2017 年 2017 巻 p. 105-110
In this paper, we propose a kind of evolutionary computation procedure for multimodal function optimization. The proposed approach is based on a partial differential equation whose solutions are analytically ensured to converge a stationary distribution depending on a given target distribution. We construct bootstrap procedure of a sampling and a target distribution estimation. Then, by numerical study, we show that our proposed algorithm has similar properties to estimation of distribution algorithms.