MLatom – Atomistic machine learning

MLatom 2 is released.

Atomistic machine learning (AML) simulations are used in chemistry at an ever-increasing pace. A large number of AML models have been developed, but their implementations are scattered among different packages, each with its own conventions for input and output.

MLatom 2 solves this problem. It is a software package that provides an integrative platform for a wide variety of AML simulations by implementing from scratch and interfacing existing software for a range of state-of-the-art models.

MLatom has been developed by Pavlo Dral and his team at the Xiamen University. Max Pinheiro Jr, from our team in Marseille, has contributed to some of the new developments. I have also been involved in the conceptualization of the NEA-ML method, included in MLatom too.

MLatom 2 has many outstanding features, including models based on Kernel methods

• KREG (native implementation),
• sGDML
• GAP-SOAP

and models based on Neural networks:

• ANI
• DeepPot-SE
• PhysNet

In this paper, we discuss the MLatom 2 implementation and overview the theoretical foundations behind all these methods.

The modular structure of MLatom allows for easy extensions to more AML model types. MLatom 2 also has many other capabilities useful for AML simulations, such as the support of custom descriptors, farthest-point and structure-based sampling, hyperparameter optimization, model evaluation, and automatic learning curve generation.

It can also be used for such multi-step tasks as Δ-learning, self-correction approach, and absorption spectrum simulation within the machine-learning nuclear-ensemble approach.

Several of these MLatom 2 capabilities are showcased in application examples.

MB

Reference

[1] P. O. Dral, F. Ge, B.-X. Xue, Y.-F. Hou, M. Pinheiro Jr., J. Huang, M. Barbatti, MLatom 2: An Integrative Platform for Atomistic Machine Learning, Top. Curr. Chem., 379, 27 (2021). DOI: 10.1007/s41061-021-00339-5