search

openforcefield / openff-benchmark

Comparison benchmarks between public force fields and Open Force Field Initiative force fields
MIT License
11 stars 2 forks source link

Speed up `analysis match-minima` #70

Open dfhahn opened 3 years ago

dfhahn commented 3 years ago

The match minima code is quite slow if you have many molecules and methods/force fields due to the calculation of the RMS matrices for each molecule. The speed could be easily improved by parallelizing it per force field or even per molecules.

In the mean time, a work around is to calculate each MM force field separately as separate tasks:

openff-benchmark report match-minima --input-path 4-compute-qm/ --input-path 4-compute-mm/gaff-2.11 --ref-method b3lyp-d3bj --output-directory 5-match_minima
openff-benchmark report match-minima --input-path 4-compute-qm/ --input-path 4-compute-mm/openff-1.0.0 --ref-method b3lyp-d3bj --output-directory 5-match_minima
.
.
.
dotsdl commented 3 years ago

Thank you for this! @BillSwope has raised this from his experience on his production set.