Optimization Problem Solving System using GridRPC

Abstract

In this paper, in order to efficiently solve a large-scaled and complicated optimization problem, the framework of applications integration on the Grid is proposed. In the proposed framework, application holders publish their own applications as a service, which provides four basic functions using the GridRPC system. These functions enable end-users to invoke applications and to exchange input/output files among services. Therefore, end-users can design arbitrary applications integration through the wide area network by just selecting services and combining the provided functions. In addition, we also propose the Application Programming Interfaces (APIs). These APIs enable the service to invoke the designed applications integration. Therefore, end-users can solve their own optimization problems by constructing the optimization system. In this paper, we implement the proposed framework and APIs using the NetSolve system, which is one of the GridRPC systems. Through computational simulation examples, it was found that the overhead time of our implementation is very short compared to the calculation time of optimization. We also found that the optimization system of our implementation can be used practically in the wide area network. Therefore, it can be expected that the proposed framework and APIs are efficient for designing the optimization problem solving system on the Grid.