2024-09-16 Science team meeting

Project updates

LW

• Sage 2.2.1 full release – flaky 3-membered molecule in HMR tests

  • AMI: CBayly mentioned issues with 3-membered ring parameters

• Sage 2.1.0 + OPC3 fit is about halfway done

  • Spend some time this week getting Evaluator on NRP

• Fit with a looser MSM force constant value

• Rest of the week:

  • Genetech, OpenFE collaborations, and polymer modelling

Project updates

BW

• Torsion fitting / restrained minimization

  • LW: maybe try blanket 1kcal/mol restraint?

• datasets

  • BW: going to submit benchmark S dataset

  • Phosphate dataset

    • Only ~7 throwing proper errors, most might be a memory issue

    • Proper errors – SCF convergence, linear angle errors, geometry optimization failed to converge

    • ~100 errors, possible segfault

    • ~56 finished

  • LW: please put up Hessian dataset of optimization set, not a clear procedure yet

• smee

  • BW – SPICE dataset run had force constants not going to 0.

  • LW – does smee900 look similar to Sage? Is that why it performs better?

    • BW – looked pretty similar

    • BW – ~10x smaller dataset in Sage than Spice

  • LW – what’s the weighting between energies and forces?

    • BW – no other targets

Project updates

AMI

• NAGL2

• Dataset test set

  • AMI: debugged PCM issues with solvent, found it was an SCF convergence issue

    •  

• Next week:

  • parameter splits