"PSE construction using software components*" Bruce Char, Jeremy Johnson, and Lakshman Y. N. Mathematics and Computer Science Drexel University Philadelphia, PA 19104 USA {bchar, jjohnson,lakshman}@mcs.drexel.edu A problem solving environment (PSE) is an integrated software system for solving problems in a specific application domain. A PSE should enable the user to input and work on problems in a manner that is natural to the problem domain without being burdened by the underlying computer software used by the PSE. Many different types of computing are typically involved in a PSE for scientific and engineering problems: numeric and symbolic computation, code generation and compilation, visualization, document processing, knowledge representation and processing, and automated reasoning. The PSE must provide support communication between the different computational components and an effective interface for the user to interact with the PSE. Since different computations may be performed on different computers (e.g. a supercomputer might be used for a computational intensive numeric simulation) and multiple users may be interacting with the PSE from different locations, the PSE must support distributed computation. This talk presents a framework for building PSEs. It is proposed that PSEs be built through the composition of software components. A software component is a configurable piece of software with a standard interface. Effective construction of PSEs will depend on having a useful set of components. We believe that a key component for PSE construction is a generic server component. The generic server serves as an application wrapper with support for control and communication. We present an initial implementation of a generic server component using java beans. This server component has been used to wrap computer algebra systems (e.g. maple and mathematica), compilers (e.g. f77), interpreters (e.g. scheme), and graphing programs (e.g. gnuplot). The use of software components in general, and the generic server component in particular, to build PSEs will be demonstrated. Three PSEs will be presented: Techtalk, a PSE for web-based mathematical collaboration, a PSE for pollution modeling, and soliton explorer, a PSE to study soliton geometry. *This work is part of the PSEware project, and was funded in part by the National Science Foundation, as grant number CCR-9527130. The PSEware project is a collaboration between researchers at Caltech, Drexel, Indiana University, Los Alamos National Laboratory, and the University of California at Irvine. Its goal is to facilitate construction of problem solving environments for science and engineering through software components.