arXiv:2507.12152 · Chinese Physics Letters

Efficient Preparation of Fermionic Superfluids in an Optical Dipole Trap through Reinforcement Learning

Yueyang Min, Zi‐Liang Li, Yi Zhong, Jia-An Xuan, Jian Lin, Lei Feng, Xiaopeng Li

Published Sep 1, 2025• 1 citation

Abstract

Abstract We demonstrate a reinforcement learning (RL)-based control framework for optimizing evaporative cooling in the preparation of strongly interacting degenerate Fermi gases of 6Li. Using a Soft Actor-Critic (SAC) algorithm, the system autonomously explores a high-dimensional parameter space to learn optimal cooling trajectories. Compared to conventional exponential ramps, our method achieves up to 130% improvement in atomic density within 0.5 second, revealing non-trivial control strategies that balance fast evaporation and thermalization. While our current optimization focuses on the evaporation stage, future integration of other cooling stages, such as gray molasses cooling, could further extend RL to the full preparation pipeline. Our result highlights the promise of RL as a general tool for closed-loop quantum control and automated calibration in complex atomic physics experiments.

DOI: 10.1088/0256-307x/42/10/100302

Related papers

High-dimensional reinforcement learning for optimization and control of ultracold quantum gases

Nicholas Milson, Arina Tashchilina, T Ooi +3 more

Abstract Machine-learning (ML) techniques are emerging as a valuable tool in experimental physics, and among them, reinforcement learning (RL) offers the potential to control high-dimensional, multistage processes in the presence of fluctuating environments. In this experimental work, we apply RL to the preparation of an ultracold quantum gas to realize a consistent and large number of atoms at microkelvin temperatures. This RL agent determines an optimal set of 30 control parameters in a dynamically changing environment that is characterized by 30 sensed parameters. By comparing this method to that of training supervised-learning regression models, as well as to human-driven control schemes, we find that both ML approaches accurately predict the number of cooled atoms and both result in occasional superhuman control schemes. However, only the RL method achieves consistent outcomes, even in the presence of a dynamic environment.

arXiv:2308.05216Machine Learning Science and TechnologyDec 1, 20239 citations

Reinforcement learning optimization of a 6 Li magneto-optical trap: DDPG versus SAC

Deok Young Lee, jaeha shin, Jae-yoon Choi

We report an experimental implementation of deep reinforcement learning (RL) for optimizing the loading stage of a 6 Li magneto-optical trap (MOT) in a high-dimensional continuous control space. An off-policy actor-critic agent observes in-situ fluorescence images at 111 ms intervals and updates seven experimental parameters over 18 decision steps. Learning is performed directly on the apparatus using a sparse terminal reward R = N × A extracted from an absorption image, where N is the atom number, and A is the peak optical depth. We benchmark Deep Deterministic Policy Gradient (DDPG) and Soft Actor-Critic (SAC) across three independent 300-episode training runs under identical alignment. Both algorithms surpass the human-optimized (HO) baseline, but the central finding concerns reliability rather than peak performance. In deterministic replay validation, the three SAC waveforms show reward coefficients of variation of 9–14%, and the minimum reward of each waveform exceeds the HO mean reward. The DDPG waveforms show up to three times the dispersion, and even the best-performing DDPG run contains a catastrophic near-zero shot. The SAC solutions are nearly time-independent and consistently converge to a blue-detuned repump laser with an elevated field gradient, reminiscent of compressed-MOT operating conditions. Our results show that off-policy actor-critic RL can autonomously optimize high-dimensional laser-cooling sequences in an operating ultracold-atom apparatus, and that SAC has a reproducibility advantage over DDPG in a real noisy environment.

Optics ExpressAug 31, 20261 citation

Adaptive Quantum Control Algorithm Optimization

Jincheng Zhang

This paper investigates the optimization of quantum control algorithms using reinforcement learning (RL). The core claim is that RL can significantly enhance the control precision and efficiency of quantum devices. We propose a framework where the quantum control algorithm is treated as an intelligent agent, interacting with the quantum environment to learn optimal control strategies. This approach allows for adaptive control, addressing the challenges inherent in designing fixed control protocols for complex quantum systems. The methodology involves defining a reward function that reflects the desired quantum state evolution, using RL algorithms to train the agent, and validating the learned control strategy through simulations. We demonstrate the potential of this approach to improve control accuracy and robustness, particularly in scenarios with noisy environments or rapidly changing system parameters. The results suggest a promising direction for leveraging machine learning techniques to unlock the full potential of quantum control.

Sep 6, 20260 citations

Multi-parameter optimization of polarization gradient cooling for ⁸⁷Rb atoms based on reinforcement learning

Changwen Liang, Shaojun Gao, Jixun Liu +5 more

Polarization gradient cooling (PGC) plays an important role in many cold atom applications including the formation of Bose-Einstein condensates (BECs) and cooling of single atom. Traditional parameter optimization of PGC usually relies on subjective expertise, faces challenges in fine manipulation, and exhibits low optimization efficiency. Here, we propose a segmented control method that differs from traditional PGC process by expanding the experiment parameters from 3 to 30. Subsequently, the conventional timing optimization problem is reformulated as a Markov Decision Process (MDP), and the experiment parameters are optimized using a reinforcement learning model. With proper settings of hyperparameters, the learning process exhibits good convergence and powerful parameter exploration capabilities. Finally, we capture ~4.3×10⁸ cold atoms, with a phase space density of ~7.1×10^(-4) at a temperature of ~3.7 μK in ~14.6 minutes. Our work paves the way for intelligent preparation of degenerate quantum gas.

Dec 23, 20240 citations

Multi-parameter optimization of polarization gradient cooling for ⁸⁷Rb atoms based on reinforcement learning

Changwen Liang, Shaojun Gao, Jixun Liu +5 more

Jul 31, 20240 citations

Adaptive Reinforcement Learning for Robust Open Quantum System Control: A Multi-Task Framework with Temporal Optimization

Haftu W. Fentaw, Steve Campbell, Simon Caton

We present a Multi-task Soft Actor-Critic (SAC) Reinforcement Learning framework designed for open-system quantum control across diverse Hamiltonians, which learns optimal pulse sequences while simultaneously discovering problem-specific evolution time T and number of control pulse segments N. Experimental results across 51 Hamiltonian variations demonstrate that the multi-task SAC model is able to generate control pulses that can drive a system, under environment noise, from its initial state to its target state with high fidelities, establishing essential foundations for universal quantum control applicable to realistic noisy quantum devices. Through progressive expansion of the training Hamiltonian set, we investigate if a single multi-task model trained using a given number of sample Hamiltonians can successfully accomplish state-transfer tasks for Hamiltonians drawn from the same Hamiltonian space but not encountered during training. In addition, our Robustness Infidelity Measure (RIM) analysis reveals that SAC trained policies exhibit superior robustness to pulse amplitude perturbations and decoherence rate variations compared to GRAPE-optimized controls.

quant-pharXiv:2605.26925May 27, 20260 citations