Deep reinforcement learning for autonomous control of hole-doped Hubbard clusters: A comparative study
APL Machine Learning · 01.03.2026 · 4 (1)
DOI: 10.1063/5.0310012 ↗Kurz gefasst
Fachartikel in der Zeitschrift APL Machine Learning, online erschienen am 23.03.2026. Verfasst von Shivanshu Dwivedi und 1 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Trinity College.
Abstract (Originalsprache)
Engineering electron correlations in quantum dot arrays demand navigation of high-dimensional, non-convex parameter spaces, where hole doping fundamentally alters the physics. We present a rigorous comparative study of two control paradigms for the 1-hole of half-filled Hubbard model: (i) systematic physics-guided design and (ii) autonomous deep reinforcement learning (RL) with geometry-aware neural architectures. While systematic analysis reveals key design principles—such as field-induced localization for trapping the mobile hole—it is computationally intractable for optimization. We demonstrate that an autonomous RL agent, benchmarked across five 3D lattices (tetrahedron to FCC), achieves human-competitive accuracy (R2 > 0.97) and 95.5% success on held-out tasks. Critically, the RL agent achieves this performance with 103−4× greater sample efficiency than grid search and outperforms other black-box optimization methods. Transfer learning demonstrates 91% few-shot generalization to unseen geometries. This work establishes autonomous RL as a viable, highly efficient framework for rapid optimization and non-obvious strategy discovery in complex quantum systems.
Förderung
- Trinity College
Zitieren
APA
Dwivedi, S., & Palandage, K. (2026). Deep reinforcement learning for autonomous control of hole-doped Hubbard clusters: A comparative study. APL Machine Learning, 4(1). https://doi.org/10.1063/5.0310012
Vancouver
Dwivedi S, Palandage K. Deep reinforcement learning for autonomous control of hole-doped Hubbard clusters: A comparative study. APL Machine Learning. 2026;4(1). doi:10.1063/5.0310012
BibTeX
@article{Dwivedi2026,
author = {Dwivedi, Shivanshu and Palandage, Kalum},
title = {Deep reinforcement learning for autonomous control of hole-doped Hubbard clusters: A comparative study},
journal = {APL Machine Learning},
year = {2026},
volume = {4},
number = {1},
doi = {10.1063/5.0310012}
}
EUVIMED