← Latest papers
🔢 mathematics

Graph Puzzles III.1: A Proof of Sabidussi's Compatibility Conjecture

This paper proves Sabidussi's compatibility conjecture by demonstrating that in any finite connected multigraph with even degrees of at least four, the edges can be partitioned into circuits (and even four-colored) such that no circuit contains two edges that appear consecutively in a given Eulerian trail.

Original authors: Nikolay Ulyanov

Published 2026-07-16
📖 1 min read🧠 Deep dive

Original authors: Nikolay Ulyanov

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

Technical Summary: A Proof of Sabidussi's Compatibility Conjecture

Problem Statement
The paper addresses Sabidussi's compatibility conjecture within the context of finite connected multigraphs. Specifically, it considers an Eulerian multigraph GG (where every vertex has even degree) with a minimum degree δ(G)4\delta(G) \geq 4. Given a closed trail TT that traverses every edge exactly once (an Euler tour), the problem asks whether the edges of GG can be partitioned into circuits (connected 2-regular subgraphs) such that no circuit contains two edges that appear consecutively in TT.

In the language of transition systems, an Euler tour induces a pairing of half-edges at each vertex. A circuit decomposition is "compatible" if no circuit pairs half-edges that are prescribed as a transition by the tour. The conjecture asserts that such a compatible decomposition always exists under the given degree constraints.

Methodology
The proof proceeds through a reduction from the graph-theoretic problem to a combinatorial problem involving cyclic words, followed by an algebraic construction using parity arguments over the field F2\mathbb{F}_2.

  1. Reduction to Cyclic Words:
    The authors define a cyclic word w=(v0,v1,,vm1)w = (v_0, v_1, \dots, v_{m-1}) representing the sequence of vertices visited by the Euler tour TT. The edges of the tour correspond to "gaps" between these letters. The problem is reformulated as finding a coloring of these gaps with elements from F22\mathbb{F}_2^2 (a 4-coloring) such that:

    • Adjacent gaps (corresponding to consecutive edges in the tour) receive different colors.
    • For every vertex vv in the graph, the colors assigned to the gaps incident to occurrences of vv satisfy a parity condition: each color appears an even number of times among the gap incidences.
  2. Algebraic Framework:
    The core of the proof relies on two lemmas established in Section 3:

    • Lemma 3.1 (Four-colour parity): A family of elements in F22\mathbb{F}_2^2 contains each element an even number of times if and only if their linear sum is zero and their quadratic sum (defined via a specific bilinear form q(x)=x1x2q(x) = x_1x_2) is zero.
    • Lemma 3.2 (Three-state balancing): A global selection principle stating that for a finite set UU and a three-element set Σ\Sigma, if certain symmetry and zero-sum conditions are met by a function β\beta, the number of assignments satisfying a system of local constraints is odd (and thus non-zero).
  3. Construction of the Coloring:
    The proof constructs the required gap coloring by:

    • Defining "local patterns" Δa,t\Delta_{a,t} for each letter aa in the cyclic word, which assign non-zero values in F22\mathbb{F}_2^2 to the occurrences of aa such that their sum is zero.
    • Defining interaction terms βab\beta_{ab} between distinct letters based on the order of their occurrences in the word.
    • Applying Lemma 3.2 to select a specific state taΩt_a \in \Omega (where Ω=F22{0}\Omega = \mathbb{F}_2^2 \setminus \{0\}) for each letter aa. This selection ensures that the interaction constraints vanish.
    • Using these selections to define a sequence yiy_i (differences between gap colors) and integrating them to recover the gap colors xix_i.
    • Verifying that the resulting coloring satisfies the even-degree condition for each color class at every vertex by showing that the sum of colors and the sum of their quadratic forms vanish, invoking Lemma 3.1.

Key Contributions and Results

  • Theorem 1.1: The paper proves that for any finite Eulerian multigraph with minimum degree at least 4 and any Euler tour TT, there exists a coloring χ:E(G)F22\chi: E(G) \to \mathbb{F}_2^2 such that consecutive edges in TT have different colors, and every vertex has even degree in each color class.
  • Corollary 1.2: As a consequence, the graph GG admits a circuit decomposition compatible with the transition system induced by TT.
  • Improvement on Cycle Double Covers: The paper notes that in the presence of a dominating circuit, the result implies that a cubic graph HH has a 5-cycle double cover containing that circuit. This improves upon the recently proved 8-cycle double cover theorem (attributed to OpenAI in the text) for graphs with a dominating circuit.
  • Formalization: The proof has been fully formalized in the Lean theorem prover.

Significance and Claims
The paper claims to provide a complete proof of Sabidussi's compatibility conjecture, a problem that had been studied since the work of Kotzig (1968) and Fleischner (1980). While previous results had established the conjecture for planar graphs, K5K_5-minor-free graphs, or specific degree constraints, this proof treats every even degree directly without restricting the graph class beyond the minimum degree requirement.

The authors explicitly state that the proof is a strengthening of the original conjecture, providing a 4-coloring with specific structural properties rather than just a decomposition. The work is presented as a definitive resolution to the conjecture, relying on a novel combination of cyclic word combinatorics and parity lemmas over finite fields.

Note on Authorship
The paper explicitly states that the proof is entirely due to "GPT 5.6 Pro," and the write-up was prepared with assistance from "GPT 5.6 Sol." The human author, Nikolay Ulyanov, acknowledges the role of AI in the generation of the mathematical argument and the exposition.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →