Is there a possibility of restarting an interrupted frequency calculation? Is it possible to utilise the previously calculated points in a finite difference hessian calculation (using file 32 perhaps) ?

Fred Bennett

Is there a possibility of restarting an interrupted frequency calculation? Is it possible to utilise the previously calculated points in a finite difference hessian calculation (using file 32 perhaps) ?

Fred Bennett

There is such a thing as running a frequency by finite difference of *energies* in “sow/reap” mode where an initial job creates many input files, you run those, then run a final analysis job. Example at https://github.com/psi4/psi4/blob/master/samples/pywrap-freq-e-sowreap/input.dat

The analogous frequency by finite difference of *gradients* procedure is not implemented. It’s not particularly difficult, just haven’t had the time and justification.

There’s nothing in place to restart from file 32. Gathering results and reformatting from output file is perhaps possible but not likely worth it unless completed jobs were very expensive.

So what nature is your job: large/small, energies/gradients, scf/ccsd(t), etc?

Sow/reap mode for frequencies by finite differences of gradients would be super useful!