Turin Lattice Meeting 2023

Session

Machine learning for Lattice Field Theory

22 Dec 2023, 11:15

Aula Magna (Dipartimento di Fisica)

5. The quest for the lost mode: Overcoming Mode-Collapse in Flow-Based Sampling for Lattice Field Theories

Kim Nicoli (University of Bonn)

22/12/2023, 11:15

Talk

Normalizing flows allow for independent sampling. For this reason, it is hoped that they can avoid the tunneling problem of local-update MCMC algorithms for multi-modal distributions. In this work, we first point out that the tunneling problem is also present for normalizing flows but is shifted from the sampling to the algorithm's training phase. Specifically, normalizing flows often suffer...

8. Flow-based sampling for Effective String Theory

Elia Cellini (Università di Torino / INFN Torino)

22/12/2023, 11:45

Talk

Effective String Theory (EST) is a powerful non-perturbative method used to study confinement in pure gauge theories through the modeling of the interquark potential in terms of vibrating strings. Due to the criticality of EST, an efficient numerical method to simulate such theories is still lacking. However, in the last years a novel class of deep generative algorithms called Normalizing...

7. Symmetry equivariant neural networks

Matteo Favoni (Swansea University)

22/12/2023, 12:15

Talk

In this talk, I will present the advantages of using neural networks that respect symmetries over their non-symmetric counterparts in lattice field theory applications. The concept of equivariance will be explained, together with the reason why it is a sufficient condition for the network to respect the desired symmetry. The benefits of equivariant networks will first be exemplified in the...