Analyzing large and high-dimensional data sets is a challenging task,
ideally carried out using sophisticated tools which automate the
analysis and allow a focus on the most relevant information. These goals
can be achieved using feature-based methods, which foster
target-oriented studies of the most important aspects of a data set.
Many classic feature-based approaches rely on numerical computation
schemes involving derivatives, which makes them prone to noise. Discrete
Morse theory, on the other hand, employs purely combinatorial
computations, which allows for robust, parameter-free, and topologically
consistent algorithms.
The talk gives a brief introduction to discrete Morse theory,
highlighting its great potential for feature-based data analysis as well
as discussing some of the intricate challenges that come with it.
Furthermore, we will demonstrate the high versatility of discrete Morse
theory by presenting its applications in computer graphics, fluid
dynamics, cell biology, and planetology.