🚀 HyperNOs: Hyperparameter Optimization for Neural Operators

[MaxGhi8]

Hi DeepXDE community,

I wanted to share an open-source project I’ve been working on: HyperNOs, a library focused on hyperparameter optimization for neural operators, now integrating examples from DeepXDE. I implemented several hyperparameter optimization ideas and adapted them to work with DeepXDE architectures, including practical examples and end-to-end usage tutorials. In particular, there are examples for DeepONet, POD-DeepONet, MIOnet, POD-MIOnet architectures.

🔗 Project links

Code: https://github.com/MaxGhi8/HyperNOs

Paper: https://link.springer.com/article/10.1007/s40574-025-00516-0

The library is currently maintained mainly by me, so there’s certainly room for improvement. I would love to hear any feedback, suggestions, or observations, and I’m happy to explore collaborations with anyone interested in experimenting with neural operators in DeepXDE.

—
Massimiliano Ghiotto

[lululxvi]

Sounds great. Maybe you can add some examples in DeepXDE?