HIGH-ORDER HOPFIELD NETWORK FOR SAT
暂无分享,去创建一个
The CNFSAT Problem is an NP-Comlete Problem. Inengineering mnyappications can be mapped into SAT Problems. So designing a fast and efficientalgorithm for solving SAT probems is valuabe. The Hopeld netWork as a classicalneural network has been widely appied in the optindzation field.In thes paper,anew mpthod tha appies the highother Hopeld neuIal network to solve the SAT problem is Presented and the thought how to map the SAT Problem into the Hopfieldnetwork is discussed in detail. The definihon of the networks. enerpy hahon isgiven and the networks' constructing method (including continuous and discretemodl) are presented. This paper also proves the networks' stahility and discussesthe strategies about how to escape from local minimal points in the network. Final-ly simulating experiments in which the discrete Hop field network is adopted aredone to solve the 3-SAT problem and their results compared with the classical localsearch algorithm are presented. It proves that the algorithm is fast and efficient,especially the number of searching is decreased prominently.