This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

On the intersection ideal graph of semigroups

Barkha Baloda, Jitender Kumar\text{Jitender Kumar}^{{}^{*}} Department of Mathematics, Birla Institute of Technology and Science Pilani, Pilani, India barkha0026@gmail.com,jitenderarora09@gmail.com
Abstract.

The intersection ideal graph Γ(S)\Gamma(S) of a semigroup SS is a simple undirected graph whose vertices are all nontrivial left ideals of SS and two distinct left ideals I,JI,J are adjacent if and only if their intersection is nontrivial. In this paper, we investigate the connectedness of Γ(S)\Gamma(S). We show that if Γ(S)\Gamma(S) is connected then diam(Γ(S))2diam(\Gamma(S))\leq 2. Further we classify the semigroups such that the diameter of their intersection graph is two. Other graph invariants, namely perfectness, planarity, girth, dominance number, clique number, independence number etc. are also discussed. Finally, if SS is union of nn minimal left ideals then we obtain the automorphism group of Γ(S)\Gamma(S).

Key words and phrases:
Semigroup, ideals, clique number, graph automorphism
* Corresponding author
2010 Mathematics Subject Classification:
05C25

1. Introduction

Literature is abound with numerous remarkable results concerning a number of constructions of graphs from rings, semigroups or groups. The intersection graph of a semigroup was introduced by Bosák [7] in 19641964. The intersection subsemigroup graph Γ(S)\Gamma(S) of SS is an undirected simple graph whose vertex set is the collection of proper subsemigroups of SS and two distinct vertices A,BA,B are adjacent if and only if ABA\cap B\neq\emptyset. In [7], it was shown that if SS is a nondenumerable semigroup or a periodic semigroup with more than two elements, then the graph Γ(S)\Gamma(S) is connected. Bosák then raised the following open problem: Does there exists a semigroup with more than two elements whose graph is disconnected? Y. F. Lin [24], answer the problem posed by Bosák, in the negative manner and proved that every semigroup with more than two elements has a connected graph. Also, B. Ponděliček [27] proved that the diameter of a semigroup with more than two elements does not exceed three.

Inspired by the work of J. Bosák , Csákány and Pollák [12] studied the intersection graphs of groups and showed that there is an edge between two proper subgroups if they have at least two elements common. Further, Zelinka [35], continued the work for finite abelian groups. R. Shen [30], characterized all finite groups whose intersection graphs are disconnected. This solves the problem posed in [12]. The groups whose intersection graphs of normal subgroups are connected, complete, forests or bipartite are classified in [16]. Tamizh et al. [31], continued the seminal paper of Csákány and Pollák to introduce the subgroup intersection graph of a finite group GG. Further, in [25], it was shown that the diameter of intersection graph of a finite non-abelian simple group has an upper bound 2828. Shahsavari et al. [29] have studied the structure of the automorphism group of this graph. The intersection graph on cyclic subgroups of a group has been studied in [14]. Further, Kayacan et al. [22] studied the conjecture given in [35], that two (noncyclic) finite abelian groups with isomorphic intersection graphs are isomorphic. In [20], finite solvable groups whose intersection graphs are not 2-connected, finite nilpotent groups whose intersection graphs are not 3-connected is classified. Further, the dominating sets of the intersection graph of finite groups is investigated in [21].

Recently, Chakrabarty et al. [8] introduced the notion of intersection ideal graph of rings. The intersection ideal graph Γ(R)\Gamma(R) of a ring RR is an undirected simple graph whose vertex set is the collection of nontrivial left ideals of RR and two distinct vertices I,JI,J are adjacent if and only if IJ{0}I\cap J\neq\{0\}. They characterized the rings RR for which the graph Γ(R)\Gamma(R) is connected and obtain several necessary and sufficient conditions on a ring RR such that Γ(R)\Gamma(R) is complete. Planarity of intersection graphs of ideals of ring with unity is described in [15] and domination number in [17]. Akbari et al. [4] classified all rings whose intersection graphs of ideals are not connected and also determined all rings whose clique number is finite. The intersection graphs of ideals of direct product of rings have been discussed in [19]. Pucanovic et al. [28] classified all graphs of genus two that are intersection graphs of ideals of some commutative rings and obtain some lower bounds for the genus of the intersection graph of ideals of a non local commutative ring. In [13], Das characterized the positive integer nn for which the intersection graph of ideals of n\mathbb{Z}_{n} is perfect. The Intersection graph for submodules of modules have been studied in [5, 6, 34]. The intersecton graph on algebraic structures have also been studied in [1, 2, 3, 18, 23, 33].

It is pertinent as well as interesting to associate graphs to ideals of a semigroup as ideals gives a lot of information about the structure of semigroups. Motivated with the work of [4, 8], in this paper, we consider the intersection ideal graph associated with semigroups. The intersection ideal graph Γ(S)\Gamma(S) of a semigroup SS is an undirected simple graph whose vertex set is nontrivial left ideals of SS and two distinct nontrivial left ideals I,JI,J are adjacent if and only if their intersection is nontrivial. The paper is arranged as follows. In Section 2, we state necessary fundamental notions and recall some necessary results. Section 3 comprises the results concerning the connectedness of intersection ideal graph of an arbitrary semigroup. In Section 4, we study various graph invariants of Γ(S)\Gamma(S) viz. girth, dominance number, independence number and clique number etc. Further, if SS is union of nn minimal left ideals then the automorphism group of Γ(S)\Gamma(S) is obtained.

2. Preliminaries

In this section, first we recall necessary definitions and results of semigroup theory from [11]. A semigroup SS is a non-empty set together with an associative binary operation on SS. The Green’s \mathcal{L}-relation on a semigroup SS defined as xx \mathcal{L} yS1x=S1yy\Longleftrightarrow S^{1}x=S^{1}y where S1x=Sx{x}S^{1}x=Sx\cup\{x\}. The \mathcal{L}-class of an element aSa\in S is denoted by LaL_{a}. A non-empty subset II of SS is said to be a left [right] ideal if SII[ISI]SI\subseteq I[IS\subseteq I] and an ideal of SS if SISISIS\subseteq I. Union of two left [right] ideals of SS is again a left [right] ideal of SS. A left ideal II is maximal if it does not contained in any nontrivial left ideal of SS. If SS has a unique maximal left ideal then it contains every nontrivial left ideal of SS. A left ideal II of SS is minimal if it does not properly contain any left ideal of SS. It is well known that every non-zero element of a minimal left ideal of SS is in same \mathcal{L}-class. If SS has a minimal left ideal then every nontrivial left ideal contains at least one minimal left ideal. If AA is any other left ideal of SS other than II, then either IAI\subset A or IA=I\cap A=\emptyset. Thus we have the following remark.

Remark 2.1.

Any two different minimal left ideals of a semigroup SS are disjoint.

Remark 2.2.

Let SS be union of nn minimal left ideals. Then each nontrivial left ideal is union of these minimal left ideals.

The following lemma is useful in the sequel and we shall use this without referring to it explicitly.

Lemma 2.3.

A left ideal KK of SS is maximal if and only if SKS\setminus K is an \mathcal{L}-class.

Proof.

First suppose that SKS\setminus K is an \mathcal{L}-class. Let if possible, KK is not maximal left ideal of SS. Then there exists a nontrivial left ideal KK^{\prime} of SS such that KKK\subset K^{\prime}. There exists aKa\in K^{\prime} but aKa\notin K. Thus, La=SKL_{a}=S\setminus K. Consequently, LaKL_{a}\subset K^{\prime} gives S=KS=K^{\prime}, a contradiction. Conversely, suppose that KK is a maximal left ideal of SS. For each aSKa\in S\setminus K, maximality of KK implies KS1a=SK\cup S^{1}a=S. Consequently, aa \mathcal{L} bb for every a,bSKa,b\in S\setminus K. Thus SKS\setminus K is contained in some \mathcal{L}-class and this \mathcal{L}-class is disjoint from KK. It follows that SKS\setminus K is an \mathcal{L}-class. ∎

We also require the following graph theoretic notions [32]. A graph Γ\Gamma is a pair Γ=(V,E)\Gamma=(V,E), where V=V(Γ)V=V(\Gamma) and E=E(Γ)E=E(\Gamma) are the set of vertices and edges of Γ\Gamma, respectively. We say that two different vertices u,vu,v are 𝑎𝑑𝑗𝑎𝑐𝑒𝑛𝑡\mathit{adjacent}, denoted by uvu\sim v or (u,v)(u,v), if there is an edge between uu and vv. We write uvu\nsim v, if there is no edge between uu and vv. The distance between two vertices u,vu,v in Γ\Gamma is the number of edges in a shortest path connecting them and it is denoted by d(u,v)d(u,v). If there is no path between uu and vv, we say that the distance between uu and vv is infinity and we write as d(u,v)=d(u,v)=\infty. The diameter diam(Γ)diam(\Gamma) of Γ\Gamma is the greatest distance between any pair of vertices. The degree of the vertex vv in Γ\Gamma is the number of edges incident to vv and it is denoted by deg(v)deg(v). A cycle is a closed walk with distinct vertices except for the initial and end vertex, which are equal and a cycle of length nn is denoted by CnC_{n}. The girth of Γ\Gamma is the length of its shortest cycle and is denoted by g(Γ){g(\Gamma)}. A subset XX of V(Γ)V(\Gamma) is said to be independent if no two vertices of XX are adjacent. The independence number of Γ\Gamma is the cardinality of the largest independent set and it is denoted by α(Γ)\alpha(\Gamma). A graph Γ\Gamma is bipartite if V(Γ)V(\Gamma) is the union of two disjoint independent set. It is well known that a graph is bipartite if and only if it has no odd cycle [32, Theorem 1.2.18]. A connected graph Γ\Gamma is Eulerian if and only if degree of every vertex is even [32, Theorem 1.2.26]. A subgraph of Γ\Gamma is a graph Γ\Gamma^{\prime} such that V(Γ)V(Γ)V(\Gamma^{\prime})\subseteq V(\Gamma) and E(Γ)E(Γ)E(\Gamma^{\prime})\subseteq E(\Gamma). A subgraph Γ\Gamma^{\prime} of Γ\Gamma is called an induced subgraph by the elements of V(Γ)V(Γ)V(\Gamma^{\prime})\subseteq V(\Gamma) if for u,vV(Γ)u,v\in V(\Gamma^{\prime}), we have uvu\sim v in Γ\Gamma^{\prime} if and only if uvu\sim v in Γ\Gamma. The chromatic number of Γ\Gamma, denoted by χ(Γ)\chi(\Gamma), is the smallest number of colors needed to color the vertices of Γ\Gamma so that no two adjacent vertices share the same color. A clique in Γ\Gamma is a set of pairwise adjacent vertices. The clique number of Γ\Gamma is the size of maximum clique in Γ\Gamma and it is denoted by ω(Γ)\omega(\Gamma). It is well known that ω(Γ)χ(Γ)\omega(\Gamma)\leq\chi(\Gamma) (see [32]). A graph Γ\Gamma is perfect if ω(Γ)=χ(Γ)\omega(\Gamma^{\prime})=\chi(\Gamma^{\prime}) for every induced subgraph Γ\Gamma^{\prime} of Γ\Gamma. Recall that the complement Γ¯\overline{\Gamma} of Γ\Gamma is a graph with same vertex set as Γ\Gamma and distinct vertices u,vu,v are adjacent in Γ¯\overline{\Gamma} if they are not adjacent in Γ\Gamma. A subgraph Γ\Gamma^{\prime} of Γ\Gamma is called hole if Γ\Gamma^{\prime} is a cycle as an induced subgraph, and Γ\Gamma^{\prime} is called an antihole of Γ\Gamma if Γ¯\overline{\Gamma^{\prime}} is a hole in Γ¯\overline{\Gamma}.

Theorem 2.4.

[10] A finite graph Γ\Gamma is perfect if and only if it does not contain hole or antihole of odd length at least 55.

A subset DD of V(Γ)V(\Gamma) is said to be a dominating set if any vertex in V(Γ)DV(\Gamma)\setminus D is adjacent to at least one vertex in DD. If DD contains only one vertex then that vertex is called dominating vertex. The domination number γ(Γ)\gamma(\Gamma) of Γ\Gamma is the minimum size of a dominating set in Γ\Gamma. A graph Γ\Gamma is said to be planar if it can be drawn on a plane without any crossing of its edges. In Γ\Gamma, a vertex zz resolves a pair of distinct vertices xx and yy if d(x,z)d(y,z)d(x,z)\neq d(y,z). A resolving set of Γ\Gamma is a subset RV(Γ)R\subseteq V(\Gamma) such that every pair of distinct vertices of Γ\Gamma is resolved by some vertex in RR. The metric dimension of Γ\Gamma, denoted by β(Γ)\beta(\Gamma), is the minimum cardinality of a resolving set of Γ\Gamma. For vertices uu and vv in a graph Γ\Gamma, we say that zz strongly resolves uu and vv if there exists a shortest path from zz to uu containing vv, or a shortest path from zz to vv containing uu. A subset UU of V(Γ)V(\Gamma) is a strong resolving set of Γ\Gamma if every pair of vertices of Γ\Gamma is strongly resolved by some vertex of UU. The least cardinality of a strong resolving set of Γ\Gamma is called the strong metric dimension of Γ\Gamma and is denoted by sdim(Γ)\operatorname{sdim}(\Gamma). For vertices uu and vv in a graph Γ\Gamma, we write uvu\equiv v if N[u]=N[v]N[u]=N[v]. Notice that that \equiv is an equivalence relation on V(Γ)V(\Gamma). We denote by v^\widehat{v} the \equiv-class containing a vertex vv of Γ\Gamma. Consider a graph Γ^\widehat{\Gamma} whose vertex set is the set of all \equiv-classes, and vertices u^\widehat{u} and v^\widehat{v} are adjacent if uu and vv are adjacent in Γ\Gamma. This graph is well-defined because in Γ\Gamma, wvw\sim v for all wu^w\in\widehat{u} if and only if uvu\sim v. We observe that Γ^\widehat{\Gamma} is isomorphic to the subgraph Γ\mathcal{R}_{\Gamma} of Γ\Gamma induced by a set of vertices consisting of exactly one element from each \equiv-class. Subsequently, we have the following result of [26] with ω(Γ)\omega(\mathcal{R}_{\Gamma}) replaced by ω(Γ^)\omega(\widehat{\Gamma}).

Theorem 2.5 ([26, Theorem 2.2]).

For any graph Γ\Gamma with diameter 22, sdim(Γ)=|V(Γ)|ω(Γ^)\operatorname{sdim}(\Gamma)=|V(\Gamma)|-\omega(\widehat{\Gamma}).

3. Connectivity of the Intersection graph Γ(S)\Gamma(S)

In this section, we investigate the connectedness of Γ(S)\Gamma(S). We show that diam(Γ(S))2diam(\Gamma(S))\leq 2 if it is connected. Also, we classify the semigroups, in terms of left ideals, such that the diameter of Γ(S)\Gamma(S) is two.

Theorem 3.1.

The intersection ideal graph Γ(S)\Gamma(S) is disconnected if and only if SS contains at least two minimal left ideals and every nontrivial left ideal of SS is minimal as well as maximal.

Proof.

First suppose that Γ(S)\Gamma(S) is not connected. Then SS has at least two nontrivial left ideals, namely I1,I2I_{1},I_{2}. Without loss of generality, assume that I1C1I_{1}\in C_{1} and I2C2I_{2}\in C_{2}, where C1C_{1} and C2C_{2} are distinct components of Γ(S)\Gamma(S). If I1I_{1} is not minimal then there exists at least one nontrivial left ideal IkI_{k} of SS such that IkI1I_{k}\subset I_{1} so that their intersection is nontrivial. Therefore, I1IkI_{1}\sim I_{k}. Now if the intersection of I2I_{2} and IkI_{k} is nontrivial then I1IkI2I_{1}\sim I_{k}\sim I_{2}, a contradiction. Therefore the intersection of I2I_{2} and IkI_{k} is trivial. If I2IkSI_{2}\cup I_{k}\neq S then I1I2IkI2I_{1}\sim I_{2}\cup I_{k}\sim I_{2}, a contradiction. Thus, IkI2=SI_{k}\cup I_{2}=S. It follows that I1I2I_{1}\sim I_{2}, again a contradiction. Thus I1I_{1} is minimal. Similarly, we get I2I_{2} is minimal.

Further assume that I1I_{1} is not maximal. Then there exists a nontrivial left ideal IkI_{k} of SS such that I1IkI_{1}\subset I_{k} so that I1IkI_{1}\sim I_{k}. If I1I2SI_{1}\cup I_{2}\neq S then I1I1I2I2I_{1}\sim I_{1}\cup I_{2}\sim I_{2}, a contradiction to the fact that Γ(S)\Gamma(S) is disconnected. It follows that I1I2=SI_{1}\cup I_{2}=S so that the intersection of IkI_{k} and I2I_{2} is nontrivial. Thus we have I1IkI2I_{1}\sim I_{k}\sim I_{2}, a contradiction. Hence I1I_{1} is maximal. Similarly, we observe that I2I_{2} is maximal. The converse follows from the Remark 2.1. ∎

Corollary 3.2.

If the graph Γ(S)\Gamma(S) is disconnected then it is a null graph (i.e. it has no edge).

Theorem 3.3.

The graph Γ(S)\Gamma(S) is disconnected if and only if SS is the union of exactly two minimal left ideals.

Proof.

Suppose first that Γ(S)\Gamma(S) is disconnected. Then by Theorem 3.1, each nontrivial left ideal of SS is minimal. Suppose SS has at least three minimal left ideals, namely I1,I2I_{1},I_{2} and I3I_{3}. Then I1I2I_{1}\cup I_{2} is a nontrivial left ideal of SS which is not minimal. Consequently, by Theorem 3.1, we get a contradiction of the fact that Γ(S)\Gamma(S) is disconnected. Thus, SS has exactly two minimal left ideals. If SI1I2S\neq I_{1}\cup I_{2}, then I1I2I_{1}\cup I_{2} is a nontrivial left ideal which is not minimal, a contradiction ( cf. Theorem 3.1). Thus, S=I1I2S=I_{1}\cup I_{2}.

Conversely, suppose S=I1I2S=I_{1}\cup I_{2} where I1I_{1} and I2I_{2} are minimal left ideals of SS. If there exists another nontrivial left ideal IkI_{k} of SS then either I1IkI_{1}\subset I_{k} or I2IkI_{2}\subset I_{k}. Without loss of generality, assume that I1IkI_{1}\subset I_{k}, we have I1IkI_{1}\sim I_{k}. Since I1I2=SI_{1}\cup I_{2}=S we get IkI2=SI_{k}\cup I_{2}=S. It follows that the intersection of I2I_{2} and IkI_{k} is nontrivial. By minimality of I2I_{2}, we can observe that I2IkI_{2}\subset I_{k}. Consequently, SIkS\subseteq I_{k}, a contradiction. Thus, by Theorem 3.1, Γ(S)\Gamma(S) is disconnected.

Theorem 3.4.

If Γ(S)\Gamma(S) is a connected graph then diam(Γ(S))diam(\Gamma(S)) \leq 22.

Proof.

Let I1,I2I_{1},I_{2} be two nontrivial left ideals of SS. If I1I2I_{1}\sim I_{2} then d(I1,I2)d(I_{1},I_{2}) = 1. If I1I2I_{1}\nsim I_{2} i.e. I1I2I_{1}\cap I_{2} is trivial then in the following cases we show that d(I1,I2)d(I_{1},I_{2})2\leq 2.

Case 1. I1I2SI_{1}\cup I_{2}\neq S. Then I1(I1I2)I2I_{1}\sim(I_{1}\cup I_{2})\sim I_{2} so that d(I1,I2)d(I_{1},I_{2}) = 2.

Case 2. I1I2=SI_{1}\cup I_{2}=S. Since Γ(S)\Gamma(S) is a connected graph, there exists a nontrivial left ideal IkI_{k} of SS such that either I1IkI_{1}\cap I_{k} is nontrivial or I2IkI_{2}\cap I_{k} is nontrivial. Now we have the following subcases.

Subcase 1. I1IkI_{1}\not\subset I_{k} and IkI1I_{k}\not\subset I_{1}. Since I1IkI_{1}\not\subset I_{k} it follows that there exists xIkx\in I_{k} but xI1x\notin I_{1} so that xI2x\in I_{2}. Consequently, I2IkI_{2}\cap I_{k} is nontrivial. Therefore, we get a path I1IkI2I_{1}\sim I_{k}\sim I_{2} of length two. Thus, d(I1,I2)=2d(I_{1},I_{2})=2.

Subcase 2. IkI1I_{k}\subset I_{1}. There exists xI1x\in I_{1} but xIkx\notin I_{k}. If I2Ik=SI_{2}\cup I_{k}=S then xI2x\in I_{2}. Thus, we get I1I2I_{1}\cap I_{2} is nontrivial, a contradiction. Consequently, I2IkSI_{2}\cup I_{k}\neq S. Further, we get a path I1(I2Ik)I2I_{1}\sim(I_{2}\cup I_{k})\sim I_{2} of length two. Thus, d(I1,I2)=2d(I_{1},I_{2})=2.

Subcase 3. I1IkI_{1}\subset I_{k}. Since I1I2=SI_{1}\cup I_{2}=S we get IkI2=SI_{k}\cup I_{2}=S. Further, the intersection of IkI_{k} and I2I_{2} is nontrivial. Consequently, I1IkI2I_{1}\sim I_{k}\sim I_{2} gives a path of length two between I1I_{1} and I2I_{2}. Thus, d(I1,I2)=2d(I_{1},I_{2})=2. Hence, diam(Γ(S))diam(\Gamma(S)) \leq 22.

Lemma 3.5.

Let SS be a semigroup having minimal left ideals. Then Γ(S)\Gamma(S) is complete if and only if SS has unique minimal left ideal.

Proof.

Suppose that SS contains a unique minimal left ideal I1I_{1}. Note that every nontrivial left ideal of SS contains at least one minimal left ideal. Since I1I_{1} is unique then it must contained in every nontrivial left ideals of SS. Thus, the graph Γ(S)\Gamma(S) is complete.

Conversely, suppose that Γ(S)\Gamma(S) is a complete graph. On contrary if SS has at least two minimal left ideals, viz. I1,I2I_{1},I_{2}. By Remark 2.1, I1I2I_{1}\nsim I_{2}, a contradiction to the fact that Γ(S)\Gamma(S) is complete. Thus SS has unique minimal left ideal. ∎

Lemma 3.6.

If graph Γ(S)\Gamma(S) is a regular if and only if either Γ(S)\Gamma(S) is null or a complete graph.

Proof.

First suppose that Γ(S)\Gamma(S) is not a null graph. Let if possible, SS has at least two minimal left ideals, namely I1,I2I_{1},I_{2}. Since Γ(S)\Gamma(S) is not a null graph then I1I_{1} and I1I2I_{1}\cup I_{2} forms a nontrivial left ideals of SS and I1(I1I2)I_{1}\sim(I_{1}\cup I_{2}). Suppose JJ is any nontrivial left ideal of SS such that JI1J\sim I_{1} then J(I1I2)J\sim(I_{1}\cup I_{2}). It follows that every nontrivial left ideal of SS which is adjacent with I1I_{1} is also adjacent with (I1I2)(I_{1}\cup I_{2}) and I2I1I2I_{2}\sim I_{1}\cup I_{2} but I2I1I_{2}\nsim I_{1} implies that deg(I1)<deg(I1I2)deg(I_{1})<deg(I_{1}\cup I_{2}), a contradiction. Therefore, Γ(S)\Gamma(S) is a complete graph. ∎

Next we classify the semigroups such that the diameter of intersection ideal graph Γ(S)\Gamma(S) is two.

Theorem 3.7.

Let SS be a semigroup having minimal left ideals. Then for a connected graph Γ(S)\Gamma(S), we have diam(Γ(S))=2diam(\Gamma(S))=2 if and only if SS has at least two minimal left ideals.

Proof.

Suppose that diam(Γ(S))=2diam(\Gamma(S))=2. Assume that I1I_{1} is the only minimal left ideal of SS. Since I1I_{1} is unique minimal left ideal then it is contained in all other nontrivial left ideals of SS. Therefore, for any nontrivial left ideals J,KJ,K, we have I1(JK)I_{1}\subset(J\cap K). Consequently, d(J,K)=1d(J,K)=1 for any J,KV(Γ(S))J,K\in V(\Gamma(S)). Therefore SS has at least two minimal left ideals. Conversely suppose that SS has at least two minimal left ideals, viz. I1,I2I_{1},I_{2}. Then by Remark 2.1, we have I1I2I_{1}\nsim I_{2}. Consequently, by Theorem 3.4, d(I1,I2)=2d(I_{1},I_{2})=2. Thus, diam(Γ(S))=2diam(\Gamma(S))=2. ∎

4. Invariants of Γ(S)\Gamma(S)

In this section, first we obtain the girth of Γ(S)\Gamma(S). Then we discuss planarity and perfectness of Γ(S)\Gamma(S). Also we classify the semigroup SS such that Γ(S)\Gamma(S) is bipartite, star graph and tree, respectively. Further, we investigate other graph invariants viz. clique number, independence number and strong metric dimension of Γ(S)\Gamma(S).

Theorem 4.1.

Let SS be a semigroup such that Γ(S)\Gamma(S) contains a cycle. Then g(Γ(S))=3g(\Gamma(S))=3.

Proof.

If Γ(S)\Gamma(S) is disconnected or a tree, then clearly g(Γ(S))=g(\Gamma(S))=\infty. Suppose that the semigroup SS has nn minimal left ideals. Now we prove the result through following cases.

Case 1. n=0n=0. If SS has no nontrivial left ideals then there is nothing to prove. Otherwise, there exists a chain of nontrivial left ideals of SS such that I1I2IkI_{1}\supset I_{2}\supset\cdots\supset I_{k}\supset\cdots. Thus, g(Γ(S))=3g(\Gamma(S))=3.

Case 2. n=1n=1. Suppose that I1I_{1} is the only minimal left ideal of SS. Since I1I_{1} is unique minimal left ideal then it is contained in all other nontrivial left ideals of SS. Therefore, for any nontrivial left ideals I,JI,J, we have I1IJI_{1}\subset I\cap J\neq\emptyset. If SS has at least three nontrivial left ideals then g(Γ(S))=3g(\Gamma(S))=3. Otherwise, g(Γ(S))=g(\Gamma(S))=\infty.

Case 3. n=2n=2. Let I1,I2I_{1},I_{2} be two minimal left ideals of SS. If I1I2=SI_{1}\cup I_{2}=S then by Theorem 3.3 and Corollary 3.2, g(Γ(S))=g(\Gamma(S))=\infty. If I1I2SI_{1}\cup I_{2}\neq S, then J=I1I2J=I_{1}\cup I_{2} is a nontrivial left ideal of SS. If SS has only these three, namely I1,I2I_{1},I_{2} and JJ, left ideals then we obtain I1JI2I_{1}\sim J\sim I_{2} so that g(Γ(S))=g(\Gamma(S))=\infty. Now suppose that SS has a nontrivial left ideal KK other than I1,I2I_{1},I_{2} and JJ. Since I1,I2I_{1},I_{2} are minimal left ideals of SS we have either I1KI_{1}\subset K or I2KI_{2}\subset K. Without loss of generality, assume that I1KI_{1}\subset K, then we get a triangle I1KJI1I_{1}\sim K\sim J\sim I_{1}. It follows that g(Γ(S))=3g(\Gamma(S))=3.

Case 4. n3n\geq 3. Let I1,I2,I3I_{1},I_{2},I_{3} be the minimal left ideals of SS. Then we have a cycle (I1I2)(I2I3)(I1I3)(I1I2)(I_{1}\cup I_{2})\sim(I_{2}\cup I_{3})\sim(I_{1}\cup I_{3})\sim(I_{1}\cup I_{2}) of length 3. Thus, g(Γ(S))=3g(\Gamma(S))=3. ∎

Let Min(S){\rm Min}(S) (Max(S){\rm Max}(S)) be the set of all minimal (maximal) left ideals of SS. For a nontrivial left ideal Ii1i2ikI_{{i_{1}}{i_{2}}\cdots{i_{k}}}, we mean Ii1Ii2IikI_{i_{1}}\cup I_{i_{2}}\cup\cdots\cup I_{i_{k}}, where Ii1,Ii2,,IikI_{i_{1}},I_{i_{2}},\cdots,I_{i_{k}} Min(S)\in{\rm Min}(S)

Theorem 4.2.

For the graph Γ(S)\Gamma(S), we have the following results:

  1. (i)

    If Γ(S)\Gamma(S) is planar then |Min(S)|3|{\rm Min}(S)|\leq 3.

  2. (ii)

    For S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}}, we have Γ(S)\Gamma(S) is planar if and only if n3n\leq 3.

Proof.

(i) Suppose that |Min(S)|=4|{\rm Min}(S)|=4 with Min(S)={I1,I2,I3,I4}{\rm Min}(S)=\{I_{1},I_{2},I_{3},I_{4}\}. Then note that the subgraph induced by the vertices I1,I12,I123,I14I_{1},I_{12},I_{123},I_{14} and I124I_{124} is isomorphic to K5K_{5}. Thus, Γ(S)\Gamma(S) is nonplanar.

(ii) The proof for Γ(S)\Gamma(S) is nonplanar for n4n\geq 4 follows from part (i). If n=2n=2 then by Corollary 3.2 and Theorem 3.3, Γ(S)\Gamma(S) is planar. For n=3n=3, Γ(S)\Gamma(S) is planar as shown in Figure 1.

Refer to caption
Figure 1. Planar drawing of Γ(S)\Gamma(S) for S=I123S=I_{123}

Theorem 4.3.

For the graph Γ(S)\Gamma(S), we have the following results:

  1. (i)

    If Γ(S)\Gamma(S) is a perfect graph then |Min(S)|4|{\rm Min}(S)|\leq 4.

  2. (ii)

    Let SS be the union of nn minimal left ideals. Then Γ(S)\Gamma(S) is perfect if and only if n4n\leq 4.

Proof.

(i) Suppose that |Min(S)|=5|{\rm Min}(S)|=5 with Min(S)={I1,I2,I3,I4,I5}{\rm Min}(S)=\{I_{1},I_{2},I_{3},I_{4},I_{5}\}. Note that I12I23I34I45I15I12I_{12}\sim I_{23}\sim I_{34}\sim I_{45}\sim I_{15}\sim I_{12} induces a cycle of length 5. Then by Theorem 2.4, Γ(S)\Gamma(S) is not perfect.

(ii) The proof for Γ(S)\Gamma(S) is not a perfect graph for n5n\geq 5 follows from part (i). If n=2n=2 then by Corollary 3.2 and Theorem 3.1, Γ(S)\Gamma(S) is disconnected. Thus, being a null graph, Γ(S)\Gamma(S) is perfect. For n{3,4}n\in\{3,4\}, we show that Γ(S)\Gamma(S) does not contain a hole or an antihole of odd length at least five (cf. Theorem 2.4). If n=3n=3, Γ(S)\Gamma(S) is perfect as shown in Figure 1. If n=4n=4 then from Figure 2 note that Γ(S)\Gamma(S) does not contain a hole or an antihole of odd length at least five.

Refer to caption
Figure 2.

Theorem 4.4.

Let SS be a semigroup having minimal left ideals such that V(Γ(S))>1V(\Gamma(S))>1. Then the following conditions are equivalent:

  1. (i)

    Γ(S)\Gamma(S) is star graph.

  2. (ii)

    Γ(S)\Gamma(S) is a tree.

  3. (iii)

    Γ(S)\Gamma(S) is bipartite.

  4. (iv)

    Either SS has exactly three nontrivial left ideals I1I_{1}, I2I_{2} and I12I_{12} such that I1I_{1} and I2I_{2} are minimal or SS has two nontrivial left ideals I1,I2I_{1},I_{2} such that I1I2I_{1}\subset I_{2}.

Proof.

We prove (ii), (iii) \Rightarrow (iv). The proof of remaining parts is straightforward. Suppose Γ(S)\Gamma(S) is a tree. Then clearly |Min(S)|2|{\rm Min}(S)|\leq 2. Otherwise, for minimal left ideals I1,I2,I3I_{1},I_{2},I_{3} we have I12I13I23I12I_{12}\sim I_{13}\sim I_{23}\sim I_{12} a cycle, a contradiction. Suppose that |Min(S)|=1|{\rm Min}(S)|=1. Let I1I_{1} be the unique minimal left ideal of SS. Consequently, I1I_{1} is contained in all other nontrivial left ideals of SS. If SS has at least three nontrivial left ideals then we get a cycle, a contradiction. Thus |V(Γ(S))|=2|V(\Gamma(S))|=2. Now we assume that |Min(S)|=2|{\rm Min}(S)|=2. Let I1,I2I_{1},I_{2} be two minimal left ideals of SS. Let if possible, S=I12S=I_{12}. Then by Corollary 3.2 and Theorem 3.3, Γ(S)\Gamma(S) is disconnected so is not a tree. Thus SI12S\neq I_{12}. Then J=I12J=I_{12} is a nontrivial left ideal of SS. Now if SS has a nontrivial left ideal KK other than I1,I2I_{1},I_{2} and JJ. Without loss of generality, assume that I1KI_{1}\subset K then we get a cycle I1I12KI1I_{1}\sim I_{12}\sim K\sim I_{1}, a contradiction. Thus, for SI12S\neq I_{12}, we have V(Γ(S))={I1,I2,I12}V(\Gamma(S))=\{I_{1},I_{2},I_{12}\}.

(iii) \Rightarrow (iv). If Γ(S)\Gamma(S) is bipartite then we have |Min(S)|2|{\rm Min}(S)|\leq 2. In the similar lines of the work discussed above, (iv) holds. ∎

Theorem 4.5.

Let SS be a semigroup with nn minimal left ideals. Then the following results hold:

  1. (i)

    If SI12nS\neq I_{12\cdots n} then γ(Γ(S))=1\gamma(\Gamma(S))=1.

  2. (ii)

    If S=I12nS=I_{12\cdots n} then γ(Γ(S))=2\gamma(\Gamma(S))=2.

Proof.

(i) Suppose that SI12nS\neq I_{12\cdots n}. It follows that J=I12nJ=I_{12\cdots n} is a nontrivial left ideal of SS. It is well known that every nontrivial left ideal of SS contains at least one minimal left ideal. Consequently, for any nontrivial left ideal KK of SS, we have JKJ\cap K is nontrivial. Thus, JJ is a dominating vertex. Hence, γ(Γ(S))=1\gamma(\Gamma(S))=1.

(ii) Suppose that S=I12nS=I_{12\cdots n}. Note that there is no dominating vertex in Γ(S)\Gamma(S) so that γ(Γ(S))2\gamma(\Gamma(S))\geq 2. Now we show that D={I1,I23n}D=\{I_{1},I_{23\cdots n}\} is a dominating set. Since SS is the union of nn minimal left ideals so any nontrivial left ideal of SS is union of these minimal left ideals (cf. Remark 2.2). Let JV(Γ(S))DJ\in V(\Gamma(S))\setminus D be any nontrivial left ideal of SS. Then JJ is union of kk minimal left ideals of SS, where 1kn11\leq k\leq n-1. If I1JI_{1}\subset J, then we are done. If I1JI_{1}\not\subset J then JJ must be union of I2,I3,,InI_{2},I_{3},\ldots,I_{n}. It follows that intersection of JJ and I23nI_{23\cdots n} is nontrivial. Consequently, JI23nJ\sim I_{23\cdots n}. Thus DD is a dominating set. This completes the proof. ∎

Theorem 4.6.

Let SS be a semigroup with nn minimal left ideals. Then α(Γ(S))=n\alpha(\Gamma(S))=n.

Proof.

Let Min(S)={Ii1:i1[n]}{\rm Min}(S)=\{{I_{i_{1}}:i_{1}\in[n]}\} be the set of all minimal left ideals of S. Then, by Remark 2.1, Min(S){\rm Min}(S) is an independent set of Γ(S)\Gamma(S). It follows that α(Γ(S))n\alpha(\Gamma(S))\geq n. Now we prove that for any arbitrary independent set UU, we have |U|n|U|\leq n. Assume that IV(Γ(S))I\in V(\Gamma(S)) such that IUI\in U. Since every nontrivial left ideal contains at least one minimal left ideal. Without loss of generality, assume that Ii1i2ikII_{{i_{1}}{i_{2}}\cdots{i_{k}}}\subseteq I for some i1,i2,,ik[n]i_{1},i_{2},\cdots,i_{k}\in[n]. Then note that |U|nk+1|U|\leq n-k+1. Otherwise, there exist at least two nontrivial left ideals which are adjacent, a contracdiction. Consequently, we have |U|n|U|\leq n. Thus, α(Γ(S))=n\alpha(\Gamma(S))=n. ∎

Lemma 4.7.

Let SS be a semigroup with n(3)n~{}(\geq 3) minimal left ideals. Then there exists a clique in Γ(S)\Gamma(S) of size nn.

Proof.

Let I1,I2,,InI_{1},I_{2},\ldots,I_{n} be nn minimal left ideals. Consider 𝒞={Ii1i2in1:i1,i2,,in1[n]}\mathcal{C}=\{I_{{i_{1}}{i_{2}}\cdots{i_{n-1}}}:i_{1},i_{2},\ldots,i_{n-1}\in[n]\}. Clearly, |𝒞|=n|\mathcal{C}|=n. Notice that for any J,K𝒞J,K\in\mathcal{C}, we have JKJ\cap K is nontrivial so that JKJ\sim K. Thus, 𝒞\mathcal{C} becomes a clique of size nn. ∎

Theorem 4.8.

Let SS be a semigroup with n(>1)n(>1) minimal left ideals. Then ω(Γ(S))=n\omega(\Gamma(S))=n if and only if one of the following holds:

  1. (i)

    S=I123S=I_{123}.

  2. (ii)

    SS has only two minimal left ideals I1I_{1} and I2I_{2} and a unique maximal left ideal I12I_{12}.

Proof.

First suppose that ω(Γ(S))=n\omega(\Gamma(S))=n. Assume that SS has n(4)n(\geq 4) minimal left ideals, namely I1,I2,,InI_{1},I_{2},\ldots,I_{n}. Then 𝒞={Ii1i2in1,Ii1i2:i1,i2,,in[n]}\mathcal{C}=\{I_{{i_{1}}{i_{2}}\cdots{i_{n-1}}},I_{{i_{1}}{i_{2}}}:i_{1},i_{2},\ldots,i_{n}\in[n]\} forms a clique of size greater than nn of Γ(S)\Gamma(S). It follows that ω(Γ(S))>n\omega(\Gamma(S))>n. If n=3n=3, assume that SI123S\neq I_{123}. Then 𝒞={I12,I13,I23,I123}\mathcal{C}=\{I_{12},I_{13},I_{23},I_{123}\} forms a clique of size four of Γ(S)\Gamma(S). It follows that S=I123S=I_{123}. For n=2n=2, we have either S=I12S=I_{12} or SI12S\neq I_{12}. For S=I12S=I_{12}, by Corollary 3.2 and by Theorem 3.3, Γ(S)\Gamma(S) is disconnected. Thus, ω(Γ(S))<n\omega(\Gamma(S))<n. Thus SI12S\neq I_{12}. If SS has a nontrivial left ideal K{I1,I2,I12}K\notin\{I_{1},I_{2},I_{12}\} then we get a clique of size three. Therefore, I12I_{12} is a unique maximal left ideal. Converse follows trivially.∎

Lemma 4.9.

If Γ(S)\Gamma(S) is connected then Max(S){\rm Max}(S) forms a clique of Γ(S)\Gamma(S).

Proof.

We prove the result by showing that if J,KMax(S)J,K\in{\rm Max}(S) then JKJ\sim K. Let if possible, JKJ\nsim K. The maximality of JJ and KK follows that JK=SJ\cup K=S. By Lemma 2.3, SJS\setminus J and SKS\setminus K are \mathcal{L}-classes of SS. It follows that JJ and KK are only nontrivial left ideals of SS. Thus, being a null graph Γ(S)\Gamma(S) is disconnected, a contradiction. ∎

Theorem 4.10.

If KK is a maximal left ideal of SS such that deg(K)deg(K) is finite, then χ(Γ(S))<\chi(\Gamma(S))<\infty.

Proof.

Let JJ be an arbitrary nontrivial left ideal of SS such that JKJ\nsim K. Note that JJ is minimal left ideal of SS. On contrary, suppose that JJ is not a minimal left ideal of SS. Then there exists a nontrivial left ideal JJ^{\prime} of SS such that JJJ^{\prime}\subset J. Since KK is maximal left ideal of SS. Consequently, JK=SJ^{\prime}\cup K=S. It follows that intersection of JJ and KK is nontrivial, a contradiction. By Remark 2.1, we can color all the vertices which are not adjacent with KK with one color. Since deg(K)deg(K) is finite, we have χ(Γ(S))<\chi(\Gamma(S))<\infty. ∎

Lemma 4.11.

For S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}}, we have ω(Γ(S))=χ(Γ(S))=2n11\omega(\Gamma(S))=\chi(\Gamma(S))=2^{n-1}-1.

Proof.

First note that SS has 2n22^{n}-2 nontrivial left ideals and every nontrivial left ideal of SS is of the form Ii1i2ikI_{{i_{1}}{i_{2}}\cdots{i_{k}}} and 1kn11\leq k\leq n-1 (cf. Remark 2.2). If nn is odd then consider 𝒞={Ij1j2jt:n2tn1}\mathcal{C}=\{I_{{j_{1}}{j_{2}}\cdots{j_{t}}}:\lceil\frac{n}{2}\rceil\leq t\leq n-1\}. Note that 𝒞\mathcal{C} forms a clique of size 2n112^{n-1}-1. We may now suppose that nn is even. Consider 𝒞1={Ij1j2jt:n2+1tn1}\mathcal{C}_{1}=\{I_{{j_{1}}{j_{2}}\cdots{j_{t}}}:\frac{n}{2}+1\leq t\leq n-1\}. Notice that 𝒞1\mathcal{C}_{1} forms a clique. Further, observe that 𝒞={Ii1i2in2:i1,i2,,in2[n]}\mathcal{C}^{{}^{\prime}}=\{I_{{i_{1}}{i_{2}}\cdots{i_{\frac{n}{2}}}}:i_{1},i_{2},\ldots,i_{\frac{n}{2}}\in[n]\} do not form a clique because for j1,j2,,jn2[n]{i1,i2,,in2}j_{1},j_{2},\ldots,j_{\frac{n}{2}}\in[n]\setminus\{i_{1},i_{2},\ldots,i_{\frac{n}{2}}\}, Ii1i2in2Ij1j2jn2I_{{i_{1}}{i_{2}}\cdots{i_{\frac{n}{2}}}}\nsim I_{{j_{1}}{j_{2}}\cdots{j_{\frac{n}{2}}}}. However, 𝒞′′={Ii1i2in2𝒞{Ij1j2jn2}:j1,j2,,jn2{i1,i2,,in2}}\mathcal{C}^{{}^{\prime\prime}}=\{I_{{i_{1}}{i_{2}}\cdots{i_{\frac{n}{2}}}}\in\mathcal{C}^{{}^{\prime}}\setminus\{I_{{j_{1}}{j_{2}}\cdots{j_{\frac{n}{2}}}}\}:j_{1},j_{2},\ldots,j_{\frac{n}{2}}\notin\{i_{1},i_{2},\ldots,i_{\frac{n}{2}}\}\} forms a clique of size |𝒞|2\frac{|\mathcal{C}^{{}^{\prime}}|}{2}. Further note that the set 𝒞1𝒞′′\mathcal{C}_{1}\cup\mathcal{C}^{{}^{\prime\prime}} also forms a clique of size 2n112^{n-1}-1. Thus, ω(Γ(S))2n11\omega(\Gamma(S))\geq 2^{n-1}-1. To complete the proof, we show that χ(Γ(S))2n11\chi(\Gamma(S))\leq 2^{n-1}-1. For I=Ii1i2ikI=I_{{i_{1}}{i_{2}}\cdots{i_{k}}} and J=Ij1j2jnkJ=I_{{j_{1}}{j_{2}}\cdots{j_{n-k}}}, where i1,i2,,ik[n]{j1,j2,,jnk}i_{1},i_{2},\ldots,i_{k}\in[n]\setminus\{j_{1},j_{2},\ldots,j_{n-k}\} we have IJI\cap J is trivial. Consequently, we can color these vertices with same color so that we can cover all the vertices with 2n112^{n-1}-1 colors. Thus χ(Γ(S))2n11\chi(\Gamma(S))\leq 2^{n-1}-1. Hence ω(Γ(S))=χ(Γ(S))=2n11\omega(\Gamma(S))=\chi(\Gamma(S))=2^{n-1}-1. ∎

Corollary 4.12.

If S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}} then Γ(S)\Gamma(S) is a weakly perfect graph.

In order to find the upper bound of the chromatic number of Γ(S)\Gamma(S), where SS is an arbitrary semigroup, first we define

X1\displaystyle X_{1} ={IV(Γ(S)):Ii1i2inI}\displaystyle=\{I\in V(\Gamma(S)):I_{{i_{1}}{i_{2}}\cdots{i_{n}}}\subseteq I\}
X2\displaystyle X_{2} ={IV(Γ(S)):IIi1i2inandIIi1i2in}\displaystyle=\{I\in V(\Gamma(S)):I\subset I_{{i_{1}}{i_{2}}\cdots{i_{n}}}~{}\text{and}~{}I\neq I_{{i_{1}}{i_{2}}\cdots{i_{n}}}\}
X3\displaystyle X_{3} =V(Γ(S))(X1X2).\displaystyle=V(\Gamma(S))\setminus(X_{1}\cup X_{2}).

Let Min(I)~\widetilde{{\text{Min}(I)}} be the set of all minimal left ideals contained in II . Further define a relation ρ\rho on X3X_{3} such that

JρKMin(J)~=Min(K)~J~{}~{}\rho~{}~{}K\Longleftrightarrow\widetilde{{\text{Min}(J)}}=\widetilde{{\text{Min}(K)}}

Note that ρ\rho is an equivalence relation.

Theorem 4.13.

Let SS be a semigroup with nn minimal left ideals and χ(Γ(S))<\chi(\Gamma(S))<\infty. Then

χ(Γ(S))|X1|+(2n11)+(2n11)m\chi(\Gamma(S))\leq|X_{1}|+(2^{n-1}-1)+(2^{n-1}-1)m,

where m=max{|C(I)|:C(I)is an equivalence class ofρ}m={\rm{max}}\{|C(I)|:C(I)\ \text{is an equivalence class of}\;\rho\}.

Proof.

Note that for any I,JX1I,J\in X_{1}, we have IJI\sim J. Since every nontrivial left ideal contains at least one minimal left ideal, consequently each element of X1X_{1} is a dominating vertex of Γ(S)\Gamma(S). Therefore, we need at least |X1||X_{1}| colors in any coloring of Γ(S)\Gamma(S). By proof of Lemma 4.11, we can color all the vertices of X2X_{2} with at least 2n112^{n-1}-1 colors so that we need at least 2n11+|X1|2^{n-1}-1+|X_{1}| colors to color X1X2X_{1}\cup X_{2}.

To prove our result we need to show that the vertices of X3X_{3} can be colored by using (2n11)m(2^{n-1}-1)m colors. Now let J,KX3J,K\in X_{3} such that Ii1i2ikJI_{{i_{1}}{i_{2}}\cdots{i_{k}}}\subset J and Ij1j2jtKI_{{j_{1}}{j_{2}}\cdots{j_{t}}}\subset K. Note that JKJ\cap K is nontrivial if and only if Ii1i2ikIj1j2jtI_{{i_{1}}{i_{2}}\cdots{i_{k}}}\cap I_{{j_{1}}{j_{2}}\cdots{j_{t}}} is nontrivial. It follows that JKJ\sim K in Γ(S)\Gamma(S) if and only if either Ii1i2ik=Ij1j2jtI_{{i_{1}}{i_{2}}\cdots{i_{k}}}=I_{{j_{1}}{j_{2}}\cdots{j_{t}}} or Ii1i2ikIj1j2jtI_{{i_{1}}{i_{2}}\cdots{i_{k}}}\sim I_{{j_{1}}{j_{2}}\cdots{j_{t}}}.

Note that the equivalence class of IX3I\in X_{3} under ρ\rho is C(I)={JX3:Min(I)~=Min(J)~}C(I)=\{J\in X_{3}:\widetilde{{\text{Min}(I)}}=\widetilde{{\text{Min}(J)}}\}. Since χ(Γ(S))<\chi(\Gamma(S))<\infty we get |C(I)|<|C(I)|<\infty. Consequently, |C(I)|m|C(I)|\leq m. Observe that C(I)C(I) forms a clique, we require maximum mm colors to color each class under ρ\rho. Note that JC(J)J\in C(J) and KC(K)K\in C(K) such that JKJ\sim K if and only if Ii1i2ikIj1j2jtI_{{i_{1}}{i_{2}}\cdots{i_{k}}}\sim I_{{j_{1}}{j_{2}}\cdots{j_{t}}} in Γ(S)\Gamma(S). Consequently, we can color the vertices in X3X_{3} by using (2n11)m(2^{n-1}-1)m colors. ∎

Theorem 4.14.

Let SS be a semigroup with nn minimal left ideals. Then

sdim(Γ(S))={|X1|+|X3|+2n12ifSIi1i2in2n11ifS=Ii1i2in\operatorname{sdim}(\Gamma(S))=\begin{cases}|X_{1}|+|X_{3}|+2^{n-1}-2&\text{\rm if}~{}S\neq I_{{i_{1}}{i_{2}}\cdots{i_{n}}}\\ 2^{n-1}-1&\text{\rm if}~{}S=I_{{i_{1}}{i_{2}}\cdots{i_{n}}}\end{cases}
Proof.

Let I,JV(Γ(S))I,J\in V(\Gamma(S)) such that Ii1i2ikII_{{i_{1}}{i_{2}}\cdots{i_{k}}}\subseteq I and Ij1j2jtJI_{{j_{1}}{j_{2}}\cdots{j_{t}}}\subseteq J. Then IJI\sim J if and only if either Ii1i2ik=Ij1j2jtI_{{i_{1}}{i_{2}}\cdots{i_{k}}}=I_{{j_{1}}{j_{2}}\cdots{j_{t}}} or Ii1i2ikIj1j2jtI_{{i_{1}}{i_{2}}\cdots{i_{k}}}\sim I_{{j_{1}}{j_{2}}\cdots{j_{t}}}. Define a relation ρ1\rho_{1} on V(Γ(S))V(\Gamma(S)) such that II ρ1\rho_{1} JJ if and only if Min(I)~=Min(J)~\widetilde{{\text{Min}(I)}}=\widetilde{{\text{Min}(J)}}. Clearly, ρ1\rho_{1} is an equivalence relation on V(Γ(S))V(\Gamma(S)). Let N[Ii1i2ik]={IV(Γ(S)):Min(I)~=Ii1i2ik}N[I_{{i_{1}}{i_{2}}\cdots{i_{k}}}]=\{I\in V(\Gamma(S)):\widetilde{{\text{Min}(I)}}=I_{{i_{1}}{i_{2}}\cdots{i_{k}}}\} be equivalence class containing Ii1i2ikI_{{i_{1}}{i_{2}}\cdots{i_{k}}}. If SIi1i2inS\neq I_{{i_{1}}{i_{2}}\cdots{i_{n}}}, then by Theorem 2.5, we have Γ(S)\mathcal{R}_{\Gamma(S)} whose vertex set V(Γ(S))={Ii1i2ik:i1,i2,,ik[n]and1kn}V(\mathcal{R}_{\Gamma(S)})=\{I_{{i_{1}}{i_{2}}\cdots{i_{k}}}:i_{1},i_{2},\cdots,i_{k}\in[n]~{}\text{and}~{}1\leq k\leq n\}. By using Lemma 4.11, note that ω(Γ(S))=2n1\omega(\mathcal{R}_{\Gamma(S)})=2^{n-1}. Then sdim(Γ(S))=|X1|+|X3|+2n12\operatorname{sdim}(\Gamma(S))=|X_{1}|+|X_{3}|+2^{n-1}-2. Next, if S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}}, then V(Γ(S))={Ii1i2ik:i1,i2,,ik[n]and1kn1}V(\mathcal{R}_{\Gamma(S)})=\{I_{{i_{1}}{i_{2}}\cdots{i_{k}}}:i_{1},i_{2},\cdots,i_{k}\in[n]~{}\text{and}~{}1\leq k\leq n-1\}. By using Lemma 4.11, note that ω(Γ(S))=2n11\omega(\mathcal{R}_{\Gamma(S)})=2^{n-1}-1. Then sdim(Γ(S))=2n11\operatorname{sdim}(\Gamma(S))=2^{n-1}-1. ∎

Now in the remaining section, we consider a class of those semigroups which are union of nn minimal left ideals. In particular, completely simple semigroups belongs to this class. In what follows, the semigroup SS is assumed to be the union of nn minimal left ideals Ii1,Ii2,,IinI_{i_{1}},I_{i_{2}},\ldots,I_{i_{n}} i.e. S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}}. The following lemma gives the lower bound of the metric dimension of Γ(S)\Gamma(S).

Lemma 4.15 ([9, Theorem 1]).

For positive integers dd and mm with d<md<m, define f(m,d)f(m,d) as the least positive integer kk such that k+dkmk+d^{k}\geq m. Then for a connected graph Γ\Gamma of order m2m\geq 2 and diameter dd, the metric dimension β(Γ)f(m,d)\beta(\Gamma)\geq f(m,d).

Theorem 4.16.

If S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}} then the metric dimension of Γ(S)\Gamma(S) is given below:

β(Γ(S))={2ifn=3nifn4\beta(\Gamma(S))=\begin{cases}2&\text{\rm if}~{}n=3\\ n&\text{\rm if}~{}n\geq 4\end{cases}
Proof.

For n=3n=3, it is easy to observe that {Ii1,Ii2}\{I_{i_{1}},I_{i_{2}}\} forms a minimum resolving set. If n4n\geq 4 then by Remark 2.2, we have |V(Γ(S))|=2n2|V(\Gamma(S))|=2^{n}-2. In view of Lemma 4.15, we get

n=f(2n2,2)β(Γ(S))n=f(2^{n}-2,2)\leq\beta(\Gamma(S)).

It is easy to observe that for k=n1k=n-1, 2k+k2n22^{k}+k\not\geq 2^{n}-2. Therefore, the least positive integer kk is nn for which k+2k2n2k+2^{k}\geq 2^{n}-2. Thus nβ(Γ(S))n\leq\beta(\Gamma(S)). To obtain upper bound of β(Γ(S))\beta(\Gamma(S)), let J=Ii1i2ikJ=I_{{i_{1}}{i_{2}}\cdots{i_{k}}} and K=Ij1j2jtK=I_{{j_{1}}{j_{2}}\cdots{j_{t}}} be distinct arbitrary vertices Γ(S)\Gamma(S). Since JKJ\neq K, there exists at least IisMin(S)I_{i_{s}}\in{\rm Min}(S) such that IisJI_{i_{s}}\sim J and IisKI_{i_{s}}\nsim K. It follows that d(J,Iis)d(K,Iis)d(J,I_{i_{s}})\neq d(K,I_{i_{s}}). Thus Min(S)={Ii1:i1[n]}{\rm Min}(S)=\{I_{i_{1}}:i_{1}\in[n]\} forms a resolving set for Γ(S)\Gamma(S) of size nn. It follows that β(Γ(S))n\beta(\Gamma(S))\leq n. This completes our proof. ∎

An automorphism of a graph Γ\Gamma is a permutation ff on V(Γ)V(\Gamma) with the property that, for any vertices uu and vv, we have ufvfuf\sim vf if and only if uvu\sim v. The set Aut(Γ)Aut(\Gamma) of all graph automorphisms of a graph Γ\Gamma forms a group with respect to composition of mappings. The symmetric group of degree nn is denoted by SnS_{n}. Now we obtain the automorphism group of Γ(S)\Gamma(S), when SS is union of nn minimal left ideal.

Lemma 4.17.

Let S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}} and let K=Ii1i2ikK=I_{{i_{1}}{i_{2}}\cdots{i_{k}}} be a nontrivial left ideal of SS. Then deg(K)=(2k2)+(2nk2)+(2nk1)(2k2)deg(K)=(2^{k}-2)+(2^{n-k}-2)+(2^{n-k}-1)(2^{k}-2).

Proof.

Let JJ be a nontrivial left ideal of SS such that JKJ\sim K. Clearly JKJ\cap K is a nontrivial left ideal. Now we discuss the following cases:

Case 1. JKJ\not\subset K and KJK\not\subset J. Since JKJ\sim K and K=Ii1i2ikK=I_{{i_{1}}{i_{2}}\cdots{i_{k}}} then note that the number of nontrivial left ideals such that JKJ\not\subset K and KJK\not\subset J is

=(i=1nk(nki))(i=1k1(ki))=(2nk1)(2k2)\displaystyle=\left(\sum_{i=1}^{n-k}\binom{n-k}{i}\right)\left(\sum_{i=1}^{k-1}\binom{k}{i}\right)=(2^{n-k}-1)(2^{k}-2)

Case 2. JKJ\subset K. The number of nontrivial left ideals of SS which are properly contained in KK are 2k22^{k}-2.

Case 3. KJK\subset J. The number of nontrivial left ideals of SS properly containing KK are 2nk22^{n-k}-2. Thus, from the above cases we have the result. ∎

Corollary 4.18.

If S=Ii1i2inS=I_{{i_{1}}{i_{2}}\cdots{i_{n}}} then the graph Γ(S)\Gamma(S) is Eulerian for n3n\geq 3.

Lemma 4.19.

For σSn\sigma\in S_{n}, let ϕσ:V(Γ(S))V(Γ(S))\phi_{\sigma}:V(\Gamma(S))\rightarrow V(\Gamma(S)) defined by ϕσ(Ii1i2ik)=Iσ(i1)σ(i2)σ(ik)\phi_{\sigma}(I_{{i_{1}}{i_{2}}\cdots{i_{k}}})=I_{\sigma({i_{1}})\sigma({i_{2}})\cdots\sigma({i_{k}})}. Then ϕσAut(Γ(S))\phi_{\sigma}\in Aut(\Gamma(S)).

Proof.

It is easy to verify that ϕσ\phi_{\sigma} is a permutation on V(Γ(S))V(\Gamma(S)). Now we show that ϕσ\phi_{\sigma} preserves adjacency. Let Ii1i2itI_{{i_{1}}{i_{2}}\cdots{i_{t}}} and Ij1j2jkI_{{j_{1}}{j_{2}}\cdots{j_{k}}} be arbitrary vertices of Γ(S)\Gamma(S) such that Ii1i2itIj1j2jkI_{{i_{1}}{i_{2}}\cdots{i_{t}}}\sim I_{{j_{1}}{j_{2}}\cdots{j_{k}}}. Then Ii1i2itIj1j2jkI_{{i_{1}}{i_{2}}\cdots{i_{t}}}\cap I_{{j_{1}}{j_{2}}\cdots{j_{k}}}\neq\emptyset. Now

Ii1i2itIj1j2jk\displaystyle I_{{i_{1}}{i_{2}}\cdots{i_{t}}}\sim I_{{j_{1}}{j_{2}}\cdots{j_{k}}} Iσ(i1)σ(i2)σ(it)Iσ(j1)σ(j2)σ(jk)\displaystyle\Longleftrightarrow I_{\sigma({i_{1}})\sigma({i_{2}})\cdots\sigma({i_{t}})}\sim I_{\sigma({j_{1}})\sigma({j_{2}})\cdots\sigma({j_{k}})}
ϕσ(Ii1i2it)ϕσ(Ij1j2jk).\displaystyle\Longleftrightarrow\phi_{\sigma}(I_{{i_{1}}{i_{2}}\cdots{i_{t}}})\sim\phi_{\sigma}(I_{{j_{1}}{j_{2}}\cdots{j_{k}}}).

Thus, ϕσAut(Γ(S))\phi_{\sigma}\in Aut(\Gamma(S)). ∎

Proposition 4.20.

For each fAut(Γ(S))f\in Aut(\Gamma(S)), we have f=ϕσf=\phi_{\sigma} for some σSn\sigma\in S_{n}.

Proof.

In view of Lemma 4.17 and Lemma 4.19, suppose that f(Ii1)=Ij1f(I_{i_{1}})=I_{j_{1}}, f(Ii2)=Ij2f(I_{i_{2}})=I_{j_{2}}, \ldots, f(Iin)=Ijnf(I_{i_{n}})=I_{j_{n}}. Consider σSn\sigma\in S_{n} such that σ(i1)=j1,σ(i2)=j2,,σ(in)=jn\sigma(i_{1})=j_{1},\sigma(i_{2})=j_{2},\ldots,\sigma(i_{n})=j_{n}. Then ϕσ(Ii1i2ik)=Iσ(i1)σ(i2)σ(ik)=Ij1j2jk\phi_{\sigma}(I_{{i_{1}}{i_{2}}\cdots{i_{k}}})=I_{\sigma({i_{1}})\sigma({i_{2}})\cdots\sigma({i_{k}})}=I_{{j_{1}}{j_{2}}\cdots{j_{k}}} (cf. Lemma 4.19). Clearly, Ii1Ii1i2ikI_{i_{1}}\sim I_{{i_{1}}{i_{2}}\cdots{i_{k}}}, Ii2Ii1i2ikI_{i_{2}}\sim I_{{i_{1}}{i_{2}}\cdots{i_{k}}}, \ldots, IikIi1i2ikI_{i_{k}}\sim I_{{i_{1}}{i_{2}}\cdots{i_{k}}}. Also note that IitIi1i2ikI_{i_{t}}\cap I_{{i_{1}}{i_{2}}\cdots{i_{k}}} is trivial for it{ik+1,ik+2,,in}{i_{t}}\in\{i_{k+1},i_{k+2},\ldots,i_{n}\} where ik+1,ik+2,,in[n]{i1,i2,,ik}i_{k+1},i_{k+2},\ldots,i_{n}\in[n]\setminus\{i_{1},i_{2},\ldots,i_{k}\}. It follows that Iik+1Ii1i2ikI_{i_{k+1}}\nsim I_{{i_{1}}{i_{2}}\cdots{i_{k}}}, Iik+2Ii1i2ikI_{i_{k+2}}\nsim I_{{i_{1}}{i_{2}}\cdots{i_{k}}}, \ldots, IinIi1i2ikI_{i_{n}}\nsim I_{{i_{1}}{i_{2}}\cdots{i_{k}}}. Thus, f(Ii1)f(Ii1i2ik)f(I_{i_{1}})\sim f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), f(Ii2)f(Ii1i2ik)f(I_{i_{2}})\sim f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), \ldots, f(Iik)f(Ii1i2ik)f(I_{i_{k}})\sim f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}) and f(Iik+1)f(Ii1i2ik)f(I_{i_{k+1}})\nsim f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), f(Iik+2)f(Ii1i2ik)f(I_{i_{k+2}})\nsim f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), \ldots, f(Iin)f(Ii1i2ik)f(I_{i_{n}})\nsim f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}). Consequently, Ij1f(Ii1i2ik)I_{j_{1}}\subset f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), Ij2f(Ii1i2ik)I_{j_{2}}\subset f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), \ldots, Ijkf(Ii1i2ik)I_{j_{k}}\subset f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}) and Ijk+1f(Ii1i2ik)I_{j_{k+1}}\not\subset f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), Ijk+2f(Ii1i2ik)I_{j_{k+2}}\not\subset f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}), \ldots, Ijnf(Ii1i2ik)I_{j_{n}}\not\subset f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}). It follows that f(Ii1i2ik)=Ij1j2jk=ϕσ(Ii1i2ik)f(I_{{i_{1}}{i_{2}}\cdots{i_{k}}})=I_{{j_{1}}{j_{2}}\cdots{j_{k}}}=\phi_{\sigma}(I_{{i_{1}}{i_{2}}\cdots{i_{k}}}). Thus, f=ϕσf=\phi_{\sigma}. ∎

Theorem 4.21.

Let SS be the union of nn minimal left ideals. Then for n2n\geq 2, we have Aut(Γ(S))SnAut(\Gamma(S))\cong S_{n}. Moreover, |Aut(Γ(S))|=n!|Aut(\Gamma(S))|=n!.

Proof.

In view of Lemma 4.19 and by Proposition 4.20, note that the underlying set of the automorphism group of Γ(S)\Gamma(S) is Aut(Γ(S))={ϕσ:σSn}Aut(\Gamma(S))=\{\phi_{\sigma}\;:\;\sigma\in S_{n}\}, where SnS_{n} is a symmetric group of degree nn. The groups Aut(Γ(S))Aut(\Gamma(S)) and SnS_{n} are isomorphic under the assignment ϕσσ\phi_{\sigma}\mapsto\sigma. Since all the elements in Aut(Γ(S))Aut(\Gamma(S)) are distinct, we have |Aut(Γ(S))|=n!|Aut(\Gamma(S))|=n!. ∎

5. Acknowledgement

The first author gratefully acknowledge for providing financial support to CSIR (09/719(0093)/2019-EMR-I) government of India. The second author wishes to acknowledge the support of MATRICS Grant (MTR/2018/000779) funded by SERB, India.

References

  • [1] H. Ahmadi and B. Taeri. Planarity of the intersection graph of subgroups of a finite group. Journal of Algebra and Its Applications, 15(03):1650040, 2016.
  • [2] S. Akbari, F. Heydari, and M. Maghasedi. The intersection graph of a group. Journal of Algebra and Its Applications, 14(05):1550065, 2015.
  • [3] S. Akbari and R. Nikandish. Some results on the intersection graph of ideals of matrix algebras. Linear and Multilinear Algebra, 62(2):195–206, 2014.
  • [4] S. Akbari, R. Nikandish, and M. J. Nikmehr. Some results on the intersection graphs of ideals of rings. Journal of Algebra and Its Applications, 12(4):1250200–13, 2013.
  • [5] S. Akbari, H. Tavallaee, and S. K. Ghezelahmad. Intersection graph of submodules of a module. Journal of Algebra and Its Applications, 11(01):1250019, 2012.
  • [6] S. Akbari, H. Tavallaee, and S. K. Ghezelahmad. Some results on the intersection graph of submodules of a module. Mathematica Slovaca, 67(2):297, 2017.
  • [7] J. Bosák. The graphs of semigroups. In Theory of Graphs and its Applications (Proc. Sympos. Smolenice, 1963). Publ. House Czechoslovak Acad. Sci., Prague, 1964.
  • [8] I. Chakrabarty, S. Ghosh, T. K. Mukherjee, and M. K. Sen. Intersection graphs of ideals of rings. Discrete Math., 309:5381–5392, 2009.
  • [9] G. Chartrand, L. Eroh, M. A. Johnson, and O. R. Oellermann. Resolvability in graphs and the metric dimension of a graph. Discrete Applied Mathematics, 105(1-3):99–113, 2000.
  • [10] M. Chudnovsky, N. Robertson, P. Seymour, and R. Thomas. The strong perfect graph theorem. Ann. of Math. (2), 164:51–229, 2006.
  • [11] A. H. Clifford and G. B. Preston. The algebraic theory of semigroups. Vol. I. Mathematical Surveys, No. 7. American Mathematical Society, 1961.
  • [12] B. Csákány and G. Pollák. The graph of subgroups of a finite group. Czechoslovak Math. J., 19 (94):241–247, 1969.
  • [13] A. Das. On perfectness of intersection graphs of ideals of n\mathbb{Z}_{n}. Discuss. Math. Gen. Algebra Appl., 37:119–126, 2017.
  • [14] E. Haghi and A. R. Ashrafi. Note on the cyclic subgroup intersection graph of a finite group. Quasigroups Related Systems, 25(2):245–250, 2017.
  • [15] S. H. Jafari and N. Jafari Rad. Planarity of intersection graphs of ideals of rings. Int. Electron. J. Algebra, 8:161–166, 2010.
  • [16] S. H. Jafari and N. Jafari Rad. On the intersection graphs of normal subgroups on nilpotent groups. An. Univ. Oradea Fasc. Mat., 20(1):17–20, 2013.
  • [17] S. H. Jafari and N. J. Rad. Domination in the intersection graphs of rings and modules. Ital. J. Pure Appl. Math, 28:19–22, 2011.
  • [18] N. Jafari Rad and S. Jafari. Results on the intersection graphs of subspaces of a vector space. arXiv: 1105.0803v1, http://arxiv.org/abs/1105.0803v1, 2011.
  • [19] N. Jafari Rad, S. H. Jafari, and S. Ghosh. On the intersection graphs of ideals of direct product of rings. Discuss. Math. Gen. Algebra Appl., 34(2):191–201, 2014.
  • [20] S. Kayacan. Connectivity of intersection graphs of finite groups. Comm. Algebra, 46(4):1492–1505, 2018.
  • [21] S. Kayacan. Dominating sets in intersection graphs of finite groups. Rocky Mountain J. Math., 48(7):2311–2335, 2018.
  • [22] S. Kayacan and E. Yaraneri. Abelian groups with isomorphic intersection graphs. Acta Mathematica Hungarica, 146(1):107–127, 2015.
  • [23] J. D. Laison and Y. Qing. Subspace intersection graphs. Discrete Math., 310(23):3413–3416, 2010.
  • [24] Y.-F. Lin. A problem of bosák concerning the graphs of semigroups. Proceedings of the American Mathematical Society, 1969.
  • [25] X. Ma. On the diameter of the intersection graph of a finite simple group. Czechoslovak Math. J., 66(141)(2):365–370, 2016.
  • [26] X. Ma, M. Feng, and K. Wang. The strong metric dimension of the power graph of a finite group. Discrete Applied Mathematics, 239:159–164, 2018.
  • [27] B. Ponděliček. Diameter of a graph of a semigroup. Časposis Pěch, 92:206–211, 1967.
  • [28] Z. S. Pucanović, M. Radovanović, and A. L. Erić. On the genus of the intersection graph of ideals of a commutative ring. Journal of Algebra and Its Applications, 13(5):1350155–20, 2014.
  • [29] H. Shahsavari and B. Khosravi. On the intersection graph of a finite group. Czechoslovak Math. J., 67(142)(4):1145–1153, 2017.
  • [30] R. Shen. Intersection graphs of subgroups of finite groups. Czechoslovak Math. J., 60(135)(4):945–950, 2010.
  • [31] T. Tamizh Chelvam and M. Sattanathan. Subgroup intersection graph of finite abelian groups. Trans. Comb., 1(3):5–10, 2012.
  • [32] D. B. West. Introduction to Graph Theory, 2nd edn. (Prentice Hall), 1996.
  • [33] F. Xu, D. Wong, and F. Tian. Automorphism group of the intersection graph of ideals over a matrix ring. Linear and Multilinear Algebra, https://doi.org/10.1080/03081087.2020.1723473, 2020.
  • [34] E. Yaraneri. Intersection graph of a module. Journal of Algebra and Its Applications, 12(05):1250218, 2013.
  • [35] B. Zelinka. Intersection graphs of finite abelian groups. Czechoslovak Math. J., 25(100):171–174, 1975.