Rectification of interleavings and a persistent Whitehead theorem

Edoardo Lanari, Luis Scoccola

TL;DR

This paper provides a theoretical foundation for comparing different "homotopy-invariant" distances used in Topological Data Analysis (TDA). The authors prove that several proposed ways to measure the distance between persistent spaces are equivalent up to a small constant factor and confirm a version of the "Persistent Whitehead Conjecture," which relates similarities in homotopy groups to the closeness of the underlying persistent spaces.

What Problem Does This Paper Address?

In Topological Data Analysis, we often study "persistent spaces"—sequences of topological shapes that grow and evolve as we change a parameter (like a distance threshold in a point cloud). To make TDA reliable, we need stability theorems: if our input data changes slightly, our topological summary should also change only slightly.

The standard tool for this is the "interleaving distance." However, the basic interleaving distance is often too sensitive because it requires strict isomorphisms between spaces. Researchers have proposed several "homotopy-invariant" versions that are more flexible, such as:

  1. Homotopy interleaving distance ($d_{HI}$): The "gold standard" but difficult to compute.
  2. Interleaving distance in the homotopy category ($d_{IHC}$): A categorical middle ground.
  3. Homotopy commutative interleaving distance ($d_{HC}$): The weakest and most computationally accessible notion.

Until now, it wasn't clear if these distances were fundamentally different or just different ways of looking at the same thing. Furthermore, a major open question—the Persistent Whitehead Conjecture—asked whether persistent spaces that have "close" homotopy groups are necessarily "close" as spaces.

Methods

The authors employ advanced tools from Category Theory and Homotopy Theory, specifically working within the framework of cofibrantly generated model categories.

Key techniques include:

  • Rectification: This is the process of taking a "weak" diagram (where paths only commute "up to homotopy") and turning it into a "strict" diagram. The authors show how to "rectify" interleavings, effectively jumping from the weakest distance ($d_{HC}$) to the strongest ($d_{HI}$).
  • Toda Brackets: The authors use these algebraic constructions to find lower bounds for their results, proving that some "weak" interleavings cannot be perfectly rectified without increasing the distance.
  • Simplicial Homotopy Theory: They use simplicial sets and Kan fibrations to prove the Persistent Whitehead theorem, allowing them to lift similarities in homotopy groups to similarities in the spaces themselves.

Datasets

Not discussed in the paper. This is a purely theoretical mathematics paper focused on proving theorems and establishing categorical relationships.

Key Insights and Ideas

  • The "Factor of Two" Rule: For 1-parameter persistence, the different homotopy-invariant distances aren't actually that different. The authors show they are all equivalent up to a multiplicative constant (specifically, a factor of 2).
  • Multi-persistence is a Different Beast: A major "aha moment" in the paper is the proof that this equivalence fails for multi-parameter persistence (e.g., bi-filtrations). In higher dimensions, you cannot simply rectify weak interleavings into strong ones; the gap between the distances can be infinite.
  • Dimension-Dependent Stability: The Persistent Whitehead theorem (Theorem B) shows that the "closeness" of two spaces depends linearly on their dimension $d$. If their homotopy groups are $\delta$-interleaved, the spaces themselves are roughly $4(d+1)\delta$-interleaved.
  • Obstructions to Rectification: By using Toda brackets, the authors demonstrate that there is a fundamental "cost" to making a diagram strict. They show that for topological spaces, the constant $c$ in their rectification theorem must be at least $3/2$.

Key Results

  • Theorem A (Rectification): For 1-persistent objects, if two spaces are $\delta$-homotopy commutative interleaved, they are $c\delta$-homotopy interleaved for any $c > 2$. This implies the chain of inequalities: $2d_{HC} \geq d_{HI} \geq d_{IHC} \geq d_{HC}$.
  • Theorem B (Persistent Whitehead): The authors prove that for $d$-dimensional persistent CW-complexes, a morphism inducing a $\delta$-interleaving in all homotopy groups implies the spaces are $(4(d+1)\delta)$-interleaved in the homotopy category.
  • Counter-example for Multi-persistence: They prove that for $m=2$ (bi-persistence), there is no constant $c$ that allows for the rectification seen in Theorem A. This highlights a significant theoretical hurdle for multi-parameter TDA.

Limitations and Future Work

  • Optimal Constants: While the authors prove $c=2$ works for rectification, they suspect the optimal constant might be $3/2$. Closing this gap remains an open problem.
  • Strictness of the Whitehead Theorem: The version of the Whitehead theorem proven here applies to "persistent CW-complexes." The authors note that the original conjecture was slightly broader (applying to "pointwise CW-complexes"), and while they believe their result covers most practical applications, the full conjecture remains technically open.
  • The $d_{HI}$ vs $d_{IHC}$ Mystery: The paper leaves as an open question whether the homotopy interleaving distance ($d_{HI}$) is exactly equal to the interleaving distance in the homotopy category ($d_{IHC}$).

Summary updated

Abstract

The homotopy interleaving distance, a distance between persistent spaces, was introduced by Blumberg and Lesnick and shown to be universal, in the sense that it is the largest homotopy-invariant distance for which sublevel-set filtrations of close-by real-valued functions are close-by. There are other ways of constructing homotopy-invariant distances, but not much is known about the relationships between these choices. We show that other natural distances differ from the homotopy interleaving distance in at most a multiplicative constant, and prove versions of the persistent Whitehead theorem, a conjecture of Blumberg and Lesnick that relates morphisms that induce interleavings in persistent homotopy groups to stronger homotopy-invariant notions of interleaving.
BibTeX
@inproceedings{lanari2023rectification,
  author = {Edoardo Lanari and Luis Scoccola},
  title = {{Rectification of interleavings and a persistent Whitehead theorem}},
  booktitle = {Algebraic},
  year = {2023},
  doi = {10.2140/agt.2023.23.803},
}

Loading related content...

Figure