Fitting Non-Maxwellian Thomson Scattering Spectra using Superpositions of Maxwellians

Thomson Scattering is a powerful diagnostic for probing plasma parameters. Current open-source PlasmaPy algorithms for fitting Thomson Scattering spectra assume a Maxwellian velocity distribution. However, many plasma phenomena, such as collisionless shocks, occur under non-equilibrium conditions where velocity distribution functions (VDFs) are non-Maxwellian. Standard fitting methods, such as differential evolution (DE), struggle to return accurate fits due to the many free parameters required to describe these VDFs. We introduce a method for fitting Thomson Scattering spectra based on automatic differentiation, a gradient-descent-based fitting approach commonly used in neural network training, which efficiently optimizes larger parameter spaces. This method is effective in fitting both Maxwellian and non-Maxwellian VDFs [1]. For broader applicability, we extend this method to fit VDFs using a superposition of Maxwellians, a more robust and generalizable approach. To assess the efficacy of this technique, we apply it to several types of synthetic, non-Maxwellian spectra.

This work was supported by the DOE NNSA under Award No. DE-NA4033, the DOE FES

under Award No. DE-SC0024566, and NASA under Grant No. 80NSSC19K0493.