Counting Barcodes with the same Betti Curve

Henry Ashley, Håvard Bakke Bjerkevik, Justin Curry, Riley Decker, Robert Green · 2026

Read Counting Barcodes with the same Betti Curve by Henry Ashley, Håvard Bakke Bjerkevik, Justin Curry, Riley Decker, Robert Green (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-09011.txt.

Introduction

Topological data analysis turns shapes into barcodes, then often compresses those barcodes into a simpler summary called the Betti curve. Henry Ashley and four co-authors ask the inverse question: how many different barcodes collapse to the same curve? The answer measures how much information the compression throws away.

They link the counting problem to two distant areas: the Kostant partition function from representation theory, and the enumerative combinatorics of magic juggling sequences studied by Ronald Graham and others. Proving equivalences between the three settings, they quantify exactly how lossy the step from persistent homology to Betti numbers can be.

Catalog record

CREATORHenry Ashley, Håvard Bakke Bjerkevik, Justin Curry, Riley Decker, Robert Green
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE33 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-09011

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Counting Barcodes with the same Betti Curve Henry Ashley, Håvard Bakke Bjerkevik, Justin Curry, Riley Decker, Robert Green Abstract This paper considers an important inverse problem in topological data analysis (TDA): How many different barcodes produce the same Betti curve? Equivalently, given a function β : [ n ] = { 1 < ⋯ < n } → ℤ ≥ 0 , \beta\colon[n]=\{1<\cdots<n\}\to\mathbb{Z}_{\geq 0}, how many different ways can we write β \beta as a sum of indicator functions supported on intervals in [ n ] [n] ? Our answer to this question is to connect persistent homology with the study of the Kostant partition function and the enumerative combinatorics for so-called “magic” juggling sequences studied by Ronald Graham and others. Specifically, we prove an equivalence between our inverse problem and corresponding statements in these other two settings. From an applications and statistics point of view, our work provides a quantification of how lossy the TDA pipeline is when moving from persistent homology to persistent Betti numbers. 1 Introduction 1.1 Context and Prior Work Topological Data Analysis (TDA) provides an impressive toolkit for the extraction and summary of shape in data. The input to the TDA pipeline can be one of several types of data: a finite subset of \mathbb{R}^{d} , a semi-algebraic set, a scalar or vector field on a manifold, and so on. The output of the TDA pipeline can also be one of several types of summary objects: a merge tree, a Reeb graph, a persistence diagram/barcode, or a piecewise constant \mathbb{Z} -valued function, such as a Betti curve or Euler curve. By fixing the class of input data and the type of output object, one naturally encounters an inverse problem : How far is the TDA pipeline from being injective for that class of data set? Since topology necessarily ignores certain aspects of a data set, it is to be expected that this mapping is not injective and typically will have high-dimensional, geometrically complex fibers 1 1 1 As a reminder, the fiber of a mapping f\colon Y\to X over a point x\in X is its pre-image f^{-1}(x) . . This is evidenced by a robust line of research into inverse problems in TDA, of which [ 21 , 23 , 14 , 27 , 12 , 33 , 18 , 16 , 30 , 29 , 22 , 31 , 38 , 7 , 6 , 15 , 37 ] is a likely incomplete sampling. This paper extends existing research [ 14 , 27 , 22 , 31 , 15 ] into inverse problems between TDA objects themselves, which tends to have a more combinatorial flavor. Those papers were mostly concerned with understanding how \pi_{0} , the set of connected components, refines H_{0} , the vector space freely generated by the set of connected components. In the persistence setting, this leads to the now well-understood mapping from merge trees to barcodes. The problem considered here is, in some aspects, similar as we consider how lossy the process of replacing a vector space with its dimension is. 1.2 Problem Statement In this paper we formulate and shed light on a hitherto unconsidered inverse problem between two well-studied TDA objects—barcodes and Betti curves—in the finitely indexed setting. To explain, recall the following: Given any functor F\colon\mathbf{[n]}=\{1<\cdots<n\}\to\mathbf{vect} from the totally ordered set on n elements to the category of finite-dimensional vector spaces, one has an associated dimension (or Hilbert) function \dim(F)\colon[n]\to\mathbb{Z}_{\geq 0}, which assigns to each i\in[n] the dimension of the vector space F(i) . Our inverse problem is the following: Given: a dimension function h\colon[n]\to\mathbb{Z}_{\geq 0} . Enumerate: all functors F\colon\mathbf{[n]}\to\mathbf{vect} , up to iso, with \dim F=h . Equivalently, a functor F\colon\mathbf{[n]}\to\mathbf{vect} is a type A_{n} quiver representation, whose isomorphism class is determined by its decomposition into interval representations. The dimension function and the multiset of intervals resulting from the decomposition are how we get our notions of Betti curves and barcodes. Definition 1.1 . A Betti curve of length n is a function \beta\colon[n]\to\mathbb{Z}_{\geq 0} , or equivalently, a finite sequence (\beta_{1},\beta_{2},\ldots,\beta_{n}) where \beta_{i}=\beta(i) . We define \textbf{Betti}(n) as the set of Betti curves of length n . Since \beta(i)\in\mathbb{Z}_{\geq 0} can be arbitrary, \textbf{Betti}(n) is necessarily an infinite set. For i\leq j , we use the notation [i,j) for the set \{i,i+1,\dots,j-1\} , and we refer to nonempty sets of this form as intervals . For instance, [n]=[1,n+1) by this convention. Definition 1.2 . A barcode is a multiset of intervals. In other words, a barcode is a set B=\{(I_{j},m_{j})\mid I_{j}\text{ is an interval in }[n]\text{ and }m_{j}\in\mathbb{Z}_{\geq 0}\text{ indicates multiplicity}\} We define \textbf{Barc}(n) as the set of barcodes whose intervals are subsets of [n] . Observe that we can associate a Betti curve \beta to a barcode B by letting \beta(i) be the number of intervals in B that contain i . More precisely, let B=\{(I_{j},m_{j})\} . We define the function \varphi\colon\textbf{Barc}(n)\to\textbf{Betti}(n) that associates a Betti curve to a barcode in the following way: \displaystyle\varphi(B) \displaystyle=\beta, \displaystyle\beta(i) \displaystyle=\sum_{j\mid I_{j}\cap\{i\}\neq\varnothing}m_{j}. Note that \varphi is surjective: for \beta=(\beta_{1},\beta_{2},\ldots,\beta_{n}) , we can construct B\in\varphi^{-1}(\beta) by letting B have \beta_{i} copies of [i,i+1) for each i . We define \textbf{Barc}(\beta)\coloneqq\varphi^{-1}(\beta). In other words, \textbf{Barc}(\beta) is the set of barcodes whose Betti curve is \beta ; it characterizes the fiber of the dimension function for all type A_{n} quiver representations. Therefore, our inverse problem is equivalent to the following: Given: a Betti curve \beta\colon[n]\to\mathbb{Z}_{\geq 0} . Enumerate: all barcodes B in \textbf{Barc}(\beta) . For a concrete example of our enumeration problem, if n=3 and one has the Betti curve \beta=(2,3,2) , then there are 13 different barcodes with this Betti curve. Figure 1: A table of all the barcodes with the Betti curve \beta=(2,3,2) . Here each bar is its own interval and bars at the same height are there only for visual convenience. 1.3 Main Contributions We provide a solution to the barcode counting problem in several ways: First, we reinterpret the problem through the lens of Lie algebras, witnessing that the answer to our question in the type A_{n} Lie algebra setting is exactly the Kostant partition function K . Theorem A . If \mu=\sum\limits_{i=1}^{n}\beta_{i}\alpha_{i} is a weight of the Lie algebra of type A_{n} , then |\mathbf{Barc}(\beta)|=K(\mu). In the above, we use \alpha_{i} to denote the simple root \mathbf{e}_{i}-\mathbf{e}_{i+1} in a root system of type A_{n} . The Kostant partition function is a famous object in representation theory and algebraic combinatorics; it counts the number of ways a weight of a Lie algebra can be partitioned into a non-negative integral linear combination of positive roots. The appearance of the Kostant partition function becomes completely intuitive when one recalls Gabriel’s Theorem, which states that the indecomposables of a Dynkin quiver representation are in one-to-one correspondence with the positive roots of the root system for the corresponding Dynkin diagram. Since a persistence module with field coefficients is an A_{n} quiver representation, this relationship between barcodes and combinations of positive roots is unsurprising. In Section 2.3 , we focus on computational aspects of this inverse problem, and provide an explicit recursive formula for enumerating the number of elements of \textbf{Barc}(\beta) . This formula was first observed by Schmidt

Read the complete text