Page Comparison

The major goal of this meering will be to identify the next datasets to submit, and at what priority

• Chaya – Torsion drives – For submission immediately

• Hyesu – Benchmark set (After may meeting)

• Maat – Nothing now

• CS – Did we decide on a level of theory for charged mols?

  • HJ – No, that benchmark is planned for after the May meeting

  • CS – Could use our standard level of theory and review results

  • BP – B3LYP-D3DJ method w/ DZVP basis set should be reasonable

  • CS – I’d previously done some charged molecules, and the results weren’t so bad

  • (General) – Current method should be fine for charged mols, though we’ll want to do a more thorough benchmark later

• Should we deprioritize 1.2.0 dataset?

  • (General) – Not yet

Getting Dotson started on QCA job maintenance

• BP and DD will coordinate to meet and transfer knowledge later this week

With any additional time, we will cover the normal topics:

1. Get updates from MolSSI

2. Perform manager maintenance if needed

3. Work on QCF integration into relevant code

• JM – Improving naming scheme for training datasets. It’s difficult when one dataset has multiple names. We should have a discussion about how to do this.

  • HJ – Would also be interested in being part of this conversation.

• TG – Is there a way to figure out if a single molecule appears in multiple datasets?

  • BP – This is currently internal data. May be hard to implement support for this.

  • DD – This might be a single JOIN in the backend database. TG, you should submit a feature request to QCFractal repo.