Temperature shoots up in long run

[mayankaditya]

Hello All,

I ran MD simulation from deepmd potential. It works pretty well up to a few picoseconds and thereafter the temperature shoots up and atoms are getting very closed( it seems like dissociate).
Should I train the model further? Do I need to include more training data or increase the cutoff?
Please suggest.

Thanks,
mayank

[njzjz]

I suggest to increase sel of your model. See #108 (comment). You can check whether it’s enough in all snapshots.