Joint Research Conference
June 24-26, 2014
His talk addresses the issue of kriging-based optimization for multi-fidelity deterministic computer experiments. Deterministic computer experiments are usually approximations of physical processes with tunable precision. For example, in finite element analysis, precision of computer experiments is controlled by mesh density. We propose an optimization algorithm to allocate experiment points and their corresponding fidelities sequentially. The algorithm balances between local refinement and global search and integrates both low- and high-accuracy simulations.