Fast and scalable computation of shape-morphing nonlinear solutions with application to evolutional neural networks

DOI10.1016/J.JCP.2023.112649MaRDI QIDQ6187618zbMATH OpenOpenAlexFDO

Authors

Publication date 31 January 2024

Published in Journal of Computational Physics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2207.13828

zbMATH Keywords

neural networkspartial differential equationsspectral methodsreduced-order modeling

Mathematics Subject Classification ID

Partial differential equations of mathematical physics and other areas of application (35Qxx) Numerical methods for partial differential equations, initial value and time-dependent initial-boundary value problems (65Mxx) Numerical analysis in abstract spaces (65Jxx)

Abstract: We develop fast and scalable methods for computing reduced-order nonlinear solutions (RONS). RONS was recently proposed as a framework for reduced-order modeling of time-dependent partial differential equations (PDEs), where the modes depend nonlinearly on a set of time-varying parameters. RONS uses a set of ordinary differential equations (ODEs) for the parameters to optimally evolve the shape of the modes to adapt to the PDE's solution. This method has already proven extremely effective in tackling challenging problems such as advection-dominated flows and high-dimensional PDEs. However, as the number of parameters grow, integrating the RONS equation and even its formation become computationally prohibitive. Here, we develop three separate methods to address these computational bottlenecks: symbolic RONS, collocation RONS and regularized RONS. We demonstrate the efficacy of these methods on two examples: Fokker-Planck equation in high dimensions and the Kuramoto-Sivashinsky equation. In both cases, we observe that the proposed methods lead to several orders of magnitude in speedup and accuracy. Our proposed methods extend the applicability of RONS beyond reduced-order modeling by making it possible to use RONS for accurate numerical solution of linear and nonlinear PDEs. Finally, as a special case of RONS, we discuss its application to problems where the PDE's solution is approximated by a neural network, with the time-dependent parameters being the weights and biases of the network. The RONS equations dictate the optimal evolution of the network's parameters without requiring any training.

Recommendations

• Neural Control of Parametric Solutions for High-Dimensional Evolution PDEs
• Evolution of nonlinear reduced-order solutions for PDEs with conserved quantities
• A comprehensive deep learning-based approach to reduced order modeling of nonlinear time-dependent parametrized PDEs
• Positional Embeddings for Solving PDEs with Evolutional Deep Neural Networks

Cites work

• A family of embedded Runge-Kutta formulae
• A method for representing periodic functions and enforcing exactly periodic boundary conditions with deep neural networks
• A minimization principle for the description of modes associated with finite-time instabilities
• A regularization parameter for nonsmooth Tikhonov regularization
• A survey of projection-based model reduction methods for parametric dynamical systems
• An adjoint-based approach for finding invariant solutions of Navier-Stokes equations
• Approximation by superpositions of a sigmoidal function
• Back in the Saddle Again: A Computer Assisted Study of the Kuramoto–Sivashinsky Equation
• Chaotic behavior of Newton's method
• Computational Methods for Inverse Problems
• Computing sensitivities in evolutionary systems: a real-time reduced order modeling strategy
• Conservative model reduction for finite-volume models
• DGM: a deep learning algorithm for solving partial differential equations
• Evolution of nonlinear reduced-order solutions for PDEs with conserved quantities
• Focusing of unidirectional wave groups on deep water: an approximate nonlinear Schrödinger equation-based model
• Fundamental limitations of ad hoc linear and quadratic multi-level regression models for physical systems
• Generalized Cross-Validation as a Method for Choosing a Good Ridge Parameter
• High order accurate vortex methods with explicit velocity kernels
• Model Reduction for Flow Analysis and Control
• Numerical linear algebra and optimization
• Numerical matrix analysis. Linear systems and least squares
• Numerical methods for Hamiltonian PDEs
• On the state space geometry of the Kuramoto-Sivashinsky flow in a periodic domain
• Reduced-order description of transient instabilities and computation of finite-time Lyapunov exponents
• Regularization independent of the noise level: an analysis of quasi-optimality
• Solving high-dimensional partial differential equations using deep learning
• Spectral/hp Element Methods for Computational Fluid Dynamics
• Symplectic model reduction of Hamiltonian systems
• The Kuramoto-Sivashinsky equation: a bridge between PDE's and dynamical systems
• The MATLAB ODE Suite
• The \(N\)-vortex problem. Analytical techniques
• The nonlinear evolution and approximate scaling of directionally spread wave groups on deep water
• The quasi-optimality criterion for classical inverse problems
• Tikhonov Regularization and Total Least Squares
• Tikhonov regularization and the L-curve for large discrete ill-posed problems
• Use of the regularization method in non-linear problems

Cited in

(1)

• Neural Control of Parametric Solutions for High-Dimensional Evolution PDEs

This page was built for publication: Fast and scalable computation of shape-morphing nonlinear solutions with application to evolutional neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6187618)