Space-Mapping-Based Modeling Utilizing Parameter Extraction with Variable Weight Coefficients and a Data Base

A new space-mapping-based surrogate modeling methodology is presented. We assume that certain fine model data, the so-called base set, is available in the region of interest. To evaluate the surrogate, we perform parameter extraction with weighting coefficients dependent on the distance between the point of interest and base points. This has advantages over standard SM modeling: (1) it can handle any base set, (2) the accuracy of the surrogate improves while the number of points in the base set grows even if the flexibility of the SM surrogate remains unchanged, (3) the model evaluation cost is roughly independent of the size of the base set. Examples confirm theoretical considerations and demonstrate robustness