Importance of information exchange in quasi-parallel genetic algorithms
Guan Wang, Degang Wu, Wenjin Chen, Kwok Yip Szeto · 2011
In this paper, we make a brief study on the effect of exchange rate in quasi-parallel genetic algorithms. The exchange rate is determined by two elements: the communication topology of the parallel populations and the communication capacity on each link. Here we formulate the communication capacity as the number of chromosomes one population exchanges with its neighbors. To study the effect of the two elements of exchange rate separately we did some tests on the minimization of the Weierstrass Function. Our results show that topology with a larger number of exchanged chromosomes generally yields better performance.