Are there specific command to print the gradient matrix?

Questions
#1

I need to calculate the gradients of the final states obtained by the DIP-EOM-CCSD method. I know that in this case only numerical calculations of the gradients is possible, the program does the necessary calculations with shifted geometries, but it does not output the gradient matrix, just the FINAL TENSOR. Are there specific command to print the gradient matrix?

Version Qchem 5.2

$rem

threads = 8

jobtype = force

method eom-CCSD

basis = cc-pVDZ

purecart = 1111

print_general_basis = true

MAX_SCF_CYCLES = 100

thresh = 14

scf_convergence = 8

n_frozen_core = 2

SYM_IGNORE = TRUE

DIP_SINGLETS = 5

CC_STATE_TO_OPT [1,1]

EOM_DAVIDSON_CONVERGENCE = 4

EOM_DAVIDSON_THRESHOLD = 4

EOM_DAVIDSON_MAX_ITER = 100

EOM_PRECONV_SD = 50

MEM_TOTAL = 25000

max_sub_file_num = 128

$end

#2

That FINAL TENSOR is the gradient. You can confirm this for yourself, e.g., by running a small CCSD force calculation (for which we have analytic gradients, JOBTYPE = FORCE) and then also forcing Q-Chem to do the same calculation by finite-difference, by setting IDERIV = 0 in $rem.

#3

Thank you for the answer. Indeed it works, we checked out. Additional question is how correct is the input given in my question above? According to the manual, we need to use SET_STATE_DERIVE = n or the $derivative group to chose the state to compute gradients, but the gradients with such commands seemse to be prodused for the initial neutral system in the ground state. Apparently, it turned out to calculate gradients for different DIP states just with the CC_STATE_TO_OPT task which is an optimization option. Is there a more correct input format for our case?

#4

With coupled-cluster methods, use CC_STATE_TO_OPT to compute forces of the specified state or transition properties with respect to the specified state. Hope that addresses your query. I’m unsure what you mean by “$derivative group”. I don’t think there is any $derivative section, if you are implying that.

#5

SET_STATE_DERIV is a CIS/TDDFT keyword (equivalent to CIS_STATE_DERIV and we should probably get rid of the former as it is older notation). I think you may be conflating different functionalities; there is a $derivative_coupling section but that is also for CIS/TDDFT.

#6

Thanks for the answers! It really helped us understand and confirm what we’re doing.

#7

FYI, I am removing SET_STATE_DERIV from future versions of the manual; it has been replaced by CIS_STATE_DERIV and we don’t need two versions of the same keyword floating around.