Load Distribution in Combinatorial Optimizations

Abstract

This chapter reports the application of the GDE and the diffusion methods in combinatorial optimizations. A combinatorial optimization is the process of finding optimal or suboptimal solutions in a defined problem space. It arises from a large variety of applications in various areas in operational research and artificial intelligence. Examples are task planning and scheduling, capital investment, layout of VLSI chips, robot motion planning and game playing. Unlike the data parallel applications discussed in Chapter 8, the combinatorial optimization problem is characterized by an unpredictably varying unstructured search space. Its parallel execution relies on load balancing strategies to distribute the problem space recursively at run-time.