[Paper Review] The Structure of $W_4$-Immersion-Free Graphs

Rémy Belmonte, Archontia C. Giannopoulou+2 more|arXiv (Cornell University)|Feb 5, 2016

Advanced Graph Theory Research18 references3 citations

TL;DR

This paper establishes a structural characterization of graphs that do not contain the wheel graph $W_4$ as an immersion, proving they can be constructed via 1-, 2-, and 3-edge-sums from subcubic graphs and graphs of bounded treewidth. The key result shows that internally 4-edge-connected graphs excluding $W_4$ as an immersion have treewidth bounded by a large but finite constant, providing a foundational decomposition theorem for $W_4$-immersion-free graphs.

ABSTRACT

We study the structure of graphs that do not contain the wheel on 5 vertices W4 as an immersion, and show that these graphs can be constructed via 1, 2, and 3-edge-sums from subcubic graphs and graphs of bounded treewidth.

Motivation & Objective

• To provide a structural characterization of graphs that exclude $W_4$ as an immersion, a containment relation less studied than minors or topological minors.
• To extend known structural results for topological minor-free graphs to the immersion setting, particularly for wheels.
• To establish a decomposition theorem using edge-sums of order at most 3, enabling algorithmic and theoretical analysis of $W_4$-immersion-free graphs.
• To identify conditions under which such graphs have bounded treewidth, especially in the presence of high connectivity and minimum degree.

Proposed method

• The authors use a decomposition approach based on $i$-edge-sums for $i \in \{1,2,3\}$, recursively breaking down $W_4$-immersion-free graphs into prime components.
• They analyze internal edge cuts and important separators of order at most 6 to control connectivity and path structures in the graph.
• Leveraging the fact that large treewidth implies the existence of a large wall as a topological minor, they derive bounds on treewidth using extremal graph theory.
• They apply a result from graph minor theory that large treewidth forces a large number of disjoint cycles with high connectivity.
• The proof relies on Lemma 3, which shows that internally 4-edge-connected graphs excluding $W_4$ as an immersion have treewidth at most $2^{2^{13} \cdot 3^6 \cdot 5^8 \cdot \log(2^6 \cdot 3^2 \cdot 5^4)}$, a finite constant.
• The decomposition is shown to be tight: 3-edge-sums are necessary, and higher connectivity does not preserve $W_4$-immersion-freeness under such operations.

Experimental results

Research questions

• RQ1 Can $W_4$-immersion-free graphs be decomposed into simpler graph classes using edge-sum operations?
• RQ2 What is the treewidth behavior of internally 4-edge-connected graphs that exclude $W_4$ as an immersion?
• RQ3 Is the decomposition via 1-, 2-, and 3-edge-sums sufficient to characterize all $W_4$-immersion-free graphs?
• RQ4 Why do higher-order edge-sums fail to preserve $W_4$-immersion-freeness, and what structural constraints are needed instead?
• RQ5 Can the treewidth bound for $W_4$-immersion-free graphs be significantly reduced by avoiding reliance on large wall minors?

Key findings

• The prime graphs in a 1-, 2-, and 3-edge-sum decomposition of any $W_4$-immersion-free graph are either subcubic or have treewidth bounded by a finite constant.
• The treewidth of internally 4-edge-connected $W_4$-immersion-free graphs is at most $2^{2^{13} \cdot 3^6 \cdot 5^8 \cdot \log(2^6 \cdot 3^2 \cdot 5^4)}$, a large but finite value.
• The decomposition is tight: 3-edge-sums are necessary, as there exist internally 3-edge-connected graphs with high-degree vertices that avoid $W_4$ as an immersion.
• Edge-sums of order 4 or higher do not preserve $W_4$-immersion-freeness, as shown by counterexamples where the sum contains $W_4$ as an immersion despite both components not containing it.
• The result provides a structural foundation for designing efficient algorithms to test $W_4$-immersion-freeness, leveraging the bounded treewidth of prime components.
• The work opens the door to similar decomposition theorems for larger wheels $W_k$ with $k \geq 5$, though such extensions face challenges due to non-preservation under edge-sums.