An e-Science problem solving environment for scientific numerical study

摘要

As increasing complexity of scientific computational applications that embrace multiple physical models and require a large range of scales on computing power, demands for interoperable infrastructures steadily have been increasing. Even though heterogeneity of computational resource including large-scale computing power, storage and networks gives difficulty of management on extending resource availability in e-Science environment, researchers, especially who have limited knowledge of managing computing resource, desire to have uniform and easy-to-use access interface to those diverse computational resources. However, no problem solving environment (PSE) is capable of solving all problems in diverse application domains as their execution requirement are so various. Depending on application characteristics, it is meaningful to provide a proper PSE to solve specific application patterns for e-Science environment and manage life cycle of an application for designing, executing, analyzing, and profiling. In this paper, we present a common and integrated framework providing a problem solving environment among multi application domains, especially, in scientific numerical study, running over e-Science environment fostering interoperability in Grid infrastructure. The proposed framework, based on UNICORE, provides seamless, secure, and intuitive access to distributed Grid resources. It also speeds up developing new numerical study with existing application patterns, which are easily shared, reusable, and extendable.