Repository navigation
openff-3.0.0-rc1 - #164
openff-3.0.0-rc1#164
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. 🚀 New features to boost your workflow:
|
chapincavender
left a comment
There was a problem hiding this comment.
Good catch on Xe parameters! Changes look good to me.
I tested the rc1 offxml on some protein systems in water boxes with Na+ and Cl- ions. Here's the total potential energies in kcal/mol using OpenMM's CUDA Platform with mixed precision.
System | no_water-3.0.0-alpha2b | 3.0.0-rc1
---------------------------------------------------
ala3 -6403.45440349 -6403.45438971
gly3 -6548.29213676 -6548.29210399
val3 -6187.65754456 -6187.65754935
gb3 -32857.87253555 -32857.87258022
bpti -47662.91126517 -47662.59997210
ubq -63929.90035637 -63929.90024801
hewl -72161.56658396 -72161.19853259
aaqaa3 -122297.21133338 -122297.21134815
Everything matches to single precision except for a 0.4 kcal/mol discrepancy in the HarmonicAngleForce for BPTI due to the arginine angle error in #144. Since we already decided that we won't patch this bug in the 3.0.0 release, I think this is good to go.
|
Per some discussion today, we're going to reduce the ions in this file to just the ones that were already present in Sage, though we'll switch them to parameters from |
This PR adds
openff-3.0.0.offxml(and the unconstrained variant), created by removing the ion parameters from 3.0.0-alpha2b (since they're carried over from the Sage line, which uses ion parameters fit for TIP3P) and replacing them with ion parameters fromopc3.offxml(just for the same ions as Sage covered, even though more are available in that file). It also adds the water model and constraints fromopc3.offxml.