What is connectivity?

Jean F. Du Plessis, Zurab Janelidze, Bernardus A. Wessels · 2025

Read What is connectivity? by Jean F. Du Plessis, Zurab Janelidze, Bernardus A. Wessels (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-2501-19226.txt.

Introduction

Du Plessis, Janelidze, and Wessels posed a basic question in this mathematics paper from twenty twenty-five: what is connectivity? Many fields use the word, but each defines it in its own way. Graphs, spaces, and logical frames all speak of connected pieces. The authors set out to build one taxonomy that covers them all.

Their approach starts with the set of connected pieces of a space, ordered by inclusion. They then study how that set sits inside the wider structure. The resulting scheme includes standard notions from point-set and point-free topology. It also covers connectivity in graphs and hypergraphs, plus stronger forms such as k-connectivity. Even elements of a frame fit the scheme. The paper offers a shared language for a scattered concept.

Catalog record

CREATORJean F. Du Plessis, Zurab Janelidze, Bernardus A. Wessels
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-19226

Browse the full library · About the Tower of Babel

Full text

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

What is connectivity? Jean F. Du Plessis, Zurab Janelidze, Bernardus A. Wessels Abstract . In this paper, we explore a taxonomy of connectivity for space-like structures. It is inspired by isolating posets of connected pieces of a space and examining its embedding in the ambient space. The taxonomy includes in its scope all standard notions of connectivity in point-set and point-free contexts, such as connectivity in graphs and hypergraphs (as well as k k -connectivity in graphs), connectivity and path-connectivity in topology, and connectivity of elements in a frame. MIT-CTP/5835 2020 Mathematics Subject Classification 54D05, 06A06, 06B23, 18A40, 06D22 Introduction Point-free topology generalizes various constructions, properties, and results about topological spaces to frames. The generalization is given by replacing the frame of open subsets of a topological space, ordered under subset inclusion, with an abstract frame — a complete lattice where finite meets distribute over infinite joins. See [ 15 ] for an overview of this theory. There is a line of work in the general study of connectivity, which takes a slightly different approach to generalizing connected sets than how point-free topology generalizes open sets. Instead of extracting the poset of connected objects and discarding the ambient structure, one abstracts the ambient structure to a complete lattice of its ‘pieces’ (whether connected or not) and isolates a subset of connected elements in the lattice. In general, this subset is not determined uniquely by the lattice. Instead, one is interested in different kinds of such subsets, each giving rise to a notion of connectivity. See [ 19 , 14 , 10 , 3 , 12 , 13 , 17 , 8 , 18 ] for such an approach to a general theory of connectivity. In most of these works, the ambient complete lattice is the powerset of the underlying set of a mathematical structure in which one is interested to study connectivity. So the existing general theory of connectivity is largely a point-set theory: connected objects are subsets of a given set whose elements represent ‘points’ of a mathematical structure. A move to a point-free theory is made by J. Serra in [ 17 ] . In his approach, the lattice no longer need be a powerset lattice; this is motivated by applications to digital imaging, where general lattices allow inclusion of applications dealing with multicolored images. Almost ten years earlier, in [ 4 ] , shortly after his pioneering work on general point-set connectivity [ 3 ] , R. Börger proposes a general theory of connectivity in categories — thus, a ‘higher’ point-free theory. However, in [ 4 ] , there is essentially a single general theorem, and the special case in which the categories are posets is not analyzed (instead, other significant categorical examples are explored). In this paper we build on Serra’s work on point-free connectivity and explore some foundational questions for a general theory whose aim is to study and compare different point-set and point-free notions of connectivity (without considering the broader, categorical situations). Our findings suggest that such a theory can be anchored at what we call a ‘connectivity lattice’ — potentially the most general notion of point-free connectivity. This context for connectivity turns out to be what you get when you suitably specialize Börger’s categorical theory to posets. Extension of our work to categories is left for the future. A connectivity lattice is defined as a pair (L,\mathcal{C}) , where L is a complete lattice and \mathcal{C} is a subset of L (called a connectivity in L ), whose elements are called connected elements of L . We impose only one axiom on this structure: every connected element below an element x\in L is always below a unique maximal connected element below x . This condition essentially says that every element can be decomposed into ‘connected components’; such decomposition gives rise to a Galois connection \mathcal{D}(\mathcal{C})\rightleftarrows L, where \mathcal{D}(\mathcal{C}) denotes the poset of totally mail-disconnected subsets S of \mathcal{C} . By mail-connectedness of a subset of \mathcal{C} we mean connectedness relative to the graph structure on \mathcal{C} where an edge between two elements of \mathcal{C} represents that these elements have a common lower bound in \mathcal{C} (see Figure 1 ). So ‘totally mail-disconnected’ refers to the case when no two distinct elements have a common lower bound. We call these Galois connections connectivity adjunctions . The left adjoint maps an element S of \mathcal{D}(\mathcal{C}) to its join \bigvee S , while the right adjoint assigns to an element x\in L the set \mathcal{C}(x) of its connected components , i.e., maximal connected elements below x . In fact, existence of such Galois connection turns out to be a characteristic property of connectivity lattices. Now, Serra’s notion of connectivity from [ 17 ] can then be characterized by further requiring that every element of L and the bottom element of \mathcal{D}(\mathcal{C}) are Galois closed for the connectivity adjunction. The approach to connectivity through these Galois connections, as far as we are aware, is new to the field of point-free connectivity. It helps to uncover what seems to be an interesting conceptual taxonomy of connectivity (see Figure 7 ). Figure 1. Venn diagram visualization of the notion of mail-connectedness in the poset of subsets of a given set. The set of regions marked by the circles is mail-connected, because any two regions can be linked by a sequence of overlapping regions. One of these taxonomic groups is given by the case when the connectivity adjunction is an isomorphism. In this case, we call the corresponding connectivity lattice (L,\mathcal{C}) an absolute connectivity lattice. A striking property of absolute connectivity lattices is that for each L there can be at most one compatible \mathcal{C} . For a frame L of open sets in a locally connected topological space, \mathcal{C} turns out to be the set of open connected sets. More generally, a frame together with its set of connected elements forms an absolute connectivity lattice. This includes all complete atomic Boolean algebras, in which case \mathcal{C} is the set of atoms. A more primitive class of absolute connectivity lattices is given by adjoining to a complete lattice a new bottom element. In this case, \mathcal{C} is the set of old elements. A culminating result in the paper is that isomorphism classes of absolute connectivity lattices are in one-to-one correspondence with isomorphism classes of posets having the following property: (C) Every mail-connected subset of the poset has a join. This property can be viewed as a form of ‘local completeness’ of a poset. We can make this precise: as we prove in this paper, a poset has the property (C) if and only if for any element x in the poset, x^{\uparrow} is a complete lattice. Thus, a poset has the property (C) if and only if the dual poset, viewed as a category, is locally complete in the sense of [ 9 ] (i.e., the subobject posets are complete lattices). The term ‘locally complete’ is used in the latter sense for preorders in [ 6 ] : a poset has the property (C) above if and only if its dual is a locally complete preorder in the sense of [ 6 ] . We avoid the use of the

Read the complete text