Measuring 3D or 4D fast-ion distribution function relies on using prior information in some form due to the ill-conditioned nature of the underlying inverse problem. Here we present a method to encode the correlations between phase-space elements caused by the neoclassical transport due to Coulomb collisions as physics-informed prior information which we compute using the orbit-following code ASCOT. We reconstruct the fast-ion distribution function from neutral beam injection (NBI) in a 4D phase-space using neoclassical anisotropic slowing-down distributions calculated with ASCOT for a JET equilibrium as basis functions. Detailed data from the NBI geometry was used to compute basis functions at full, half, and one-third injection energies. The fast-ion distribution at each injection energy was divided into 10 separate basis functions based on flux surfaces of the ionized neutrals. Reconstructions based on synthetic data were computed by solving for basis function coefficients with Tikhonov regularization. Reconstructions of the phase-space distribution from one NBI match well with the true solution, but an addition of another NBI reduced the quality of the reconstructions significantly.
This paper presents a method to incorporate neoclassical transport physics as prior information in reconstructing the 3D or 4D fast-ion distribution function from neutral beam injection data. The authors use the ASCOT code to compute anisotropic slowing-down distributions, which are then used as basis functions in the reconstruction process. This approach helps address the ill-conditioned nature of the underlying inverse problem and improves the accuracy of the reconstructions.