---
title: Twisted Droms Theorem for T-RAAGs
url: https://www.emergentmind.com/topics/twisted-droms-theorem
type: topic
---

# Twisted Droms Theorem for T-RAAGs

The Twisted Droms Theorem, in the sense formalized for twisted right-angled Artin groups, characterizes exactly which twisted right-angled Artin groups are closed under taking finitely generated subgroups: for a mixed graph $\Gamma$ and its twisted right-angled Artin group $T(\Gamma)$, every finitely generated subgroup of $T(\Gamma)$ is again a twisted right-angled Artin group if and only if $\Gamma$ is a Droms mixed graph [2504.21196]. This extends Droms’ classical 1987 subgroup theorem for right-angled Artin groups (RAAGs), preserves the classical chordality criterion for coherence, and yields a rigidity theory for defining mixed graphs in a substantial subclass through the notions of sinkholes, satellites, abelianization, and cone decompositions [2504.21196].

## 1. Classical Droms background

The classical point of departure is the RAAG attached to a finite simplicial graph $\bar\Gamma=(V,E)$,
$$
A(\bar\Gamma)=\langle V \mid [v_i,v_j]=1 \text{ whenever } \{i,j\}\in E\rangle,
$$
where $[x,y]=xyx^{-1}y^{-1}$. In this presentation, adjacency in $\bar\Gamma$ records commutation and nothing else. The foundational results of Droms identify graph-theoretic conditions governing subgroup closure and coherence for this class [2504.21196].

Droms’ subgroup theorem states that, for a RAAG $A(\bar\Gamma)$, the following are equivalent: every finitely generated subgroup of $A(\bar\Gamma)$ is again a RAAG, and $\bar\Gamma$ contains neither the path $P_4$ on four vertices nor the $4$-cycle $C_4$ as an induced subgraph. Graphs satisfying this forbidden-subgraph condition are called Droms simplicial graphs. Droms also proved that $A(\bar\Gamma)$ is coherent if and only if $\bar\Gamma$ is chordal, meaning that $\bar\Gamma$ has no induced cycles of length at least $4$ [2504.21196].

These two theorems provide the exact baseline for the twisted theory. In the untwisted setting, subgroup-closedness is controlled by forbidding induced $P_4$ and $C_4$, whereas coherence is controlled by chordality. The twisted theory retains the second criterion unchanged, but the first requires genuinely new mixed-graph obstructions [2504.21196].

## 2. Mixed graphs and twisted right-angled Artin groups

The twisted theory is formulated on a mixed graph $\Gamma=(V,E,D,o,t)$, consisting of a simplicial graph $(V,E)$ together with a distinguished subset $D\subseteq E$ of directed edges and maps $o,t:D\to V$ assigning origin and terminus. An undirected edge $\{x,y\}\in E\setminus D$ is denoted $[x,y]$, while a directed edge $\{x,y\}\in D$ with origin $x$ and terminus $y$ is denoted $[x,y\rangle$ [2504.21196].

A vertex is **negative** if it occurs as the terminus of some directed edge. A **sinkhole** is a vertex $v$ such that every edge incident to $v$ points into $v$. A **signature** is a map $\theta:V\to \mathbb{Z}^\times=\{\pm1\}$ such that $\theta(x)=-1$ for every negative vertex $x$, and $\theta(y)=1$ for every non-isolated vertex that is not the terminus of a directed edge; isolated vertices may be assigned either sign. A mixed graph is **special** if every negative vertex is a sinkhole; equivalently, it avoids the three minimal non-special induced subgraphs listed in equations (1.1)–(1.3) of the paper [2504.21196].

The associated twisted right-angled Artin group is
$$
T(\Gamma)=\langle V \mid ab=ba \text{ if } [a,b]\in E\setminus D,\ \ aba=b \text{ if } [a,b\rangle\in D\rangle.
$$
The relations $aba=b$ are referred to as Klein relations and encode the twist carried by directed edges. When $D=\varnothing$, the construction reduces to the RAAG on the underlying simplicial graph $\bar\Gamma=(V,E)$, so $T(\Gamma)=A(\bar\Gamma)$ [2504.21196].

Several structural maps are fundamental. If $\Delta$ is an induced subgraph of $\Gamma$, then the natural map $T(\Delta)\to T(\Gamma)$ is injective. The underlying RAAG $A(\bar\Gamma)$ also injects into $T(\Gamma)$ via the squaring map $v\mapsto v^2$. More generally, if $\Gamma_u$ is obtained by turning all directed edges $[v,u\rangle$ into undirected edges $[v,u]$, then the map $T(\Gamma_u)\to T(\Gamma)$ sending $u\mapsto u^2$ and fixing the remaining generators is injective [2504.21196].

The abelianization provides a coarse but powerful invariant:
$$
T(\Gamma)^{ab}\cong \mathbb{Z}^r\times (\mathbb{Z}/2\mathbb{Z})^s,
$$
where $s=|o(D)|$ is the number of vertices serving as origins of directed edges. This decomposition is repeatedly used to recover directed structure and to distinguish non-isomorphic mixed graphs that could otherwise have closely related subgroup structure [2504.21196].

A further essential operation is the **cone**. Given a mixed graph $\Gamma$ and a signature $\theta$, the cone $\nabla_\theta(\Gamma)$ adjoins a new vertex $w$ adjacent to every existing vertex, with $[w,v\rangle$ directed precisely when $v$ is negative. Algebraically,
$$
T(\nabla_\theta(\Gamma))=\langle w\rangle \rtimes_\theta T(\Gamma),
$$
with
$$
x^{-1}wx=w^{\theta(x)} \qquad \text{for every } x\in T(\Gamma).
$$
Iterating this construction yields towers of semidirect products over $\mathbb{Z}$ [2504.21196].

## 3. Statement of the theorem and constructive content

The decisive mixed-graph notion is that of a **Droms mixed graph**. Such a graph $\Gamma$ must satisfy three conditions: it is special; its underlying simplicial graph $\bar\Gamma$ is a Droms simplicial graph, so it has no induced $P_4$ and no induced $C_4$; and it contains no induced copy of the special mixed graph $\Lambda_s$, consisting of three vertices $a_1,b,a_2$ with directed edges $[a_1,b\rangle$ and $[a_2,b\rangle$ and no edge between $a_1$ and $a_2$ [2504.21196].

The Twisted Droms Theorem states that for $G=T(\Gamma)$ the following are equivalent: every finitely generated subgroup of $G$ is again a twisted right-angled Artin group; and $\Gamma$ is a Droms mixed graph [2504.21196].

The theorem is not merely the classical RAAG criterion with orientations appended. A common oversimplification is to retain only the forbidden induced subgraphs $P_4$ and $C_4$ from the untwisted setting. In the twisted setting this is insufficient: non-special mixed graphs already fail subgroup-closure, and even among special graphs one must exclude the additional obstruction $\Lambda_s$ [2504.21196].

Its constructive content is encoded by a recursion on induced subgraphs. A mixed graph is Droms mixed if and only if every induced subgraph is either a disjoint union of two non-empty proper subgraphs or a cone $\nabla_{\theta_0}(\Lambda_0)$ for some induced subgraph $\Lambda_0$ and signature $\theta_0$. Consequently, $T(\Gamma)$ is an elementary T-RAAG, built recursively from infinite cyclic groups by free products and semidirect products with $\mathbb{Z}$ determined by signatures [2504.21196].

The proof splits along the two implications. For subgroup-closedness, the recursion is combined with closure under free products and cones. Kurosh theory controls finitely generated subgroups of free products, while in the coned case one projects a subgroup $H\le \langle w\rangle\rtimes_\theta T(\Gamma)$ to $T(\Gamma)$, lifts the projection by a section of the form $\sigma(v)=vw^{n_v}$, and reconstructs $H$ as a semidirect product with induced signature. For the converse, the paper proceeds contrapositively through “poisonous” induced subgraphs: if $\bar\Gamma$ contains $P_4$ or $C_4$, then the corresponding RAAGs inject into $T(\Gamma)$ via squaring, and Droms’ classical non-RAAG finitely generated subgroups survive as obstructions; if $\Gamma$ is not special, or if $\Lambda_s$ occurs, explicit finitely generated subgroups such as $\langle x,yz\rangle$ have subgroup structure incompatible with any T-RAAG on two vertices [2504.21196].

The examples in the introduction sharpen necessity. The graph $\Gamma_3$ is special, has Droms underlying graph, and avoids $\Lambda_s$; correspondingly, $T(\Gamma_3)$ is subgroup-closed and admits the form $\mathbb{Z}^2\rtimes F_2$. By contrast, $\Gamma_4$ fails the $\Lambda_s$ condition, while $\Gamma_1$ and $\Gamma_2$ are not special, and all three fail subgroup-closure [2504.21196].

## 4. Coherence and the persistence of chordality

For T-RAAGs, coherence is classified exactly as in the RAAG case: $T(\Gamma)$ is coherent if and only if the underlying simplicial graph $\bar\Gamma$ is chordal [2504.21196]. This is one of the most striking formal parallels between the twisted and untwisted settings.

The non-chordal direction is reduced immediately to the classical RAAG obstruction. If $\bar\Gamma$ is not chordal, then $A(\bar\Gamma)$ is non-coherent by Droms’ theorem, and the injective squaring map $A(\bar\Gamma)\hookrightarrow T(\Gamma)$ transfers this non-coherence to $T(\Gamma)$. Twisting therefore does not remove the classical pathologies created by induced cycles of length at least $4$ [2504.21196].

The chordal direction is proved by decomposition along complete subgraphs. One writes
$$
T(\Gamma)=T(\Gamma_1)\ast_{T(C)}T(\Gamma_2),
$$
where $C$ is complete and $T(C)$ is virtually $\mathbb{Z}^k$, because complete mixed graphs produce virtually free abelian T-RAAGs. Induction together with the Karrass–Solitar coherence lemma yields coherence of the amalgamated product. The role of the twist is therefore absorbed into the algebra of the complete pieces, while the global criterion remains purely a property of $\bar\Gamma$ [2504.21196].

This yields a clean taxonomy. Complete mixed graphs, trees, and cones over chordal bases are coherent. Any mixed graph whose underlying simplicial graph contains $C_n$ for $n\ge 4$ or $P_4$ is non-coherent. The resulting principle is that subgroup-closedness is sensitive to the directed structure, whereas coherence is governed entirely by the untwisted chordality of the underlying graph [2504.21196].

## 5. Rigidity, satellites, and the isomorphism problem

The same paper also addresses rigidity for a notable subclass of T-RAAGs, that is, whether the abstract isomorphism type of $T(\Gamma)$ determines the defining mixed graph $\Gamma$. The key obstruction is the notion of a **satellite**. In a special mixed graph, a vertex $v\ne w$ is a satellite of a sinkhole $w$ if $v$ is not adjacent to $w$, some third vertex is joined to both, and every neighbor of $v$ is also a neighbor of $w$ [2504.21196].

The rigidity theorem has three parts. First, if a special graph $\Gamma$ is rigid, then it has no satellites. Second, if $\Gamma$ has a sinkhole joined to every other vertex, then $\Gamma$ is rigid. Third, in the Droms mixed subclass, rigidity is equivalent to the absence of satellites. Hence, for Droms mixed graphs, the isomorphism class of $T(\Gamma)$ determines $\Gamma$ exactly when satellites are absent [2504.21196].

The proof uses two group-theoretic invariants in a decisive way. The first is the abelianization
$$
T(\Gamma)^{ab}\cong \mathbb{Z}^r\times (\mathbb{Z}/2\mathbb{Z})^s,
$$
whose $2$-torsion rank records the number of origins of directed edges and constrains the possible sinkhole pattern. The second is the maximal abelian normal subgroup $N$ in the connected Droms case. The paper shows that $N\cong \mathbb{Z}^r$ is unique and arises from “central” vertices joined to all others: either $N=\langle w_1,\dots,w_r\rangle$ when all cone tips are positive, or $N=\langle u^2,w_1,\dots,w_{r-1}\rangle$ when there is a unique negative central vertex $u$. In either case $\Gamma$ is obtained by $r$ iterated conings over an induced subgraph $\Gamma_0$, and $T(\Gamma)/N\cong T(\Gamma_0)$, so the cone layers are visible from the group alone [2504.21196].

The non-rigid direction is realized by explicit graph surgery. If a positive satellite $v$ of a sinkhole $w$ exists, one can construct a distinct mixed graph $\Gamma'$ by turning $v$ into a sinkhole $v'=vw$ while preserving the isomorphism type of the group. This shows that satellites are genuine obstructions rather than merely combinatorial nuisances. Conversely, when a sinkhole is joined to all vertices, projection onto the cyclic factor generated by that sinkhole, together with underlying RAAG rigidity, forces coincidence of defining graphs [2504.21196].

The resulting decision procedure is graph-theoretic rather than presentation-theoretic. In the Droms case, one verifies specialness, checks for satellites around sinkholes, and concludes rigidity exactly when no satellites occur. The examples recorded in the paper reflect each possibility: $\Gamma_3$ is Droms but not rigid because $a_1$ and $a_2$ are satellites of each other; $\Gamma_2$ and $\Gamma_4$ are special with a sinkhole joined to all vertices and are therefore rigid; the example $\Upsilon$ is Droms with no satellites and is rigid [2504.21196].

## 6. Related usages, scope, and limitations

The phrase “Twisted Droms Theorem” is used in recent arXiv literature in more than one setting. The mixed-graph theorem for T-RAAGs is the direct theorem bearing that name in the title [2504.21196], but closely related “Droms-type” classification phenomena also appear for Coxeter systems and for subdirect products of limit groups over Droms RAAGs [1708.00960] [2104.14849].

| Setting | Core statement | arXiv id |
|---|---|---|
| Twisted right-angled Artin groups | Finitely generated subgroup-closure is equivalent to $\Gamma$ being a Droms mixed graph | [2504.21196] |
| Coxeter systems of type FC with only $\mathbb{Z}_2$ twists | Angle-compatible Coxeter generating sets differ by elementary twists and conjugation; under these hypotheses the defining graph is unchanged | [1708.00960] |
| Full subdirect products of limit groups over Droms RAAGs | Type $FP_s(\mathbb{Q})$ implies virtual surjection onto every $s$-tuple of coordinates | [2104.14849] |

In the Coxeter-group usage, the relevant background is Mühlherr’s Twist Conjecture. Under the assumptions that the defining graph is of type FC and admits only twists in $\mathbb{Z}_2$, angle-compatible Coxeter generating sets are related by elementary twists and conjugation; because $\mathbb{Z}_2$-twists do not change the defining graph, one obtains a strong rigidity phenomenon analogous to Droms’ theorem for RAAGs [1708.00960]. In the limit-group usage, the phrase is analogical rather than official: the paper does not itself use the term, but its main theorem is presented as a twisted analogue of Droms’ subgroup theorem, replacing subgroup closure by a virtual-surjection theorem for full subdirect products of type $FP_s(\mathbb{Q})$ [2104.14849].

Within the T-RAAG setting proper, the main scope and limitations are explicit. The theorem extends Droms’ subgroup-closure criterion by replacing the sole obstruction pair $P_4/C_4$ with the mixed-graph package of specialness, Droms underlying graph, and exclusion of $\Lambda_s$. It preserves the RAAG coherence classification exactly through chordality of the underlying simplicial graph. It further solves the isomorphism problem for the notable subclass of Droms mixed graphs via the satellite criterion, abelianization, and cone-layer analysis [2504.21196].

At the same time, the theory is not complete outside the special and Droms subclasses. Rigidity may fail beyond these classes, and the paper records the conjectural expectation that satellites characterize rigidity for all special graphs. No complexity bounds are given for the decision procedures, which remain structural and graph-theoretic rather than algorithms from arbitrary group presentations. Finally, the twisted setting introduces Klein relations and torsion phenomena absent from RAAGs, even though subgroup-closure and coherence continue to mirror the classical Droms theorems with remarkable fidelity [2504.21196].

Source: https://www.emergentmind.com/topics/twisted-droms-theorem