Extraction of Salient Structures for Analysis and Visualization of Scientific Data
As they strive to deepen their understanding of multifaceted natural phenomena scientists and engineers leverage ever increasing computational resources to study, alongside with experimental data, predictive models of growing size and complexity through numerical simulations. The corresponding information explosion brings about the urgent need for effective visualization and analysis techniques to bridge the widening gap between the sheer amount of available data and the resulting insight. In particular, one of the major challenges for the corresponding methods is to convey the high-level structure of the data by characterizing its inherent coherence across spatial and temporal scales.
In this talk I will give an introduction to the topological approach in visualization, which provides a compelling and mathematically sound framework to design depiction and analysis algorithms for the vector and tensor fields that are ubiquitous in practical problems. I will explain how topological methods permit the automatic extraction of a schematic graph that captures the global structure of the associated mathematical flow and lend themselves to automatic processing and synthetic representations. I will present several applications of this general approach in computational fluid dynamics and fusion research. I will then proceed by showing how some practical limitations of the topological formalism can be overcome in the context of transient turbulent flows and brain imaging while resting the analysis upon similar underlying principles of spatial coherence. I will conclude my talk by pointing out exciting new fields of application for this general methodology.
Bio: Xavier Tricoche is a Research Assistant Professor in the School of Computing at the University of Utah. He is a member of the Scientific Computing and Imaging Institute that he joined in 2004 as a postdoctoral fellow. He studied computer science and applied mathematics in Grenoble, France and obtained his PhD in 2002 from the University of Kaiserslautern in Germany for his work on topological methods in vector and tensor visualization. His current research focuses on the structural analysis and effective visual representation of large-scale scientific data in applications including cardiovascular research, neuroscience, computational fluid dynamics, and fusion research.