Visual Exploration of Simulated and Measured Blood Flow - Scientific Visualization: Uncertainty, Multifield, Biomedical, and Scalable Visualization - page 307

Part IV
Scalable Visualization
We live in the era of the data tsunami. Data is generated faster than our ability to
digest the deluge of information. Visualization researchers and domain experts
typically think of scalability in terms of the development of high-performance
computational resources systems and of powerful new scientific instruments
collecting vast amounts of data. These have lead to an unprecedented rate of
growth of scientific data. The potential impacts to science of being able to
effectively analyze these abundant sources of both simulation and measured/
scanned data are tremendous. For effective data analysis, scalable visualization
methods are required. Scalability takes on many forms from algorithmic
scalability, the diversity of types of data, scalability across devices, scalable
functional representations of data, and the integration into high-performance
computational environments. This Part of the topic addresses these issues.
In the first chapter, Garth and Gaither discuss the visualization and analysis,
using integration-based methods, of large-scale vector fields on parallel architec-
tures. They provide an overview and describe parallelization over seeds verses
parallelization over blocks. The chapter concludes with a discussion of future
directions.
As data sizes increase, one approach to scalability is to use feature-based
techniques to represent the data with a smaller feature space. Bennet, Gyulassy,
Pascucci, and Bremer discuss in Chap. 27 a feature hierarchy framework with two
examples: the merge tree and the Morse-Smale complex. They describe how to
perform interactive exploration of feature-based statistics and apply these concepts
to a computational combustion example.
Scalability includes the number of different data sources as well as interactions
between physical scales. In Chap. 28 , Ebert, Gaither, and Lasher-Trapp explore
system-of-systems and cross-scale issues and opportunities. The vast variety of
data poses interesting scalability issues including visual scalability, software
scalability, and information scalability. After discussing such issues, the authors
conclude with an assessment of technology needs to address these.