Overview of the SciDAC EFIT-AI Machine-Learning (ML) and Artificial-Intelligence (AI) Enhanced Equilibrium Reconstruction Project

Equilibrium reconstruction is fundamental to tokamak research and operation. The goal of the SciDAC EFIT-AI project is to harness novel ML / AI algorithms to enhance equilibrium reconstruction for modeling and real-time applications. This includes development of a ML-enhanced Bayesian framework to automate and maximize information from measurements and Model-Order-Reduction (MOR)-based ML models to efficiently guide the search of solution vector. A device-independent portable core equilibrium solver has been created to ease adaptation of ML enhanced reconstruction algorithms. An EFIT database comprising of DIII-D magnetic, MSE, and kinetic reconstruction data is being constructed for developments of EFIT-MOR surrogate models to speed up the search of solution vector. Approaches to improve portability between the OpenMP and GPU EFIT versions are being explored on Linux GPU clusters and the new NERSC Perlmutter to create a performance-portable GPU implementation for further optimization. Other progress includes development of a Gaussian-Process Bayesian framework to improve processing of input data, and construction of a 3D perturbed equilibrium database for developments of 3D-MOR surrogate models.