Machine learning techniques in magnetic levitation problems

Abstract

We present a method for calculating the stability region of a perfect diamagnet levitated in a magnetic field created by a circular current loop making use of the machine learning techniques. As an application we compute stability regions, points of stable equilibrium and stable oscillatory motions in two chip-based superconducting trap architectures used to levitate superconducting particles. Our procedure is an alternative to a full numerical scheme based on finite element methods which are expensive to implement for optimizing experimental parameters.