Learning data-driven discretizations for partial differential equations
Top Cited Papers
Open Access
- 16 July 2019
- journal article
- research article
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 116 (31) , 15344-15349
Abstract
The numerical solution of partial differential equations (PDEs) is challenging because of the need to resolve spatiotemporal features over wide length- and timescales. Often, it is computationally intractable to resolve the finest features in the solution. The only recourse is to use approximate coarse-grained representations, which aim to accurately represent long-wavelength dynamics while properly accounting for unresolved small-scale physics. Deriving such coarse-grained equations is notoriously difficult and often ad hoc. Here we introduce data-driven discretization, a method for learning optimized approximations to PDEs based on actual solutions to the known underlying equations. Our approach uses neural networks to estimate spatial derivatives, which are optimized end to end to best satisfy the equations on a low-resolution grid. The resulting numerical methods are remarkably accurate, allowing us to integrate in time a collection of nonlinear equations in 1 spatial dimension at resolutions 4× to 8× coarser than is possible with standard finite-difference methods.Keywords
All Related Versions
This publication has 22 references indexed in Scilit:
- Equation-Free Multiscale Computation: Algorithms and ApplicationsAnnual Review of Physical Chemistry, 2009
- Holistic discretization ensures fidelity to Burgers' equationApplied Numerical Mathematics, 2001
- Identification of distributed parameter systems: A neural net based approachComputers & Chemical Engineering, 1998
- Essentially non-oscillatory and weighted essentially non-oscillatory schemes for hyperbolic conservation laws1998
- Approximate inertial manifolds for the Kuramoto-Sivashinsky equation: Analysis and computationsPhysica D: Nonlinear Phenomena, 1990
- On approximate Inertial Manifolds to the Navier-Stokes equationsJournal of Mathematical Analysis and Applications, 1990
- Approximate inertial manifolds for reaction-diffusion equations in high space dimensionJournal of Dynamics and Differential Equations, 1989
- Inertial manifolds for nonlinear evolutionary equationsJournal of Differential Equations, 1988
- Uniformly high order accurate essentially non-oscillatory schemes, IIIJournal of Computational Physics, 1987
- Interaction of "Solitons" in a Collisionless Plasma and the Recurrence of Initial StatesPhysical Review Letters, 1965