NeuralOperator
NeuralOperator is the reference implementation library for neural operators, providing Fourier Neural Operator (FNO), DeepONet and other mainstream architectures with training pipelines under the MIT license (free for commercial use and modification). It standardizes learning 'function-to-function' maps and is the core tool for building physical-field surrogates (replacing CFD solves in milliseconds).
Use cases: Training fast surrogates for flow / temperature fields; replacing slow numerical solves for design iteration.
Tags: Neural OperatorFNODeepONetMIT