Roman Dominating Functions in Graph Theory

• Roman dominating functions are graph labelings that assign values from {0,1,2} such that every vertex labeled 0 is adjacent to a vertex labeled 2, ensuring defense constraints.
• They quantify graph stability by defining parameters like the Roman domination number and the NP-hard Roman bondage number, which assess structural vulnerability.
• Applications include delineating stability classes (e.g., R_UVR), constructing critical graphs with unique independent dominating sets, and establishing differential invariants.

A Roman dominating function is a fundamental structure in graph theory, combining elements of classical domination with combinatorial labeling governed by military-style defense constraints. This concept underlies a rich theory with connections to criticality, vulnerability, differential stability, and precise graph constructions.

1. Formal Definition and Key Parameters

A Roman dominating function (RDF) for a finite simple graph $G=(V,E)$ is a labeling $f: V \to \{0,1,2\}$ such that every vertex $v$ with $f(v)=0$ has at least one neighbor $u$ with $f(u)=2$ (Samodivkin, 2015). The weight of $f$ is $f(V)=\sum_{v\in V} f(v) = |V_1^f| + 2|V_2^f|$, with $V_i^f = \{ v \in V : f(v) = i \}$ partitioning $V$.

The Roman domination number $\gamma_R(G)$ is the minimum weight of an RDF on $G$. Cockayne et al. established the sharp bounds $\gamma(G)\leq \gamma_R(G)\leq 2\gamma(G)$, where $\gamma(G)$ is the ordinary domination number. An RDF attaining this minimum is called a $\gamma_R$-function.

The Roman bondage number $b_R(G)$ measures edge vulnerability: it is the minimum cardinality of all sets $F\subseteq E$ such that $\gamma_R(G-F)>\gamma_R(G)$. Computing $b_R(G)$ is NP-hard, even for bipartite graphs.

2. Structural Properties and Special Classes

For any $v\in V$, one may compare $\gamma_R(G-v)$ and $\gamma_R(G)$. The graph $G$ is in the stability class $\mathcal{R}_{UVR}$ (“vertex deletion unchanging”) if $\gamma_R(G-v)=\gamma_R(G)$ for all $v\in V$ (Samodivkin, 2015, Samodivkin, 2017). This is equivalent to: in no $\gamma_R$-function does any vertex receive label $1$ (“no labelling with value $1$ is allowed for any vertex in any optimal RDF”). Consequently, every $G\in\mathcal{R}_{UVR}$ is a “Roman” graph in the original sense, and satisfies $\gamma_R(G)=2\gamma(G)$.

Graphs in $\mathcal{R}_{UVR}$ exhibit the property that every vertex $v\in V_2^{f}$ (vertices labeled $2$) has at least three private neighbors; $V_2^f$ forms an independent dominating set, and every $v\in V_2^f$ is “defendable” by exactly those private neighbors.

3. Upper Bounds and Extremal Properties

For $G\in\mathcal{R}_{UVR}$ connected of order $n$, $\gamma_R(G)\leq 2n/3$ holds (Samodivkin, 2015). Equality occurs if and only if every $\gamma_R$-function yields an efficient dominating set with each relevant vertex having degree $2$.

The Roman bondage number in this class admits $b_R(G)\leq \delta(G)$, the minimal degree. Deleting all edges incident to a minimum-degree vertex isolates it, forcing an increase in the Roman domination number through necessary use of label $1$.

In the context of Mycieleskian graphs, the Roman domination number of the $m$-Mycieleskian $\mu_m(G)$ for a special Roman graph is given precisely by a period-$4$ formula, depending on $\gamma_R(G)$ and $m$ (Kazemi, 2011).

4. Characterizations and Construction Methods

For trees, $G\in\mathcal{R}_{UVR}$ if and only if $G$ admits a labeling using statuses $A,B,C$ constructed via four specific operations starting from $K_{1,2}$ (O1–O4) (Samodivkin, 2015). In such trees, the set $S_B$ of status-$B$ vertices is the unique independent dominating set, each with three private neighbors.

Tree Characterization Equivalences:

• (i) Construction via O1–O4 yields all trees in the class $\mathcal{T}$.
• (ii) $T\in\mathcal{R}_{UVR}$.
• (iii) Unique $\gamma_R$-function with $V_1^f=\emptyset$, $V_2^f$ independent, and each $v\in V_2^f$ has three private neighbors.
• (iv) Unique independent dominating set with three private neighbors per member.
• (v) Differential stability: the graph differential $\partial(G)$ remains unchanged under vertex deletion.

Minimum-edge cases: For $n\notin\{4,5,8\}$, the minimum-edge graphs in $\mathcal{R}_{UVR}$ are precisely the trees constructed above (with $n-1$ edges).

5. Connections to Criticality and Differential Stability

The stability class $\mathcal{R}_{UVR}$ coincides with graphs that are “differential-stable”: for these graphs, the quantity $\partial(G)$ (maximum $|B(S)| - |S|$ over all subsets $S \subset V$) remains unchanged when any vertex is deleted. A direct relation $\gamma_R(G)+\partial(G)=|V|$ holds (Samodivkin, 2015).

Lemmas underpin the combinatorial logic: deleting a vertex $v$ lowers $\gamma_R$ precisely when $v$ is labeled $1$ in some optimal assignment; the absence of any label-$1$ in all $\gamma_R$-functions implies differential and Roman domination stability.

6. Proof Techniques and Combinatorial Invariants

Key mechanisms include:

• Partitioning $V$ into $V_0^f$ and $V_2^f$, and counting private neighborhoods to derive inequalities.
• Recursive and constructive labeling schemes (O1–O4) for trees, yielding full control over degree and neighborhood structure in each step.
• Use of differential arguments to relate the Roman domination number to broader combinatorial invariants.
• Inductive proofs showing extension of unique $\gamma_R$-functions through tree-growing operations, maintaining stability throughout.

7. Significance and Research Directions

Roman dominating functions encapsulate a fusion of local combinatorial defense and global optimality. Their properties under vertex and edge deletion have yielded structural graph classes, sharp bounds for domination parameters, and full constructive characterizations for critical families such as $\mathcal{R}_{UVR}$-trees. The differential-stability equivalence frames Roman domination in terms of classical combinatorial optimization.

Current research continues to refine extremal bounds, extend the concept to weighted graphs, higher domination thresholds, and study computational complexity—most notably in characterizations of critical and stable graphs, and discoveries of new invariants connecting domination, differential, and Roman parameters (Samodivkin, 2015, Kazemi, 2011).