Bayesian inference using JET’s microwave diagnostic system

Abstract

At the JET tokamak, three electron cyclotron emission (ECE) diagnostics (two Martin-Puplett interferometers and a heterodyne radiometer) and a reflectometer form the basic microwave diagnostic system. The standard analysis approaches deduce electron density and temperature profiles independently of each diagnostic measurement. Via the Bayesian framework Minerva, the microwave diagnostic system is modelled, and electron temperature and density profiles are inferred jointly for an Ohmic JET plasma. Furthermore, profile length-scales for different plasma domains, a magnetic field correction, distinct reflection properties of the high-field side and low-field side walls, and radiometer sensitivities are estimated together. This inference scheme can use one of two models to predict broadband ECE spectra; one is less accurate but fast for a single prediction, and a more accurate model, relying on the ray-tracer SPECE parallelised via web services. The faster model allows the investigation of correlations between parameters and the execution of a numerical marginalisation, i.e. an uncertainty propagation.

Machine learning and Bayesian inference in nuclear fusion research: an overview

“…Other applications. Numerous detailed models of ECE diagnostic measurements can be found in the literature: in [18], the authors infer temperature and density profiles simultaneously from multiple diagnostics at ASDEX Upgrade; in [20], the authors employ a Bayesian model of three ECE diagnostics, including ray-tracing, Martin-Puplett IFs, a heterodyne radiometer and a reflectometer at JET to infer temperature and density profiles, discussing also local posterior correlations between the profiles; in [21], the authors describe the model of a general radiometer calibration that can be used to model the radiation temperature of the ECE of W7-X plasmas using a 3D ray tracing model; in [22], the authors apply Bayesian inference to investigate higher harmonics of the ECE obtained with a Michelson IF at W7-X. Langenberg et al [23] show a Bayesian forward model of an x-ray imaging crystal spectrometer system, which allows one to infer kinetic profiles and study impurity transport at W7-X.…”

A Gaussian process guide for signal regression in magnetic fusion

“…In [3,[23][24][25] GPRs are used to fit experimental core temperature and density profiles, to infer second order derivative information from these profiles, and to propagate uncertainties, while in [26] GPs are explored for use in quasi-coherent noise suppression. At JET, GPs have been used to infer electron cyclotron emission spectra [27], and high resolution Thomson scattering and far infrared interferometer data in [28], while work in [29] is focused on using GPRs to quantify edge plasma evolution from experimental data. A general overview of Bayesian inference in fusion can be found in [10,30].…”

From minimum-viable-products to full models: a step-wise development of diagnostic forward models in support of design, analysis and modelling on the ST40 tokamak

“…Given the wealth of tools previously developed by the fusion community for this purpose (see e.g. [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17]) it is vital to make the most of this wealth of knowledge, avoid reinventing the wheel, and contribute to these widely used, developed, and maintained projects. This of course applies to those tools that are either open-source or for which collaboration frameworks are already in place, which includes e.g.…”