search

stfc / janus

collection of scripts to train and generate data for machine learnt interatomic potentials
BSD 3-Clause "New" or "Revised" License
0 stars 0 forks source link

Describe and cleanup input and output files #8

Open ElliottKasoar opened 1 year ago

ElliottKasoar commented 1 year ago

Throughout the workflows, more description of the expected input and output files would be useful.

Many of the steps also create a large number of files: nnp-x.log, sf.xxx.xxxx.histo, neuron-stats, test..., train..., ...lammpstrj, ...lmp etc. It would be useful to clarify when these are generated, what information they contain, and if any can be deleted.

It may also be useful as part of this to automatically clean up files, either by deleting any which have no further purpose or creating directories and moving files that have limited use.

dave452 commented 1 year ago

We should add a function called something along the lines of cleanup_n2p2_output that deletes alll except those requied to for the running of n2p2, i.e. the select weights and the scaling data. One thing I did when runnning n2p2 without this was to put all of these type of files in a folders i.e. train/ scaling/ prune/ which could help to better understand whats important for future use and which is important for analysis.