Zakari Denis, EPFL

Leuchs-Russell Auditorium, A.1.500, Staudtstr. 2

Location details

Abstract:

Investigating and simulating quantum-optical systems is essential both to further our fundamental understanding of quantum many-body physics as to advance emerging quantum technologies. These systems exhibit rich nonequilibrium dynamics due to strong interactions, infinite-dimensional local Hilbert spaces, and coupling to structured environments, making their simulation a long-standing challenge. Neural quantum states (NQS) offer a promising path forward, as they are not constrained by entanglement nor affected by the sign problem. Indeed, such neural-network representations of the quantum many-body wave function have consistently shown remarkable accuracy across a variety of systems. Besides spin systems, where they stand as the state-of-the-art method for frustrated lattices, NQS have been successfully applied to other degrees of freedom with applications reaching well beyond ground-state calculations. However, their application to driven-dissipative quantum systems presents key obstacles: (i) efficiently representing interacting bosons in an infinite local Hilbert space and (ii) capturing the non-Hermitian dynamics induced by system-environment interactions.

In this talk, I will present recent advances addressing these challenges. In particular, I will highlight ongoing efforts to develop variational neural representations and algorithms suited for bosonic many-body systems and nonequilibrium settings, and I will explore their potential for capturing driven-dissipative dynamics in open quantum systems. While this direction is yet barely explored, recent progress suggests that the field is at a turning point, with clear and exciting pathways for further investigation.