fairchem is the FAIR Chemistry's centralized repository of all its data, models, demos, and application efforts for materials science and quantum chemistry.
⚠️ If you cloned the repo before January 17, 2025 and are actively developing, please read this announcement.
If you are looking for Open-Catalyst-Project/ocp, it can now be found at fairchem.core. Visit its corresponding documentation here.
The repository is organized into several directories to help you find what you are looking for:
fairchem.core: State of the art machine learning models for materials science and chemistryfairchem.data: Dataset downloads and input generation codesfairchem.demo: Python API for the Open Catalyst Demofairchem.applications: Follow up applications and works (AdsorbML, CatTSunami, etc.)
Packages can be installed in your environment by the following:
pip install -e packages/fairchem-{fairchem-package-name}
fairchem.core requires you to first create your environment
Pretrained models can be used directly with ASE through our OCPCalculator interface:
from ase.build import fcc100, add_adsorbate, molecule
from ase.optimize import LBFGS
from fairchem.core import OCPCalculator
# Set up your system as an ASE atoms object
slab = fcc100("Cu", (3, 3, 3), vacuum=8)
adsorbate = molecule("CO")
add_adsorbate(slab, adsorbate, 2.0, "bridge")
calc = OCPCalculator(
model_name="EquiformerV2-31M-S2EF-OC20-All+MD",
local_cache="pretrained_models",
cpu=False,
)
slab.calc = calc
# Set up LBFGS dynamics object
dyn = LBFGS(slab)
dyn.run(0.05, 100)If you are interested in training your own models or fine-tuning on your datasets, visit the documentation for more details and examples.
Since many of our repositories rely heavily on our other repositories, a single repository makes it really easy to test and ensure consistency across repositories. This should also help simplify the installation process for users who are interested in integrating many of the efforts into one place.
fairchem is available under a MIT License.
