Interatomic Potential Machine learning Interface
Project description
FitSNAP3
A Python Package For Training SNAP Interatomic Potentials for use in LAMMPS Molecular Dynamics
Copyright (2016) Sandia Corporation. Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation, the U.S. Government retains certain rights in this software. This software is distributed under the GNU General Public License
A short description of each parent class of the FitSNAP package will be provided below along with an overall workflow of the code.
This folder should be on your PYTHONPATH, and if you change the name of this folder (fitsnap3) then your command line execution will have to reflect this change: python3 -m newfoldername [inputfile] [options]
scrapers/
- Parent class that parses training data from stored files.
calculators/
- Parent class where feature vectors (descriptors) for the learned model are generated based on configurations parsed from scraper.
solvers/
- Parent class that handles how regression (fitting) is carried out such that weights on feature vectors are found to minimize the loss function with respect to ground truth data (i.e. from DFT).
io/
- Parent class that handles all input and output functions.
parallel_tools.py
- This is the real star of the show in terms of functionality, all of the parent classes will lean on this to distribute tasks and collect data across ranks/nodes. Pull requests that modify this will be given heavy scrutiny, plenty of additional functionality can be more easily made within scrapers/calculators/solvers/io.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
fitsnap3-2.0.1.tar.gz
(40.5 kB
view hashes)
Built Distribution
fitsnap3-2.0.1-py3-none-any.whl
(69.4 kB
view hashes)