RQMP (CPM) Seminar

Quantum state representation with artificial and spiking neural networks

Stefanie Czischek

University of Ottawa

Artificial neural networks (ANNs) have found promising applications in various fields over the last few years. Beyond text prediction, game playing, and traffic sign recognition, ANNs have demonstrated great powers in simulating quantum many-body systems. Groundbreaking works have shown that ANNs provide a general and efficient wave function ansatz that can be tuned to reconstruct quantum states from a limited amount of experimental measurement data or to find ground states of given Hamiltonians via variational energy minimization.

In this talk, I will introduce the concepts of ANN-based quantum state representations using different network architectures. I will discuss the powers and limitations of the models and show how they can overcome the limitations of conventional simulation methods. I will further introduce spiking neural networks as biologically-inspired network architectures which can run efficiently on analog neuromorphic hardware. I will show that these network models are promising candidates to advance quantum many-body physics.