X

Search Results

Searching....

7.10 Coupled-Cluster Excited-State and Open-Shell Methods

7.10.22 EOM-CCSDT method for Exclusively High Accuracy (CCMAN2 only)

(September 1, 2024)

Equation-of-motion singles, doubles and triples (EOM-CCSDT) method can be viewed as an extension of EOM-CCSD method in which full triples are employed at CC level (in addition to singles and doubles) and at the EOM level.Conceptually, EOM-CCSD and EOM-CCSDT are same. The CC Hamiltonian matrix being non-hermitian, the left and right eigenvectors are not hermitian conjugates of each other but can be chosen to form a mutually biorthogonal set. Using Eq 7.77, the right eigenvectors can be obtained and this is sufficient for obtaining energies/excitation energies of the corresponding target states. For gradients and properties, one needs to additionally solve for the of left eigenvectors of the respective states.

In EOM-CCSDT, the CCSDT Hamiltonian matrix is constructed to triples block so as to give one-, two- and three- body eigenvectors of the CC Hamiltonian. As a result, the EOM-CCSDT computed energies and properties are highly accurate as compared to the EOM-CCSD ones – almost by an order of magnitude. For example, the EOM-CCSDT provides the accurcy of 0.01eV for singly excited states/single bond-breaking and of 0.1-0.2eV for doubly excited states, as opposed to EOM-CCSD, for which, the errors in these states are of the order of 0.1-0.2eV and >1eV, respectively. EOM-CCSDT has N8 scaling as opposed to N6 scaling of EOM-CCSD and, unlike EOM-CC(2,3), it is rigorously size-intensive (as is the EOM-CCSD). This exclusive method may be used with careful memory management. Currently, the method is implemented only for single point energy computation of excited states (SF and EE variants) and can be availed by choosing METHOD=EOM-CCSDT. Currently, the EOM-CCSDT has been developed within the new coupled-cluster suite of methods (CCMAN2) only. Therefore, CCMAN2 must not be disabled in order to use this method. The other job-control variables may be appropriately chosen. Illustrative examples employing EOM-CCSDT method are given below.

Example 7.132  Computation of first excited singlet B1 and first excited triplet B2 energies of BH using EOM-EE-CCSDT/STO-3G in single-precision setup

$comment
Computation of energies singlet and a triplet excited states of BH
using EOM-EE-CCSDT/STO-3G
$end

$molecule
0 1
B
H  1  1.234347
$end

$rem
basis=sto-3g
job_type=sp
method eom-ccsdt
cc_single_prec=1
eom_single_prec=1
n_frozen_core fc
ee_singlets=[0,0,1,0]
ee_triplets=[0,0,0,1]
$end

Example 7.133  Computation of singlet ground state and lowest triplet state energies of BH using EOM-SF-CCSDT/STO-3G in double precision setup

$comment
Computation of spin-flipping excitation energies of
BH using EOM-SF-CCSDT/STO-3G, UHF reference
$end

$molecule
0 3
B
H 1 1.191857
$end

$rem
jobtype=sp
method eom-ccsdt
basis=sto-3g
sf_states=[1,0,0,1]
eom_nguess_singles=4
n_frozen_core fc
$end