DeepHall framework for QMC study of fractional quantum Hall system

Strong correlation drives exotic emergent phenomena, challenging to model—fractional quantum Hall systems being a key example. Traditional methods often neglect Landau level mixing due to its complexity, even though this mixing can significantly alter the ground state. We propose the DeepHall deep learning framework and apply it to FQH systems at 1/3 and 2/5 fillings, naturally including higher Landau levels and achieving lower energies than lower Landau level exact diagonalization, matching state-of-the-art methods like fp-DMC. This highlights deep learning’s potential to advance studies of strongly correlated systems.

• Y. Qian et al. Phys. Rev. Lett. 134, 176503 (2025)
• [News] 成果 | 陈基课题组与合作者开发分数量子霍尔态的深度学习量子蒙特卡洛算法