search

openmm / spice-dataset

A collection of QM data for training potential functions
MIT License
153 stars 9 forks source link

Update spec #46

Closed pavankum closed 2 years ago

pavankum commented 2 years ago

Explicitly mention spice_default for downloading datasets to avoid database inconsistencies. Resolves #44 .

peastman commented 2 years ago

Rather than hardcoding the name of a particular subset, will it work if we just fix the values of method and basis and keep the others as they are? Or can we somehow loop through the records to find the one with the correct method? Anything that specifies subsets by name will be brittle and lead to problems in the future.

peastman commented 2 years ago

I created a different implementation in #47. Can you take a look and see if everything looks correct? I already tested downloading a couple of small subsets and the results matched the 1.1 version of the dataset.

pavankum commented 2 years ago

Anything that specifies subsets by name will be brittle and lead to problems in the future.

Yeah, I agree.