Information Exchange in Multi Colony Ant Algorithms

Multi colony ant algorithms are evolutionary optimization heuristics that are well suited for parallel execution. Information exchange between the colonies is an important topic that not only in uences the parallel execution time but also the optimization behaviour. In this paper di erent kinds of information exchange strategies in multi colony ant algorithms are investigated. It is shown that the exchange of only a small amount of information can be advantageous not only for a short running time but also to obtain solutions of high quality. This allows the colonies to pro t from the good solutions found by other colonies and also to search in di erent regions of the search space by using di erent pheromone matrices.

• Type: Conference
• Year: 2000
• Where: IPPS
• Authors: Martin Middendorf, Frank Reischle, Hartmut Schmeck
• Keywords: Colony Ant Algorithms, Distributed And Parallel Computing, IPPS 2000, Multi Colony, Parallel Execution