Home » Technology » Scientific Computing » Are You Solving Your Problems in the Right Environment?

Are You Solving Your Problems in the Right Environment?

© Adam Hughes

As the arena of scientific computing continues to grow, both in the number of practitioners and the number and sophistication of software available, it is important that efforts are made to keep software accessible to a large number of interested researchers. This set of circumstances has led to the interest in developing Problem Solving Environments, or PSE's. One interesting effort in this field is the PSEware (pronounces SEAware) out of Indiana University and encompassing researchers at a variety of institutions.

Shifting their focus from the PSE's developed for problems involving differential equations, scientists involved with PSEware have focused on symbolic problem definitions and graphical use interfaces, parallel program templates, and collaboration technologies, while employing object-oriented design techniques. In particular, PSEware strives to allow the user to define his problem symbolically in language that is specific to their field of research and well-understood by practitioners in that field. By developing these symbolic interfaces in an object-oriented manner, PSEware encourages their reuse in larger and more diverse applications to come.

Because much scientific study is done in groups, PSEware endeavors to establish an environment that can be used by all members of a research project. By employing various internet technologies, this sharing can even become a reality over large geographic expanses. To further enhance the usefulness of the PSE, PSEware is developed in a general way, so as to provide the basic tools necessary to allow researchers to build their own, customized PSE's.

As can be seen by the goals outlined above, the PSEware project is more concerned with laying out a set of tools that will allow scientists to build PSE's rather than just supplying the PSE. That said, the project is basing taking its direction from work in a number of specific fields. Among these are soliton exploration, code generation, sparse linear systems, and others. A list of these contributing technologies, and links pertaining to them can be found at

http://www.extreme.indiana.edu/pseware/c...

In the coming weeks we will look at how the PSEware project is integrating ideas from symbolic computations, parallel languages and parallel runtime systems, the internet, and computational software tools into a powerful system for developing PSE's that scientists will want to use.