NN-OpInf

NN-OpInf is a PyTorch based approach to operator inference that utilizes composable, structure preserving neural networks to represent nonlinear operators. Operator inference, as made popular by Peherstorfer, Willcox, and co-authors, is an approach for inferring low-dimensional systems from data. Classically, OpInf infers polynomial models for the system dynamics. Numerous systems of interest, however, do not admit such polynomial structure. NN-OpInf addresses this challenge by parameterizing operators with neural networks.

Getting Started

Install the package and train your first model.

API Reference

Browse operators, models, steppers, and training utilities.

Examples

Follow end-to-end workflows and compare ROM outputs.