---
title: 'H-Planar Treewidth: A Hybrid Graph Parameter'
url: https://www.emergentmind.com/topics/h-planar-treewidth
type: topic
---

# H-Planar Treewidth: A Hybrid Graph Parameter

to=arxiv_search 不中反json_string code  天天爱彩票怎么x? Let's see format.
to=arxiv_search  天天种彩票ង់json code  天天中彩票不ార్క?
to=arxiv_search  天天中彩票上json 凤凰大参考 code 平台开号{"query":"\"H-Planarity and Parametric Extensions: when Modulators Act Globally\"","max_results":5}
to=search_arxiv  天天中彩票派奖json 
\(H\)-planar treewidth is a hybrid width parameter introduced in the modulator/target framework of "H-Planarity and Parametric Extensions: when Modulators Act Globally" [2507.08541]. It measures whether a graph can be organized around a torso of bounded planar treewidth while every connected component outside the modulator belongs to a fixed graph class \(H\). The parameter is paired with \(H\)-planarity and \(H\)-planar treedepth, and its defining feature is that the modulator may act globally: the remaining core may be an arbitrary planar graph and therefore need not have bounded ordinary treewidth.

## 1. Definition and formalism

For a graph \(G\) and a vertex set \(X \subseteq V(G)\), the torso of \(X\) in \(G\), denoted \(\operatorname{torso}(G,X)\), is obtained from the induced subgraph \(G[X]\) by turning \(N_G(V(C))\) into a clique for every connected component \(C\) of \(G-X\). Equivalently, two vertices \(u,v \in X\) are adjacent in \(\operatorname{torso}(G,X)\) if either \(uv \in E(G)\), or some connected component of \(G-X\) is adjacent to both \(u\) and \(v\) [2507.08541].

Given a graph class \(H\), a set \(X \subseteq V(G)\) is a planar \(H\)-modulator if \(\operatorname{torso}(G,X)\) is planar and every connected component of \(G-X\) belongs to \(H\). A graph is \(H\)-planar if it admits such a modulator. The associated decision problem is \(H\)-PLANARITY: given \(G\), decide whether \(G\) admits a planar \(H\)-modulator.

The paper places this inside a general operator scheme. For a union-closed graph class \(H\) and a minor-monotone parameter \(p\),
\[
H\text{-}p(G)=\min \bigl\{ k \mid \exists X \subseteq V(G):\ p(\operatorname{torso}(G,X)) \le k \text{ and } G-X \in H \bigr\}.
\]
Here \(G-X \in H\) is interpreted componentwise because \(H\) is union-closed. In this language, \(H\)-planar treewidth is \(H\text{-}\operatorname{ptw}\), where \(\operatorname{ptw}\) denotes planar treewidth.

Planar treewidth itself is defined as follows. The planar treewidth of a graph \(G\), denoted \(\operatorname{ptw}(G)\), is the minimum \(k\) such that \(G\) admits a tree decomposition in which each bag either has size at most \(k+1\) or has a planar torso. Consequently,
\[
H\text{-}\operatorname{ptw}(G)=\min\bigl\{ k \mid \exists X\subseteq V(G):\ \operatorname{ptw}(\operatorname{torso}(G,X))\le k \text{ and } G-X\in H \bigr\}.
\]

With \(\mathrm{PT}_k\) denoting the class of graphs of planar treewidth at most \(k\), the notation \(\mathrm{PT}_k > H\) denotes exactly the class of graphs of \(H\)-planar treewidth at most \(k\). This operator viewpoint is central: \(H\)-planar treewidth is not an isolated definition, but one instance of a general “torso parameter plus target class” construction.

## 2. Decomposition viewpoint

The intended decomposition-theoretic meaning is explicit. A graph has \(H\)-planar treewidth at most \(t\) if it admits a tree decomposition such that every leaf-bag belongs to \(H\), and all other bags are either of size at most \(t+1\) or have a planar torso [2507.08541]. This is the analogue of \(H\)-treewidth, but with planar torsos allowed in place of uniformly bounded-width bags.

The paper develops a second formulation through sphere decompositions. In Section 6 it defines a local cell property \(\Pi_{H,k}\): roughly, after cliquifying the boundary of a cell, the resulting graph must admit a \(\mathrm{PT}_k > H\)-modulator containing that boundary. If a graph has a sphere decomposition all of whose cells satisfy \(\Pi_{H,k}\), then the graph has \(H\)-planar treewidth at most \(k\). Conversely, on the unbreakable instances used in the recognition algorithm, bounded \(H\)-planar treewidth implies the existence of such sphere decompositions.

This decomposition viewpoint is paired with \(H\)-planar treedepth. Planar treedepth \(\operatorname{ptd}\) is defined recursively: \(\operatorname{ptd}(G)=0\) if \(G\) is empty, and \(\operatorname{ptd}(G)\le k\) if there exists \(X\subseteq V(G)\) with \(\operatorname{torso}(G,X)\) planar and \(\operatorname{ptd}(G-X)\le k-1\). Then \(H\)-planar treedepth is \(H\text{-}\operatorname{ptd}\). The pair \((H\text{-}\operatorname{ptd}, H\text{-}\operatorname{ptw})\) provides recursive and decomposition-based extensions of \(H\)-planarity.

A basic normalization follows immediately from the definition: planar graphs have planar treewidth \(0\), since a single bag containing all vertices has planar torso. This is one reason the framework is described as a parametric extension of planarity rather than merely another bounded-width formalism.

## 3. Relation to adjacent width notions

The paper states that both \(H\)-planar treedepth and \(H\)-planar treewidth are more general than \(H\)-treedepth or \(H\)-treewidth [2507.08541]. The reason is structural: planar torsos may have arbitrarily large ordinary treewidth, so bounded \(H\)-planar treewidth does not force bounded \(H\)-treewidth.

It also differs from deletion distance. On highly connected graphs, however, the distinction collapses: if \(G\) is \((\max\{4,k\}+2)\)-connected, then a graph \(G^\ast \in H\) can be obtained from \(G\) by at most \(k\) vertex deletions if and only if the \(H\)-planar treewidth of \(G\) is at most \(k\). Thus, in high connectivity, the hybrid parameter becomes equivalent to a deletion parameter.

A common source of confusion is the distinction between \(H\)-planar treewidth and planar-treewidth. In "5-Approximation for \(\mathcal H\)-Treewidth Essentially as Fast as \(\mathcal H\)-Deletion Parameterized by Solution Size" [2306.17065], planar-treewidth means \(\mathcal H\)-treewidth for \(\mathcal H=\mathsf{planar}\): large planar induced subgraphs are placed in leaf-local base components, and only the nonplanar interfaces count toward the width. By contrast, \(H\)-planar treewidth is defined through planar torsos and components in \(H\); its controlling object is a planar torso inside the modulator/torso formalism, not a planar base component in a tree \(\mathcal H\)-decomposition.

A second nearby notion is the existence of tree decompositions whose bags themselves induce low-treewidth graphs. "Optimal Tree-Decompositions with Bags of Bounded Treewidth" proves that every planar graph has an optimal tree-decomposition in which every bag has treewidth at most \(3\) [2511.22196]. That result is adjacent in spirit, but it concerns induced bag subgraphs at optimal decomposition width, whereas \(H\)-planar treewidth is defined by planar torsos and \(H\)-components. This suggests that \(H\)-planar treewidth belongs to a wider family of planar-hybrid decomposition parameters, but it is not reducible to bounded induced bag treewidth.

## 4. Recognition and parameterized tractability

The first algorithmic layer is \(H\)-PLANARITY itself. If \(H\) is hereditary, CMSO-definable, and decidable in polynomial time, then \(H\)-PLANARITY is solvable in polynomial time [2507.08541]. This gives polynomial-time recognition of graphs that admit a planar \(H\)-modulator.

The main parameterized transfer theorem concerns \(H\)-planar treedepth and \(H\)-planar treewidth. Let \(p\in\{\operatorname{ptd},\operatorname{ptw}\}\), and let \(H\) be hereditary, CMSO-definable, and union-closed. If \(H\)-DELETION parameterized by solution size \(k\) is solvable in time \(O_k(n^c)\), then there is a non-uniform algorithm deciding whether \(H\text{-}p(G)\le k\) in time
\[
O_k(n^4+n^c\log n).
\]
For \(p=\operatorname{ptw}\), this yields fixed-parameter tractability of bounded \(H\)-planar treewidth from fixed-parameter tractability of deletion into \(H\).

A central restricted-instance theorem underlies this transfer. For hereditary union-closed \(H\), and integers \(a,k\), there is an algorithm that, given an \((a,3)\)-unbreakable graph \(G\), checks whether \(G\) has \(H(a-1)\)-planar treewidth at most \(k\) in time
\[
O_{k,a}(n(n+m)).
\]
The general theorem then follows from the CMSO reduction to unbreakable graphs.

The framework is also constructive at the decomposition level. Under the assumptions of the main theorem, if \(G\) has \(H\)-planar treewidth at most \(k\), one can construct a \(\mathrm{PT}_k > H\)-modulator \(S\) together with a tree decomposition of \(\operatorname{torso}_G(S)\) of planar width at most \(k\).

## 5. Structural techniques and global modulators

The technical novelty of the framework is the treatment of modulators that act globally. Earlier modulator/treewidth meta-theorems typically require modulators of low bidimensionality, so that the modulator does not spread throughout a large wall or grid minor. Here the planar torso itself may spread across the whole wall, and this breaks the usual locality-based irrelevant-vertex strategy [2507.08541].

To overcome this, the paper combines planar \(H\)-modulators with sphere decompositions, well-linked and ground-maximal renditions, and a new irrelevant-vertex argument. Large apex grids serve as obstructions, while flat walls are used positively. In the wall case, one removes a central vertex and proves irrelevance by comparing decompositions inside the compass with decompositions outside the central region.

For \(H\)-planar treewidth, the key irrelevance statement is Lemma 17. Let \(H\) be hereditary, let \(G\) be an \((a,3)\)-unbreakable graph, let \((W,\mathcal A)\) be a flatness pair of height \(r\), and let \(v\) be a central vertex of \(W\), where \(r \ge \max\{a+3,k+1,7\}\) is odd. Then \(G\) has \(H(a-1)\)-planar treewidth at most \(k\) if and only if both the \(\mathcal A\)-compass of \(W\) and \(G-v\) have \(H(a-1)\)-planar treewidth at most \(k\). This is the analogue of the classical irrelevant-vertex lemma, redesigned for globally acting modulators.

Planarity of the torso is not a cosmetic condition. It is used to bound attachment neighborhoods, to obtain sphere embeddings and renditions, to invoke flat-wall structure, and to make the gluing of local decompositions topologically consistent. The paper also notes a hardness barrier in the background: without the planarity requirement on the torso, the corresponding partition problem becomes NP-hard even for \(H=\{K_4\}\).

## 6. Applications, examples, and limitations

The framework applies to graph classes \(H\) satisfying the theorem assumptions, and the paper explicitly points to examples such as edgeless graphs, bipartite graphs, and perfect graphs [2507.08541]. It also identifies target classes on which the downstream applications are meaningful, including bounded clique-width classes, classes excluding a shallow-vortex minor, chain graphs, co-chain graphs, threshold graphs, perfect graphs, and \(P_6\)-free graphs.

For coloring, if \(H\) is hereditary, CMSO-definable, polynomial-time decidable, and \(\chi(H)\) is polynomial-time computable on graphs in \(H\), then every \(H\)-planar graph can be colored in polynomial time with at most \(\chi(G)+4\) colors. In the bounded-parameter setting, if \(H\text{-}\operatorname{ptw}(G)\le k\), then \(G\) can be colored in polynomial time using at most
\[
\chi(G)+\max\{4,k+1\}
\]
colors.

For counting perfect matchings, if \(H\) is hereditary, CMSO-definable, polynomial-time decidable, and counting weighted or unweighted perfect matchings is polynomial-time solvable on graphs in \(H\), then the same counting problem is polynomial-time solvable on \(H\)-planar graphs. If \(H\text{-}\operatorname{ptw}(G)\le k\), the weighted or unweighted number of perfect matchings can be computed in time
\[
n^{O(k)}.
\]

For Maximum Independent Set, if \(H\) is hereditary, CMSO-definable, polynomial-time decidable, and Maximum Independent Set is polynomial-time solvable on \(H\), then there is an algorithm that, given \(\varepsilon>0\) and an \(H\)-planar graph \(G\), computes in time
\[
2^{O(1/\varepsilon)}\cdot |V(G)|^{O(1)}
\]
an independent set of size at least \((1-\varepsilon)\alpha(G)\). The paper presents this as an EPTAS consequence of the planar-torso framework.

The main limitations are also explicit. The recognition theorems are non-uniform because they rely on the CMSO-to-unbreakable reduction theorem. Heredity is essential: \(H\)-PLANARITY can be NP-hard without it, and the paper proves NP-hardness even for \(H=\{K_4\}\). CMSO-definability is part of the current theorem statement, but whether it is necessary is left open.

In summary, \(H\)-planar treewidth is a hybrid parameter in which the controlling torso may be planar rather than bounded-treewidth, and the components outside the modulator belong to \(H\). Its significance lies in making “global” planar modulators algorithmically tractable while preserving access to tree-decomposition methods. The resulting theory sits between planarity, \(H\)-deletion, and hybrid-width parameters, and it supports fixed-parameter recognition together with concrete algorithms for coloring, counting perfect matchings, and approximation schemes.

Source: https://www.emergentmind.com/topics/h-planar-treewidth