Creation and manipulation of quantized vortices in Bose-Einstein condensates using reinforcement learning

Abstract

We apply the technique of reinforcement learning to the control of nonlinear matter waves. In this method, an agent controls the position, strength, and shape of an external Gaussian potential to create and manipulate quantized vortices in a Bose-Einstein condensate (BEC) trapped in a harmonic potential. The density and velocity distributions of the BEC at each moment obtained by the Gross-Pitaevskii evolution are directly input into a convolutional neural network to determine the next action of the agent. We demonstrate that a stationary single-vortex state can be produced in a two-dimensional system, and a stationary vortex-ring state can be produced in a three-dimensional system.