Erlangen 2022 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
Q: Fachverband Quantenoptik und Photonik
Q 61: Quantum Optics (Miscellaneous)
Q 61.9: Poster
Donnerstag, 17. März 2022, 16:30–18:30, P
Machine learning Lindblad dynamics — •Francesco Carnazza1, Federico Carollo1, Sabine Andergassen1, Dominik Zietlow2, Georg Martius2, and Igor Lesanovsky1 — 1Institut für Theoretische Physik and Center for Quantum Science, Universität Tübingen, Auf der Morgenstelle 14, 72076 Tübingen, Germany — 2Max Planck Institute for Intelligent Systems, Max-Planck-Ring 4, 72076 Tübingen, Germany
Even if full knowledge of the wave-function of a quantum system is unattainable, important information can still be retrieved by observing local degrees of freedom. Typically, it is possible to single out and measure just a subsystem, while regarding the rest as a bath. In the simplest case, the evolution of the reduced quantum state obtained by tracing out the environment is governed by a Markovian, i.e. time independent, quantum master equation, also known as Lindblad master equation. Here we investigate if it is possible to train a fully interpretable neural network which learns the parameters of a Lindblad generator [1]. We test this idea in a class of spin models, and investigate in which certain situations the network can indeed provide good predictions.
[1] P. Mazza et al. Phys. Rev. Research 3, 023084 (2021)