Machine learning and quantum phase transition

Abstract

The study of quantum phase transitions is an aspect of great interest in the scientific community. Despite the efforts to detect the transitions and classify the phases, currently, this is still an enormous problem given the computational effort required in the diagonalization of high-dimensional matrices. Our goal, in this sense, is to understand how Machine Learning (ML) techniques can be useful in detecting and classifying quantum phase transitions. We will work with different models, both spin-1 and spin-1/2, testing the effectiveness of the ML methods. We intend, using classical ML techniques, to develop a "universal classifier" algorithm so that given an arbitrary and unknown physical model, it could describe the phase diagram of the system. (AU)