Using neural networks to accelerate the solution of the Boltzmann equation

Abstract

One of the biggest challenges for simulating the Boltzmann equation is the evaluation of fivefold collision integral. Given the recent successes of deep learning and the availability of efficient tools, it is an obvious idea to try to substitute the evaluation of the collision operator by the evaluation of a neural network. However, it is unlcear whether this preserves key properties of the Boltzmann equation, such as conservation, invariances, the H-theorem, and fluid-dynamic limits. In this paper, we present an approach that guarantees the conservation properties and the correct fluid dynamic limit at leading order. The concept originates from a recently developed scientific machine learning strategy which has been named "universal differential equations". It proposes a hybridization that fuses the deep physical insights from classical Boltzmann modeling and the desirable computational efficiency from neural network surrogates. The construction of the method and the training strategy are demonstrated in detail. We conduct an asymptotic analysis and illustrate its multi-scale applicability. The numerical algorithm for solving the neural network-enhanced Boltzmann equation is presented as well. Several numerical test cases are investigated. The results of numerical experiments show that the time-series modeling strategy enjoys the training efficiency on this supervised learning task.

20 pages, 11 figures

References in corpus (4)

• Physics-Constrained Deep Learning for High-dimensional Surrogate Modeling and Uncertainty Quantification without Labeled Data
• DiffEqFlux.jl - A Julia Library for Neural Differential Equations
• A stochastic kinetic scheme for multi-scale plasma transport with uncertainty quantification
• A Well-Balanced Unified Gas-Kinetic Scheme for Multiscale Flow Transport Under Gravitational Field

Cited by in corpus (11)

• 2022 Review of Data-Driven Plasma Science
• Combining physics-based and data-driven models: advancing the frontiers of research with Scientific Machine Learning
• Machine learning moment closure models for the radiative transfer equation I: directly learning a gradient based closure
• RelaxNet: A structure-preserving neural network to approximate the Boltzmann collision operator
• opPINN: Physics-Informed Neural Network with operator learning to approximate solutions to the Fokker-Planck-Landau equation
• Predicting continuum breakdown with deep neural networks
• Solving the QCD effective kinetic theory with neural networks
• Solving continuum and rarefied flows using differentiable programming
• FourierSpecNet: Neural Collision Operator Approximation Inspired by the Fourier Spectral Method for Solving the Boltzmann Equation
• Learning invariance preserving moment closure model for Boltzmann-BGK equation
• Learning collision operators from plasma phase space data using differentiable simulators