Availability and License

On April 15, 2023, the MLIP-2 code went open-source. It is now available under the BSD license essentially requiring the users only acknowledging the use of the software. If you publish paper based on the MLIP-2 software, please cite

[1] I.S. Novikov, K. Gubaev, E.V. Podryabinkin, A.V. Shapeev.
The MLIP package: Moment tensor potentials with mpi and active learning.
Machine Learning: Science and Technology, 2(2):025002, 2020. (doi: 10.1088/2632-2153/abc9fe)

Additional references that may be cited (at the discretion of the user) are:

Original MTP potentials:
[2] A.V. Shapeev, Moment tensor potentials: A class of systematically improvable interatomic potentials,
Multiscale Modeling and Simululation, volume 14, pages 1153-1173, 2016. (doi: 10.1137/15M1054183)

Active learning of MTP:
[3] E.V. Podryabinkin, A.V. Shapeev, Active Learning of Linearly Parametrized Interatomic Potentials,
Computational Materials Science, volume 140, pages 171-180, 2017. (doi: 10.1016/j.commatsci.2017.07.010)

Multicomponent MTP:
[4] Konstantin Gubaev, Evgeny V. Podryabinkin, Gus L. W. Hart, Alexander V. Shapeev,
Accelerating high-throughput searches for new alloys with active learning of interatomic potentials,
Computational Materials Science, volume 156, pages 148-156, 2019. (doi: 10.1016/j.commatsci.2017.08.031)

How to download

You can download the software by simply cloning repository, e.g., by running

git clone

Documentation and Materials

The materials available for you to get started are

  1. The tutorial paper describing the use of MLIP:
  2. and files of the main repository
  3. Documentation and comprehensive examples available in the doc/ folder of the main repository
  4. Elementary examples available in the test/examples/ folder of the main repository
  5. A couple of simple tutorials available at

We strive to have clear documentation and examples to help everyone with using MLIP on their own. We will happily fix issues in the documentation and examples should you find any, however, we will not be able to offer extensive user support and training, except for our collaborators.