FAIR Chemistry's library of machine learning methods for chemistry

Read about the UMA model and dataset release.

If you want to explore model capabilities check out our educational demo

Install fairchem-core using pip,

pip install git+https://github.com/facebookresearch/fairchem.git@fairchem_core-2.0.0#subdirectory=packages/fairchem-core

The PyPI install (pip install fairchem-core) is not available right now as we are waiting for a few dependencies to release their PyPI packages, will update this soon when it's available!

The easiest way to use pretrained models is via the ASE FAIRChemCalculator. A single uma model can be used for a wide range of applications in chemistry and materials science by picking the appropriate task name for domain specific prediction.

Make sure you have a Hugging Face account, have already applied for model access to the UMA model repository, and have logged in to Hugging Face using an access token.

• oc20: use this for catalysis
• omat: use this for inorganic materials
• omol: use this for molecules
• odac: use this for MOFs
• omc: use this for molecular crystals

Relax adsorbate on a catalytic surface,

from ase.build import fcc100, add_adsorbate, molecule
from ase.optimize import LBFGS
from fairchem.core import FAIRChemCalculator

calc = FAIRChemCalculator(hf_hub_filename="uma_sm.pt", device="cuda", task_name="oc20")

# Set up your system as an ASE atoms object
slab = fcc100("Cu", (3, 3, 3), vacuum=8, periodic=True)
adsorbate = molecule("CO")
add_adsorbate(slab, adsorbate, 2.0, "bridge")

slab.calc = calc

# Set up LBFGS dynamics object
opt = LBFGS(slab)
opt.run(0.05, 100)

Or relax an inorganic crystal,

from ase.build import bulk
from ase.optimize import FIRE
from ase.filters import FrechetCellFilter
from fairchem.core import FAIRChemCalculator

calc = FAIRChemCalculator(hf_hub_filename="uma_sm.pt", device="cuda", task_name="omat")

atoms = bulk("Fe")
atoms.calc = calc

opt = LBFGS(FrechetCellFilter(atoms))
opt.run(0.05, 100)

Run molecular MD,

from ase import units
from ase.io import Trajectory
from ase.md.langevin import Langevin
from ase.build import molecule
from fairchem.core import FAIRChemCalculator

calc = FAIRChemCalculator(hf_hub_filename="uma_sm.pt", device="cuda", task_name="omol")

atoms = molecule("H2O")
atoms.calc = calc

dyn = Langevin(
    atoms,
    timestep=0.1 * units.fs,
    temperature_K=400,
    friction=0.001 / units.fs,
)
trajectory = Trajectory("my_md.traj", "w", atoms)
dyn.attach(trajectory.write, interval=1)
dyn.run(steps=1000)

A single repository simplifies testing and ensures consistency across our interconnected core, data and application packages. The repo 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 chemistry
• fairchem.data: Dataset downloads and input generation codes
• fairchem.demo: Python API for the Open Catalyst Demo
• fairchem.applications: Follow up applications and works (AdsorbML, CatTSunami, etc.)

You can still use models from fairchem version 1, by installing version 1,

pip install fairchem-core==1.10

And using the OCPCalculator

from fairchem.core import OCPCalculator

calc = OCPCalculator(
    model_name="EquiformerV2-31M-S2EF-OC20-All+MD",
    local_cache="pretrained_models",
    cpu=False,
)

fairchem is available under a MIT License.