WSの分散処理を用いた線形計画問題の分割法について

概要

論文の詳細を見る
Recently, the remarkable developments of workstations, such as the high perfomance and the mark down of the costs, have rapidly popularized the use of workstations. Also, the problem solving style constituted by most of the workstations is well utilitized. By using the network communication ability, distributed processing can be considered. It is possible to have systems that take much less processing time by using distributed processing than concentrated processing. Moreover, many algorithms have been proposed to solve the problems of large scale linear programming. Usually, the solving of large scale linear programming takes a tremendous amount of time. However, this kind of problem usually consists of many zero coefficient variables. Thus, the original problem breaks down into many smaller, independent problems. The original problem become to be structuralized. This kind of problem can be well solved by using Dantzig-Wolfe's decomposition principle. Decomposition principle takes the original problem and divides it one into central problem and many sub-problems. Therefore, distributed Processing is a good application of the decomposition principle. This reseach develops the system that solves the structuralized problem, and the algorithm of solving large scale linear programming problems is also discussed.
岡山理科大学の論文