The Paired Domination Number of Cubic Graphs

If \mathcal{C}=S_u is a solo A pair, then f(S_u)\geq 2, because S_u have at least 2 private neighbors.

If \mathcal{C}=S_u is a solo BC pair, then f(S_u)\geq 1+ 1/6*3=3/2, because u or \overline{u} has a private neighbor, and each edge in [B\cup C,V\setminus S] has weight at least 1/6.

If \mathcal{C} consists of two BC pair S_u and S_v with uv\in E(G), then according to Lemma 2, there is a vertex x\in V\setminus S, such that N_S(x)=\{\overline{u}, \overline{v}\}. Thus f(S_u)+f(S_v)\geq f(u')+f(v')+f(x)=3.

If \mathcal{C} consists of a(\geq 1) A pairs and b(\leq 2) BC pairs, then the total weight of it is at least 2a+b+b/3\geq 3(a+b)/2. Here we use the fact that each vertex in C has two edges to V\setminus S, which totally have weight 1/3.

If S_u is an A pair in which u has no private neighbor, then \overline{u} has at least two private neighbors. Let \overline{u}' be a private neighbor of \overline{u}. If S_u is not a solo pair, then d_S(u)\geq 2. It follows that replacing \{u\overline{u}\} with \{\overline{u}\overline{u}'\} gives a PDS with smaller \lambda, and so S_u is a solo pair.

Let S_u be a D pair such that d_A(u)=d_B(u)=d_D(u)=1. Suppose uv, uw\in E(G), where v\in B and w\in A. Then according to Lemma 2, there is a vertex x\in V\setminus S, such that N_S(x)=\{\overline{u},\overline{v}\}. Replacing \{u\overline{u}, v\overline{v}\} with \{vv', x\overline{u}\} gives a PDS with smaller \lambda, a contradiction.

Note that \mathcal{C} contains no edge between D pairs. If d\geq 3, then \mathcal{C} contains three D pairs S_u, S_v, S_w, such that there are two A pairs S_p, S_q with \{up, \overline{p}\overline{v}, \overline{q}\overline{v}, qw\}\subseteq E(G) and every vertex in \{p,\overline{p},q,\overline{q}\} has at least one private neighbor, according to Lemma 5. Then replacing \{u\overline{u}, p\overline{p}, v\overline{v}\} with \{up,\overline{p}\overline{v}\} gives a smaller PDS, or there is a vertex t_1\in V\setminus S, such that N_S(t_1)=\{\overline{u},v\}, according to Lemma 2. Similarly, there is a vertex t_2\in V\setminus S, such that N_S(t_2)=\{v,\overline{w}\}. Then replacing \{u\overline{u},v\overline{v}\} with \{t_1\overline{u}\} gives a smaller PDS, a contradiction. Therefore, the number of D pairs in \mathcal{C} is at most 2.

According to Lemma 1, replacing \{u\overline{u},w\overline{w}\} with \{tw\} gives a smaller PDS, or there is a vertex t_1\in V\setminus S, such that N_S(t_1)=\{u,\overline{w}\}.

According to Lemma 1, replacing \{ u\overline{u},w\overline{w},v\overline{v}\} with \{rw,x\overline{v}\} gives a smaller PDS, or there is a vertex vertex t_1\in V\setminus S, such that N_S(t_1)\subseteq \{\overline{u},v,\overline{w}\}.

Observe that replacing \{ u\overline{u},w_1\overline{w_1},w_2\overline{w_2} \} with \{xw_1,yw_2 \} gives a smaller PDS, or there is a vertex t\in V\setminus S, such that N_S(t)\subseteq \{u,\overline{u},\overline{w_1},\overline{w_2}\}.

Suppose on the contrary, \{w_1, w_2\} is a constraint pair. Then there is a vertex t\in V\setminus S, such that N_S(t)\subseteq \{ u,\overline{u}, \overline{w_1}, \overline{w_2} \}. If \overline{w_1} already has degree 3, then N_S(t)\subseteq \{u,\overline{u},\overline{w_2}\}. But u and \overline{u} have no common neighbor, it is only possible that N_S(t)\subseteq \{u,\overline{w_2}\} or N_S(t)\subseteq \{\overline{u},\overline{w_2}\}. As edges in [S_u,V\setminus S] have weight not equal 1/2, it is not possible that N_S(t)= \{u,\overline{w_2}\} or N_S(t)=\{\overline{u},\overline{w_2}\}. Moreover, t is not a private neighbor of any vertex in \{u,\overline{u},\overline{w_2}\}, which is a subset of D. Above all, it is not possible for the existence of t. The argument also works for the case when \overline{w_2} already has degree 3. According to Lemma 10, \{w_1, w_2\} can not be a constraint pair.