I have one set of variables A which are the numerical solution of one group of PDEs, say the atmospheric model, and another set of variables B coming from the other PDE, for example, X, X’, X” w.r.t a parameter theta.
So how can I use some fancy techniques to build an ML from A to B and study its mechanisms? Or since the input A comes from PDEs, can I do some feature engineering using these PDEs?
I knew that from Stanford’s paper (https://advances.sciencemag.org/content/5/12/eaay6946) that the computational fluid dynamics equations follow a similar form with RCN. But how can I apply it to our field?