B . Hnich and K . Stergiou
نویسندگان
چکیده
In this paper, we investigate a method to decompose the search space explored by a Constraint Programming algorithm to allow a parallel solution of a Constraint Satisfaction or Optimization Problem (CSP or COP). The static tree decomposition used is based on domain partitioning according to a value ordering heuristic and a discrepancybased search strategy. This work is a parallel adaptation of the Decomposition Based Search introduced by van Hoeve and Milano. DBS is a good candidate for parallel computing since it implies a decomposition of the problem into independent subproblems. The concepts are applied to the solution of Partial Latin Squares and can be generalized to Constraint Optimization Problems.
منابع مشابه
Boosting Partial Symmetry Breaking by Local Search
S. D. Prestwich, B. Hnich, H. Simonis, R. Rossi, and S. A. Tarim Cork Constraint Computation Centre, University College Cork, Ireland Faculty of Computer Science, Izmir University of Economics, Turkey Logistics, Decision and Information Sciences Group, Wageningen UR, the Netherlands Operations Management Division, Nottingham University Business School, Nottingham, UK [email protected], brah...
متن کاملScheduling internal audit activities: a stochastic combinatorial optimization problem
The problem of finding the optimal timing of audit activities within an organisation has been addressed by many researchers. We propose a stochastic programming formulation with Mixed Integer Linear Programming (MILP) and Constraint Programming (CP) certainty-equivalent models. In experiments neither approach dominates the other. However, the CP approach is orders of magnitude faster for large ...
متن کاملTemplate Design under Demand Uncertainty by Integer Linear Local Search
Production planning under uncertain demands leads to optimisation problems that are hard both to model and to solve. We describe an integer linear model for a template design problem under uncertainty, and investigate its solution by a general-purpose local search algorithm for integer linear programs. Several such algorithms have previously been proposed as tools for solving large combinatoria...
متن کاملSymmetry Breaking by Nonstationary Optimisation
We describe a new partial symmetry breaking method that can be used to break arbitrary variable/value symmetries in combination with depth first search, static value ordering and dynamic variable ordering. The main novelty of the method is a new dominance detection technique based on local search in the symmetry group. It has very low time and memory requirements, yet in preliminary experiments...
متن کامل