Continuity of Magnitude at Skew Finite Subsets of ℓ 1 N \ell_{1}^{N}
Sara Kališnik x, Davorin Lešnik · 2026
Read Continuity of Magnitude at Skew Finite Subsets of ℓ 1 N \ell_{1}^{N} by Sara Kališnik x, Davorin Lešnik (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-2603-04271.txt.
Introduction
Kalisnik and Lesnik show magnitude stays continuous at skew finite sets in finite dimensional l1 space. Skew means every coordinate projection takes distinct values on the set. The authors study cubical thickenings of such sets and find explicit formulas for the associated weight measures. Those formulas give each thickening a magnitude that converges to the magnitude of its underlying set. Because skew sets are open and dense among finite subsets, continuity holds on an open dense part of the space. The result covers a large part of the space.
Catalog record
| CREATOR | Sara Kališnik x, Davorin Lešnik |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 72660 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.04271
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Continuity of Magnitude at Skew Finite Subsets of ℓ 1 N \ell_{1}^{N} Sara Kališnik x, Davorin Lešnik Abstract Magnitude is an isometric invariant of metric spaces introduced by Leinster. Although magnitude is nowhere continuous on the Gromov–Hausdorff space of finite metric spaces, continuity results are possible if we restrict the ambient space. In this paper, we focus on ℓ 1 N \ell_{1}^{N} and prove that magnitude is continuous at every skew finite subset of ℓ 1 N \ell_{1}^{N} , that is, at every finite set whose coordinate projections are injective. For such sets, we analyze cubical thickenings and derive an explicit formula for their weight measures. This yields a formula for the magnitude of these thickenings, which we use to prove that their magnitude converges to that of the underlying finite set. Since skew finite subsets of ℓ 1 N \ell_{1}^{N} form an open and dense subset of the space of all finite subsets, magnitude is continuous on an open dense subset of the space of finite subsets of ℓ 1 N \ell_{1}^{N} . 1 Introduction Magnitude is a cardinality-like invariant of metric spaces introduced by Leinster in the 2010s [ 8 ] . Originally arising in category theory as an extension of Euler characteristic to enriched categories, it was soon recognized as a geometric invariant of compact metric spaces [ 8 , 6 , 17 , 16 , 11 ] . For example, it has been shown to encode volume, capacity, dimension, and intrinsic volumes [ 8 , 10 , 17 , 6 ] . It has also found applications in diverse areas: in ecology as a quantitative measure of biodiversity [ 15 , 4 ] , in machine learning as a descriptor of intrinsic diversity in latent representations [ 9 ] , and in topological data analysis through connections with persistent homology [ 12 , 13 , 1 ] . From both theoretical and applied perspectives, understanding the continuity properties of magnitude is of central importance. When one considers the space of isometry classes of finite metric spaces equipped with the Gromov–Hausdorff metric, magnitude fails to be continuous [ 8 , 14 ] . In fact, it is shown in [ 3 ] that magnitude is nowhere continuous on this space. Nevertheless, there are indications that when restricting to suitable subclasses of metric spaces, magnitude is continuous. In this paper, the subclass we focus on are subspaces of \ell_{1}^{N} . There are two reasons for this. The first is that this is an interesting and important class of spaces in its own right, especially in the context of magnitude which interacts particularly nicely with the 1 -metric (for example, the magnitude of the 1 -product of two metric spaces is the product of their magnitudes [ 8 , Proposition 2.3.6] ). The second is that a proof of continuity of magnitude in spaces \ell_{1}^{N} may allow for generalization to a far greater class of spaces. Recall that in [ 7 , Theorem 3.1] , Leinster and Meckes prove the one-point property (essentially a special case of continuity) for all metric spaces that can be isometrically embedded into a finite-dimensional vector subspace of L_{1} . Their result covers many of the one-point property results that appear in the literature for specific classes of metric spaces (such as Euclidean spaces). The proof proceeds by reducing the problem to the one-point property of subspaces of \ell_{1}^{N} . It is our hope that something similar could be done for continuity in general, i.e. that proving (a sufficiently strong form of) continuity of magnitude in all \ell_{1}^{N} would imply continuity in all finite-dimensional subspaces of L_{1} . Our first contribution is a general criterion that reduces continuity of magnitude at a compact set to a one-sided estimate along metric thickenings. More precisely, in any tractable metric space M (in the sense of [ 2 ] ), continuity of magnitude at compact K\subseteq M is equivalent to the convergence \lim_{r\mathop{\text{\tiny$\searrow$}}0}\mathsf{mag}\big(\overline{\mathrm{th}}_{M}\!\left(K,r\right)\big)=\mathsf{mag}(K) and, equivalently, to the existence of arbitrarily small r>0 with \mathsf{mag}\big(\overline{\mathrm{th}}_{M}\!\left(K,r\right)\big) arbitrarily close to \mathsf{mag}(K) from above (Lemma 3.2 ). We then specialize to M=\ell_{1}^{N} , where it is technically convenient to work with cubical thickenings in the \infty -metric: for p\in\ell_{1}^{N} and r>0 we write \overline{C}_{p}(r):=\overline{B}_{\infty}\!\left(p,r\right) and, for a finite set F\subseteq\ell_{1}^{N} , we set \overline{C}_{F}(r):=\bigcup_{p\in F}\overline{C}_{p}(r) . Our main technical result is an explicit description of the weight measure of \overline{C}_{F}(r) when F is skew , i.e. when the coordinate projections \mathrm{pr}_{k}|_{F} are injective. In this case, for all sufficiently small r the cubes in \overline{C}_{F}(r) have pairwise disjoint coordinate projections, and we prove that \overline{C}_{F}(r) admits a weight measure of the form \omega_{\overline{C}_{F}(r)}\ =\ \Big(\tfrac{1}{2^{N}}\sum_{D=0}^{N}\lambda_{\overline{C}_{F}(r)^{(D)}}^{D}\Big)\ -\!\!\sum_{\begin{subarray}{c}p\in F\\ s\in\left\{{-1,1}\right\}^{N}\end{subarray}}\!\!\!\!\alpha_{p,s}(r)\,\delta_{p+rs}, where the coefficients \alpha_{p,s}(r) are determined by a linear system of equations indexed by the vertices of \overline{C}_{F}(r) (Theorem 5.1 ), \overline{C}_{F}(r)^{(D)} denotes the union of the D -dimensional faces of the cubes in \overline{C}_{F}(r) , \lambda_{\overline{C}_{F}(r)^{(D)}}^{D} is the D -dimensional Lebesgue measure restricted to these faces, and \delta_{p+rs} denotes the Dirac measure at the vertex p+rs . This yields an explicit formula for \mathsf{mag}\big(\overline{C}_{F}(r)\big) . Using this cube weight measure formula, we deduce that \lim_{r\mathop{\text{\tiny$\searrow$}}0}\mathsf{mag}\big(\overline{C}_{F}(r)\big)=\mathsf{mag}(F)\qquad\text{for every skew finite $F\subseteq\ell_{1}^{N}$}. Hence, magnitude in \ell_{1}^{N} is continuous at every skew finite subset (Theorem 6.4 ). Since skew finite subsets of \ell_{1}^{N} form an open and dense subset in the set of all finite subsets of \ell_{1}^{N} , the finite space version of magnitude in \ell_{1}^{N} is continuous “almost everywhere”. 1.1 Notation We denote the set of natural numbers by \mathbb{N} . We treat 0 as a natural number, so \mathbb{N}=\left\{{0,1,2,3,\ldots}\right\} . We denote the set of real numbers by \mathbb{R} , and the set of extended real numbers (with infinities included) by \overline{\mathbb{R}} . That is, \overline{\mathbb{R}}=\mathbb{R}\cup\left\{{-\infty,\infty}\right\} . Subsets, given by a relation, are denoted by that relation in the index; for example, \mathbb{R}_{\geq 0} is the set of non-negative real numbers. Intervals between two numbers are denoted by these two numbers in brackets and in the index. Round, or open, brackets (\ ) denote the absence of the boundary in the set, and square, or closed, brackets [\ ] its presence. For example, {\mathbb{R}}_{[0,1]}=\left\{{x\in\mathbb{R}}\;\middle|\;{0\leq x\leq 1}\right\} is the usual closed unit interval, and {\mathbb{N}}_{[5,10)}=\left\{{n\in\mathbb{N}}\;\middle|\;{5\leq n<10}\right\}=\left\{{5,6,7,8,9}\right\} . For any set A , we denote its cardinality by \mathnormal{\#}{A} . The powerset of a set A (the set of all subsets of A ) is denoted by \mathcal{P}(A) . We denote the set of finite subsets of A by \mathrm{Fin}(A) , and the set of non-empty finite subsets by \mathrm{Fin}_{+}(A) . If A is also equipped with a topology, then \mathrm{Cmp}(A) denotes the set of compact subspaces of A , and \mathrm{Cmp}_{+}(A) the set of non-empty compact subspaces of A . A function f mapping from a set A to a set B is denoted as f\colon A\to B . If f is merely a partial map (not necessarily defined on the whole A ), we write this as f\colon A\mathrel{\rightharpoonup}B . Following the notation from [ 7 ] , for any p\in\overline{\mathbb{R}}_{\geq 1} and N\in\mathbb{N} , let \ell_{p}^{N} denote the Banach space \mathbb{R}^{N} , equipped with the p -norm (and the induced p -metric), and let \ell_{p} denote the infinite-dimensional version of this space.