1989 年 4 巻 3 号 p. 330-339
The critical problem of search is the amount of time and space necessary to find a solution. For example, exhaustive search is not feasible for Rubik's Cube, because examining all sequences of means would require operation in a search space in which the number of nodes grows exponentially with the number of means. Such a phenomenon is called a combinatorial explosion. To prevent it from happening, one must reduce the size of the search space. The reduction of the search space, in this paper, is accomplished by generating hypotheses for a solution. At first, hypothesis is generated to limit the search space. Then the process is followed by the verification of hypotheses, namely, a solution is found if the hypothesis is true. It has a possibility, however, that the number of hypotheses grows too large, since a hypothesis is generated from a necessary condition for a solution. If that happens it results that the amount of time and space necessary to find a solution would reach to the same level as the exhaustive search. In this paper, we introduce a distributed system that has no restrictions on the number of processes and we assign the verifications to the processes of the system in order for all of the verifications to be done concurrently. As a result, the efficiency of problem solving is greatly improved. In order to examine the improvement acquired in the present system, the paper describes the results of simulation how the system solves 2×2×2 Rubik's Cube.
