Machine Learning Enhanced Multiscale Methods

Doctoral Researcher
Name Role at KCDS
KCDS Fellow
KCDS Supervisors
Name Role at KCDS
MATH Supervisor
1 additional person visible within KIT only.

Abstract

Physical processes often involve effects on multiple scales. Examples for this are composite materials or groundwater flow in heterogeneous media. The underlying differential equations, these different scales pose a problem due to their requirements on the resolution of the method. Specialized methods solve these problems in a parallel manner by exploiting locality properties. In my project, the goal is to develop such methods and enhance them with machine learning techniques to speed up the computations or improve results.