Dear Q-Chem developers,
In EOM-CCSD calculation, Q-Chem by default prints T and R amplitudes with abs > 0.05 (or something like that) to the log file on 4 decimals. Can I anyhow make it print them on more decimals and with tighter threshold for abs. value? Or alternatively, is there any other file from which I could read them?
Thank you in advance.
Tomislav
It seems these numbers, unfortunately, can’t be controlled with REM variables. Is there a specific reason you want to control these printouts?
We need to generate (EOM-)CCSD wavefunctions in the Slater-determinant representation, so we can calculate some quantities we need in the surface-hopping dynamics and related calculations, such as overlaps of states on two different geometries, 1RDMs, Dyson orbitals etc.
In particular, we’re trying to do this in the SHARC code (https://sharc-md.org/), through SHARC-QCHEM interface that we’re implementing. SHARC has a general module that can calculate all quantities from the CI vectors, MO coefficients and the AO overlap matrix.
I know qchem can calculate 1RDMs and Dyson orbitals with full correlation and print them to fchk, but this requires more than one call of qchem (e.g., it cannot generate cccd->eom and eom->eom 1TDMs in a single calculation, state_analysis and cc_do_dyson together cause qchem to freeze,… etc, see my previous question Getting whole set of OPDMs and Dyson orbitals), so it’s not convenient and is more expensive than needed.
Also, in this way, the states calculated in the different calls of qchem get different random signs, so the sign-sensitive quantities (e.g., ccsd->eom and eom->eom 1tdms) are not matching.
In any case, it’s only the amplitudes from which we can calculate the wavefunction overlaps between two different geometries, so we need them on a few more decimals and a few more of them. Although, it seems to work fine with this precision, but it’s far from perfect.
I suppose this is something that might be easily changed in the code, so if you (qchem developers) could do it for some of the new releases of qchem, we from the SHARC community would be grateful :))
Also, the same feature for MP/ADC would be amazing, since adcmann also prints on the same precision and with the same threshold, plus it doesn’t print the MP “amplitudes” for the ground state.