Automatic Learning and Autonomous Control Agents toward Grid Environments
 
Mika Koganeyama1 Yuki Matsuoka2 Hitomi Matsuyama2 Yutaka Ueshima3 and Kazuki Joe2
 
1Nara Women’s University
Graduate School of Human Culture
2Nara Women’s University
Faculty of Science, Department of Information and Computer Sciences
3JAERI, KANSAI Research Establishment
Advanced Photon Research Center
 
 
Abstract
 
 
Efficient operations of an integrated management system for simulation cycles are required for the simulation users because they have to know about the simulations in detail and operate the system complexly. Moreover, data sharing and large-scale simulations for many fields on Grid have been investigated as an important research topic. Considering the above background, we began research for the development of an autonomous agent system for large-scale simulation cycles independent from the users. Moreover, we aim at standardization of an agent interface which allows each agent on Grid to communicate with the other agents cooperatively and share the knowledge each other.

As the first step, we design and implement an agent system for large-scale simulations at JAERI KANSAI Research Establishment, Advanced Photon Research Center. In this poster, we pick up two of the problems of large-scale simulations at JAERI, which are input parameter setting and the output data saving to data servers, and implement the agents to solve the two problems.