Anderson localization and spin-orbit induced emergent physics in elemental tellurium

Luis Martin-Moreno
(CSIC-Universidad de Zaragoza, Espanha)

Resumo:

Multidimensional data is notoriously troublesome to visualise; thus, it may contain symmetries that are difficult to detect. It is thus desirable to have methods that can detect them. Recently, machine learning methods have started to be applied to these problems. For instance, a machine learning technique was developed for finding continuous symmetries in fields that are solutions of partial differential equations [1]. In this talk, I will present a method for finding discrete symmetry in a general dataset (not necessarily representing a field) based on multi-evaluated neural networks. I will show how this technique works using some examples extracted from physics. This method may have applications not only to symmetry detection but also to optimal inverse design.

[1] Z. Liu and M. Tegmark, PHYSICAL REVIEW LETTERS 128, 180201 (2022)

Departamento de Física dos Sólidos
Instituto de Física UFRJ