Minimized structures from JACS CDK2 set with:
Openff_unconstrained_1.3.0
- in cyan
Openff_1.3.1a
with the parameters a30
and a31
rolled back to smirnoff99frost values - in purple
Almost all of the 1.3.1a geometries overlap (only a single color is seen) with 1.3.0 or closely match. From Hyesu’s observation the discrepancy in tetrahedral angles is seen in MD simulations rather than in geometry optimizations.
Files: