search

atomistic-machine-learning / schnetpack

SchNetPack - Deep Neural Networks for Atomistic Systems
Other
782 stars 214 forks source link

Does trained schnet potential be able to use in LAMMPS? #307

Closed Roy-Kid closed 1 year ago

Roy-Kid commented 3 years ago

Hi, I want to use Schnetpack to fit a polymer potential, and embed to the lammps then do simulation. So is there any way to export the trained potential/Neural network to use in the LAMMPS, like what deepmd do?

ktschuett commented 3 years ago

I wanted to look into this, also for OpenMM. It seems that you need to write some interface code in C++. Do you have any experience with this?

Roy-Kid commented 3 years ago

I have a little experiment to extend LAMMPS and write a book which I think is the best tutorial for now. You can find it at the official website of LAMMPS https://lammps.org/books.html and can download ebook from http://libgen.rs/search.php?req=LAMMPS&open=0&res=25&view=simple&phrase=1&column=def.

And I am learning how to write OpenMM plugin because We are developing an auto differentiate multipole forcefield. We can exchange the experiment more often.

mgastegger commented 1 year ago

We added a LAMMPS interface in PR #501. Documentation can be found in the How-To section of the docs.