| Three studies toward adaptive and evolutionary systems: Emergent Computation Model, Coevolutionary Model and Cooperating robots (1995) | |||||||||||||||
Abstract | |||||||||||||||
| M) [Kan 94b]. FAM is a method similar to simulated annealing but only uses local information. The performance is comparable to GSAT in large-scale graph coloring problems [Kan 95c]. Secondly, CCM-based computation has been proved to be accelerated by parallel processing by the following two methods. One is the independent parallel search method [Kan 94c, Kan 95b]. A parallel computer with M processors is used in this method. The same set of reaction rules and LODs are used in each processor, and the same or different initial data is used. Even if the same data is used, the execution speed is improved nearly M times under certain conditions. The performance of the N queens problem is improved on the Cray Superserver 6400. The other method of parallel processing is the parallel reaction method [Kan 95c]. This method enables parallel processing with little mutual exclusion, which causes performance degradatio | |||||||||||||||
Publication details | |||||||||||||||
| |||||||||||||||