Abul K.M Fahimuddin*, Markus Krosche, Hermann G. Matthies
Institute of Scientific Computing, Technical University of Braunschweig, Hans-Sommer-Strasse 65, Braunschweig, 38106, Germany
ABSTRACT
We describe an abstract interface for surrogate-model based optimization in the PLATON framework. The abstract interface encapsulates data structure and ensures easy integration of various third party implementation of surrogate methods. Kriging, a statistical method originating from the field of geostatistics, is used here to replace the expensive simulation code. The paper concentrates on the software development issues related to the integration of surrogate modeling in PLATON. In particular, PLATON's flexible structure for easy incorporation of third-party implementations is portrayed. At the end, a numerical experiment is provided for justifying the realization of the PLATON-Surrogate framework.
Keywords:
Component-based software engineering; Numerical optimization; Abstract interface; PLATON; Surrogate modeling; Kriging