A GLOBAL FUNCTION OPTIMIZATION METHOD BASED ON BRANCHING SEARCH
Changqin Liu · Chinese Journal of Computers · 1997
This paper considers global optimization problem, proposes a simple global search method, branching search. The basic algorithm is like binary search.The search region of the objective function be partitioned into two subregions of equal size (the case of multi-subregion may be considered). Sampling function values from each subregion, the subregion containing the best of the known values is retained as the promising subregion and the other is excluded, then, the retained subregion is further partitioned. A group of domain-independent branching search algorithms are proposed, though their performances are investigated in the context of function optimization problems. The theoretical and empirical studies in this paper demonstrate that this embarrassingly simple method is practical effective.