2025-10-01 Cole Lab Check-In (Aus) Meeting Notes

Recent Bespoke/ Transferable Valence Fitting Progress

FC

• LW: Have you looked at adding neighbours to SMIRKS for transferable fits?

  • FC: Have played around with it but never fully trained and benchmarked FF with neighbour information.

• DC: When you look at the bespokefit dynamics, do you observe lack of sampling in poorly-fitted regions

  • FC: Yes, but likely that issues with metad protocol are due to irrelevant high energy conformers.

• DC: Comparing to original bespokefit results on fragments risks loss of accuracy in going from full molecule (used for new bespoke fitting) to fragment

Meeting timings

• Switch to 11 AM UK time next meeting, then 12 AM.