OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Mattergen

MatterGen is a generative model and Python implementation for designing inorganic materials across the periodic table, with pretrained models for unconditional and property-conditioned candidate generation.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Mattergen — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use MatterGen when a materials-design workflow needs generated inorganic structures, either without a specified property condition or guided by a supported pretrained condition.

Who it is for

  • Materials researchers exploring inorganic crystal candidates
  • Python users seeking to generate structures under material-property constraints

Documented uses

  • Generate candidate inorganic crystal structures with a pretrained base model.
  • Generate candidates conditioned on properties such as magnetic density, band gap or bulk modulus.
  • Save generated structures as CIF and extended XYZ files for further research workflows.
Before you choose

Requirements & limits

The repository is public and specifies an MIT license. The documented installation path assumes Linux, Python 3.10 or later and a CUDA GPU; Apple Silicon use is marked experimental.

  • Pretrained checkpoints cover distinct conditions, including chemical system, space group, magnetic density, band gap and bulk modulus.
  • Generation guidance trades off stronger adherence to target values against sample diversity and realism.
  • Generated structures and optional denoising trajectories are written to local output files.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Generate candidates conditioned on magnetic density

Use the pretrained magnetic-density model to generate inorganic material candidates for a target value.

Before you start

  • Python 3.10 or later
  • Linux and a CUDA GPU for the documented installation path
  • MatterGen installed and the dft_mag_density pretrained checkpoint available

Steps

  1. Install MatterGen using the repository's documented environment or package installation instructions.

  2. Set MODEL_NAME to dft_mag_density and RESULTS_PATH to a chosen output directory.

  3. Run mattergen-generate with the pretrained model, a batch size, the target dft_mag_density value, and a diffusion guidance factor.

Expected outcome

Generated crystal structures are saved in the selected results directory, including CIF and extended XYZ outputs.

Guide limitations

  • Increasing guidance tends to improve adherence to the requested property value at the expense of diversity and realism.
  • Exact reproducibility is only expected with the same software and hardware configuration because some CUDA operations are nondeterministic.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
microsoft/mattergen · Official homepage title · retained in its source language
How we review sourcesSuggest a correction