Through the availability of increasingly powerful computers with increasing amounts of internal and external memory, it is possible to investigate incredibly complex dynamics by means of ever more realistic simulations. However, this brings with it vast amounts of data . To analyze these data it is imperative to have software tools which can visualize these multi-dimensional data sets. Comparing this with experiment and theory it becomes clear that visualization of scientific data is useful yet difficult. For complicated, time-dependent simulations, the running of the simulation may involve the calculation of many time steps, which requires a substantial amount of CPU time , and memory resources are still limited, one cannot save the results of every time step. Hence, it will be necessary to visualize and store the results selectively in `real time' so that we do not have to recompute the dynamics if we want to see the same scene again. `Real time' means that the selected time step will be visualized as soon as it has been calculated.

The main reasons for scientific visualization are the following ones : it will compress a lot of data into one picture (data browsing), it can reveal correlations between different quantities both in space and time, it can furnish new space-like structures beside the ones which are already known from previous calculations, and it opens up the possibility to view the data selectively and interactively in `real time'. By following the formation and the deformation as well as the motions of these structures in time, one will gain insight into the complicated dynamics. As was mentioned before, we also want to integrate our simulation codes into a visualization environment in order to analyze the data 'real time' and to by-pass the need to store every intermediate result for later analysis. This is possible by means of processing in which the simulation is distributed over a set of high-performance computers and the actual visualization is done on a graphical distributive workstation. It is also very useful to have the possibility to interactively change the simulation parameters and immediately see the effect of this change through the new data. This process is called computational steering and it will increase the effective use of CPU time.