Open calvinp0 opened 3 months ago
I agree it's good to begin with a larger number than we have today. Maybe also change the Gaussian input file template we have?
I agree it's good to begin with a larger number than we have today.
Maybe also change the Gaussian input file template we have?
I agree we should but I was thinking it would require a bigger overhaul - whereby we define in input_dict keys that are related to variables So for example if we have scf Input_dict['scf']=['maxcycle=512','xqc'] And then in the template we set the section for {scf}
Commonly, our fine opt fails due to number of steps exceeded. Whilst we may troubleshoot this after an initial failure, I believe it is best to set the maxcycles at 100 to begin with so that we increase the chance of convergence prior to troubleshooting. This can reduce our creation of opt jobs.