Session A4 - Computational Geometry and Topology
July 10, 18:00 ~ 18:25
Stabilizing the unstable output of persistent homology computations
University of Florida, United States - email@example.com
For data filtered by time or scale, persistent homology provides a succinct summary, the persistence diagram or bar code, which encodes the births and deaths of topological features. Crucially, this summary is stable with respect to perturbations of the data. Persistent homology computations may also provide the locations of these topological features, something of particular interest to practitioners, but these are unfortunately unstable. I will present a general framework for providing stable versions of such calculations.
Joint work with Paul Bendich (Duke University) and Alexander Wagner (University of Florida).