Bayesian modeling of charge exchange recombination spectroscopy in fusion plasmas

A diagnostic simulation code for charge exchange recombination spectroscopy has been developed, aiming at performing a Bayesian inference of the velocity distribution function from the measured charge exchange emission spectrum. A code has been written to reproduce the emission spectrum based on the input velocity distribution function and other experimental information, such as magnetic equilibrium, neutral beam density distribution, line of sight geometry, and others. A set of demonstrations of Bayesian inference has been accomplished for the setting of a simple problem: the parametric inference of an ion temperature profile. An uncertainty estimation based on a Bayesian model evidence is presented by finding the optimum hyperparameter.