Gaussian Process Regression for Equilibrium Reconstruction of DIII-D Plasmas

Gaussian Process Regression is a Bayesian method for inferring profiles based on input data. The technique is increasing in popularity in the fusion community for its many advantages over traditional fitting techniques. For kinetic EFIT reconstructions on DIII-D, data from magnetic probes and flux loops, motional Stark effect diagnostics, and plasma density and temperature profile measurements from Thomson scattering and charge exchange recombination diagnostics are critical for obtaining accurate reconstructions. Each of these data sources contain unique challenges for proper analysis that can be aided by using the GPR techniques within the magnetic equilibrium reconstruction process. Here, we review our progress on applying these GPR techniques to experimental data within the EFIT workflow, and contrast with similar efforts within the fusion community.