2024-02-28 BespokeFit Meeting notes

benchmarks of QM theory, implicit solvent models &QCA interface

• (CA presents)

• (see recording)

• PB – Is the rinicker model DASH? I see charge sloshing with that for larger mols.

  • DC – I don’t think it’s DASH, but we can double check.

• CA – Have other people had issues with psi4 and implicit solvent and DZVP? We’ve had trouble with PCM.

  • DC – Yeah, this was released in psi4 a few months ago, but it seems to be a bit buggy.

  • WZW – PCM has a well known issue where it rolls a probe along the molecular surface, but this means that small changes in geo can lead to big changes in retults. ORCA solved this by describing surface using gaussians.

  • CA – …

  • (see recording ~22 mins)

  • CA – SCF seemed to go wrong with larger basis sets.

  • DC – Should check whether there are any recommended basis sets.

• JW – I’m not sure that we have implicit solvent work happening at OpenFF. Can anyone speak to whether the stack even currently works?

  • (no response)

  • JH – I’m doing some related work to support implicit solvent in QCEngine . Yeah, we can submit this as a test.

Implicit solvent and electrostatics datasets

DC –