The purpose of this paper is to show tour-distance Voronoi diagrams which represent catchment areas of facilities under the assumption that users use more than one facility. Bisectors composing tour-distance Voronoi diagrams are generally curved lines, although those of higher-order Voronoi diagrams are composed of straight lines. It is observed that contiguous facilities have larger joint catchment areas than scattered facilities. Applying them to locational selection of competitive facilities, it is found out that locational concentration tends to be superior for facilities which maximize their catchment areas, if users make tours to move around from one facility to another.
In this paper, we show a performance evaluation of distributed database in a hierarchical network in the two cases. In one case, database is managed by one supervisor agent and in the other case, it is distributed among all agents. We investigate the average expectation values of the time required by an agent to obtain a required data. In addition, how the time transition of the average expectation values of the time require by an agent to obtain a searched data in a distributed management systems changes with a queue or without queue is also analyzed.
We present an efficient algorithm based on the Legendre-Galerkin approximation for solving the Poisson equation on a fan-shaped domain. The key to the efficiency of our algorithm is to map the domain to a rectangular domain by using a new type of the double exponential formula. Some numerical results show that the convergence rate of our algorithm is nearly exponential. As an application, we also discuss the approach for solving the Poisson equation on triangular domains by using the Legendre-Galerkin method in conjunction with the Schwarz Alternating method.
The polynomial root-finding algorithm that uses the errors of numerical integration of the logarithmic derivative was announced. In this algorithm, a new approximate expression of zeros was proposed. We call the method NIEM (Numerical Integration Error Method). In general, the accuracy of the multiple zero is worse than that of the simple one. The reason for this deterioration of accuracy is that the polynomial and its differentiation is estimated in a neighborhood of the multiple zero. NIEM can avoid this deterioration by evaluating them away from the multiple zero. In this paper we propose an approach for multiple or clustered zeros using NIEM.