Skip to main content

MLatom

PyPI version Downloads Downloads per month License: Apache 2.0 Documentation

MLatom

Version 3.27.0

MLatom is an open-source package for atomistic simulations with machine learning and quantum chemical methods — DFT, wavefunction-based, and semi-empirical approximations. Use it as a Python library, through input files, or from the command line — run it locally, or online with no installation.

Website: http://mlatom.com · Documentation: http://mlatom.com/docs · GitHub: https://github.com/dralgroup/mlatom

Run online — no installation

Run MLatom in your browser on either online platform — both powered by Protomia, with an AI assistant for autonomous atomistic simulations:

Installation

MLatom is developed and tested on Linux — the only platform it is verified on — and requires Python 3.9 or newer. It is often usable on macOS but nothing there is tested; Windows is not supported. Install and upgrade via pip:

pip install -U mlatom

That pulls the core, including the ML backends. A few features need one extra package — install only what you use:

for install
PySCF methods: single point, TDDFT, frequencies and thermochemistry, densities, Raman pip install pyscf
AIMNet2 models pip install "aimnet==0.0.1" (the version MLatom's interface targets; newer aimnet releases are a rewrite that does not work with the shipped models)
format conversions (SMILES to xyz and back) conda install -c conda-forge openbabel
the ASE interface (ASE optimizers and dynamics) and MACE pip install ase
KREG_API backend (ml_program='MLatomF' needs none of this) conda install -c conda-forge mkl mkl-service
MDtrajNet pip install "e3nn==0.5.0" (with e3nn 0.5.1 or newer, the published MDtrajNet-1 model produces different trajectories)
MLTPA / DMC / hyperparameter search pip install rdkit / pyvibdmc / hyperopt

The AIQM2 quick start below additionally needs the DFT-D4 program, which is distributed via conda:

conda install -c conda-forge 'dftd4==3.6.0'   # the version MLatom is tested against
export dftd4bin=$(which dftd4)

See the installation guide for optional interfaces and other methods. UAIQM and other unpublished methods are available through the Aitomic add-ons.

Quick start

Optimize the geometry of a water molecule with AIQM2 — an AI-enhanced quantum-mechanical method (native to MLatom, CHNO elements) that reaches beyond-DFT accuracy at semi-empirical cost:

import mlatom as ml

mol = ml.data.molecule.from_xyz_string('''3

O    0.00000    0.00000    0.11779
H    0.00000    0.75545   -0.47116
H    0.00000   -0.75545   -0.47116
''')
aiqm2 = ml.methods(method='AIQM2')
opt = ml.optimize_geometry(model=aiqm2, initial_molecule=mol).optimized_molecule
print(opt.energy)          # optimized energy in hartree (≈ -76.3838)

Prefer zero setup? Run it online on the Aitomistic Hub or Aitomistic Lab@XMU — no installation needed.

Features

  • Methods — universal ML potentials (ANI, AIMNet2); AI-enhanced QM methods (AIQM1/AIQM2/AIQM3, UAIQM) approaching coupled-cluster accuracy at semi-empirical cost; and DFT, semi-empirical, and wavefunction methods via interfaces (PySCF, Gaussian, ORCA, xtb, MNDO, Turbomole, DFTB+, Sparrow, Columbus).
  • Simulations — geometry optimization, transition-state search, IRC, frequencies and thermochemistry, molecular dynamics, surface-hopping nonadiabatic dynamics, IR/Raman/UV–vis spectra, and periodic boundary conditions.
  • ML models — train and use KREG, GAP-SOAP, ANI, MACE, and more, with active learning, Δ-learning, transfer learning, and self-correction.

Full manuals and tutorials are at mlatom.com/docs.

Recent releases

Full release notes.

  • 3.27 (upcoming) — AIQM3 included directly in open-source MLatom; single-point calculations save molecular properties to JSON.
  • 3.26 — periodic-material simulations with MACE-MP-0 and MACE-MPA-0; improved numerical gradients and Hessians for optimization and frequency calculations.

Using with AI agents

MLatom can be driven by AI agents. Protomia is the main assistant and runs online with nothing to install; Aitomia is an agentic system academic users can install locally; and MLatom Skills is an open, agent-agnostic collection of skills for your own AI coding assistant (Protomia, Claude, Cursor, GitHub Copilot, ...). The MLatom repository also ships an AGENTS.md to orient your agent. See Using MLatom with AI agents for an overview.

License and citations

License

MLatom is open-source software under the Apache License 2.0.

Copyright 2013-2026 Pavlo O. Dral (http://dr-dral.com/)

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this software except in compliance with the License. You may obtain a copy of the License at https://www.apache.org/licenses/LICENSE-2.0.

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Citations

If you use MLatom in scientific work, please cite it. For convenience, the citations are provided below in BibTeX format.

@article{MLatom 3,
author = {Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Zheng, Peikun and Chen, Yuxinxin and Barbatti, Mario and Isayev, Olexandr and Wang, Cheng and Xue, Bao-Xin and Pinheiro Jr, Max and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Shuang and Zhang, Lina and Ullah, Arif and Zhang, Quanhao and Ou, Yanchi},
title = {MLatom 3: A Platform for Machine Learning-Enhanced Computational Chemistry Simulations and Workflows},
journal = {J. Chem. Theory Comput.},
volume = {20},
number = {3},
pages = {1193--1213},
DOI = {10.1021/acs.jctc.3c01203},
year = {2024},
type = {Journal Article}
}

@misc{MLatomProg,
author = {Dral, Pavlo O. and Ge, Fuchun and Hou, Yi-Fan and Chen, Yuxinxin and Zheng, Peikun and Xue, Bao-Xin and Martyka, Mikolaj and Zhang, Lina and Martinka, Jakub and Zhang, Quanhao and Tong, Xin-Yu and Ullah, Arif and Pios, Sebastian V. and Kumar, Vignesh B. and Ou, Yanchi and Jr, Max Pinheiro and Su, Yuming and Dai, Yiheng and Chen, Yangtao and Zhang, Shuang and Hu, Jinming and Bispo, Matheus O.},
title = {MLatom: A Package for Atomistic Simulations with Machine Learning},
note = {version 3.27.0, Xiamen University, Xiamen, China},
year = {2013--2026},
type = {Computer Program}
}

Metadata

Release files for mlatom 3.27.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for mlatom 3.27.0
File Size Uploaded Provenance Actions
mlatom-3.27.0.tar.gz 61.9 MB Without provenance Details for mlatom-3.27.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for mlatom 3.27.0
File Interpreter ABI Platform Actions
mlatom-3.27.0-py3-none-any.whl Python 3 none any Details for mlatom-3.27.0-py3-none-any.whl

Total release size: 124.0 MB

Release files / mlatom-3.27.0.tar.gz

Download URL mlatom-3.27.0.tar.gz
Size 61.9 MB
Tags Source
SHA-256 checksum
How to use checksums
d6eb66a450e9b0f29f925b4d1a8abf05dbd1115dd9dbb52f53dde01986d7ba29
BLAKE2b-256 checksum
How to use checksums
ba2bc6b4304cab481444827d507fd9000fdafdd687d8748298312426a1fc6a62
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.18

Release files / mlatom-3.27.0-py3-none-any.whl

Download URL mlatom-3.27.0-py3-none-any.whl
Size 62.1 MB
Tags Python 3
SHA-256 checksum
How to use checksums
b14d532131291525041e326896634d6a4251380892ffb0c551c122de65835009
BLAKE2b-256 checksum
How to use checksums
6c2dfad52e6a7eb71a2e5c5d853d105b9dafaaf77ca63ca2b196e96dbf843d24
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.13.18

Security for mlatom 3.27.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

3.27.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page