Project D1
Project D1 - Carsten Eden, Peter Korn, Thomas Ludwig, Stephan Olbrich and Thomas Rung: Scalable In-situ and post-processing data analysis and visualization in simulation sciences
Scientific Background and Motivation
Data analysis and visualization of results from scientific simulations is mostly done by a post-processing approach. This involves conventional software that is designed for the application on 3D graphics workstations with little degree of parallelization, providing result file readers and tools to generate 3D graphics, 2D images, or movies. In the context of challenging high-resolution, massively-parallel simulations a paradigm shift is required in order to visualize complete multi-dimensional data sets instead of by now commonly processed data parts, such as slices or average values. One possible approach to solve the problems regarding huge amounts of temporary data (depending on application: 1 to 500 PB) that cannot be stored at all and to avoid sequential bottlenecks is the so-called in-situ data extraction and visualization.
Aims and Objectives
In this project, we establish a cross-sectional data analysis visualization lab. Parallel data extraction will be integrated in several applications that can take advantage of large data analysis, in integrated simulation scenarios (in-situ) and in post-pro-cessing scenarios (analysis of archived data). We focus on disciplines where high-performance scientific computing and visualization are key instruments to increase knowledge, and to mediate it in multi-media-enhanced lectures.
Contact: Prof. Dr. Stephan Olbrich, Director of the Rechenzentrum Uni Hamburg, email: