| 主权项 |
1. A search method of a pertinent solution using a genetic algorithm that performs genetic operation including selection, mutation, and crossover of a plurality of individuals each having an element of a candidate solution to a problem in the form of a gene sequence, the method comprising:
initial creating a plurality of initial individuals and using the plurality of initial individuals as a plurality of target individuals that are a target of processing; storing, in an information storage portion, genetic information about the plurality of target individuals as a plurality of existing target individuals; generating a plurality of new target individuals from the plurality of existing target individuals by performing at least one genetic operation among selection, mutation, and crossover for at least one of the plurality of existing target individuals; writing, to the information storage portion, the genetic information about the plurality of new target individuals generated in the at least one genetic operation; calculating a fitness value corresponding to the candidate solution to the problem on the basis of the gene sequence of each of the plurality of initial individuals and the plurality of new target individuals; a reprocessing step of repeating the generating and the calculating for the plurality of new target individuals; and setting at least one value with a highest adaptability among plural fitness values obtained in the calculating that is repeated plural times, as one of the candidate solutions to the problem, wherein the information storage portion includes, as one of the items of the genetic information about the plurality of target individuals, information about whether all individuals are in which state among a state that is a target of genetic operation in the generating and a target of the calculation of the fitness value in the calculating, and a dead state that is not the target of genetic operation nor the target of the calculation of the fitness value, wherein in the initial creating and the generating, a plurality of layers, where the plurality of target individuals can be retained up to a predetermined number, are provided, wherein a breeding area is composed of the plurality of layers, and at least one of the plurality of layers has a relatively high ranking and at least another one of the plurality of layers has a relatively low ranking, and a predation of leading an individual belonging to a lower layer to a dead state due to predation by an individual belonging to a higher layer in the breeding area constitutes an aspect of the generating, wherein in the calculating, a check of an optimal fitness value is performed for all of the plurality of layers constituting the breeding area respectively, and further comprising: completely replacing the individuals belonging to one layer with individuals belonging to another layer under predetermined conditions. |
