Uniform Property (S

William B. Johnson, Tomasz Kania · 2026

Read Uniform Property (S by William B. Johnson, Tomasz Kania (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-2602-09106.txt.

Introduction

William B. Johnson and one co-author introduce a quantitative version of Steinhaus' property (S) for Banach spaces, called uniform property (S). A space has the property when, for any two distinct unit vectors and any perturbation size, the largest gap in perturbed norms stays above a positive function. That function depends on perturbation size and vector separation. The modulus is computed exactly for atomless L-one spaces.

The class proves stable under ultrapowers and Bochner L-one constructions, and includes all Gurariĭ spaces and Banach lattices of almost universal disposition. Each Banach space fits isometrically inside a non-strictly convex space of the same density with uniform property (S). An explicit renorming of l-one spaces is given, settling several open questions in ZFC about property (S) geometry.

Catalog record

CREATORWilliam B. Johnson, Tomasz Kania
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-09106

Browse the full library · About the Tower of Babel

Full text

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

Uniform Property (S) Thanks: IM CAS (RVO 67985840). William B. Johnson, Tomasz Kania Abstract . We introduce and investigate a quantitative version of Steinhaus’ property ( S ) (S) for Banach spaces, called the uniform property ( S ) (S) . A Banach space X X is said to have uniform ( S ) (S) if for every pair of distinct unit vectors x , y ∈ X x,y\in X and every a > 0 a>0 , the difference of the perturbed norms sup ‖ z ‖ ⩽ a | ‖ x + z ‖ − ‖ y + z ‖ | \sup_{\|z\|\leqslant a}\big|\|x+z\|-\|y+z\|\big| is bounded below by a positive function of a a and ‖ x − y ‖ \|x-y\| . We compute this modulus exactly for the spaces L 1 ​ ( μ ) L_{1}(\mu) with atomless measure μ \mu , U L 1 ​ ( μ ) ​ ( d , a ) = ( 4 ​ a 2 + d ∧ 1 ) ​ d , U_{L_{1}(\mu)}(d;a)=\Big(\tfrac{4a}{2+d}\wedge 1\Big)d, The class of spaces with uniform ( S ) (S) is stable under ultrapowers, Bochner- L 1 L_{1} constructions, and contains all Gurariĭ spaces as well as Banach lattices of almost universal disposition. In particular, every Banach space embeds isometrically into a non-strictly convex Banach space of the same density having uniform ( S ) (S) . We further exhibit an explicit equivalent renorming of ℓ 1 ​ ( Γ ) \ell_{1}(\Gamma) , ‖ x ‖ S = ( ‖ x ‖ 1 2 + ‖ x ‖ 2 2 ) 1 / 2 , \|x\|_{S}=\big(\|x\|_{1}^{2}+\|x\|_{2}^{2}\big)^{1/2}, which endows ℓ 1 ​ ( Γ ) \ell_{1}(\Gamma) and all its ultrapowers with uniform ( S ) (S) . These results settle, in ZFC, several open questions about the quantitative geometry of property ( S ) (S) posed by Kochanek and the second-named author. 2010 Mathematics Subject Classification 46B04, 46B20 (primary), 11H06 (secondary). 1. Introduction It is a familiar observation from plane geometry that for every positive integer n , it is possible to encircle precisely n points with integer coordinates in the Euclidean plane (this seems to be first mentioned by Steinhaus in 1957 as problem 498 in volume 10 of Matematyka: Czasopismo dla Nauczycieli , a journal for Mathematics teachers in Poland, and later also as Problem 24 in [ 21 , p. 17] . In May 1967, Steinhaus put in the New Scottish Book (that revived the traditions of the Lvov Scottish Book in Wrocław) problem 782 of whether every measurable subset of the plane that has infinite Lebesgue measure can be moved so that the intersection of the result set with the integer lattice is infinite. Worth mentioning is also a result from plane geometry by Fast and Świerczkowski [ 10 ] , which asserts that for any open bounded region G in the Euclidean plane with a simple closed curve as its boundary and an integer-valued area, there exists an (isometric copy of the) integral lattice sharing precisely |G| points with G . The possibility of encircling any number of lattice points, mentioned by Steinhaus, was generalised by Zwoleński [ 23 ] to arbitrary Hilbert spaces as follows: for every infinite subset A in a Hilbert space X with a finite intersection with each open ball, there exists a dense subset Y of X such that for all y\in Y and n\in\mathbb{N} there is an open ball B centred at y with |A\cap B|=n . Kochanek and the second-named author [ 16 ] expanded beyond Hilbert spaces, characterising geometrically Banach spaces for which Zwoleński’s assertion remains valid; these were termed spaces having property (S) after Steinhaus. In this paper, we adopt one of the equivalent geometric conditions enumerated in [ 16 , Theorem A] as the definition of spaces with property (S) . Definition 1.1 . A normed space X has property (S) if, for every distinct points x and y in the unit sphere of X and every a>0 , there exists a vector z in X with \|z\|<a such that \|x+z\|\neq\|y+z\| . This definition includes possibly incomplete spaces; however, completeness is necessary to establish the equivalence between the various properties enumerated in [ 16 , Theorem A] . (In [ 16 ] , the above-introduced property was called property (S^{\prime\prime}) , while a metric space X was said to have property (S) if, for any infinite subset A of X with a finite intersection with every ball, there exists a dense subset Y of X such that for all y\in Y and n\in\mathbb{N} there is a ball B centred at y with |A\cap B|=n .) The goal of this paper is to quantify property (S) , specifically to introduce uniform property (S) , and to demonstrate that various well-known Banach spaces possess uniform property (S) , or at least can be equivalently renormed to have uniform property (S) . In the process of doing this, we solve a problem about property (S) that was raised in [ 16 ] . Let X be a normed space. For x,y\in X and a>0 , we define the (S) - modulus by (1) U_{X}(x,y;a)=U(x,y;a):=\sup_{\|z\|\leqslant a}\big|\|x+z\|-\|y+z\|\big|. For 0<d<2 and a>0 we define (2) U_{X}(d;a)=U(d;a):=\inf_{\overset{x,y\in S_{X},}{\|x-y\|\geqslant d}}U(x,y;a). Definition 1.2 . A Banach space X has uniform property (S) provided that for all 0<d<2,a>0 one has U_{X}(d;a)>0. It is evident that uniform property (S) implies property (S) , and a standard compactness argument shows that if X is finite-dimensional and has property (S) , then it also has uniform property (S) . Uniform property (S) is arguably more manageable than property (S) because the (S) -modulus, U(d;a) , is stable under passing to completions and ultrapowers. Moreover, U(x,y;a) changes only slightly if x and y are perturbed, provided the perturbation is small relative to d and a . The stability of U(d;a) will be used in the sequel without further comment. The modulus of uniform property (S), as defined, may be interpreted as the p=1 case of the whole scale of moduli whose strict positivity is equivalent to uniform property (S). Indeed, for a normed space X and p\in[1,\infty) we set (3) U_{X}^{(p)}(d;a)=U^{(p)}(d;a):=\inf_{\begin{subarray}{c}x,y\in S_{X}\\ \|x-y\|\geqslant d\end{subarray}}U^{(p)}(x,y;a)\quad(d\in(0,2),\ a>0), where (4) U_{X}^{(p)}(x,y;a)=U^{(p)}(x,y;a):=\sup_{\|z\|\leqslant a}\big|\|x+z\|^{p}-\|y+z\|^{p}\big|\quad(x,y\in X,\ a>0). Certainly, U_{X}^{(1)}(d;a)=U_{X}(d;a) and for p>1 we have U_{X}^{(p)}(d;a)>0 if and only if U_{X}(d;a)>0 , which follows (for small enough a ) from the elementary fact: Lemma 1.3 . Fix p>1 and a\in(0,1) . If r,s\in[1-a,1+a] , then by the mean value theorem p(1-a)^{p-1}\,|r-s|\ \leqslant\ |r^{p}-s^{p}|\ \leqslant\ p(1+a)^{p-1}\,|r-s|. Consequently, for every normed space X and every a\in(0,1) we have, for 0<d<2 , p(1-a)^{p-1}\,U_{X}(d;a)\ \leqslant\ U_{X}^{(p)}(d;a)\ \leqslant\ p(1+a)^{p-1}\,U_{X}(d;a). In particular, for any a\in(0,1) and p>1 , U_{X}^{(p)}(d;a)>0 if and only if U_{X}(d;a)>0 . Proof. Apply the one-variable mean value theorem to t\mapsto t^{p} on [1-a,1+a] and note that the intermediate value \xi lies in [1-a,1+a] . ∎ Let us illustrate it briefly with inner-product spaces. Proposition 1.4 . Let X be an inner-product space of dimension at least 2 . Then U_{X}^{(2)}(d;a)=2ad for a>0 and d\in(0,2) . Proof. Let x,y\in S_{X} and z with \|z\|\leqslant a . We have \|x+z\|^{2}=\|x\|^{2}+\|z\|^{2}+2\langle x,z\rangle and \|y+z\|^{2}=\|y\|^{2}+\|z\|^{2}+2\langle y,z\rangle . Since x and y are unit vectors, we have \big|\|x+z\|^{2}-\|y+z\|^{2}\big|=2\big|\langle x-y,z\rangle\big|. Maximising over all

Read the complete text