Inferring the Isotropic-Nematic Phase Transition with Generative Machine Learning

ER Beyerle and P Tiwary, PHYSICAL REVIEW LETTERS, 135, 068102 (2025).

DOI: 10.1103/1wdj-ym3s

Generative machine learning models are capable of learning the phase behavior in condensed matter systems such as the Ising model. We utilize a score-based modeling procedure called thermodynamic maps to describe the isotropic-nematic phase transition in a melt of Gay-Berne ellipsoids. When trained on samples from a single temperature on either side of the phase transition, this generative machine learning approach infers effectively the nematic order parameter at intermediate temperatures. These results demonstrate score-based models' ability to learn the physics of a nontrivial liquid crystal phase transition.

Return to Publications page