Where symmetries between Machine Learning and Quantum Mechanics improve simulations

towards-data-science

This post was originally published by Therese Paoletta at Towards Data Science - Medium Tagged

UT Austin physics grad student Chris Roth tackles exotic quantum problems with the power of symmetry.

The laws of quantum mechanics, infamous for being unintuitive, predict a litany of strange effects. Many exotic materials, such as superconductors, have such complicated behavior that even the most powerful computers cannot handle their calculations [1]. As a result, some systems must be conquered through innovative, large-scale simulations [2]. UT Austin researcher Chris Roth has developed a machine-learning algorithm that uses two symmetries to make this problem more tractable [3]. First, the periodic system finds an analog in the input structure. Second, the forces between the particles conveniently obey a type of dependence characteristic of the output of the algorithm.


Figure 1. A section of a two-dimensional crystal. The pattern of an orange dot in the bottom left corner and green dot in top right corner repeats itself for every translation in the vertical or horizontal direction of length a. In this case, the unit cell is the square of length a with one orange and one green atom. Therefore, the view from point P will be the same as long as you move in these types of discrete displacements. Image by author.

Figure 2. A recurrent neural network with three inputs and outputs. The output is always weighted more by recent input than earlier ones. Image by author.
Spread the word

This post was originally published by Therese Paoletta at Towards Data Science - Medium Tagged

Related posts