Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
QI: Fachverband Quanteninformation
QI 34: Quantum Control I
QI 34.3: Vortrag
Donnerstag, 13. März 2025, 15:00–15:15, HS II
Classical surrogates of quantum control landscapes — •Martino Calzavara1,2, Tommaso Calarco1,2,3, and Felix Motzoi1,2 — 1Peter Grünberg Institute - Quantum Control (PGI-8), Forschungszentrum Jülich GmbH, Wilhelm-Johnen-Straße, 52428 Jülich, Germany — 2Institute for Theoretical Physics, University of Cologne, Zülpicher Straße 77, 50937 Cologne, Germany — 3Dipartimento di Fisica e Astronomia, Università di Bologna, 40127 Bologna, Italy
Since the introduction of the GRAPE algorithm for efficiently computing fidelity gradients, piecewise-constant controls have become a widely adopted ansatz for studying Quantum Optimal Control problems. The time evolution for this class of time-dependent Hamiltonians can be represented as a Parametrized Quantum Circuit, allowing us to analyze the properties of the fidelity as a function of the control pulses - the so-called Quantum Control Landscape - by employing concepts and techniques borrowed from Quantum Machine Learning (QML) and Variational Quantum Algorithms (VQA). Among these techniques are classical surrogate models, which represent the output of a quantum circuit as a linear combination of non-linear feature maps, providing valuable insights into the representational power of QML models and the structure of VQA landscapes. In this work, we employ classical surrogate models as a theoretical tool to investigate the properties of Quantum Control Landscapes, and to learn approximate representations of such landscapes using supervised learning.
Keywords: Variational Quantum Algorithms; Quantum Machine Learning; Parametrized Quantum Circuits; Supervised learning; Quantum Optimal Control