Neural-network QoI surrogate#
Method. A neural-network surrogate approximates quantities of interest as functions of model parameters. The implementation can constrain the model’s Lipschitz constant.
Example. The example constructs a PyTorch QoI surrogate, including a configuration with explicit Lipschitz constraints.
Install optional support with pip install 'romtools[WithTorch]'.
For algorithm details, settings, and full source listings, see neural network surrogate — detailed notes.