A case study of knowledge engineering in the application area of elliptic partial differential equations (PDEs)

The importance of problem solving environments (PSEs) for scientific computing is becoming increasingly apparent. The main components of such a PSE are the user interface and the expert system. The authors have proposed a task-level architecture called NAIS which attempts to capture the common nucleus of particularities of expert systems in scientific computing. They report on the application of NAIS to the very important area of elliptic PDEs (partial differential equations), initially using the VAX-II system as the target architecture for the PDE library. As the resulting expert system is divided to an architecture dependent and independent part, it can be easily adapted to other architectures. Finally, this development constitutes an interesting effort of knowledge engineering in scientific computing using a task-level architecture.<>