Implementation of a scalable parallel genetic algorithm for multiple architectures
There is a lack of a programming free solution which can run a distributed genetic algorithm in parallel with the only requirement that the end-user make some changes in the configuration files to tune it to the required specifications. ^ The objective of this research work is to build such a system, this system will be run on multiple computer architectures supporting parallel execution to test the performance over a variety of platforms. The parallel program based on the island Genitor model was implemented. The program is scalable enough to use any specified number of servers and can run on a variety of architectures ranging from the Beowulf class clusters, shared memory machines and network of workstations. The program is also configurable enough to read parameters from configuration files and adjust its operation based on the parameter values specified. The program can also use any fitness function containing any number of variables also specified through the configuration files, this feature allows this program to be used in a wide variety of scientific applications without any need for programming or recompiling. (Abstract shortened by UMI.)^
Engineering, Electronics and Electrical
Dayal, Yash, "Implementation of a scalable parallel genetic algorithm for multiple architectures" (2005). ETD Collection for University of Texas, El Paso. AAI1427726.