Non-Intrusive MOR using Neural Networks

[dynamic-queries]

The idea is simple enough :

• This learns the forcing function in the reduced space. (POD space for instance)
• System is allowed to evolve in the reduced space and projected back.

https://www.sciencedirect.com/science/article/pii/S0021999118301190

If I understand correctly, Lift and Learn from #3 and Operator inference (for the lack of a better term) could be bundled with this in a single module, with NN being a generic version of it. Am I missing something?