[Paper Review] Twin-width and permutations

Édouard Bonnet, Jaroslav Nešetřil+3 more|arXiv (Cornell University)|Feb 13, 2021

Limits and Structures in Graph Theory4 citations

TL;DR

This paper establishes that a class of binary relational structures has bounded twin-width if and only if it is a first-order transduction of a proper permutation class. The key contribution is a model-theoretic characterization linking twin-width to permutation classes via transductions, showing that all such classes admit at most $2^{O(n)}$ non-isomorphic $n$-vertex graphs.

ABSTRACT

Inspired by a width invariant on permutations defined by Guillemot and Marx, Bonnet, Kim, Thomassé, and Watrigant introduced the twin-width of graphs, which is a parameter describing its structural complexity. This invariant has been further extended to binary structures, in several (basically equivalent) ways. We prove that a class of binary relational structures (that is: edge-colored partially directed graphs) has bounded twin-width if and only if it is a first-order transduction of a~proper permutation class. As a by-product, we show that every class with bounded twin-width contains at most $2^{O(n)}$ pairwise non-isomorphic $n$-vertex graphs.

Motivation & Objective

• To characterize classes of binary relational structures with bounded twin-width using model-theoretic tools.
• To establish a precise logical connection between twin-width and permutation classes through first-order transductions.
• To show that any class with bounded twin-width contains at most $2^{O(n)}$ non-isomorphic $n$-vertex graphs.
• To introduce and formalize the concept of ranked twin-models as a structural tool for analyzing twin-width.
• To unify structural graph theory with permutation pattern theory via transduction equivalence.

Proposed method

• Define twin-width for binary relational structures (edge-colored partially directed graphs) using contraction sequences with red edges tracking errors.
• Introduce ranked twin-models—rooted trees with transversal edges satisfying minimality and consistency—whose optimal width equals the twin-width.
• Use first-order transductions to encode binary structures within permutation classes, leveraging logical interpretations.
• Prove that every class of bounded twin-width arises as a first-order transduction of a proper permutation class.
• Leverage known results: proper permutation classes have bounded twin-width, and transductions preserve bounded twin-width.
• Apply model-theoretic tools (e.g., monadic dependence, transduction equivalence) to analyze structural and logical properties.

Figure 3. From a graph $G$ to a permutation $\sigma$ , and back.

Experimental results

Research questions

• RQ1 What logical and structural conditions characterize classes of binary relational structures with bounded twin-width?
• RQ2 Can all classes of bounded twin-width be expressed as first-order transductions of permutation classes?
• RQ3 What is the maximum number of non-isomorphic $n$-vertex graphs in a class of bounded twin-width?
• RQ4 How do twin-models and their rankings relate to the twin-width of a structure?
• RQ5 What is the relationship between twin-width and permutation pattern avoidance in structural graph theory?

Key findings

• A class of binary relational structures has bounded twin-width if and only if it is a first-order transduction of a proper permutation class.
• Every class with bounded twin-width contains at most $2^{O(n)}$ pairwise non-isomorphic $n$-vertex graphs.
• The twin-width of a structure coincides with the optimal width of a ranked twin-model of that structure.
• Proper permutation classes—defined as those avoiding at least one permutation—have bounded twin-width.
• Transductions preserve bounded twin-width, so any first-order transduction of a proper permutation class also has bounded twin-width.
• The connection between twin-width and permutation classes reveals a deep link between structural graph theory and permutation pattern theory.

Figure 4. A contraction sequence, a so-called block representation of the contractions, and a twin-model.