Authors:
Sven Löffler
;
Ilja Becker
and
Petra Hofstedt
Affiliation:
Brandenburg University of Technology Cottbus, Senftenberg, Konrad-Wachsmann-Allee 5, 03046 Cottbus, Germany
Keyword(s):
Constraint Satisfaction, Planning and Scheduling, Hybrid Intelligent Systems, Traveling Salesman Problem, Greedy Search, Clustering.
Abstract:
Constraint optimization problems offer a means to obtain at a global solution for a given problem. At the same time the promise of finding a global solution, often this comes at the cost of significant time and computational resources. Greedy search and cluster identification methods represent two alternative approaches, which can lead fast to local optima. In this paper, we explore the advantages of incorporating greedy search and clustering techniques into constraint optimization methods without forsaking the pursuit of a global solution. The global search process is designed to consider clusters and initially behave akin to a greedy search. This dual strategy aims to achieve two key objectives: firstly, it accelerates the attainment of an initial solution, and secondly, it ensures that this solution possesses a high level of optimality. This guarantee is generally elusive for constraint optimization problems, where solvers may struggle to find a solution, or find one of adequate q
uality in acaptable time. Our approach is an improvement of the general Bunch-and-Bound approach in constraint programming. Finally, we validate our findings using the Traveling Salesman Problem as a case study.
(More)