Metric functionals and weak convergence

Armando W. Gutiérrez, Olavi Nevanlinna · 2025

Read Metric functionals and weak convergence by Armando W. Gutiérrez, Olavi Nevanlinna (2025) 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-2506-04154.txt.

Introduction

Armando Gutierrez and Olavi Nevanlinna define a new notion of weak convergence for plain distance spaces. In standard spaces weak convergence is a core idea, but distance spaces lack the usual linear tools. Their key tools are special real valued maps tied to distance.

They test a sequence against all such maps to say if it converges weakly. For bounded sequences in normed spaces, their notion matches the classic one. That match shows the new idea is a true extension, not a rival. It opens weak methods to a far wider class of spaces.

Catalog record

CREATORArmando W. Gutiérrez, Olavi Nevanlinna
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-04154

Browse the full library · About the Tower of Babel

Full text

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

Metric functionals and weak convergence Armando W. Gutiérrez, Olavi Nevanlinna Abstract We introduce a notion of weak convergence in arbitrary metric spaces. Metric functionals are key in our analysis: weak convergence of sequences in a given metric space is tested against all the metric functionals defined on said space. When restricted to bounded sequences in normed linear spaces, we prove that our notion of weak convergence agrees with the standard one. 1 Introduction A natural question to ask is whether there exists a notion of weak convergence which may be valid in all metric spaces. If such a concept exists, one may naturally expect that it agrees with the standard weak convergence in all normed linear spaces. While investigating certain fixed point problems, T.-C. Lim [ Lim77 ] introduced an interesting notion of convergence in metric spaces, the so-called \Delta- convergence. This concept, however, does not agree with the standard weak convergence in some normed linear spaces. This is discussed in the end of Section 3 . Lim’s work has nevertheless motivated some investigations on weak topologies in CAT( 0 ) spaces. In these spaces, one can define metric projections on compact geodesics. This has led researchers to consider a notion of weak convergence that agrees with the standard one in Hilbert spaces. See the works [ KP08 ] , [ EFL09 ] , [ Bač23 ] , and [ LP23 ] . In this note, we present a novel notion of weak convergence in metric spaces. Our approach uses metric functionals. Definition 1.1 . Let (X,d) be a metric space. We say that a sequence of points (x_{n})_{n\geq 1} in X converges d- weakly to a point z in X if we have \liminf_{n\to\infty}\,\h(x_{n})\geq\h(z), (1.1) for all metric functionals \h on X . We use the notation x_{n}\xrightharpoonup{\;d\;}z . Our concept of d- weak convergence can be applied to nets. We discuss here d- weak convergence of sequences only to simplify the exposition of our ideas. We also want to emphasize that d- weak convergence fully depends on the given metric d . The metric d determines all the corresponding metric functionals \h that are essential to test the inequality ( 1.1 ). The concept of a metric functional is fairly known in some mathematical disciplines. Nonetheless, we recall such a concept and provide new properties in Section 2 . We proceed now to provide evidence that supports our choice of Definition 1.1 as a suitable candidate for weak convergence in metric spaces. The following nontrivial statements concern d- weak convergence of sequences in normed linear spaces. Theorem 1.2 . Let (X,\lVert\cdot\rVert) be a normed linear space. Let d be the metric on X induced by the norm \lVert\cdot\rVert . If a sequence of points (x_{n})_{n\geq 1} in X converges d- weakly, then its limit is unique. The following result shows the connection of d- weak convergence with the standard weak convergence in normed linear spaces. Namely, they agree for bounded sequences in all normed linear spaces. Theorem 1.3 . Let (X,\lVert\cdot\rVert) be a normed linear space. Let d be the metric on X induced by the norm \lVert\cdot\rVert . Suppose that (x_{n})_{n\geq 1} is a bounded sequence in X and z is a point in X . Then, (x_{n})_{n\geq 1} converges d- weakly to z if and only if (x_{n})_{n\geq 1} converges weakly to z . One may wonder whether unbounded sequences in normed linear spaces converge d- weakly. We can prove that this does not happen in some normed linear spaces. Theorem 1.4 . Suppose that X is \ell_{1} , or C[0,1] , or a normed linear space whose dual is strictly convex. If d is the corresponding induced metric, then unbounded sequences in X do not converge d- weakly. Conjecture 1.5 . There is no normed linear space where unbounded sequences converge d- weakly. Without a proof of the above conjecture, it is unclear how d- weak convergence behaves on linear combinations of two d- weakly convergent sequences. Nevertheless, we can offer the following. Theorem 1.6 . Let (X,\lVert\cdot\rVert) be a normed linear space. Let d be the metric on X induced by the norm \lVert\cdot\rVert . Suppose that (x_{n})_{n\geq 1} and (y_{n})_{n\geq 1} are two sequences in X . If there are two vectors u and v in X such that x_{n}\xrightharpoonup{\;d\;}u and d(y_{n},v)\rightarrow 0 , then we have sx_{n}+ty_{n}\xrightharpoonup{\;d\;}su+tv, for all real numbers s and t . All the results presented above examine d- weak convergence of sequences in the whole normed linear space. Proper subsets become legit metric spaces when we equip them with the subspace topology. With this in mind, we can produce simple examples of metric spaces where unbounded sequences converge d- weakly, or where d- weakly convergent sequences have more than one limit. An example where both situations happen is the following. Equip the set X=\{0,e_{1},2e_{2},3e_{3},\dots\} with the metric d that is induced by the norm of \ell_{1} . We can verify that the unbounded sequence (ne_{n})_{n\geq 1} in X converges d- weakly to every point in X . More examples are presented in Section 3 . The following result states that the set of all d- weak limits of a sequence in a W- convex metric space is also W- convex. The concept of W- convexity was introduced by W. Takahashi in [ Tak70 ] . We recall this concept in Definition 2.7 . Briefly we mention that every metric space that admits a conical geodesic bicombing is W- convex. Theorem 1.7 . Let (X,d) be a W- convex metric space. Then, the set of all d- weak limits of a sequence in X is W- convex and closed. 2 Metric functionals 2.1 Where do they appear? Metric functionals have become valuable tools in investigations of iterative processes in metric structures. Fixed point problems in metric spaces are studied by means of metric functionals. Briefly, instead of looking directly for a fixed point in the space, one rather seeks a metric functional that remains (sub)invariant by the iterations of the mapping or family of mappings involved in the problem. These ideas have been put into action in [ GV12 ] , [ LLN18 ] , [ GK21 ] , [ Lin22 ] , [ Kar24 ] , [ GN24 ] . Metric functionals also have been applied to study noncommuting random products. There is always a metric functional that determines the direction along which a random product grows [ GK20 ] . Motivated by this result, A. Avelin and A. Karlsson [ AK22 ] have used metric functionals to study a cut-off phenomenon associated with the dynamics of deep neural networks. Metric functionals have appeared recently in studies on Lipschitz-free spaces [ Meg24 ] and Hermitian symmetric spaces [ CCAL24 ] . 2.2 What are they like? Let (X,d) be a metric space. We choose a basepoint o in X and for each point w in X we consider the funtional \h_{w}(x)=d(x,w)-d(o,w),\;\text{for all}\;x\;\text{in}\;X. (2.1) We equipped the product space \mathbb{R}^{X} with the topology of pointwise convergence and denote by X^{\vee} the subset of \mathbb{R}^{X} containing all the functionals of the form ( 2.1 ). The closure of X^{\vee} is denoted by X^{\diamondsuit} and each element \h in it is

Read the complete text