← Papers
Operators

Neural Operator

Neural operators: learning maps between function spaces

Paper title Neural Operator: Learning Maps Between Function Spaces

Kovachki et al. · General PDEs · Public: code

Paper arXiv 2108.08481PDF

A network that maps functions to functions: each layer is a learned integral operator followed by a pointwise nonlinearity, so the same weights apply to any discretisation of the input. The integral is computed in four ways: on a graph (GNO), on multipole graphs, with a low-rank kernel, and in Fourier space (FNO).

Published
JMLR

Share on LinkedIn, X or Bluesky, or or .

Know related work that belongs here?

Something wrong on this page?

More papers

All papers
  1. OperatorsTurtLES — A 3D turbulent-wake benchmark where operators fail over long rollouts
  2. OperatorsFunction-space certification — Domain-wide guarantees for neural operators, not only at grid points
  3. OperatorsL-SplineFNO — Learnable splines take the FNO to unstructured meshes