MatterSim

MatterSim is a deep learning atomistic model across elements, temperatures and pressures.

Installation

Install from PyPI

TODO: To be released to PyPI.

Install from source code

Requirements:

• Python == 3.9

To install the package, run the following command under the root of the folder:

conda env create -f environment.yaml
conda activate mattersim
pip install -e .

Obtain the model checkpoints

You may obtain our model checkpoints from

# link to somewhere

Usage

A minimal test

from mattersim.forcefield.potential import Potential
from mattersim.datasets.utils.build import build_dataloader

potential = Potential.load(load_path="/path/to/checkpoint", device="cuda:0")
from ase.build import bulk
si = bulk("Si", "diamond", a=5.43)
dataloader = build_dataloader([si], only_inference=True, model_type=model_name)
predictions = potential.predict_properties(dataloader, include_forces=True, include_stresses=True)
print(predictions)

Reference

If you use MatterSim, please cite our preprint on arXiv:

@article{yang2024mattersim,
  title={Mattersim: A deep learning atomistic model across elements, temperatures and pressures},
  author={Yang, Han and Hu, Chenxi and Zhou, Yichi and Liu, Xixian and Shi, Yu and Li, Jielan and Li, Guanzhi and Chen, Zekun and Chen, Shuizhou and Zeni, Claudio and others},
  journal={arXiv preprint arXiv:2405.04967},
  year={2024}
}

Researcher and Developers

MatterSim is currently in active development. If you have any specific research interests related to this model or encounter any issues, please don't hesitate to reach out to us.