Bayesian Analysis of Conventional and Ultrafast Spectroscopy Data for Investigating Detachment in the MAST-Upgrade Super-X

TL;DR

This paper introduces a new adaptive Bayesian inference method for analyzing spectroscopic data in fusion experiments, significantly reducing analysis time while maintaining accuracy in investigating divertor detachment physics.

Contribution

The paper presents a novel adaptive Bayesian analysis technique that enhances computational efficiency and accuracy in spectroscopic data interpretation for fusion divertor studies.

Findings

Achieved up to 1000x reduction in analysis time.

Demonstrated strong agreement with previous analysis methods.

Validated the approach with synthetic and real data.

Abstract

This paper presents the application, testing and first results of a new adaptive Bayesian inference analysis which utilises conventional and ultrafast spectroscopic measurements made in the divertor chamber to investigate the divertor physics during detachment. Validation of this software is performed prior and during analyses of results, demonstrated by compelling reproductions of ideal test cases and synthetic spectroscopic measurements. Application on real diagnostic data shows strong agreement with results from previous analysis methods. We identify unprecedented success in significant advances in time and computational efficiencies. We demonstrate a $\lesssim$1000$\times$ reduction in analysis time for spectroscopic measurements from simulated and real Super-X configurations, with the analysis technique presented in this report completing in <3 minutes. Analysis of synthetic and real diagnostic measurements identifies detachment physics in agreement with previous literature.