Abstract
The subject of this paper is the technology (the ‘how’) of constructing machine-learning interatomic potentials, rather than science (the ‘what’ and ‘why’) of atomistic simulations using machine-learning potentials. Namely, we illustrate how to construct moment tensor potentials using active learning as implemented in the MLIP package, focusing on the efficient ways to automatically sample configurations for the training set, how expanding the training set changes the error of predictions, how to set up ab initio calculations in a cost-effective manner, etc. The MLIP package (short for Machine-Learning Interatomic Potentials) is available at https://mlip.skoltech.ru/download/.
Highlights
Machine-learning interatomic potentials have recently been a subject of research and they are turning into a tool of research
In this manuscript we focus on the methodology of applying moment tensor potentials (MTP) and active learning to performing atomistic simulations
We introduce the notion of extrapolation grade γ(cfg)—a feature of a configuration that correlates with the prediction error but does not require ab initio information to be calculated prior to its evaluation; its precise definition is given below
Summary
Machine-learning interatomic potentials have recently been a subject of research and they are turning into a tool of research. The concept of machine-learning potentials was pioneered in 2007 by Behler and Parrinello [5], who, motivated by the success of approximating potential energy surfaces of small molecules with neural networks, proposed the use of neural networks as a functional form of interatomic potentials, going beyond the small molecular systems by employing the locality of interaction. MTPs, like SOAP-GAP, are not based on solely two- and three-body descriptors and can provably approximate an arbitrary local interaction, and so are ACE [18] and PIP [75] Because of this MTP together with GAP showed excellent accuracy in recent cheminformatics benchmark test [54] and interatomic potential test [82]; in the latter MTP showed a very good balance between accuracy and computational efficiency when compared against other machine-learning potentials. The description of and references to the MLIP code is kept to minimum, yet retained for the purpose of easier preproduction of the described results
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
