Bayesian Inference of the 1D Electron Density Profile within the WEST Tokamak using Interferometry

The electron density profile is essential information for the control of tokamak plasmas to avoid instabilities. Interferometry is a key diagnostic for the Inference of the electron density profile although it does not provide enough information for a full reconstruction of the profile. Bayesian Inference is thus required to obtain the most probable density profile given the data available. The profile is modelled as a Gaussian process and various methods have been tested for handling the hyperparameters involved. These include Maximum A Posteriori of the hyperparameter posterior and the full Bayesian technique where the hyperparameters are marginalised out of the inferative posterior. These are compared and contrasted on both synthetic data and real data from the WEST tokamak.