Syncretic Evolutionary Algorithm Based On ACO And GA With Its Applications In Composition Of Investment

Evolutionary Computing (EC), which has been on the upgrade in the recent 40 years, is a new type of computing enlightened by the biological evolution thought. It is an optimized method of imitating the biological evolution process, belonging to the category of imitating human intelligence. EC expresses different kinds of complicated structures with simple code technology, instructs study and ensures search direction according to simple heredity operation and"the superior win the inferior"law of nature. Since EC searches with the method of the population (namely a group), it can search many regions in space at the same time, especially useful to the massive parallel computation. EC is typical of self-organization, self-adaptation and self-study. It neither is confined by the limits of the searching space (for example being differentiable, Single protruding and so on), nor needs any other auxiliary information (for example differential coefficient). This enables the EC work more efficiently, much simpler and more universal. The three branches of the EC include: Genetic Algorithm (GA), Evolution Strategy (ES) and Evolutionary Programming (EP). Although there are subtle differences in the algorithm realization, the common feature of the three major branches is to solve actual problems with biological evolutionary thoughts and theories.This paper starts from the knowledge of the intelligence evolutionary computing, analyzes the basic principles of the ant group algorithm and work-flow, studies the improvement, expansion and developing situation of the aut group algorithm, and also researches the mechanism of heredity algorithm and the inspiration thought, as well as the securities investment theory. Based on the research of ant group computing and heredity computing, this paper summarizes the advantages and disadvantages of the two computing methods, puts forward the basic ideas of the heredity ant group computing, works out the model of the heredity ant group algorithm, and makes up a concrete model in view of the actual application by combining the heredity ant group computing and security investment. With the analysis of cases, this paper confirms the feasibility and actuality of the heredity ant computing.Based on the ant group algorithm, the paper starts researching from the basic ant group algorithm to study the working principle of the basic ant group algorithm; improves with the basic ant group algorithm as the center to make a more real and more intelligent ant group system; grasps the fundamental principles of the heredity computing and the basic securities investment process; builds the intelligent mixed computing by combining heredity computing and ant group computing; enables the heredity ant group computing to be applied by combining it with the securities investment; compares the operational result of the heredity ant group computing with that of the ant group computing and with that of the heredity computing respectively to make a conclusion.This paper views the situation as a whole, analyzes the details one by one and argues according to the method of"planning overall, penetrating re- spectively, breaking first and establishing later, inducing comprehensively". By planning overall, this paper formulates the train of thought; by penetrating respectively, the performance of the ant group computing and the heredity computing will be improved; by breaking first and establishing later, the ant group computing and the heredity computing will be combined together to obtain a totally new mixed heredity ant group computing, then the heredity ant computing will be applied to the securities investment; by inducing comprehensively, the paper summarizes the heredity ant computing.This paper analyzes the securities investment with the ant group computing and the heredity computation. Compared to the former application, the basic ant group computing has been improved in five aspects. (1)Each ant exchanges information with the other ants and carries on the intelligent study after searching once and returning. (2)Ants can freely unify for a group after several times of searching. The group information is more abundant and the instruction searching ability of the group is much stronger. Here the group is a hypothesized association, whose behavior can instruct single ant without absolute control over its members, that is, no matter the ant joins a group or not, it remains independent. The principle of setting up the group is to group the ants with the same disposition (parameter), to group the strong ants together, to group the weak ants together; to group the strong and the weak in one category, and to assemble some ants at random to form a group. (3)Since ants are organisms, their descendants are possibly more intelligent and work much harder than their father's generation due to heredity. Thus after searching several times, the ants will perform hereditary computing. This is mainly considered in the following aspects: doing overlapping operation, which adopts the method of bi-search, to 1/3 ants at random; doing variation operation, including free variation and inducing variation influenced by the seed ants, to another 1/3 ants at random; doing the heredity computing to the last 1/3 ants tacitly. (4)Each ant possesses simple memory, and carries on the search according to the routine that it has passes through as well as intelligent study. (5)Use multithreads in the realization of procedure, let all ants work synchronically and search parallel, immediately update the information element and each parameter index of the ants in the dynamic renewal route. Dynamically change the proportion relations between alpha and beta in probability to diversify the performance of the ant and to find the best (or the secondary) solution. There are three advancements in terms of the basic heredity computing. Firstly, variation consists of self-stochastic variation and inducing variation. Secondly, the overlapping computing is chiefly applied to the ant and the searching results of the ant rather than to the final outcome. Thirdly, recoding is not carried on to any object. This paper directly applies the heredity principle to adjust the performance and the searching results of the ant, in order to simplify the process.