ˆ3 1ˆk
 T
Exponential operator
ˆ
Tˆ ˆ
e 1T + T  T  generates excited Slater
2 6 k 0 k! determinants (Taylor series)

ˆ T
T ˆ  Tˆ  Tˆ   Tˆ N
Cluster Operator
1 2 3 N is the number of electrons

T1 is the operator of all single excitations,

where T2 is the operator of all double excitations and so forth
 occ vir
ˆ     t a a
T Being the one-particle excitation operator and the
1 0 i i two-particle excitation operator, T1 and T2 convert
i a
the HF reference function Φ0 into a linear
occ vir
ˆ     t ab ab
T
combination of the singly and doubly excited
2 0 ij ij Slater determinants, respectively.
i j a b

tia
tab
ij
 Expansion coefficients are called amplitudes;
equivalent to the ai’s in the general multi-
determinant wavefunction.
HF ref.
ˆ 1 ˆ 2  ˆ ˆ  1 Tˆ 3   T ˆ  1 Tˆ 2  1 Tˆ T ˆ 4  
ˆ2 1 T
e  1ˆ  Tˆ1  T ˆ T ˆ  Tˆ T
Tˆ
 T  T  T
2
2 1   3 2 1
6 1   4 3 1
2 2 2 2 1 24 1 
singles doubles triples Quadruple excitations

• The way that Slater determinants are generated is rather different…

 1ˆ HF reference
ˆ
T Singly excited states
1

ˆ 1 ˆ 2  ˆ
T Connected doubles (true doubly by T2 by T1T1)
2
T 
 2 2 T1  ˆ2
T Dis-connected doubles
1

ˆ
T 3 Connected triples, ‘true’ triples
ˆ ˆ 3 
ˆ Tˆ  1 T
T
 3  T ˆ Tˆ
T
2 1
6 1  2 1
‘Product’ Triples, disconnected triples
ˆ
T 3
1

ˆ ˆ  1 Tˆ 2 
T4  Tˆ 3 T ˆ
 1
2 2  T 4
True quadruples - four electrons interacting
 1 1 
 Tˆ 2 T
ˆ  2 ˆ 
T 4
ˆ 2
 2 1
24 1  T2
Product quadruples - two non-interacting pairs

ˆ Tˆ , T
T ˆ Tˆ 2, Tˆ 4 Product quadruples, and so on.
3 1 2 1 1
 If all cluster operators (CC) up to TN are included, the method yields
energies that are essentially equivalent to Full CI.

 In practice, only the singles and doubles excitation operators are used
forming the Coupled Cluster Singles and Doubles model (CCSD).

e
Tˆ 1  Tˆ 2 ˆ  T
 1ˆ  T ˆ 2  Tˆ T
ˆ 1T ˆ  1 Tˆ 3   1 Tˆ 2  1 Tˆ T ˆ 4 
ˆ2 1 T
1  2 2 1   2 1 6 1  2 2 2 2 1 24 1 

 The result is that triple and quadruple excitations also enter into the
energy expression (not shown) via products of single and double
amplitudes.

 It has been shown that the connected triples term, T3 , is important. It

can be included perturbatively at a modest cost to yield the CCSD(T)
model. With the inclusion of connected triples, the CCSD(T) model
yields energies close to the Full CI in the given basis, a very accurate
wavefunction.
 The energy is given by:

  
occ vir
Ecc  E0    tijab  tia t bj  tibt aj  i  j |  a  b   i  j |  b  a
i  j a b

So, the coupled cluster correlation energy is determined completely by the

singles and doubles amplitudes and the two-electron MO integrals

Truncated Coupled Cluster Methods

 If all TN operators are included in T the CC wave-function is equivalent to

full CI wave-function, but this is possible only for the smallest systems.

Truncation of T

 Including only the T1 operator there will be no improvement over HF, the
lowest level of approximation being T=T2 ( CCD=Coupled Cluster Doubles)

If T=T1+T2  CCSD  scales as K6 the only generally applicable model

If T=T1+T2+T3  CCSDT  scales as K8
Types of coupled-cluster methods

 The classification of traditional coupled-cluster methods rests on the

highest number of excitations allowed in the definition of T.

 The abbreviations for coupled-cluster methods usually begin with the

letters "CC" (for coupled cluster) followed by
• S - for single excitations (singles in CC)
• D - for double excitations (doubles)
• T - for triple excitations (triples)
• Q - for quadruple excitations (quadruples)

 Thus the T operator in CCSDT has the form T = T1+ T2 +T3

 Terms in round brackets indicate that these terms are calculated

based on perturbation theory. For example, a CCSD(T) approach
simply means:
• A coupled-cluster method
• It includes singles and doubles fully
• Triples are calculated non-iteratively (without repetition )
 Binding energy of N2

HF 122.2
MP2 239.6
MP3 215.2
MP2
MP4

Be / kcal mol-1
MP4 231.5 MP3 CCSD
CCSD(T)
CCSD 217.2
MRCI
CCSD(T) 226.4 HF

MRCI 227.2
AK Wilson, T van Mourik, TH Dunning, Jr.,
Exptl. 228.4 THEOCHEM 368, 339 (1996)