Information-Analytical System to Support the Solution of Compute-Intensive Problems of Mathematical Physics on Supercomputers

Abstract

The paper presents an approach to the development of an information-analytical system to support the solution of compute-intensive problems of mathematical physics on supercomputers. The basis of this system is a knowledge base built on the basis of the problem domain ontology. This system provides effective information and analytical support to users thanks to detailed systematized descriptions of (a) the methods and algorithms designed for solving problems on a supercomputer, (b) software components implementing parallel algorithms and fragments of a parallel code, and (c) parallel architectures and devices used in them. Moreover, the system contains information about publications and information resources on this subject. These capabilities saves considerably the time required for mastering the methods for solving compute-intensive problems of mathematical physics on supercomputers since all the necessary information is structured and collected in one place, namely, in the knowledge base of the information-analytical system.