Neural Operators for Solving PDEs and Inverse Design
- Creators
- Anandkumar, Anima
Abstract
Deep learning surrogate models have shown promise in modeling complex physical phenomena such as photonics, fluid flows, molecular dynamics and material properties. However, standard neural networks assume finite-dimensional inputs and outputs, and hence, cannot withstand a change in resolution or discretization between training and testing. We introduce Fourier neural operators that can learn operators, which are mappings between infinite dimensional spaces. They are discretization-invariant and can generalize beyond the discretization or resolution of training data. They can efficiently solve partial differential equations (PDEs) on general geometries. We consider a variety of PDEs for both forward modeling and inverse design problems, as well as show practical gains in the lithography domain.
Additional Information
© 2023 Copyright is held by the owner/author(s).
Attached Files
Published - 3569052.3578911.pdf
Files
Name | Size | Download all |
---|---|---|
md5:48104c8791550947831a81f1552b170c
|
710.9 kB | Preview Download |
Additional details
- Eprint ID
- 120427
- DOI
- 10.1145/3569052.3578911
- Resolver ID
- CaltechAUTHORS:20230327-853994000.2
- Created
-
2023-03-30Created from EPrint's datestamp field
- Updated
-
2023-06-21Created from EPrint's last_modified field