https://github.com/picocentauri/plode-code
Auxiliary Code Repository for 'Physics-inspired Equivariant Descriptors of Non-bonded Interactions'
https://github.com/picocentauri/plode-code
Last synced: 8 months ago
JSON representation
Auxiliary Code Repository for 'Physics-inspired Equivariant Descriptors of Non-bonded Interactions'
- Host: GitHub
- URL: https://github.com/picocentauri/plode-code
- Owner: PicoCentauri
- Created: 2023-09-26T13:57:54.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2023-10-03T09:20:53.000Z (about 2 years ago)
- Last Synced: 2025-02-17T06:25:41.617Z (10 months ago)
- Size: 1.2 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Physics-inspired Equivariant Descriptors of Non-bonded Interactions - Auxiliary Code Repository
[](https://arxiv.org/abs/2308.13208)
[](https://doi.org/10.24435/materialscloud:23-99)
The repository includes codes used for the work "Physics-inspired Equivariant
Descriptors of Non-bonded Interactions" available at https://arxiv.org/abs/2308.13208.
Besides the provided code this repository brings together multiple software project
written in lab-cosmo:
- https://github.com/lab-cosmo/metatensor as a data storage format for atomistic machine
learning;
- https://github.com/luthaf/rascaline to compute LODE-based representations;
- https://github.com/lab-cosmo/equisolve to computing machine learning models based on
metatensor objects.
## Installation
You'll need a C++ compiler, CMake, and a [Rust](htpps://rustup.rs/) compiler installed
on your machine. Then, in a fresh Python environment (virtualenv or conda), run the
following command to install the code and all dependencies:
```bash
pip install -r requirements.txt
pip install -r requirements_rascaline.txt
```
## Usage
The [examples](examples) folder contains Python source code to train and evaluate the
linear as well as neural network (nn) potential based on LODE descriptors. For the
examples in this repository we take a small subset of the dimer dataset from the main publication.
The complete data set is available for download at [Materials
Cloud](https://doi.org/10.24435/materialscloud:23-99).
Additionally, the [examples](examples) folder contains Python classes to generate
splined radial integrals for
[rascaline](https://luthaf.fr/rascaline/latest/references/api/python/utils/splines.html)