Application Research On Numerical Methods Based On Evolution Strategy

Evolution computing is an auto-adapted global optimize search algorithms which simulates the genetic and forms of the natural environment. The method is different from the traditional numerical method. People have realized increasingly that the research about soft computing are very important and take it to be a novel method which settles the problems,this method studies the individuals by reorganization, mutation, selection in the evolution process and approaches to the optimal solution.The numerical method is a mathematics branch, whose object of study is to solve the numerical methods of each mathematics questions by using the computer. The content includes digital approximation (interpolation and fitting), numerical integration and numerical differentiation, numerical solution of linear equation(group) function, numerical solution of ordinary differential equation group and partial differential equation group and so on. They had been proposed in 1990s, even earlier. The modern computers have created conditions for large scale numerical computing, it is urgent and necessary to research assembly and systematic which suit computer numerical method. In this object uses Evolution Strategy to research numerical computing, in order to settle the numerical computing problems transfer the numerical computing problems to functional optimization problems and proposed new hybrid algorithm by combining the three intelligent algorithms: Evolution Strategy, Differential Evolution Algorithm and Functional Networks at the same time. Using the method to solve the matrix's eigenvalues and eigenvectors; balance chemical equations; compute the complex functions and so on. These problems can be settled by the traditional numerical method but there exist some disadvantage such as select of the initial value sensitively, the speed convergent slowly, the accuracy lowly, even not convergent and so on.In view of questions about traditional numerical methods, prime task of this article uses the characteristics of Evolution Strategy, such as, parallel search, global convergence and robustness and so on, to solve the problems of traditional numerical methods. Let the three methods Evolution Strategy, Differential Evolution Algorithm and Functional Networks be combined with each other and exert each advantages, it can obtain best answer when deal with some problems. So, researching numerical computing by Evolution Strategy, Differential Evolution Algorithm and Functional Networks have higher theory value and practical significance.