The solution of quantum field theories is one of the most challenging topics in modern theoretical physics. Recently, promising advances have relied on the use of variational approaches that exploit machine learning techniques [1]. These approaches encode the gauge symmetries in the architecture of neural networks [2], which provides advantages with respect to other sampling techniques [3]. Variational solutions to quantum field theories can be then exploited to find numerical solutions to these problems [4].
In this project, you will employ machine learning techniques to address a simplified quantum field theories, and solve it variationally. The results you will obtain will be benchmarked with other methods, including lattice field theory. You will study and analyze the quality of the variational ansatz depending on different architectures. You will look into the efficiency of these new variational techniques, as well as different sampling and optimization strategies.