The convergence strategies and pause matter for evolutionary modeling
Ni He, Gang Chen, Fengrui Sun · Chinese Control Conference · 2010
As the application of genetic programming in mathematic modeling, evolutionary modeling method provided an effective means for the describing of the higher-order or nonlinear systems. Although evolutionary modeling method displayed strong aptitude and self-learning ability in applications, its academic groundwork is instable, one of the reasons is the evolutionary arithmetic, which this method adopted, is a sort of stochastic optimize arithmetic, its convergence theory wants strict mathematic demonstrate. Studying the convergence abilities of evolutionary modeling based on the works of other researchers, and deduced a recurrence formula of the probability of groups containing satisfying solutions by analyzing the diagnostic parameters of algorithmic operators. A sufficient term of group convergence is educed consequently out of this formula, and thereby the operable convergence strategies for several familiar evolutionary patterns are provided. The pause time of evolutionary modeling are also included, which can guide the design of the modeling arithmetic.