Tatiana Kostiuchenko participated in MMM2018 with a report “High-entropy alloys investigation using machine-learning potentials” at Osaka, Japan.
We won a grant from the Russian Foundation for Fundamental Research (18-32-00736) with our collaborators from the Moscow Institute of Physics and Technology. This project is dedicated to the development of an atomistic model for evaluating the thermodynamic stability of high-entropy alloys. This work is based on the interatomic potentials, that are being developed in our group.
We have developed a methodology for a machine-learning acceleration of high-throughput computational alloy discovery in collaboration with the research group of Gus Hart from BYU. The approach relies on our machine-learning interatomic potentials, MTP, which are trained using our active learning approach. This allows for predicting new stable alloys with geometrically new structures, in contrast to the commonly used on-lattice models (such as the cluster expansion approach) that assume a fixed host lattice. We achieve a 1000-fold speed-up in comparison to the ab initio-based screening (e.g., with density functional theory) .
The corresponding article “Accelerating high-throughput searches for new alloys with active learning of interatomic potentials” was recently accepted to Computational Materials Science and now is available at arXiv.
Our group participated in the CSP-2018 with the following reports:
We investigate the extent to which a charge-equilibration (QEq) model can improve the accuracy of interatomic potentials in general, and a machine-learning potential in particular. The corresponding paper is available at arXiv.
Together with our colleagues from Artem Oganov lab we delivered a talk “Crystal Structure Prediction using Machine Learning Potentials” at X International Scientific Conference “Kinetics and Mechanisms of Crystallization. Crystallization and materials of the future”.
Together with Yury Suleimanov from the Cyrpus Institute we published the preprint of the paper “Automated Calculation of Thermal Rate Coefficients using Ring Polymer Molecular Dynamics and Machine-Learning Interatomic Potentials with Active Learning: Application to Thermally Activated Gas Phase Chemical Reactions”, which is now available at arXiv.
With the help of machine-learning potentials, a new superconducting allotropy of boron was predicted. It contains 54 atoms in the unit cell and has Im-3 symmetry group.
Our group participated in the All-Russian conference with international participation “Solid State Chemistry and Functional Materials” with the following reports:
Our group won a 6 mln rub/year Russian Science Foundation (РНФ) grant.
Our article “Machine Learning of Molecular Properties: Locality and Active Learning” was published in the Journal of Chemical Physics.
Konstantin Gubaev delivered a talk “Machine Learning Interatomic Potentials for Multicomponent Systems” in the Second Physics Informed Machine Learning Conference at Santa Fe, USA.
Konstantin Gubaev delivered a talk “Machine Learning and Multiscale Modeling” in the Second All-Russian Scientific and Practical Conference of Young Scientists at Moscow, Russia.
Our group participated in the FAMMS-2018 with reports dedicated to machine learning for materials.
Our research group took part in the Skoltech-MIT conference with the following poster reports:
We have open PhD student/research scientist positions in the project of development of machine learning algorithms to molecular modeling. The developed algorithms will be applied, together with our Russian and international collaborators, to computational materials discovery, investigation and development of alloys, organic and inorganic semiconductors, etc. More info here.