Machine learning moment closure models for the radiative transfer equation I: directly learning a gradient based closure
Numerical Analysis
2022-02-09 v2 Machine Learning
Numerical Analysis
Computational Physics
Abstract
In this paper, we take a data-driven approach and apply machine learning to the moment closure problem for radiative transfer equation in slab geometry. Instead of learning the unclosed high order moment, we propose to directly learn the gradient of the high order moment using neural networks. This new approach is consistent with the exact closure we derive for the free streaming limit and also provides a natural output normalization. A variety of benchmark tests, including the variable scattering problem, the Gaussian source problem with both periodic and reflecting boundaries, and the two-material problem, show both good accuracy and generalizability of our machine learning closure model.
Cite
@article{arxiv.2105.05690,
title = {Machine learning moment closure models for the radiative transfer equation I: directly learning a gradient based closure},
author = {Juntao Huang and Yingda Cheng and Andrew J. Christlieb and Luke F. Roberts},
journal= {arXiv preprint arXiv:2105.05690},
year = {2022}
}