Análise tomográfica de redes neurais aplicada a problemas quânticos de muitos corpos

Abstract

The use of artificial neural networks to represent many-body quantum states has spar- ked great interest in recent years, offering a promising approach to addressing complex problems arising in this context. In this work, we propose a tomography of the weights of a neural network applied to the problem to obtain the ground state of 1D spin chains with neighbor interaction. One of the objectives is to understand how unsupervised machine learning methods can identify regimes in 1D spin chains, such as ferromagnetic, antifer- romagnetic, and frustrated regimes. Our work seeks to contribute to the understanding of how neural networks incorporate physical characteristics of physical models, such as entanglement, into their synaptic weights.