Solving the discretised neutron diffusion equations using neural networks

DOI10.1002/NME.7321MaRDI QIDQ6082621zbMATH OpenFDO

Authors Claire Heaney, Bo-Yang Chen, A. G. Buchan, C. C. Pain

Publication date 30 November 2023

Published in International Journal for Numerical Methods in Engineering (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/2301.09939

zbMATH Keywords

convolutional neural networkfinite difference methodfinite volume methodmultigrid solverneutron diffusion equationnumerical solution of partial differential equationsreactor physicsU-net

Mathematics Subject Classification ID

Computational issues in computer and robotic vision (65D19) Iterative numerical methods for linear systems (65F10) Finite difference methods for boundary value problems involving PDEs (65N06) Finite volume methods for boundary value problems involving PDEs (65N08) Finite element, Rayleigh-Ritz and Galerkin methods for boundary value problems involving PDEs (65N30) Multigrid methods; domain decomposition for boundary value problems involving PDEs (65N55) Learning and adaptive systems in artificial intelligence (68T05) Nuclear reactor theory; neutron transport (82D75)

Abstract: This paper presents a new approach which uses the tools within Artificial Intelligence (AI) software libraries as an alternative way of solving partial differential equations (PDEs) that have been discretised using standard numerical methods. In particular, we describe how to represent numerical discretisations arising from the finite volume and finite element methods by pre-determining the weights of convolutional layers within a neural network. As the weights are defined by the discretisation scheme, no training of the network is required and the solutions obtained are identical (accounting for solver tolerances) to those obtained with standard codes often written in Fortran or C++. We also explain how to implement the Jacobi method and a multigrid solver using the functions available in AI libraries. For the latter, we use a U-Net architecture which is able to represent a sawtooth multigrid method. A benefit of using AI libraries in this way is that one can exploit their power and their built-in technologies. For example, their executions are already optimised for different computer architectures, whether it be CPUs, GPUs or new-generation AI processors. In this article, we apply the proposed approach to eigenvalue problems in reactor physics where neutron transport is described by diffusion theory. For a fuel assembly benchmark, we demonstrate that the solution obtained from our new approach is the same (accounting for solver tolerances) as that obtained from the same discretisation coded in a standard way using Fortran. We then proceed to solve a reactor core benchmark using the new approach.

Recommendations

• Solving the discretised multiphase flow equations with interface capturing on structured grids using machine learning libraries
• Three ways to solve partial differential equations with neural networks — A review
• Solving parametric PDE problems with artificial neural networks
• Sweep-Net: an artificial neural network for radiation transport solves
• DeepXDE: a deep learning library for solving differential equations

Cites work

• A neural network multigrid solver for the Navier-Stokes equations
• A TensorFlow simulation framework for scientific computing of fluid flows on tensor processing units
• Assembly of finite element methods on graphics processors
• Deep UQ: learning deep neural network surrogate models for high dimensional uncertainty quantification
• Finite difference computing with exponential decay models
• Generation of large finite-element matrices on multiple graphics processors
• GPU-accelerated discontinuous Galerkin methods on hybrid meshes
• Hadamard inverses, square roots and products of almost semidefinite matrices
• scientific article; zbMATH DE number 48448 (Why is no real title available?)
• scientific article; zbMATH DE number 961607 (Why is no real title available?)
• MgNet: a unified framework of multigrid and convolutional neural network
• Opencl-based implementation of an unstructured edge-based finite element convection-diffusion solver on graphics hardware
• Optimized implementation of the lattice Boltzmann method on a graphics processing unit towards real-time fluid simulation

Cited in

(5)

• Sweep-Net: an artificial neural network for radiation transport solves
• Using neural networks to accelerate the solution of the Boltzmann equation
• Solving multiscale steady radiative transfer equation using neural networks with uniform stability
• Solving the discretised multiphase flow equations with interface capturing on structured grids using machine learning libraries
• NP4Buoyancy: an end-to-end differentiable neural-physics framework for buoyancy-driven flows

This page was built for publication: Solving the discretised neutron diffusion equations using neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6082621)