geometry optimization: questions on overwriting, Hessian and DFT convergence criteria


Click here for full thread
Just Got Here
NWCHEM users,

I have completed a geometry optimization followed by a frequency calculation and I have a few questions:

1) In the logfile, I get these overwrite warnings, is this typical for a frequency calculation? Can I do something to prevent the warnings/overwrites?



NWChem Analytic Hessian




HESSIAN: the one electron contributions are done in 7.5s
HESSIAN: 2-el 1st deriv. term done in 171.2s
HESSIAN: 2-el 2nd deriv. term done in 285.9s
stpr_wrt_fd_from_sq: overwrite of existing file:./nnk-diazonium-ion-hf-321Gs-b3lyp-321Gs-631Gs-6311Gss-opt-freq.hess
stpr_wrt_fd_dipole: overwrite of existing file./nnk-diazonium-ion-hf-321Gs-b3lyp-321Gs-631Gs-6311Gss-opt-freq.fd_ddipole
HESSIAN: the two electron contributions are done in 793.9s
SCF residual: 5.3512747886381211E-006
HESSIAN: the CPHF contributions are done
stpr_wrt_fd_from_sq: overwrite of existing file:./nnk-diazonium-ion-hf-321Gs-b3lyp-321Gs-631Gs-6311Gss-opt-freq.hess
stpr_wrt_fd_dipole: overwrite of existing file./nnk-diazonium-ion-hf-321Gs-b3lyp-321Gs-631Gs-6311Gss-opt-freq.fd_ddipole



2) Since the default optimization algorithm is using an estimate of the Hessian and I did a frequency calculation, does NWChem check for the convergence of the optimization using the Hessian calculated with the frequency calculation? If not, is it possible to restart the optimization and instruct NWChem to use the Hessian calculated in my frequency calculation (instead of the estimate)? Can I instruct NWChem to use a calculated Hessian during the optimization instead of the estimated one?

3) The first 6 frequencies correspond to translational and rotational degrees of freedom, is this correct? I just want to make sure that when I look for a minimum, I need to make sure that all but the first six projected frequencies are positive.

4) Do you recommend the tightening of the DFT Convergence criteria (energy, density, gradient) during a geometry optimization?

Thank you,

Christos