2020-12-08 QCFractal Users Meeting notes

Updated from MolSSI

• BP – No major updates, working through infrastructure issues. No downtime planned.

Queue status/Manager maintenance

• DD – Some dataset priority rearrangement. No manager updates needed. Many datasets at each stage of submission pipeline but more are welcome. No manager updates needed.

User questions

• HJ – Question about qca-dataset-submission PR #169. Basis set not being found so it’s automatically failing validation.

  • BP – These “heavy” basis sets that are automatically generated are hard to handle, since there’s so many of them. Problem is that, since these are “modifications” to existing datasets, adding them all would multiply the current 600 basis sets to thousands.

  • (General) – Should have validation code in QCSubmit strip “modifications” to basis set names and search the basis set name without it.

    • BP – Could strip “heavy”, month names, d from “daug”, t from “taug”

    • JH – I’ll add this to QCSubmit

• HJ – Question about PR #137 – “draft of first official standards” – How do I add more molecules to an existing dataset?

  • TG – This would be a minor version bump (eg. 3.0 to 3.1)

  • HJ – Which procedure should I follow?

  • TG – You’d make a changelog, and then submit a new dataset with all of the original inputs, as well as the new inputs.

  • JH – This could be a good time to combine the theory levels that are being benchmarked into a single dataset.

• PB – Regarding deduplication using InChI

  • JW – I think we should change this isomorphism check to have formal_charge_matching=False, bond_order_matching=False, aromatic_matching=False

  • JH – Agree

•