This seminar will emphasize four areas of Scientific Visualization:

• Volume Modeling :
The development of techniques for representing 2D surface models, which started in the early seventies and continues today, has been one of the most influential successes in graphics and CAGD research. The results of this research have influenced how we design cars and toasters and the way we are entertained (for example, the movies Toy Story and Jurassic Park). Research is just beginning on extending surface model techniques to volume models. That is, the dimension of the independent data is increased from 2D to 3D. To date, there has been very little research on volume models, but the potential benefits are tremendous. Presently, most visualization techniques deal directly with the volume data and only produce an image. The image is, of course, only 2D and so nothing that is 3D is modeled or extracted form the data. Rather than apply techniques (visualization, queries, feature extraction) directly to the data, volume modeling is concerned with the development of techniques for producing a 3D volume model and then applying these techniques to the model. A volume model will allow for the generation of multiresolution models which are extremely useful for many of the operations commonly needed in a scientific investigation. The volume model also serves as a means of compressing the original volume data and this has benefits with regards to archiving and transmission of large volumetric data sets. Volume modeling is a very important emerging topic in scientific visualization which will be emphasized at this seminar.
• Information Visualization:
An increasing number of automated processes and scanning devices are producing voluminous data that cannot be represented using conventional 3-D computer graphics techniques. These include single monitoring systems, LAN (local area networks) with attached scanning devices and, of course, the WWW. Research into new methods for the representation, archiving, perusal and understanding of this data is currently underway. The need for new methods for handling these abstract data sets is extremely widespread and the potential impact is tremendous. Two particularly fruitful areas are emerging which will be emphasized at this workshop. These are "smart spaces" and "visual interaction". Wireless technology and mobile computing will soon allow users to move within a variety of distributed environments while declining prices of computing power will enable an increasing number of common objects to possess an intelligence heretofore reserved for single purpose computing devices. The confluence of these two trends is allowing researchers the opportunity to explore innovative ways of connecting mobile users and computers with intelligent objects embedded within smart spaces. A smart space may be stationary, such as an office or conference room with attending aids or it may be mobile such as a vehicle with sensing and scanning devices. The possibilities and opportunities are very exciting and potentially rewarding. The new area of visual interaction is based upon recent advances in immersed visualization and gestural input and new algorithms for analyzing and processing data consisting of images (also 3D), video and audio signals. This is allowing the possibility of new and more effective paradigms for interaction between users and the simulated and real worlds contained within or accessible through computer systems.
• Flow Visualization:
Flow and vector fields are common data types in many scientific applications. The need and benefits of understanding 3D flows is well accepted. Visualization techniques are providing assistance in this area. New and exciting research is occurring in flow visualization. This is due in part to the tremendous success of the some recently developed and highly useful techniques which have revealed the importance of this research area. Emerging topics which will be included in this seminar include: multiresolution and wavelet methods for curvilinear grids, the identification, classification and representation of higher(than linear)order critical points and associated topological graphs for vector fields, extension to 3D of the surface based, highly successful LIC (line integral convolution), and real-time, interactive methods applied in virtual environments.
• Volume Visualization:
The topic of volume visualization has been one of the main research topics of scientific visualization from the very beginning days of the late 1980's. The importance and benefits of this technology has been the driving force for a high level of quality research in this area. Volume visualization is rapidly maturing to a very pervasive technology which is used on an every- day basis in many application areas. Still it is an active research area as the frontiers are pushed further and further. This seminar will focus on Volume Graphics, which is the subfield of volume visualization that deals with modeling, manipulating, and rendering of 3D scenes that are represented by volumetric data. The emphasis in Volume Graphics is on the ability to use volume data to achieve or even surpass many of the abilities traditionally supported by surface-based approaches. Recent trends in the industry and the academia predict that volume graphics is poised to leap forward in the near future. It is the goal of this symposium to reflect and promote this trend. The seminar will focus on algorithmic and complexity aspects arising in the development of efficient solution techniques for computationally difficult optimization problems. A general technique for efficient approximation algorithms is to formulate an optimization problem as an integer linear program and then to relax the integrality conditions. Recently, there has been striking success in obtaining also approximation algorithms on more general mathematical programming such as semidefinite programming. In this and other context, randomization has proved to be a powerful algorithmic method: It yields to simple and easy to analyze algorithms for many optimization problems, and it leads to a better performance guarantee. The seminar is intended to bring together researchers from different areas in combinatorial optimization and from applications. It will support the collaboration between researchers in computer science, mathematics and operations research. The workshop will have the following aims: extending of randomization and semidefinite programming techniques to other optimization problems, improved approximation algorithms and structural insights by studying linear programming, semidefinite programming and randomization, development of approaches to solve (approximatively) large linear and semidefinite programs, complexity questions for randomization and semidefinite programming, practical implementation of the used techniques (randomization, semidefinite programming), exchange of informations on recent research and stimulation of further research.