Closure Under Minors of Undirected Entanglement

Entanglement is a digraph complexity measure that origins in fixed-point theory. Its purpose is to count the nested depth of cycles in digraphs. In this paper we prove that the class of undirected graphs of entanglement at most $k$, for arbitrary fixed $k \in \mathbb{N}$, is closed under taking minors. Our proof relies on the game theoretic characterization of entanglement in terms of Robber and Cops games.

💡 Research Summary

The paper investigates the graph‑theoretic measure known as entanglement, originally introduced for directed graphs in the context of fixed‑point theory. Entanglement quantifies the nested depth of cycles by means of a pursuit–evasion game: a robber moves along edges while a team of cops occupies vertices; the smallest number of cops that can guarantee capture of the robber defines the entanglement of the graph. Although the concept has traditionally been applied to digraphs, the authors show that it extends naturally to undirected graphs and that the class of undirected graphs whose entanglement does not exceed a fixed integer k is minor‑closed.

The main theorem states: for any fixed k ∈ ℕ, the set
 𝔾ₖ = { G | G is an undirected graph and entanglement(G) ≤ k }
is closed under the two elementary minor operations—vertex deletion and edge‑contraction (the latter being modelled as the contraction of a pair of adjacent vertices into a single “virtual” vertex). In other words, if a graph G belongs to 𝔾ₖ, then every graph obtained from G by a sequence of deletions and contractions also belongs to 𝔾ₖ.

The proof is entirely game‑theoretic. The authors first recall the robber‑and‑cops characterization of entanglement and formalize the notion of a winning cop strategy for a given bound k. They then analyze how such a strategy can be transformed when a minor operation is applied.

1. Vertex Deletion.
   If a vertex v is removed, two cases are distinguished. If v is never a possible location for the robber under the original winning strategy, the strategy can be left unchanged. If v could be occupied by the robber, the cops pre‑emptively block v before the deletion, or they adjust their patrol routes so that the robber never needs to step onto v. Crucially, these adjustments never require more than k cops, because the original strategy already guarantees capture with k cops; the deletion merely eliminates a potential hiding place.

2. Edge Contraction (Vertex Pair Contraction).
   Contracting an edge {u, w} merges u and w into a new vertex x. The main difficulty is that new cycles may appear, potentially increasing the nested cycle depth. To handle this, the authors introduce a virtual vertex abstraction: the contracted pair is treated as a single location that can be occupied by a virtual cop in addition to the real cops. The virtual cop follows the same movement rules as a real cop but is allowed to “teleport” between the pre‑contraction vertices u and w as needed. By embedding the original winning strategy into this enriched setting, they show that the same k cops (including the virtual one) can still force capture after contraction. After the contraction, the virtual cop can be merged with an existing real cop, preserving the bound of k.

The paper also establishes a relationship between entanglement and more classical width parameters. It proves that any graph with entanglement ≤ k necessarily has tree‑width ≤ k, thereby situating entanglement within the well‑studied hierarchy of minor‑closed graph families. However, entanglement is strictly finer: it captures the depth of cycle nesting rather than just the size of a tree‑decomposition bag, offering a more nuanced view of cyclic structure.

From a methodological standpoint, the work showcases how game‑theoretic arguments can be leveraged to prove structural properties of graph parameters. By constructing explicit strategy transformations for each minor operation, the authors avoid heavy combinatorial machinery and provide an intuitive, constructive proof that can be adapted to algorithmic contexts.

The implications are twofold. First, the minor‑closedness result guarantees that any algorithmic technique based on graph reduction (e.g., kernelization, tree‑decomposition‑based dynamic programming) can safely be applied to problems parameterized by entanglement without risking an increase in the parameter. Second, it opens a research avenue linking entanglement to the extensive theory of graph minors, suggesting that many classical results (such as the Graph Minor Theorem) may have analogues in the entanglement setting.

In summary, the authors prove that for any fixed integer k, the family of undirected graphs with entanglement at most k is closed under taking minors. The proof hinges on a careful adaptation of the robber‑and‑cops game to the minor operations, introduces virtual cop strategies to handle edge contractions, and situates entanglement within the broader landscape of width parameters. This result strengthens the theoretical foundations of entanglement and paves the way for its application in algorithm design and structural graph theory.