Relative Expected Improvement in Kriging Based Optimization

TL;DR

This paper introduces an extended Expected Improvement criterion for Kriging models, including those with derivatives, aimed at optimizing computationally expensive functions like CFD problems.

Contribution

It presents a novel extension of the Expected Improvement criterion tailored for complex Kriging models with derivatives, applicable to expensive optimization tasks.

Findings

Enhanced optimization efficiency for CFD problems

Applicable to other fields with expensive function evaluations

Extension of EI criterion for derivative-based Kriging models

Abstract

We propose an extension of the concept of Expected Improvement criterion commonly used in Kriging based optimization. We extend it for more complex Kriging models, e.g. models using derivatives. The target field of application are CFD problems, where objective function are extremely expensive to evaluate, but the theory can be also used in other fields.