What is Implicit Parallelism

Performance of genetic algorithms (GAs) is mainly determined by several factors. Not only the genetic operators affect the performance of a GA with varying degrees, but also the parameter settings for genetic operators interact in a complicated manner with each other in influencing a GA’s performance. Though many studies have been conducted for this cause, they failed to converge to consistent conclusions regarding the importance of different genetic operators and their parameter settings on the performance of GAs. Actually, optimizing the combinations of different strategies and parameters for different problem types is an NPcomplete problem in itself, and is still an open research problem for GAs (Mitchell, 1996). Recognizing the intrinsic difficulties in finding universally optimal parameter configurations for different classes of problems, we advocate the experience-based approach to discovering generalized guiding rules for different problem domains. To this end, it is necessary for us to gain a better understanding about how different genetic operators and their parameter combinations affect a GA’s behavior. In this research, we systematically investigate, through a series of experiments, the effect of GA operators and the interaction among GA operators on the performance of the GA-based batch selection system as proposed in Deng (2007). This paper intends to serve as an initial inquiry into the research of useful design guidelines for configuring GA-based systems.