Topological Analysis for the Exascale

11^th June 2024, 13:00 Ashton Lecture Theatre
Prof. Hamish Carr
School of Computing, University of Leeds

Abstract

Scientific simulations are now capable of generating an exabyte of data representing a physical phenomena. While our computer systems have kept pace with the demands of computing and storing data at this scale, our ability as humans to absorb and understand this data has not. As a result, formal mathematical tools such as topological analysis are increasingly important for extracting knowledge from our data. This requires developing algorithms for computing and exploiting topological abstractions, user interface metaphors for supporting scientific interrogation of the data, and application-specific tools. Finally, all of these tools must be scaled up from the single-machine, serial context where the concept is proven, to the supercomputers where the interpretative power they provide is most needed. This talk will map out the entire process for one of the principal tools - the contour tree, from initial development both mathematically and computationally, through to recent work that has made it feasible to apply contour tree analysis at supercomputer scale.

Biography

Hamish Carr received his Ph.D in Computer Science from the University of British Columbia in 2004, and is now a Professor at the University of Leeds. His areas of interest include computer graphics, scientific visualisation and computational topology. His work ranges from the foundational mathematics of visualisation through rendering problems and user interfaces to algorithmic development, topological analysis and domain-specific applications. Most recently, he has been working on scaling visualisation tools, and in particular topological tools, to modern hybrid exascale architectures.