Bayesian Inference of Equilibrium Magnetic Field Geometry on the MAST Experiment

Abstract

Inference of plasma equilibrium geometry in tokamak fusion plasmas constitutes a challenging inference problem, given intrinsic difficulties surrounding the making of direct measurements in such physical systems. Traditionally, this problem has been handled by codes that attempt to reconcile solutions of the Grad-Shafranov (GS) equation with external magnetic diagnostics. Due to this inference being an intrinsically ill-posed problem, these codes suffer from numerical difficulties that require experiment-specific algorithms to handle. Here, we present a method to directly infer plasma equilibrium structure based on Bayesian analysis, which does not require solving the GS equation nor the use of any experiment-specific numerics.