Model reduction, centering, and the Karhunen-Loeve expansion

Abstract

We propose a new computationally efficient modeling method that captures a given translation symmetry in a system. To obtain a low order approximate system of ODEs, prior to performing a Karhunen Loeve expansion, we process the available data set using a "centering" procedure. This approach has been shown to be efficient in nonlinear scalar wave equations.

Files

Additional details