The information approach to multiextremal optimization problems

Abstract

A survey of some practical new global search algorithms is presented. These algorithms are derived as optimal statistical decision functions within the framework of a stochastic model representing the function to be optimized as a sample of some random function. This statistical language allows the elaboration of an analytical technique for efficiently deriving simple and powerful algorithms using the a priori assumptions on the function behaviour.

Keywords:

• Global search
• multiextremal optimization
• decision function