A Caltech Library Service

Bayesian learning of thermodynamic integration and numerical convergence for accurate phase diagrams

Ladygin, V. and Beniya, I. and Makarov, E. and Shapeev, A. (2021) Bayesian learning of thermodynamic integration and numerical convergence for accurate phase diagrams. Physical Review B, 104 (10). Art. No. 104102. ISSN 2469-9950. doi:10.1103/physrevb.104.104102.

[img] PDF - Published Version
See Usage Policy.

[img] PDF - Submitted Version
See Usage Policy.


Use this Persistent URL to link to this item:


Accurate phase diagram calculation from molecular dynamics requires systematic treatment and convergence of statistical averages. In this work we propose a Gaussian process regression based framework for reconstructing the free-energy functions using data of various origins. Our framework allows for propagating statistical uncertainty from finite molecular dynamics trajectories to the phase diagram and automatically performing convergence with respect to simulation parameters. Furthermore, our approach provides a way for automatic optimal sampling in the simulation parameter space based on a Bayesian optimization approach. We validate our methodology by constructing phase diagrams of two model systems, the Lennard-Jones and soft-core potential, and compare the results with the existing studies and our coexistence simulations. Finally, we construct the phase diagram of lithium at temperatures above 300 K and pressures below 30 GPa from a machine-learning potential trained on ab initio data. Our approach performs well when compared to coexistence simulations and experimental results.

Item Type:Article
Related URLs:
URLURL TypeDescription Paper
Ladygin, V.0000-0002-5697-6956
Beniya, I.0000-0002-7095-194X
Makarov, E.0000-0001-5879-5131
Shapeev, A.0000-0002-7497-5594
Additional Information:© 2021 American Physical Society. (Received 3 May 2021; revised 19 August 2021; accepted 27 August 2021; published 7 September 2021) A.S. thanks Richard Otis (Caltech) for extensive discussions that led to the creation of this work. This work was supported by the Russian Foundation for Basic Research under Grant No. 20-53-12012.
Funding AgencyGrant Number
Russian Foundation for Basic Research20-53-12012
Issue or Number:10
Record Number:CaltechAUTHORS:20210927-213256207
Persistent URL:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:111063
Deposited By: George Porter
Deposited On:27 Sep 2021 22:29
Last Modified:27 Sep 2021 22:29

Repository Staff Only: item control page