Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
QI: Fachverband Quanteninformation
QI 23: Quantum Control
QI 23.9: Vortrag
Donnerstag, 21. März 2024, 11:45–12:00, HFT-FT 131
Accurate Quantum Feedback Control via Conditional State Tomography with Reinforcement Learning — •Sangkha Borah1, 2 and Bijita Sarma2 — 1Max Planck Institute for the Science of Light, Staudtstraße 2, 91058 Erlangen, Germany — 2Friedrich-Alexander-Universitat Erlangen-Nurnberg, Staudtstraße 7, 91058 Erlangen, Germany
The efficacy of measurement-based feedback control (MBFC) protocols faces challenges due to the presence of measurement noise, impacting the accurate inference of the underlying dynamics of a quantum system from noisy continuous measurement records. This, in turn, hinders the determination of precise control strategies. To address these limitations, this study investigates a real-time stochastic state estimation approach facilitating noise-free monitoring of conditional dynamics, encompassing the complete density matrix of the quantum system. Referred to as 'conditional state tomography,' this method allows for leveraging noisy measurement records within a single quantum trajectory. Consequently, it empowers the development of refined MBFC strategies, effectively overcoming the constraints posed by measurement noise. The proposed approach holds promise for diverse feedback quantum control scenarios and proves particularly advantageous for reinforcement-learning (RL)-based control. In RL applications, the agent can be trained using arbitrary conditional averages of observables or the full density matrix as input, enabling the rapid and accurate learning of control strategies.
Keywords: Quantum control; Tomography; Reinforcement leraning; Machine learning; Feedback control