A Concept of Effectively Global Search and a Global Adaptation Strategy for Local Search Heuristics
Manabu Hasegawa · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 2004
A concept of effectively global search is introduced and used for setting an adaptive stopping criterion for local search heuristics. By applying this criterion to the detection of equilibrium, a new adaptive cooling schedule is developed for the simulated annealing algorithm. To examine the function and optimization performance of the presented method, it is numerically tested on the Euclidean traveling salesman problem. The proposed strategy realizes an adaptive search to the global landscape structure of the cost function. As a result, the present adaptive cooling shows good ability in yielding better solutions stably than the conventional non-adaptive one.