Abstract
We study distances on zigzag persistence modules from the viewpoint of derived categories and Auslander--Reiten quivers. The derived category of ordinary persistence modules is derived equivalent to that of arbitrary zigzag persistence modules, depending on a classical tilting module. Through this derived equivalence, we define and compute distances on the derived category of arbitrary zigzag persistence modules and prove an algebraic stability theorem. We also compare our distance with the distance for purely zigzag persistence modules introduced by Botnan--Lesnick and the sheaf-theoretic convolution distance due to Kashiwara--Schapira.
Introduction
In Topological Data Analysis (TDA), persistent homology is a primary tool for extracting "barcodes" that represent the lifespan of topological features within a dataset. While standard persistence assumes a linear filtration where data only grows (e.g., increasing a radius), real-world applications—such as time-series analysis or moving point clouds—often necessitate zigzag persistence. In these scenarios, data can be both added and removed at each step, leading to more complex mathematical structures.
A fundamental requirement for any TDA tool is an Algebraic Stability Theorem (AST). This theorem ensures that small perturbations in input data (noise) result in only small changes in the output barcode. While stability is well-established for ordinary persistence, providing a general and computable framework for arbitrary zigzag orientations has remained a significant challenge. This paper addresses this gap by leveraging advanced tools from representation theory and homological algebra.
Theoretical Framework: Derived Categories and Tilting
The authors address the stability problem by shifting the focus from individual modules to their derived categories. This high-level algebraic abstraction treats complexes of modules as the primary objects of study.
A key innovation is the proof of Derived Equivalence: the authors demonstrate that the derived category of ordinary persistence modules is equivalent to the derived category of any zigzag persistence module. This equivalence is bridged by a "classical tilting module," a concept from representation theory that allows for the "transport" of well-understood stability results from the ordinary setting to the more complex zigzag setting.
To facilitate explicit calculations, the authors utilize Auslander–Reiten (AR) quivers. These are graphical representations of the relationships between indecomposable modules, effectively providing a map of the "space" of persistence modules. By using these quivers, the researchers can define and compute new "derived" versions of interleaving and bottleneck distances, which are the standard metrics for comparing barcodes.
Key Findings and Stability Results
The central result of the paper is the Algebraic Stability Theorem, which formally proves that the derived bottleneck distance ($d_B^D$) between two zigzag barcodes is bounded by the derived interleaving distance ($d_I^D$). Furthermore, the authors establish an Isometry Theorem, showing that $d_B^D = d_I^D$. This means the bottleneck distance, which is more intuitive and easier to visualize as a matching between barcode intervals, perfectly captures the algebraic interleaving of the modules.
The study also reveals that this new distance metric is incomparable to the "block distance" previously introduced by Botnan and Lesnick. Depending on the specific data and zigzag orientation, one metric may be more sensitive than the other, suggesting that no single distance metric is universally superior for zigzag persistence. Additionally, the paper connects TDA to microlocal sheaf theory by demonstrating that their derived interleaving distance is essentially a version of the convolution distance used in the work of Kashiwara and Schapira. This connection links TDA to deep results in algebraic geometry and topology.
Conclusion and Future Directions
This work provides a unified mathematical framework for the stability of zigzag persistence modules, moving the theory closer to practical algorithmic implementation. By showing that zigzag modules are essentially "rotated" versions of ordinary modules within a larger derived category, the authors simplify a previously intractable problem.
However, challenges remain regarding computational complexity. While the mathematical formulas for these distances are now established, optimized algorithms for calculating them on large-scale datasets are still needed. Future research could focus on implementing these metrics in software libraries like Hera or Gudhi to evaluate their performance on real-world time-series data. Additionally, further work is needed to interpret "derived barcodes" in the context of physical data features, such as specific homology groups at given time points.
Summary updated