Parallel branch-and-bound is an optimization technique which renders more efficient the solution of some hard problems such as the puzzle of colored blocks and the traveling-salesman problem. In a functional language such as Sisal 2.0, it is difficult for the programmer to describe a parallel version of this technique due to the lack of imperative features in the language. In this paper, we propose a version of Sisal 2.0 extended with user-declared mutable global variables in order to enable Sisal programmers to apply the parallel branch-and-bound technique. In a simple example (the puzzle of colored blocks), we show that this approach yields better performance than either conventional functional programs or imperative programs. It is easy to see that the same strategy can be used to solve a number of hard problems such as the traveling-salesman problem.
Parallel Branch and Bound Algorithm with the Hierarchical Master-Worker Paradigm on the Grid
