Zoom (Please
contact
amug@uwaterloo.ca
for
meeting link) 
Speaker
Annabelle Bohrdt, Universität Regensburg
Title
Machine learning methods for quantum many-body systems
Abstract
Predicting and understanding the phenomena emerging from the interplay of many strongly correlated quantum particles is inherently challenging. The Hilbert space dimension of a quantum many-body system grows exponentially with the number of particles, thus inhibiting exact numerical simulations for extended systems. One possibility to investigate such systems consists in quantum simulation, where a well controlled and highly tunable quantum system, such as neutral atoms in optical lattices, is used to emulate the model of interest. Another approach are approximate numerical methods, which aim to take the most relevant degrees of freedom into account. In this talk, I will give an overview of three different routes my research group takes to make use of machine learning methods to investigate and understand strongly correlated quantum many-body systems: (i) Using optimization methods for state preparation and readout in quantum simulation experiments to engineer and probe desired quantum many-body states; (ii) using neural networks to represent quantum states (neural quantum states, NQS), in particular for ground states, dynamics, and finite temperature states; (iii) an interpretable analysis of quantum data, as obtained from quantum simulators or numerical simulations, using tailored machine learning techniques. The combination of these three approaches allows us to simulate quantum systems of interest and obtain physical insights through interpretable analysis of the resulting data.