Bayesian Model Comparison of Basis Function Systems for Particle-Reflection-Distribution Data

Roland Preuss, Udo von Toussaint

2026 · Zenodo · Zenodo · 10.5281/zenodo.20028948

Abstract

Numerical simulations for particles impinging on the first wall of fusion devices yield angle- and energy-distributions of reflected particles. Common practice is to summarize this data in extensive tables for input to plasma wall interaction codes. Our approach is to summarize the multivariate distributions by a suitable basis function system and reduce the amount of data to be stored to the coefficients of the function system under consideration. In this paper we employ MCMC within a Bayesian model comparison for calculating the evidence in order to determine which of a selection of hemispherical basis function systems performs best.

BibTeX

@article{https://doi.org/10.5281/zenodo.20028948,

  doi = {10.5281/ZENODO.20028948},

  url = {https://zenodo.org/doi/10.5281/zenodo.20028948},

  author = {Preuss, Roland and von Toussaint, Udo},

  title = {Bayesian Model Comparison of Basis Function Systems for Particle-Reflection-Distribution Data},

  publisher = {Zenodo},

  year = {2026},

  copyright = {Creative Commons Attribution 4.0 International}
}