🔢 mathematics

What is Connectivity?

This paper presents a comprehensive taxonomy of connectivity for space-like structures, inspired by the study of posets of connected pieces and their embeddings, which unifies standard notions of connectivity across point-set and point-free contexts including graphs, topology, and frames.

Original authors: Jean F. Du Plessis, Zurab Janelidze, Bernardus A. Wessels

Published 2026-05-11

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Jean F. Du Plessis, Zurab Janelidze, Bernardus A. Wessels

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Question: What Does It Mean to Be "Connected"?

Imagine you are looking at a map of a city. Some parts are neighborhoods where you can walk from any house to any other house without leaving the area. We call these connected. Other parts are islands separated by water; you can't walk between them. We call these disconnected.

In mathematics, specifically in a field called topology (the study of shapes and spaces), mathematicians have spent decades trying to define exactly what "connected" means. They have definitions for:

Graphs: Can you walk from one dot to another along lines?
Topological Spaces: Can you draw a continuous line between two points without lifting your pen?
Digital Images: Are the pixels of the same color touching each other?

The problem is that all these definitions feel slightly different. This paper asks: Is there one single, universal rule that covers all of them?

The authors say "Yes," but to find it, they had to invent a new way of looking at the problem.

The New Tool: The "Chainmail"

To solve this, the authors introduce a concept they call a Chainmail.

The Analogy:
Imagine a medieval knight's armor made of interlocking metal rings.

If you pull on one ring, it might tug on its neighbors.
If you have a group of rings that are all linked together in a chain (Ring A touches Ring B, Ring B touches Ring C), they form a single, sturdy unit.
If you try to pull that whole group apart, you can't separate them without breaking a link.

In the paper's math language:

A "Mail" is a group of things that all share a common "bottom" or connection point.
A "Mail-Connected" group is a chain of these things where you can walk from any item to any other item by hopping through shared connections.
A "Chainmail" is a system where, if you have a group of things linked together like this, their "sum" (or union) is guaranteed to exist and be a single, valid thing.

Why is this cool?
The authors prove that every standard way of defining connectivity (in graphs, in maps, in digital images) creates a structure that behaves exactly like a Chainmail. If you take all the "connected" pieces of a system and look at how they fit together, they always form a Chainmail.

The Two Sides of the Coin: The "Exterior" and the "Lattice"

The paper sets up a relationship between two worlds:

The Chainmail (The Pieces): This is the collection of all the "connected" islands.
The Lattice (The Whole Space): This is the entire universe of shapes, including the broken ones, the disconnected ones, and the connected ones.

The authors discovered a magical bridge between these two worlds called a Galois Connection (don't worry, it's just a fancy word for a perfect matching system).

Going Up (The Join): If you take a bunch of disconnected islands and smash them together, you get a bigger shape.
Going Down (The Components): If you take a big, messy shape, you can break it down into its fundamental "connected components" (the islands that make it up).

The paper shows that for a system to be a valid "Connectivity System," this breaking-down process must work perfectly. You must always be able to take a shape, break it into its unique connected islands, and then put those islands back together to get the exact same shape.

The "Absolute" Truth

The paper identifies a special, perfect type of connectivity called an Absolute Connectivity Lattice.

The Analogy:
Imagine a puzzle.

In a normal puzzle, you might have a picture where two pieces look like they fit, but when you put them together, the picture changes slightly.
In an Absolute puzzle, the pieces are so perfect that the picture is exactly the sum of its parts, and the parts are exactly what make up the picture. There is no loss of information.

The authors prove a stunning result:
There is a perfect, one-to-one match between "Absolute Connectivity Lattices" and "Chainmails."

This means:

If you have a list of all possible "connected islands" (a Chainmail), you can automatically build the perfect "universe" (the Lattice) that contains them.
If you have a perfect "universe" of shapes, you can automatically extract the list of all "connected islands."

They are two sides of the same coin.

The Taxonomy (The Family Tree)

The paper doesn't just stop at the perfect case. It creates a Taxonomy (a family tree) of all the different ways connectivity can work.

Think of it like sorting animals:

Mammals: All have hair and milk.
Reptiles: All have scales and lay eggs.

The authors sort all the different mathematical definitions of connectivity into groups based on how "perfect" they are:

Pre-connectivity: The loosest group. Just a collection of things that might be connected.
Connectivity: Things that can be broken down into components.
Typical/Serra Connectivity: Things where the components can be put back together perfectly to rebuild the original.
Absolute Connectivity: The "Gold Standard" where the system is perfectly reversible and unique.

They show that famous examples fit into specific branches:

Graphs (dots and lines) fit in one branch.
Topological spaces (rubber sheet geometry) fit in another.
Digital images (pixels) fit in yet another.

The Answer to "What is Connectivity?"

The title asks, "What is Connectivity?"

The paper's answer is: Connectivity is simply a subset of a system that behaves like a Chainmail.

If you have a collection of "connected" things, and that collection follows the rules of a Chainmail (where linked groups always have a valid sum), then you have a valid definition of connectivity.

Summary in One Sentence

The authors built a universal mathematical "skeleton" (the Chainmail) that fits perfectly inside any system of connected things, proving that whether you are studying a map, a computer image, or a mathematical shape, the rules of "being connected" are all variations of the same fundamental structure.

Title: What is Connectivity?
Authors: Jean F. Du Plessis, Zurab Janelidze, and Bernardus A. Wessels
Source: MIT-CTP/5835 (arXiv:2501.19226v2)

Problem Statement

The paper addresses the need for a unified, general theory of connectivity that encompasses both point-set and point-free contexts. While point-free topology generalizes topological spaces to frames (complete lattices), existing general theories of connectivity largely remain point-set based, treating connected objects as subsets of a given set. Although R. Börger proposed a categorical theory of connectivity and J. Serra introduced a point-free approach using general lattices (motivated by digital imaging), a comprehensive framework that systematically compares and classifies various notions of connectivity (such as graph connectivity, path-connectivity, and $k$ -connectivity) within a single lattice-theoretic structure was lacking. The authors aim to establish a foundational taxonomy for connectivity that does not rely on the broader categorical machinery of Börger but specializes it to posets.

Methodology

The authors develop a theory anchored in the concept of a connectivity lattice, defined as a pair $(L, \mathcal{C})$ , where $L$ is a complete lattice and $\mathcal{C} \subseteq L$ is a subset of "connected elements." The methodology proceeds through the following steps:

Axiomatic Definition: The authors impose a single axiom on $(L, \mathcal{C})$ : every connected element below an element $x \in L$ must be below a unique maximal connected element below $x$ . This ensures that every element can be decomposed into "connected components."
Galois Connections: They analyze the structure via a Galois connection (termed a connectivity adjunction) between the lattice $L$ and the poset $D(\mathcal{C})$ of "totally mail-disconnected" subsets of $\mathcal{C}$ . The left adjoint maps a set to its join, while the right adjoint maps an element to its set of connected components.
Introduction of Chainmails: The authors introduce the concept of a chainmail, a poset where every "mail-connected" subset has a join. A "mail" is a non-empty set of elements with a common lower bound; a set is mail-connected if any two elements can be linked by a sequence of overlapping mails.
Taxonomic Classification: By varying the properties of the connectivity adjunction (e.g., whether it is an isomorphism, whether it preserves specific elements like the bottom element, or whether it satisfies specific closure conditions), the authors classify different types of connectivity lattices (e.g., kernel, typical, Serra, separated, absolute).

Key Contributions and Results

1. The Concept of Chainmails and Property (C)
The paper defines a chainmail as a poset where every mail-connected subset has a join. This is shown to be equivalent to the condition that for every element $x$ , the upper set $x^\uparrow$ is a complete lattice. The authors prove that the poset of connected elements in any connectivity lattice forms a chainmail.

2. Characterization of Connectivity via Chainmails
A central result is that a subset $\mathcal{C}$ of a complete lattice $L$ is a connectivity if and only if $\mathcal{C}$ is a subchainmail of $L$ (closed under joins of mail-connected subsets). This provides a concise answer to the title question: "Connectivity is the property of a subset being a subchainmail."

3. The Connectivity Adjunction and Classification
The paper establishes that the existence of a specific Galois connection between $L$ and $D(\mathcal{C})$ is a characteristic property of connectivity lattices. Based on the behavior of this adjunction, the authors define a taxonomy:

Kernel Connectivity: The bottom element $0$ is in $\mathcal{C}$ .
Typical Connectivity: $0 \notin \mathcal{C}$ , but elements decompose into components.
Serra Connectivity: A typical connectivity where every element is the join of its components (satisfying condition CL2).
Separated Connectivity: The right adjoint is a left inverse of the left adjoint.
Absolute Connectivity: The connectivity adjunction is an isomorphism.

4. Correspondence with Chainmails
The paper proves a one-to-one correspondence between isomorphism classes of absolute connectivity lattices and isomorphism classes of chainmails.

The mapping $(L, \mathcal{C}) \mapsto \mathcal{C}$ sends an absolute connectivity lattice to its chainmail of connected elements.
The inverse mapping sends a chainmail $\Gamma$ to its exterior $D(\Gamma)$ (the lattice of totally mail-disconnected sets), which forms an absolute connectivity lattice.
This result implies that chainmails serve as "universal posets of connected objects."

5. Absolute Connectedness and Frames
The authors define absolutely connected elements via four equivalent conditions (E1–E4) in the context of frames (distributive lattices). They show that in a frame, the set of absolutely connected elements coincides with the set of connected elements in the usual topological sense. Consequently, a frame is locally connected if and only if it is an absolute connectivity lattice.

6. Examples and Scope
The theory successfully unifies standard notions:

Graphs and Hypergraphs: The poset of connected subsets forms a chainmail.
Topology: Connected and path-connected sets in topological spaces form chainmails.
Digital Imaging: The framework accommodates non-binary images via general lattices, extending Serra's work.
Counter-examples: The paper provides examples of connectivity lattices that are not frames (e.g., certain non-distributive lattices derived from specific chainmails), demonstrating that the theory is strictly more general than local connectedness in frames.

Significance and Claims

The paper claims to provide a foundational taxonomy for connectivity that bridges point-set and point-free approaches without requiring the full machinery of category theory. By isolating the structural properties of the "ambient" lattice and the "connected" subset, the authors demonstrate that:

Connectivity can be rigorously defined via the existence of a Galois connection (connectivity adjunction).
Chainmails are the fundamental building blocks; any connectivity lattice is essentially an embedding of a chainmail into a complete lattice.
Absolute connectivity represents a "universal" form of connectivity, where the structure of the lattice is entirely determined by its connected components.

The authors explicitly state that their work is a specialization of Börger's categorical theory to posets and that extending their results to the full categorical context is left for future work. They do not claim to solve open problems in topology or computer science directly but rather to provide the structural language and classification system necessary to analyze such problems across different mathematical domains. The paper concludes by noting that the taxonomy of connectivity lattices forms a complete lattice itself, offering a structured way to compare different notions of connectivity.