ÖPG-CMD Joint Meeting 2026

Machine learning latent orders from the two-particle vertex

Sep 22, 2026, 4:00 PM

30m

HS 12.11 (University of Graz)

4) Invited talk M22 - Compressing complexity: machine learning in hard and soft condensed matter physics Mini-Colloquium

Speaker

Sabine Andergassen (TU Wien)

Description

We explore the ability of machine learning models to extract information encoded in the two-particle vertex Γ that generalizes across different quantum phases. To steer the model away from relying on global phase-specific patterns we employ a sub-sampling strategy that encourages the model to learn general features tied to the phase specific competition between kinetic energy and Coulomb repulsion. We show that an autoencoder trained only on data from antiferromagnetic and ferromagnetic phases is able to reconstruct samples from a previously unseen superconducting phase. This demonstrates that the model captures essential aspects of the underlying many-body physics.