---
title: 'Quasidominance Condition: Controlled Domination'
url: https://www.emergentmind.com/topics/quasidominance-condition
type: topic
---

# Quasidominance Condition: Controlled Domination

The expression **quasidominance condition** does not denote a single standardized concept across the arXiv literature. In the cited papers it is used, or explicitly identified as a useful label, for several distinct formal conditions: a bounded-overlap domination requirement in graph theory, a Schreier-type ideal factor-lifting property in integral domains, an intermediate near-unanimity condition in universal algebra, singular-value criteria equivalent to dominated splitting for matrix sequences, and several dominance or tail-control conditions in matrix theory, probability, and potential theory [1411.7818] [1112.0132] [2105.14041] [2501.15940]. This suggests a common organizing idea—controlled domination subject to upper constraints—but the technical content depends entirely on the surrounding discipline.

## 1. Terminological scope and formal patterns

In the supplied literature, the same label is attached to conditions that are formally unrelated. What unifies them is not a shared definition, but a recurring schema: one asks that an object dominate another while preserving a bounded multiplicity, an exact factorization, an eventual tail pattern, or an integrable majorant.

| Area | Ambient object | Formal core |
|---|---|---|
| Graph theory | $S\subseteq V(G)$ | $\forall v\in V\setminus S,\ 1\le |N(v)\cap S|\le k$ |
| Ideal theory | nonzero ideals of $D$ | $I\supseteq AB \Rightarrow I=A'B'$ or $I^*=(HJ)^*$ with enlarged factors |
| Universal algebra | $(n+2)$-ary term $u$ | identities (Eq. 3.1)–(Eq. 3.3) |
| Matrix cocycles | bounded $B:\mathbb Z\to M(2,\mathbb C)$ | (SVG) + (FI) |
| qpr-sequences | $q_1\cdots q_n\in\{A,S,N\}^n$ | once $N$ appears, the tail is all $N$ |
| Potential theory | family $\mathcal F_K$ | $u\le F_K$ with $\psi(F_K)\in L^1_{\mathrm{loc}}$ plus structural hypotheses |

Because these notions live in different subjects, direct transfer of results between them is generally impossible. The same phrase therefore functions as a cross-disciplinary label rather than a universally fixed term.

## 2. Graph-theoretic quasidominance

In graph theory, the relevant notion is **$k$-quasiperfect domination**. For a finite, undirected, simple, connected graph $G=(V,E)$ and $k\ge 1$, a set $S\subseteq V$ is a $k$-quasiperfect dominating set if it is dominating and satisfies the **quasidominance condition**
\[
\forall v\in V\setminus S,\qquad 1\le |N(v)\cap S|\le k.
\]
The minimum cardinality of such a set is the $k$-quasiperfect domination number $\gamma_{1,k}(G)$, also written in the paper as $\gamma_{1k}(G)$ [1411.7818].

This condition interpolates between two classical extremes. When $k=1$, every vertex outside $S$ has exactly one neighbor in $S$, so one recovers **perfect domination**:
\[
\gamma_{1,1}(G)=\min\{|S|: S\subseteq V,\ \forall v\in V\setminus S,\ |N(v)\cap S|=1\}.
\]
When $k=\Delta(G)$, the upper bound is vacuous for dominating sets, and ordinary domination is recovered:
\[
\gamma_{1,\Delta(G)}(G)=\gamma(G).
\]
Accordingly, the parameters form the decreasing chain
\[
n\ge \gamma_{1,1}(G)\ge \gamma_{1,2}(G)\ge \cdots \ge \gamma_{1,\Delta(G)}(G)=\gamma(G).
\]
The paper interprets this chain as measuring how far a graph is from being perfectly dominated, and calls it **short** when $\gamma_{1,2}(G)=\gamma(G)$ [1411.7818].

Several structural facts govern these parameters. If $\gamma(G)\le \Delta(G)$, then the tail stabilizes:
\[
\gamma_{1,\gamma(G)}(G)=\gamma_{1,\gamma(G)+1}(G)=\cdots=\gamma_{1,\Delta(G)}(G)=\gamma(G).
\]
There are also local constraints that are frequently used in proofs: if a vertex $v$ satisfies $|N(v)\cap S|>k$, then necessarily $v\in S$; if $K$ is a clique and $|V(K)\cap S|>k$, then $V(K)\subseteq S$. The paper further gives a $0$–$1$ integer programming formulation with the adjacency matrix $N$, characteristic vector $X$, and all-ones vector $1_n$:
\[
NX\ge 1_n-X,\qquad NX\le k\cdot 1_n+(n-k-1)X,
\]
with objective $\min\sum_i x_i$ [1411.7818].

The principal structural theorem identifies broad graph classes in which the chain is short: if $G$ satisfies any of the following, then $\gamma_{1,2}(G)=\gamma(G)$: $\Delta(G)\ge n-3$, $\Delta(G)\le 2$, $G$ is a cograph, or $G$ is claw-free. The paper develops these cases in detail. For connected cographs, explicit values of $\gamma_{1,1}(G)$ are obtained via the join decomposition $G=G_1\vee G_2$; for claw-free graphs, broad realizability results for pairs $(\gamma(G),\gamma_{1,1}(G))$ are established; and for extremal maximum degree families, the paper shows that the perfect domination number may vary widely even when the chain is short. Canonical examples include paths, cycles, complete graphs, stars, complete bipartite graphs, and wheels, all of which are analyzed explicitly [1411.7818].

## 3. Ideal-theoretic quasidominance: sharp and *-sharp domains

In multiplicative ideal theory, the phrase refers to a **Schreier-type factor-lifting property**. For an integral domain $D$, the paper "A Schreier Domain Type Condition" introduces the class of **sharp domains**: $D$ is sharp if for all nonzero ideals $I,A,B$,
\[
I\supseteq AB \Rightarrow \exists\, A',B' \text{ ideals with } A'\supseteq A,\ B'\supseteq B,\ \text{and } I=A'B'.
\]
The paper explicitly identifies this condition with what is being called the quasidominance condition, and emphasizes that it strengthens the pre-Schreier and quasi-Schreier properties by requiring factor-lifting for **all** nonzero ideals, not merely principal or invertible ones [1112.0132].

A central characterization is the **colon-factorization criterion**:
\[
D \text{ is sharp } \Longleftrightarrow I=[I:(I:H)]\cdot (I:H)
\]
for every two nonzero ideals $I,H$. This gives canonical enlargements from any inclusion $AB\subseteq I$ by taking
\[
A' = I:(I:A),\qquad B' = I:A.
\]
The valuation-theoretic description is equally strong: for a valuation domain $V$, sharpness is equivalent to pseudo-Dedekindness and to the value group $\Gamma(V)$ being a complete subgroup of $\mathbb R$. From this the paper derives the global structural theorem: if $D$ is sharp, then each localization $D_M$ at a maximal ideal is a valuation domain whose value group is a complete subgroup of $\mathbb R$; consequently, $D$ is a Prüfer domain of Krull dimension at most $1$. Further consequences include
\[
\text{sharp} \Rightarrow \text{pseudo-Dedekind} \Rightarrow \text{completely integrally closed} \Rightarrow \text{GGCD},
\]
and under additional hypotheses one gets Dedekind behavior, for example in TV, Noetherian, Krull, or countable settings [1112.0132].

The companion paper extends this framework to star operations. Let $*$ be a star operation on an integral domain $D$. The domain is ***-sharp** if for all nonzero ideals $I,A,B\subseteq D$ with $I\supseteq AB$, there exist nonzero ideals $H,J$ such that
\[
I^*=(HJ)^*,\qquad A\subseteq H^*,\qquad B\subseteq J^*.
\]
This is presented as the star-operation analogue of the same quasidominance pattern [1112.1236].

The *-sharp theory inherits many of the structural features of the sharp case, but now filtered through the star operation. If $*_1\le *_2$ and $D$ is $*_1$-sharp, then $D$ is $*_2$-sharp. If $*$ is stable and of finite character and $D$ is *-sharp, then every localization $D_M$ at a maximal $*$-ideal is a valuation domain with value group a complete subgroup of the reals; in particular, $D$ is a Prüfer $*$-multiplication domain and the $*$-dimension is at most $1$. The paper also identifies important special cases: $v$-sharp is equivalent to complete integral closure, $t$-sharp is equivalent to $w$-sharp, every Krull domain is $t$-sharp, and under TV or countability hypotheses one obtains *-Dedekind or Krull conclusions. Polynomial and Nagata-type extensions are analyzed as well: $D$ is $t$-sharp if and only if $D[X]$ is $t$-sharp, and this is also equivalent to sharpness of the localization $D[X]_{N_v}$ [1112.1236].

## 4. Quasidominance as an intermediate near-unanimity condition

In universal algebra, the term is used for the condition that the paper equates with **Lipparini’s $n+\frac12$-near-unanimity term**. Fix $n\ge 3$. A term $u$ of arity $n+2$ witnesses quasidominance if it satisfies the identities
\[
u(z,z,x,x,\dots,x)\approx x,
\]
\[
u(\underbrace{x,\dots,x}_{i-1},z,\underbrace{x,\dots,x}_{n+2-i})\approx x\qquad (2\le i\le n+2),
\]
and
\[
u(x,x,x,z,z,\dots,z)\approx u(x,z,z,z,z,\dots,z).
\]
The paper proves that this condition is strictly intermediate between having an $n$-ary near-unanimity term and having an $(n+1)$-ary near-unanimity term, with the precise implications
\[
NU(n)\Rightarrow \text{quasidominance}\Rightarrow NU(n+1),
\]
and both implications are strict [2105.14041].

Strictness is witnessed in two directions. For the failure of the reverse implication on the right, the variety $N_{n+1}$ has an $(n+1)$-ary near-unanimity term but is not $2n-3$-distributive, whereas quasidominance would force that level of congruence distributivity. For the failure of the reverse implication on the left, the paper constructs a Boolean term-reduct with operation
\[
u(x_1,x_2,\dots,x_{n+2}) := (x_1 + x_2') \cdot \prod_{2\le i<j\le n+2} (x_i + x_j),
\]
which satisfies the quasidominance identities but admits no $n$-ary near-unanimity term [2105.14041].

The main structural payoff is exact control of congruence identities. Every variety with quasidominance is $CD(2n-3)$ and $CM(2n-2)$, where $CD(k)$ denotes $k$-distributivity and $CM(k)$ denotes $k$-modularity. These bounds are sharp: the paper constructs a family $N_{n\frac12}$ for which quasidominance holds but $CD(2n-4)$ fails and $CM(2n-3)$ fails. The proofs proceed by extracting directed Jónsson/Gumm chains from the identities above and then converting them into Day terms. The paper also relates the condition to other Maltsev and Taylor notions: quasidominance implies $NU(n+1)$ and therefore implies Taylor, with the standard finite-idempotent CSP consequences that accompany near-unanimity polymorphisms [2105.14041].

## 5. Singular-value quasidominance for $M(2,\mathbb C)$-sequences

In the study of bounded matrix sequences $B:\mathbb Z\to M(2,\mathbb C)$, the paper "Equivalent Conditions for Domination of $\mathrm{M}(2,\mathbb{C})$-sequences" does not formally name a quasidominance condition, but the supplied data identifies the pair **(SVG)+(FI)** as functioning in precisely that role. Write
\[
A^{(n)}(j)=B(j+n-1)\cdots B(j),\qquad s_1(A)\ge s_2(A)\ge 0
\]
for the singular values. The relevant conditions are:

\[
\sup_{j\in\mathbb Z}\max\left\{\frac{s_2(A^{(n)}(j))}{s_1(A^{(n+1)}(j))},\frac{s_2(A^{(n)}(j+1))}{s_1(A^{(n+1)}(j))}\right\}\le C\mu^{-n}\qquad (n\ge 0),
\]
the **singular-value gap** condition (SVG), and
\[
\sup_{j\in\mathbb Z}\max\left\{\frac{s_1(A^{(n)}(j))}{s_1(A^{(n+1)}(j))},\frac{s_1(A^{(n)}(j+1))}{s_1(A^{(n+1)}(j))}\right\}\le C\mu^{(1-\varepsilon)n}\qquad (n\ge 1),
\]
the **fast invertibility** condition (FI). The main theorem states that a bounded sequence $B$ has dominated splitting if and only if it satisfies both (SVG) and (FI) [2501.15940].

Dominated splitting is formulated for possibly singular sequences by requiring one-dimensional subspaces $E^u(j),E^s(j)\subseteq \mathbb C^2$ with $B$-invariance, domination of the form
\[
\|A^{(N)}(j)u(j)\|\ge \lambda \|A^{(N)}(j)s(j)\|,
\]
uniform separation $\inf_j d(E^u(j),E^s(j))>0$, and uniform nondegeneracy of blocks. Under dominated splitting, one gets a uniform exponential singular-value gap:
\[
s_1(A^{(n)}(j))\ge c\mu^n s_2(A^{(n)}(j)).
\]
Conversely, from (SVG)+(FI) the paper constructs invariant line fields as limits of most contracted and most expanded directions, proves uniform separation, and recovers the domination estimate [2501.15940].

This formulation clarifies the relationship with classical invertible settings. For bounded $\mathrm{SL}(2,\mathbb C)$-sequences, domination reduces to uniform hyperbolicity and (SVG) is equivalent to uniform exponential growth. For bounded $\mathrm{GL}(2,\mathbb C)$-sequences with determinants uniformly bounded away from zero, domination is equivalent to (SVG) alone; in the general $M(2,\mathbb C)$ setting, however, (FI) is necessary, and the paper gives an example where (SVG) holds but uniform separation fails. The same work also proves an **Avalanche Principle** for possibly singular matrices: under local gap and local alignment assumptions, the sequence has dominated splitting and the quantity $\log s_1(A^{(n)}(j))$ is approximated by a telescoping combination of one-step and two-step terms with error bounded by $C n\mu^{-2}$ [2501.15940].

## 6. Other specialized meanings

A further usage appears in the theory of **quasi principal rank characteristic sequences** of symmetric matrices. If $qpr(A)=q_1q_2\cdots q_n$ with symbols in $\{A,S,N\}$, the paper proves the **N Theorem**: if $q_k=N$ for some $k$, then $q_j=N$ for all $j\ge k$. Equivalently, once $N$ appears, the sequence is an $N$-tail, so the substrings $NA$ and $NS$ cannot occur. Over fields of characteristic $0$, this tail property together with the terminal constraint $q_n\in\{A,N\}$ completely characterizes attainable qpr-sequences [1711.01003].

In probability theory, the paper on strong laws for pairwise positively quadrant dependent random variables does not use the term directly, but the supplied data identifies a **quasi-dominance requirement** in the form of stochastic domination or necessary tail control. Under pairwise PQD and the dependence summability condition
\[
\sum_{1\le k<j<\infty}(kj)^{-1/p}G_{X_k,X_j}\big(k^{1/p},j^{1/p}\big)<\infty,
\]
if
\[
\frac{1}{n^{1/p}}\sum_{k=1}^n (X_k-c)\xrightarrow{\text{a.s.}}0
\]
for some finite $c$ and $1<p<2$, then necessarily
\[
\sum_{k=1}^\infty \mathbb P\big(|X_1|>k^{1/p}\big)<\infty,
\]
equivalently $\mathbb E|X_1|^p<\infty$. In the sufficiency results cited there, the relevant domination hypothesis is that the sequence is **stochastically dominated** by a random variable $X\in L^p$:
\[
\sup_{n\ge 1}\mathbb P(|X_n|>t)\le C\,\mathbb P(|X|>t)\qquad (t>0).
\]
Here the notion is not algebraic or combinatorial; it is a tail-comparison condition controlling large deviations in dependent sequences [2004.02949].

In potential theory, Riihentaus extends domination results of Domar and Rippon from subharmonic to quasinearly subharmonic functions. Let $\mathcal F_K$ be a family of $K$-quasinearly subharmonic functions with $u(x)\le F_K(x)$, where $F_K$ is Lebesgue measurable. Under structural hypotheses on increasing functions $\varphi,\psi$—including the inequality
\[
2K(\psi\circ\varphi)(s-s_0)\le (\psi\circ\varphi)(s)\qquad (s\ge s_1)
\]
and the convergence condition
\[
\int_{s_1}^{+\infty}\frac{ds}{[\varphi(s-s_0)]^{1/(n-1)}}<+\infty,
\]
together with local integrability of $\psi(F_K)$ on compact sets, the family is locally uniformly bounded and the upper semicontinuous regularization of
\[
w(x)=\sup_{u\in\mathcal F_K}u^+(x)
\]
is $K$-quasinearly subharmonic. In this setting the quasidominance idea is a domination-by-majorant principle strong enough to guarantee regularity of the envelope [1102.1136].

Taken together, these usages show that **quasidominance condition** functions as a mathematically portable label rather than a single doctrine. In graph theory it bounds the number of dominators; in ideal theory it forces exact factor-lifting; in universal algebra it isolates a half-step between two near-unanimity arities; in matrix dynamics it becomes quantified singular-value separation; in qpr theory it is an irreversible tail phenomenon; and in probability and potential theory it appears as tail or majorant control. The phrase therefore names a family of analogous strategies—domination with additional quantitative restraint—rather than a unique formalism.

Source: https://www.emergentmind.com/topics/quasidominance-condition