Linear Arrangement of Halin Graphs

We study the Optimal Linear Arrangement (OLA) problem of Halin graphs, one of the simplest classes of non-outerplanar graphs. We present several properties of OLA of general Halin graphs. We prove a lower bound on the cost of OLA of any Halin graph, and define classes of Halin graphs for which the cost of OLA matches this lower bound. We show for these classes of Halin graphs, OLA can be computed in O(n log n), where n is the number of vertices.

💡 Research Summary

The paper investigates the Optimal Linear Arrangement (OLA) problem for Halin graphs, a simple class of non‑outerplanar planar graphs formed by taking a plane tree T and connecting its leaves with a cycle C. While OLA is known to be NP‑hard for general graphs and even for several restricted families, the authors aim to identify subclasses of Halin graphs for which OLA can be solved in polynomial time.

The work begins by recalling the definition of a linear arrangement as a bijection φ from the vertex set V to {1,…,|V|} and the OLA objective as minimizing the sum of absolute label differences over all edges. They restate known properties of optimal arrangements for trees: the extreme vertices (labels 1 and n) are leaves, the “spinal path” connecting these leaves is monotone in the ordering, and subtrees hanging off the spinal path can be treated independently.

Extending these ideas to Halin graphs, the authors define a spinal path that lies entirely within the underlying tree T. By removing the spinal path and the leaf‑connecting cycle C, the graph decomposes into a collection of subtrees rooted at the vertices of the spinal path. Lemma 3.3 shows that in any optimal arrangement the vertices of each such subtree occupy a contiguous block of labels, mirroring the tree case. Lemma 3.5 further proves that two branches attached to the same spine vertex and placed on the same side of that vertex never interleave; they appear as whole blocks.

Theorem 3.6 characterizes the possible extreme vertices of an optimal Halin arrangement: either both are leaves of T, or a non‑leaf extreme must have degree 3 and be adjacent to exactly two leaf vertices. Consequently, Corollary 3.7 establishes a universal lower bound for any Halin graph H with n vertices:
 LA(φ*, H) ≥ 2·(n − 1) + LA(φ*_T, T),
where φ*_T is an optimal arrangement of the underlying tree. The term 2·(n − 1) corresponds to the minimum contribution of the cycle C, because each of its n edges must span at least one unit of distance in any linear ordering.

The second part of the paper focuses on a subclass of Halin graphs whose underlying tree T is “recursively balanced” (RBT). An RBT is defined recursively: after removing the root, all resulting subtrees are isomorphic, and each of those subtrees is itself recursively balanced. This structure guarantees the existence of a central vertex (the root) satisfying the “central‑vertex theorem” (Theorem 4.2), which bounds the size of each component after removal of the root by ⌊n/2⌋.

Using earlier results on optimal arrangements for trees (Theorem 4.3 and related lemmas), the authors show that for an RBT one can construct an optimal ordering by placing half of the equal‑size subtrees to the left of the root and the other half to the right, preserving the order within each subtree. Lemma 4.5 proves that any swapping of whole subtrees (operator σ) does not affect the total arrangement cost, allowing flexibility in arranging the subtrees while keeping optimality. Lemma 4.6 guarantees that for any two leaves belonging to different subtrees there exists an optimal arrangement where those leaves occupy the extreme positions (labels 1 and n), i.e., the spinal path can be chosen as the leaf‑to‑leaf path through the root.

The authors then present a linear‑time algorithm (Theorem 4.8) to compute an optimal arrangement for an RBT by recursively applying the subtree‑swap strategy and labeling the root appropriately; each vertex is visited a constant number of times.

Finally, Theorem 4.9 ties the tree result back to Halin graphs: if the underlying tree T of a Halin graph H is an RBT, then starting from an optimal tree arrangement φ⊛ one can insert the cycle vertices in O(|V| log |V|) time to obtain an arrangement φ* for H whose cost equals LA(φ⊛, T) + 2·(n − 1). Since this matches the lower bound from Corollary 3.7, φ* is optimal for H. The construction proceeds by placing the cycle vertices in a contiguous block between the two halves of the tree ordering and performing a series of subtree swaps that never increase the tree cost while strictly decreasing the total cost contributed by the cycle edges.

In summary, the paper contributes three main results: (1) a general lower bound on the OLA cost of any Halin graph; (2) a structural characterization of optimal arrangements for Halin graphs, showing that extreme vertices can be assumed to be leaves or degree‑3 vertices adjacent to two leaves; (3) a polynomial‑time (specifically O(n log n)) algorithm for computing OLA on the subclass of Halin graphs whose underlying tree is recursively balanced. This work expands the limited set of graph families for which OLA is tractable and provides concrete algorithmic techniques that may inspire further extensions to broader classes of planar graphs.