A Framework for Coordinating Parallel Branch and Bound Algorithms

  • Authors:
  • Andries Stam

  • Affiliations:
  • -

  • Venue:
  • COORDINATION '02 Proceedings of the 5th International Conference on Coordination Models and Languages
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

Branch and bound algorithms can be used for a variety of optimization problems. They are known to be very well suited for parallelization, which is a useful property to investigate in the light of coordination. This paper presents a general framework for parallel branch and bound algorithms, implemented using the coordination language MANIFOLD. Within this framework, the code for the optimization problem is separated from the generic branch and bound algorithm and the coordination strategy is separated from the coordinated components. The framework is an example of how the use of a coordination language can lead to a clean, comprehensible and flexible software architecture for complex parallel systems.