Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2412.11793 · Chinese Physics B

Accelerated Bayesian optimization in deep cooling atoms

Xiaoxiao Ma, Changwen Liang, Rong Sha, Chao Zhou, Guochao Wang, Jixun Liu, Shuhua Yan, Lingxiao Zhu

Published Nov 27, 2025• 0 citations
arXivHTMLPDF

Abstract

Abstract Laser cooling, which cools atomic and molecular gases to near absolute zero, is the crucial initial step for nearly all atomic gas experiments. However, fast achievement of numerous sub-μK cold atoms is challenging. To resolve the issue, we propose and experimentally validate an intelligent polarization gradient cooling approach enhanced by an optical lattice, utilizing maximum hypersphere compensation sampling Bayesian optimization (MHCS-BO). MHCS-BO demonstrates a twofold increase in optimization efficiency and superior prediction accuracy compared to conventional Bayesian optimization. Finally, approximately 10 8 cold atoms at a temperature of 0.4±0.2 μK can be achieved given the optimal parameters within 15 minutes. Our work provides an intelligent protocol, which can be generalized to other high-dimensional parameter optimization problems, and paves the way for the preparation of ultracold atoms in quantum experiments.

DOI
10.1088/1674-1056/ae24ef

Related papers

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.

•Jul 31, 2024•0 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, 2024•0 citations

Multi-parameter optimization of polarization gradient cooling for 87 Rb 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.

Optics Express•Oct 11, 2024•6 citations

Machine-learning-accelerated Bose-Einstein condensation

Zachary Vendeiro, Joshua Ramette, Alyssa Rudelis +5 more

Machine learning is emerging as a technology that can enhance physics experiment execution and data analysis. Here, we apply machine learning to accelerate the production of a Bose-Einstein condensate (BEC) of $^{87}\mathrm{Rb}$ atoms by Bayesian optimization of up to 55 control parameters. This approach enables us to prepare BECs of $2.8\ifmmode\times\else\texttimes\fi{}{10}^{3}$ optically trapped $^{87}\mathrm{Rb}$ atoms from a room-temperature gas in 575 ms. The algorithm achieves the fast BEC preparation by applying highly efficient Raman cooling to near quantum degeneracy, followed by a brief final evaporation. We anticipate that many other physics experiments with complex nonlinear system dynamics can be significantly enhanced by a similar machine-learning approach.

arXiv:2205.08057•Physical Review Research•Dec 27, 2022•35 citations

Strong Light — Matter Interactions in a Bow-Tie Cavity and Machine Learning-Enhanced Cooling

Matthew L. Peters

Neutral atoms in optical tweezers have emerged as a highly promising platform, offering long intrinsic coherence times, the ability to arrange qubits in flexible, reconfigurable geometries, and the capacity for fast entangling gates mediated by strong Rydberg interactions. However, their advancement towards large-scale quantum processing is often impeded by several key limitations. These include slow or destructive qubit readout methodologies, probabilistic array preparation which necessitates time-consuming atom rearrangement, and cooling techniques that can be dependent on specific atomic internal structures or require complex experimental geometries. This thesis addresses these critical challenges by detailing the first applications of a novel experimental system that integrates an array of neutral cesium atoms with a high-cooperativity (𝜂 > 20) four-mirror optical bow-tie cavity. We first leverage our optical resonator to demonstrate fast, non-destructive, and number-resolved readout of individual atoms. This capability enables the real-time observation of light-assisted collisional dynamics and underpins an adaptive feedback protocol that achieves quasi-deterministic single-atom loading with a 92(2)% success probability. Building on this robust control, we utilize the cavity’s unique properties—including a narrow linewidth and intrinsic running-wave modes—to realize efficient three-dimensional cooling near the ground state and report the first demonstration of programmable few-atom Bragg scattering in a bow-tie resonator. To address the overarching challenge of optimizing complex, high-dimensional experimental systems, we then pioneer the use of machine learning for atomic cooling. We introduce a deep reinforcement learning agent, trained via a sim-to-real protocol, that learns to actively cool a single atom more rapidly than conventional methods. In a separate demonstration, we apply Bayesian optimization to the multi-parameter landscape of magneto-optical trap compression and polarization gradient cooling to produce a Bose-Einstein condensate directly from polarization gradient cooling. Collectively, these results establish the integration of bow-tie cavities and machine intelligence as a powerful paradigm for building the next generation of robust, scalable, and high-fidelity neutral-atom quantum processors.

•Sep 1, 2025•0 citations

Bayesian optimization of Bose-Einstein condensation via evaporative cooling model

Jihao Ma, Ruihuan Fang, Chengyin Han +8 more

To achieve Bose-Einstein condensation, one may implement evaporative cooling by dynamically regulating the power of laser beams forming the optical dipole trap. We propose and experimentally demonstrate a protocol of Bayesian optimization of Bose-Einstein condensation via the evaporative cooling model. Applying this protocol, pure Bose-Einstein condensate of 87Rb with 2.4X10e4 atoms can be produced via evaporative cooling from the initial stage when the number of atoms is 6.0X10e5 at a temperature of 12μK. In comparison with Bayesian optimization via blackbox experiment, our protocol only needs a few experiments required to verify some close-to-optimal curves for optical dipole trap laser powers, therefore it greatly saves experimental resources.

arXiv:2303.05358•Mar 9, 2023•1 citation
Abstract Laser cooling, which cools atomic and molecular gases to near absolute zero, is the crucial initial step for nearly all atomic gas experiments. However, fast achievement of numerous sub-μK cold atoms is challenging. To resolve the issue, we propose and experimentally validate an intelligent polarization gradient cooling approach enhanced by an optical lattice, utilizing maximum hypersphere compensation sampling Bayesian optimization (MHCS-BO). MHCS-BO demonstrates a twofold increase in optimization efficiency and superior prediction accuracy compared to conventional Bayesian optimization. Finally, approximately 10 8 cold atoms at a temperature of 0.4±0.2 μK can be achieved given the optimal parameters within 15 minutes. Our work provides an intelligent protocol, which can be generalized to other high-dimensional parameter optimization problems, and paves the way for the preparation of ultracold atoms in quantum experiments.
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.
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.
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.
Machine learning is emerging as a technology that can enhance physics experiment execution and data analysis. Here, we apply machine learning to accelerate the production of a Bose-Einstein condensate (BEC) of 87Rb^{87}\mathrm{Rb}87Rb atoms by Bayesian optimization of up to 55 control parameters. This approach enables us to prepare BECs of $2.8\ifmmode\times\else\texttimes\fi{}{10}^{3}$ optically trapped 87Rb^{87}\mathrm{Rb}87Rb atoms from a room-temperature gas in 575 ms. The algorithm achieves the fast BEC preparation by applying highly efficient Raman cooling to near quantum degeneracy, followed by a brief final evaporation. We anticipate that many other physics experiments with complex nonlinear system dynamics can be significantly enhanced by a similar machine-learning approach.
Neutral atoms in optical tweezers have emerged as a highly promising platform, offering long intrinsic coherence times, the ability to arrange qubits in flexible, reconfigurable geometries, and the capacity for fast entangling gates mediated by strong Rydberg interactions. However, their advancement towards large-scale quantum processing is often impeded by several key limitations. These include slow or destructive qubit readout methodologies, probabilistic array preparation which necessitates time-consuming atom rearrangement, and cooling techniques that can be dependent on specific atomic internal structures or require complex experimental geometries. This thesis addresses these critical challenges by detailing the first applications of a novel experimental system that integrates an array of neutral cesium atoms with a high-cooperativity (𝜂 > 20) four-mirror optical bow-tie cavity. We first leverage our optical resonator to demonstrate fast, non-destructive, and number-resolved readout of individual atoms. This capability enables the real-time observation of light-assisted collisional dynamics and underpins an adaptive feedback protocol that achieves quasi-deterministic single-atom loading with a 92(2)% success probability. Building on this robust control, we utilize the cavity’s unique properties—including a narrow linewidth and intrinsic running-wave modes—to realize efficient three-dimensional cooling near the ground state and report the first demonstration of programmable few-atom Bragg scattering in a bow-tie resonator. To address the overarching challenge of optimizing complex, high-dimensional experimental systems, we then pioneer the use of machine learning for atomic cooling. We introduce a deep reinforcement learning agent, trained via a sim-to-real protocol, that learns to actively cool a single atom more rapidly than conventional methods. In a separate demonstration, we apply Bayesian optimization to the multi-parameter landscape of magneto-optical trap compression and polarization gradient cooling to produce a Bose-Einstein condensate directly from polarization gradient cooling. Collectively, these results establish the integration of bow-tie cavities and machine intelligence as a powerful paradigm for building the next generation of robust, scalable, and high-fidelity neutral-atom quantum processors.
To achieve Bose-Einstein condensation, one may implement evaporative cooling by dynamically regulating the power of laser beams forming the optical dipole trap. We propose and experimentally demonstrate a protocol of Bayesian optimization of Bose-Einstein condensation via the evaporative cooling model. Applying this protocol, pure Bose-Einstein condensate of 87Rb with 2.4X10e4 atoms can be produced via evaporative cooling from the initial stage when the number of atoms is 6.0X10e5 at a temperature of 12μK. In comparison with Bayesian optimization via blackbox experiment, our protocol only needs a few experiments required to verify some close-to-optimal curves for optical dipole trap laser powers, therefore it greatly saves experimental resources.