Oleksandr Markovichenko’s thesis presentation on: Persistent cohomology of cover refinements

March 10, 2022 @ 3:00 pm – 4:00 pm

Topological data analysis (TDA) is a new approach to analyzing complexdata which often helps reveal otherwise hidden patterns by highlightingvarious geometrical and topological features of the data. Persistenthomology is a key in the TDA toolbox. It measures topological featuresof data that persist across multiple scales and thus are robust withrespect to noise. Persistent homology has had many successfulapplications, but there is room for improvement. For large datasets,computation of persistent homology often takes a significant amount oftime. Several approaches have been proposed to try to remedy this issue,such as witness complexes, but those approaches present their owndifficulties.

In this work, we show that one can leverage a well-known data structurein computer science called a cover tree. It allows us to create a newconstruction that avoids difficulties of witness complex and canpotentially provide a significant computational speed up. Moreover, weprove that the persistence diagrams obtained using our novelconstruction are actually close in a certain rigorously defined way topersistent diagrams which we can get using the standard approach basedon Čech complexes. This quantifiable coarse computation of persistentdiagrams has the potential to be used in many applications wherefeatures with a low persistence are known to be less important.