High-dimensional reinforcement learning for optimization and control of ultracold quantum gases
arXiv:2308.05216 · doi:10.1088/2632-2153/ad1437
Abstract
Machine-learning techniques are emerging as a valuable tool in experimental physics, and among them, reinforcement learning offers the potential to control high-dimensional, multistage processes in the presence of fluctuating environments. In this experimental work, we apply reinforcement learning to the preparation of an ultracold quantum gas to realize a consistent and large number of atoms at microkelvin temperatures. This reinforcement learning agent determines an optimal set of thirty control parameters in a dynamically changing environment that is characterized by thirty 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 machine learning approaches accurately predict the number of cooled atoms and both result in occasional superhuman control schemes. However, only the reinforcement learning method achieves consistent outcomes, even in the presence of a dynamic environment.
References in corpus (28)
- Adam: A Method for Stochastic Optimization
- Practical Bayesian Optimization of Machine Learning Algorithms
- Gaussian Error Linear Units (GELUs)
- All Models are Wrong, but Many are Useful: Learning a Variable's Importance by Studying an Entire Class of Prediction Models Simultaneously
- A high-bias, low-variance introduction to Machine Learning for physicists
- Reinforcement Learning in Different Phases of Quantum Control
- Flow Navigation by Smart Microswimmers via Reinforcement Learning
- Fast machine-learning online optimization of ultra-cold-atom experiments
- Rapid production of Rb BECs in a combined magnetic and optical potential
- Bandit Problems with Side Observations
- Making, probing and understanding Bose-Einstein condensates
- Multiparameter optimisation of a magneto-optical trap using deep learning
- Basic protocols in quantum reinforcement learning with superconducting circuits
- Computer-inspired Quantum Experiments
- Unsupervised machine learning of topological phase transitions from experimental data
- Remote optimization of an ultra-cold atoms experiment by experts and citizen scientists
- Machine Learning for Achieving Bose-Einstein Condensation of Thulium Atoms
- Machine learning assisted readout of trapped-ion qubits
- Applying machine learning optimization methods to the production of a quantum gas
- Machine-learning-accelerated Bose-Einstein condensation
- Multiqubit and multilevel quantum reinforcement learning with quantum technologies
- Storing short single-photon-level optical pulses in Bose-Einstein condensates for high-performance quantum memory
- Active Learning Approach to Optimization of Experimental Control
- Observing a topological phase transition with deep neural networks from experimental images of ultracold atoms
- Bi-color atomic beam slower and magnetic field compensation for ultracold gases
- Maximizing the capture velocity of molecular magneto-optical traps with Bayesian optimization
- Reinforcement Learning in Ultracold Atom Experiments
- Bayesian optimization of Bose-Einstein condensation via evaporative cooling model