A Unified Method for Inference of Tokamak Equilibria and Validation of Force-Balance Models Based on Bayesian Analysis

TL;DR

This paper introduces BEAST, a Bayesian analysis method that infers tokamak plasma equilibria and validates force-balance models by quantifying their consistency using evidence calculations, enhancing understanding of plasma behavior.

Contribution

The paper presents a unified Bayesian framework with observation splitting for inferring plasma equilibria and assessing force-balance models, implemented in the BEAST code.

Findings

Initial results on MAST discharges demonstrate the method's effectiveness.

Evidence calculations help compare the accuracy of different force-balance models.

The approach enables quantification of model-data consistency in tokamak experiments.

Abstract

A new method, based on Bayesian analysis, is presented which unifies the inference of plasma equilibria parameters in a Tokamak with the ability to quantify differences between inferred equilibria and Grad-Shafranov force-balance solutions. At the heart of this technique is the new method of observation splitting, which allows multiple forward models to be associated with a single diagnostic observation. This new idea subsequently provides a means by which the the space of GS solutions can be efficiently characterised via a prior distribution. Moreover, by folding force-balance directly into one set of forward models and utilising simple Biot-Savart responses in another, the Bayesian inference of the plasma parameters itself produces an evidence (a normalisation constant of the inferred posterior distribution) which is sensitive to the relative consistency between both sets of models. This evidence can then be used to help determine the relative accuracy of the tested force-balance model across several discharges/times. These ideas have been implemented in a code called BEAST (Bayesian Equilibrium Analysis and Simulation Tool), which uses a special implementation of Skilling's nested sampling algorithm [Skilling, Bayesian Analysis 1(4), 833--859 (2006)] to perform sampling and evidence calculations on high-dimensional, non-Gaussian posteriors. Initial BEAST equilibrium inference results are presented for two high-performance MAST discharges.