emvalbuena / romtime

Unsteady Finite Element Reduced Order Model for Time Moving Domains
1 stars 1 forks source link

PERF : MDEIM for nonlinear term is sooooo slow #38

Closed emvalbuena closed 2 years ago

emvalbuena commented 3 years ago

I wonder what is going on there ...

It's true that there are three nested loops:

Perhaps I could combine the latter two and use multiprocessing.


That being said, it has more or less acceptable times (30 mins por 10 parameters for each ROM/DEIM/MDEIM) to get some results and pour them in a document.

emvalbuena commented 2 years ago

It was slow because the basis was orthogonal and the trilinear form well ... linear haha.

Hence the SVD could not compress.