Brian J. d'Auriol, Ph.D.

Student Research Details
Return to d'Auriol's Students Supervised Page


Rupa Policherla, Scientific Visualization Techniques for Program Visualization, Department of Computer Science, The University of Texas at El Paso, May 2004. Advisor: Brian J. d'Auriol

Software built over a span of a few years becomes complex and hard to maintain. Often, it is time consuming and expensive for software programmers and researchers to understand such programs. One way that has been looked at is program visualization. Program visualization has ongoing exploration in the area of visualizing program code, especially, to identify an organized structure of the program. However, another approach to address this is to analyze and investigate the applicability of scientific visualization techniques for program visualization. Scientific visualization comprises various advanced visualization techniques and tools for a better comprehension of the system, phenomenon or an application. These techniques enable data to be presented in a distinguished structural form. Therefore, this thesis looks at the feasibility of applying these techniques to define a structure of the program data and, thereby, to promote program comprehension. The hypothesis considered in this thesis is that a thorough insight into the program structure can be achieved by utilizing techniques that are applied to large scale scientific data.

The first part of this work discusses scientific visualization in terms of its pipeline, techniques and applications. This is followed by the analysis of program visualization with respect to program visualization systems. To understand the code structure of an existing program, scientific visualization techniques such as glyphs and orthoslice are investigated in the creation of a graphical model of a program. The second part of this work systematically investigates the characterization of scientific and program data. Characterizing scientific data eases visualization, especially, in choosing appropriate techniques for operating on specific data organizations. Program data characterization includes program artifacts such as procedures, dependencies, function calls, data attributes and other data organizations. A Program Scientific Visualization (PSV) model, the primary contribution of this thesis, is developed to reflect a framework for the adopted systematic approach of characterizing scientific and program data. This model defines the process of mapping program data to scientific data, which is further complemented with a formalized notation. The approach considered in this thesis is novel. The third part of this work is the application of scientific visualization techniques to program data. The approach is innovative in applying mature scientific visualization techniques to visualize the structure of programs. Doing so allows programmers to gain insight into program design

Last Updated: August 3, 2007