SAMOP 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
Q: Fachverband Quantenoptik und Photonik
Q 42: Poster III
Q 42.13: Poster
Mittwoch, 8. März 2023, 16:30–19:00, Empore Lichthof
Machine-learning optimized entanglement in photonic topological insulators — •Saipavan Vengaladas1,2, Armando Pérez-Leija4, Kurt Busch1,3, and Konrad Tschernig4 — 1Max-Born-Institut, 12489 Berlin, Germany — 2Department of Physics, Freie Universität Berlin, 14195 Berlin, Germany — 3Humboldt-Universität zu Berlin, AG Theoretische Optik & Photonik, 12489 Berlin, Germany — 4CREOL/College of Optics & Photonics, University of Central Florida, Orlando, 32816 FL, USA
Photonic Floquet topological insulators feature so-called edge states, and wavepackets built from edge states are topologically protected. As a result, they naturally resist the scrambling effects of disorder and retain their shape during propagation. This useful property gives rise to the intriguing possibility of protecting two-photon entangled states. While the necessary conditions to protect two-photon entanglement in topological insulators (TIs) have been well established[1], it is yet unclear how to construct optimally entangled two-photon states. This is no trivial task since the degree of entanglement, and the amount of bulk scattering are naturally competing properties inside TIs. In this work, we take a more general approach by defining a black-box optimization problem, which is tackled using a machine-learning based Latent Action Monte Carlo Tree Search (LA-MCTS) meta-algorithm[2]. We present the optimized states that we obtain by exploring the space of all possible states and discuss their properties. [1] K.Tschernig et al., Nat. Commun.12, 1974 (2021). [2] arxiv.org/abs/2007.00708