Publication Details
Genetic Algorithm Optimization of Group Communications
Genetic algorithms, parallel computing, message passing, group communication, network topologies
This paper presents a genetic algorithm application in area of parallel computing. An algorithm which optimizes routing of messages in message passing parallel computers is introduced. Initial implementation has been tested for regular architectures with known reachable lower bounds of communication complexity. A real application area of the algorithm includes especially irregular interconnection networks, where regular communication patterns cannot be applied. The algorithm is intended to be run especially on parallel computers which makes searching for solutions of larger scale problems feasible. Therefore, a parallel version of the algorithm is also proposed.
@INPROCEEDINGS{FITPUB7222, author = "Ji\v{r}\'{i} Staroba and V\'{a}clav Dvo\v{r}\'{a}k", title = "Genetic Algorithm Optimization of Group Communications", pages = "47--52", booktitle = "9th International Conference on Soft Computing Mendel 2003", year = 2003, location = "Brno, CZ", publisher = "Faculty of Mechanical Engineering BUT", ISBN = "80-214-2411-7", language = "english", url = "https://www.fit.vut.cz/research/publication/7222" }