Large scale computer simulations require a special class of tools for calibration, debugging and interpretation of results. Computer graphics and scientific visualization are currently finding increasing use in many large scale modeling efforts. Visualization provides an effective means of dealing with many aspects of numerical models involving large scale computation. Applications of visualization in this area include data set preparation, model calibration, code debugging, code performance analysis, and interpretation and presentation of model results. Visualization is especially useful in allowing modelers to interrogate the data sets produced by their models. Often a modeler will need to recalibrate a computer model and recompute to expose certain details or features in the underlying physical problem. Visualization helps the modeler to ensure that all of the detail available from a particular set of computational results is exposed before undertaking the often expensive task of recomputing.
This chapter discusses some recent advances in the application of scientific visualization software tools in the area of computational science. Specifically, we illustrate how visualization tools are being used in three areas of mathematical simulation modeling
In each of these areas, we will first outline some general concepts and techniques and then discuss their implementation. Following a general overview, some implementations of scientific visualization techniques will be illustrated with examples. We believe that, in the area of scientific visualization, it is especially important to strike a balance between visualization concepts and techniques and the operational issues surrounding their implementation on specific hardware using specific software. In moving from one graphics workstation to another, a computational scientist encounters a great deal of diversity in graphics hardware and software capabilities. Also, graphics hardware and software are advances that are following a very steep growth curve. To remain current, the user of scientific visualization must endeavor to track this growth curve. A computational scientist who becomes implementation oriented (recipe bound), rather than concept oriented, is sure to fall behind.
Having argued that visualization concepts and techniques are foremost, we acknowledge that no scientific visualization gets done without implementation. For purposes of this chapter, we will assume that the implementation or actual visualization activity, takes one of two forms:
To act as an end user of scientific visualization software on a UNIX workstation, requires the computational scientist to be able to
Each workstation vendor offers its own windowing systems and graphics library. Even though skills and codes developed on one graphics workstation usually transfer to another fairly easily, this transfer can sometimes be tedious and time consuming. Unless one requires some hardware or software feature that is vendor specific, the issue of skill and code transfer can often be avoided. X-Windows provides a ``standard'' environment that is common to a broad range of UNIX workstations. To the software end user, the X-Window environment is portable and robust and provides access to a large number of high quality, public domain visualization software packages. In addition, the X11 library contains functions that enable a computational scientist to manage windows and to develop custom graphical interfaces.
In many universities and laboratories there are not enough seats at UNIX graphics workstations to meet everyone's needs. For those scientists who are developing custom visualization software, a PC (Macintosh or IBM clone) equipped with at least VGA quality graphics has proven to be a valuable tool in combating a shortage of workstations. Much of the early design and ``proof of concept'' testing can be done on the PC. With the advent of PCs with several megabytes of memory and with increased graphics resolution, such as the SVGA at 1024x768 pixels, the modeler can now use a DOS based PC to perform many of the tasks of the UNIX workstation.