SKM 2023 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
DY: Fachverband Dynamik und Statistische Physik
DY 22: Machine Learning in Dynamics and Statistical Physics II
DY 22.5: Talk
Tuesday, March 28, 2023, 15:00–15:15, ZEU 160
Phase Diagram of the J1-J2 Ising Model from Unsupervised Learning: Neural Networks vs Image Comparison — •Burak Çivitcioğlu1, Andreas Honecker1, and Rudolf A. Römer2 — 1Laboratoire de Physique Théorique et Modélisation, CNRS UMR 8089, CY Cergy Paris Université, Cergy-Pontoise, France — 2Department of Physics, University of Warwick, Coventry, CV4 7AL, United Kingdom
Machine learning methods have been shown to be one of the novel approaches in identifying the phases and phase transitions in models of statistical physics. Here, we study the performance of unsupervised learning in the J1-J2 Ising model. We benchmark the results for phase diagram reconstruction using variational autoencoders (VAEs) against straightforward image comparison. We show that such image comparison can result in accuracies that are akin to that of VAEs.