2025-10-28 FF Fitting Meeting

BM: water mixture updates

Evaluator 0.3. vs 0.5

• MS: I’m impressed by how different the results are

• LW: I wonder how much is from Sage 2.1 vs 2.2.1, and how much is from Evaluator updates

parsley 1.3 vs 2.2

• LW: looks like tip3p r2 got worse for dhmix?

• DM: Would it be trivial to add kendall tau? I am not a huge fan of R^2 because it’s so sensitive to extrema.

  • MS: should be easy

Sage 2.3 vs Sage 2.2.1

• LW: rc1 vs rc2?

  • BM: probably rc1, I started working with this more than 2 weeks ago

  • (eventually) probably rc1 by date

• MS: OpenFF alcohol + water mixtures are basically ideal mixtures with tip3p with excess molar volume

  • CI: Enthalpy of Mixing and Liquid Structure for Mixtures of Primary and Secondary Alcohols: Molecular Simulations and COSMO–SAC Calculations attempts to fix this

• MS: the question is what to do next with FFs. This should motivate moving to OPC3. Does seem to be motivation for 4-site model.

  • BM: would want to look into rc2. Also, running 2.1.0, also looking more into alcohols.

• LW: whichi s the best 3-site? We could start form that as a starting point.

  • General: TIP3P-FB does better on dhmix, comparable on density.

  • MS: could be useful/interesting to start from both. WE need to balance constraints on water model vs solvation properties.