Efficient initialization of a pseudo-Bayesian global optimization hybrid

The authors present an initialization procedure for a pseudo-Bayesian global optimization hybrid which allows initialization by a single point. The search space is modeled by a constrained electrostatic field. The resulting field is described by Laplace´s equation in n-dimensions which can be solved numerically to approximate critical function parameters in the initialization process. Results are given for the performance of the new algorithm on a series of standard test functions and compared to the performance of Bayesian as well as other classes of global optimization algorithms