Abstract: Persistent homology is a key method in topological data analysis, a young but rapidly growing field at the intersection of computational geometry and algebraic topology. Persistence is used to describe the shape of data in a way that generalizes clustering: besides considering what connected components (clusters) are present in the data, it also describes their topology.

 

This talk will introduce persistence and some of its applications in materials science, neuroscience, and cosmology. We will also discuss computation of persistence at scale, and how lock-free and distributed data structures help solve this problem.

Short Bio: Dmitriy Morozov is a Staff Scientist at the Lawrence Berkeley National Laboratory. He received B.Sc. degrees in Applied Mathematics and Computer Science from NC State University in 2003, followed by a Ph.D. in Computer Science from Duke University in 2008. After a postdoctoral appointment in Departments of Mathematics and Computer Science at Stanford University, he moved to the Lawrence Berkeley National Laboratory in 2011. His main interests are computational topology and geometry as well as high-performance computing, and how the interplay between these fields helps data analysis.