Applying machine learning optimization methods to the production of a quantum gas

Abstract

We apply three machine learning strategies to optimize the atomic cooling processes utilized in the production of a Bose-Einstein condensate (BEC). For the first time, we optimize both laser cooling and evaporative cooling mechanisms simultaneously. We present the results of an evolutionary optimization method (Differential Evolution), a method based on non-parametric inference (Gaussian Process regression) and a gradient-based function approximator (Artificial Neural Network). Online optimization is performed using no prior knowledge of the apparatus, and the learner succeeds in creating a BEC from completely randomized initial parameters. Optimizing these cooling processes results in a factor of four increase in BEC atom number compared to our manually-optimized parameters. This automated approach can maintain close-to-optimal performance in long-term operation. Furthermore, we show that machine learning techniques can be used to identify the main sources of instability within the apparatus.

19 pages, 7 figures, 1 table

References in corpus (14)

• Deep Learning in Neural Networks: An Overview
• Quantum phase transition from a superfluid to a Mott insulator in a gas of ultracold atoms
• Berezinskii-Kosterlitz-Thouless Crossover in a Trapped Atomic Gas
• Emergence of a Turbulent Cascade in a Quantum Gas
• Fast machine-learning online optimization of ultra-cold-atom experiments
• Time-averaged adiabatic ring potential for ultracold atoms
• Machine Learning techniques for state recognition and auto-tuning in quantum dots
• Multiparameter optimisation of a magneto-optical trap using deep learning
• Machine learning assisted readout of trapped-ion qubits
• Ultracold atoms in multiple-radiofrequency dressed adiabatic potentials
• Trapping Ultracold Atoms in a Time-Averaged Adiabatic Potential
• Stochastic optimization of a cold atom experiment using a genetic algorithm
• Evolutionary optimization of an experimental apparatus
• Optimizing Quantum Gas Production by an Evolutionary Algorithm

Cited by in corpus (21)

• Machine-learning-accelerated Bose-Einstein condensation
• Single-exposure absorption imaging of ultracold atoms using deep learning
• Single-site-resolved imaging of ultracold atoms in a triangular optical lattice
• Perspective on new implementations of atomtronic circuits
• Smooth bang-bang shortcuts to adiabaticity for atomic transport in a moving harmonic trap
• Deep learning based quantum vortex detection in atomic Bose-Einstein condensates
• Counterdiabatic control of transport in a synthetic tight-binding lattice
• Active Learning Approach to Optimization of Experimental Control
• Bose-Einstein-condensate source on an optical-grating-based atom chip for quantum sensor applications
• Thermometry of one-dimensional Bose gases with neural networks
• Optimizing persistent currents in a ring-shaped Bose-Einstein condensate using machine learning
• Vortex detection in atomic Bose-Einstein condensates using neural networks trained on synthetic images
• High-dimensional reinforcement learning for optimization and control of ultracold quantum gases
• Quantum Reservoir Computing for Speckle-Disorder Potentials
• Creation and manipulation of quantized vortices in Bose-Einstein condensates using reinforcement learning
• Quantum Simulations with Bilayer 2D Bose Gases in Multiple-RF-dressed Potentials
• Review and experimental benchmarking of machine learning algorithms for efficient optimization of cold atom experiments
• Combining machine learning with physics: A framework for tracking and sorting multiple dark solitons
• Inspiration from machine learning on example of optimization of the Bose-Einstein condensate of thulium atoms in a 1064-nm trap
• Multimode feedback cooling of the collective modes of a Bose-Einstein condensate
• Efficient production of sodium Bose-Einstein condensates in a hybrid trap