# ACEsuit/mace

MACE - Fast and accurate machine learning interatomic potentials with higher order equivariant message passing.

Repository: https://github.com/ACEsuit/mace
Canonical: https://ross.abutalabs.com/products/acesuit-mace
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-08-21T21:37:12+00:00

## Health v2 (maintenance only)
Score: 89/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 71, longevity 100
- inputs: {"age_days": 1534, "days_push": 12, "days_rel": 115, "gap_med": 45, "n_releases_24m": 10}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1324, forks 469 (observed 2026-08-28T04:04:22.360346+00:00)

## What it is
MACE is a Python library implementing fast and accurate machine learning interatomic potentials using higher-order equivariant message passing neural networks. It includes training and evaluation tooling plus pretrained foundation models (MACE-MP, MACE-OFF, MACE-Polar) usable with ASE.

## Use cases
- train machine learning interatomic potentials for molecular simulations
- run molecular dynamics with a neural network force field
- predict energies and forces for atomic structures
- simulate materials with a pretrained universal force field
- model organic molecules with transferable force fields
- finetune a foundation model on my own DFT data

## When to choose
- you need accurate ML interatomic potentials with equivariant message passing
- you want pretrained universal force fields for materials or organic molecules
- you work with ASE and want GPU-accelerated potential evaluation

## When to avoid
- you need classical fixed-form force fields rather than learned potentials
- you need a non-Python or JAX-first training workflow (see mace-jax instead)
- your project requires a permissively documented stable API - documentation is partial

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, simulation, gpu-computing
- domain: machine-learning, chemistry, simulation
- platform: python
- tags: interatomic-potentials, molecular-dynamics, equivariant-neural-networks, force-fields, graph-neural-networks, computational-chemistry, materials-science, linux, macos, gpu

## Member repositories
- ACEsuit/mace (main) score 89

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:22.360346+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:46:40.049840+00:00, confidence not recorded.
  - readme: https://github.com/ACEsuit/mace (fetched 2026-08-28T04:04:22.360346+00:00, sha ad3116789ca1)
- Data as of 2026-08-30T08:39:29.467469+00:00.