2020-04-02 QCFractal Meeting notes

Get updates from MolSSI

• Ben Pritchard will be taking over for Daniel Smith as OpenFF contact for QCA.

• Probably few upcoming releases as Ben comes onboard

• New server is very high-capacity. Should be main server for next several years

Perform manager maintenance if needed and review queue status/timing

• @Jeffrey Wagner will ping @John Chodera to restart Lilac workers

• Roughly 100 torsions + 2000 hessians left. Probably a week left.

• @Jessica Maat (Deactivated) 's PR will add about another week of compute when PR is merged

• @David Cerutti (Deactivated) will be submitting amino acid-like molecules soon in large quantities, but these will be lower priority

Work on QCF integration into relevant code

• Pritchard is an expert on basis sets. Would be interested to follow along with project progress and potentially be involved in discussions.

• @Matt Thompson should look into the CMILES tagging of molecule datasets, and whether they’re complete and accessible. Can talk to @Joshua Horton , who has experience pulling data down final molecules

• @Joshua Horton Running Hessians on a final geometry is a bit painful, since we need to run it as a separate job.

• Ben (and maybe others) should review Josh’s QCSubmit tool (changes not pushed yet, but should be near end of this week):

• BP – What’s OpenFF’s tolerance for QCA errors/failures, and how do we distinguish between the two?

• HJ – Sometimes it’s our fault – like, we try to rotate about a linear torsion. Other times, we get issues around molecule size. In the recent sets, that’s been a big problem. So I resubmitted the sets with smaller molecules.

• BP – So generally, you check out the errors, and then bring some subset of them to MolSSI’s attention, and then resubmit as appropriate

• HJ – Yes