Quality assessment of coarse models and surrogates for space mapping optimization (Q1024093)

In space mapping, the objective function to be optimized is constructed from the response of the so called ``fine model''. It is also assumed that there is an alternative set of functions available, from the so called ``coarse model'', not so accurate but much faster to evaluate. These coarse and fine models can be used to develop an appropriate surrogate model that is almost as cheap as the coarse model and (locally at last) almost as accurate as the fine model. Estimation methods are proposed to assess the quality of surrogate/coarse model. These methods are based on convergence results for space mapping algorithms. Moreover, two engineering design optimization problems are presented to illustrate the proposed assessment technology.