Font Size:
Deep Generative Models Approaches the Ising Model
Building: Cero Infinito
Room: Posters hall
Date: 2024-12-10 04:30 PM – 06:30 PM
Last modified: 2024-11-19
Abstract
In this work we present Deep Generative Models (DGM) applications in the Ising Model.
Our goal is to analyze huge amounts of data and get insights about it. We use different
types of DGM to study the Ising model. It was possible to make our machine learning
models learn the intrinsic behavior of our statistical systems. We could, after the learning
process, reproduce system configurations with physical observables in accordance with
our input data, obtained from importance sampling. Also we could compare the results
of our models from the well established simulation methods. We could also have a bet-
ter understanding of how machine learning models “learn” physics. In the literature of
statistical physics of machine learning this type of work can be found [1, 2, 3].
[1] Torlai, G., Melko, R., “Learning thermodynamics with Boltzmann machines”. Phys. Rev. B
94, 165134 (2016).
[2] Carrasquilla, J., Melko, R., “Machine learning phases of matter”. Nature Phys 13, 431–434
(2017).
[3] Morningstar, A., Melko, R., “Deep Learning the Ising Model Near Criticality”. Journal of
Machine Learning Research 18 (2017)
Our goal is to analyze huge amounts of data and get insights about it. We use different
types of DGM to study the Ising model. It was possible to make our machine learning
models learn the intrinsic behavior of our statistical systems. We could, after the learning
process, reproduce system configurations with physical observables in accordance with
our input data, obtained from importance sampling. Also we could compare the results
of our models from the well established simulation methods. We could also have a bet-
ter understanding of how machine learning models “learn” physics. In the literature of
statistical physics of machine learning this type of work can be found [1, 2, 3].
[1] Torlai, G., Melko, R., “Learning thermodynamics with Boltzmann machines”. Phys. Rev. B
94, 165134 (2016).
[2] Carrasquilla, J., Melko, R., “Machine learning phases of matter”. Nature Phys 13, 431–434
(2017).
[3] Morningstar, A., Melko, R., “Deep Learning the Ising Model Near Criticality”. Journal of
Machine Learning Research 18 (2017)