betaFeedback

WorksPhysics and Astronomy

Adaptive Hamiltonian Identification via Quantum Reinforcement Learning under Non-Stationary Dynamics

Abstract

Accurate identification of the Hamiltonian governing a quantum system is important for quantum-system characterization, device calibration, simulation, and reliable operation of quantum hardware. In practical systems, however, the governing dynamics may vary over time because of parameter drift, changing control conditions, or transitions between operating regimes. In this work, we formulate Hamiltonian identification as a quantum reinforcement learning (QRL) problem in which a quantum agent processes sequential quantum-state observations and estimates the governing Hamiltonian parameters. We first consider stationary dynamics using a Vanilla-QRL agent and then introduce a recurrent GRU-QRL architecture to track non-stationary dynamics. The framework is evaluated on transverse-field Ising model systems under stationary dynamics, abrupt Hamiltonian switching, rapidly changing Hamiltonians, random switch locations over longer trajectories, and gradual Hamiltonian drift. We further compare the proposed approach with a Quantum Graph Recurrent Neural Network (QGRNN) baseline and evaluate frozen policies on unseen trajectories. Results show that both approaches accurately reconstruct stationary Hamiltonians, while GRU-QRL provides substantially improved tracking under non-stationary dynamics. These results demonstrate the potential of recurrent QRL as an adaptive system-identification tool for monitoring time-varying quantum systems.

Paper details

vlnc:W_yw9h47py
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
arXiv nonexclusive license
Usage

views · 5

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…