Coloring of n-inordinate invariant intersection graphs

S. Madhumitha1, Sudev Naduvath1
1Department of Mathematics, Christ University, Bangalore, India

Abstract

In the literature of algebraic graph theory, an algebraic intersection graph called the invariant intersection graph of a graph has been constructed from the automorphism group of a graph. A specific class of these invariant intersection graphs was identified as the n-inordinate invariant intersection graphs, and its structural properties has been studied. In this article, we study the different types of proper vertex coloring schemes of these n-inordinate invariant intersection graphs and their complements, by obtaining the coloring pattern and the chromatic number associated.

Keywords: Intersection graphs, Permutation group, Fix of a permutation, Invariant intersection graph, n-inordinate invariant intersection graphs

1. Introduction

For terminology in group theory, we refer to [7]. For basic definitions and results in graph theory, see [17] and for further concepts in algebraic graph theory, refer to [6]. We refer the reader to [9], for the fundamentals in combinatorics.

An automorphism of a graph G of order n is an isomorphism of G to itself. The automorphism group of G, written as Aut(G), is the set of all automorphisms defined on V(G) and is a subgroup of the symmetric group Sn, which is the set of all permutations of a set with n elements. The fix of a permutation πSn is the set of all elements whose image is invariant, and is given by fix(π)={xS:π(x)=x}, where S is the set of n elements on which the permutations are defined.

Research in algebraic graph theory involves constructing graphs based on algebraic structures and investigating their properties (see [11,12]). In this direction, an algebraic intersection graph based on the automorphism group of a graph, called the invariant intersection graph ΛG of a graph G was defined in [10] as the graph with the vertex set V(ΛG)=Aut(G), and any two distinct vertices vπi,vπjV(ΛG) corresponding to the automorphisms πi, πjAut(G) are adjacent in ΛG when fix(πi)fix(πj).

In [13], the invariant intersection graphs of graphs with automorphism group as the symmetric group Sn, were termed as the n-inordinate invariant intersection graphs and the structural properties of these graphs were studied extensively, along with the properties of their complements, called the n-inordinate invariant non-intersection graphs, where it was found that ΛKnΛKnρ(n)K1, where ρ(i); iN, is the number of derangements of i elements, and ΛKn is the non-trivial component of ΛKn having n!ρ(n) vertices (see [13]).

An illustration of the n-inordinate invariant intersection graph is given in Figure 1. Throughout the study, we denote the identity permutation by π0 and the vertex corresponding to any permutation πAut(G) by vπV(ΛG).

Figure 1 The 4-inordinate invariant intersection graph

In [13], it was proved that the graphs ΛKn and Λ¯Kn are weakly perfect, for all n; that is, χ(ΛKn)=ω(ΛKn)=(n1)! and χ(Λ¯Kn)=ω(Λ¯Kn)=n+ρ(n); whereas, are perfect only when n4. This instills the curiosity to study different proper coloring schemes of these graphs, and to compare the chromatic numbers associated with these colorings with the clique number. Note that a graph G is weakly perfect if χ(G)=ω(G) and a graph G is perfect if χ(H)=ω(H), for all induced subgraphs H of G (see [17]).

2. Vertex colorings of n-inordinate invariant intersection graphs

Graph coloring is an assignment of colors to the vertices, edges or faces of a graph according to certain predefined rules. A proper coloring of a graph G is the assignment of colors to the entities of G such that no two adjacent entities receive the same color. Beginning with the chromatic coloring of a graph G, several types of vertex colorings have been defined, based on the requirements of the assignment or scheduling problems that are to be represented as graph coloring problems.

The investigation of different coloring problems in algebraic graphs is an inquisitive topic of research as many algebraic graphs in the literature were constructed to satisfy certain coloring protocols (c.f. [3,4]). Therefore, in this section, we study different proper vertex coloring schemes of the n-inordinate invariant intersection graphs and the n-inordinate invariant non-intersection graphs, for which the following notions and notations defined in [13] are used.

The group Sn is partitioned into n sets such that, for 1kn, Rk={πSn:|fix(π)|=k} and for each 1kn, the subgraph induced by Rk in ΛKn and Λ¯Kn are denoted by ΛRk and Λ¯Rk, respectively. The vertex set of ΛKn is divided into n non-disjoint subsets Vi={vπ:ifix(π)}; for 1in and for each 1kn, Vik={πVi:πRk};1in.

2.1. Coloring of n-inordinate invariant intersection graphs inducing color pairs

In this section, we discuss the coloring schemes for the graphs ΛKn and Λ¯Kn, in which the assignment of colors to the vertices are defined based on the color-pairs that are induced by vertex pairs possessing some property. We begin the study by investigating the complete coloring of the graphs ΛKn and Λ¯Kn.

A complete coloring of a graph G is a proper vertex coloring in which every pair of colors must appear on at least one pair of adjacent vertices and the maximum number of colors that can be used to obtain a complete coloring of G is the achromatic number of G, denoted by χach(G) (c.f. [14]).

Theorem 2.1. For nN, χach(ΛKn)=χach(ΛKn)=(n1)!.

Proof. Note that ω(ΛKn)=χ(ΛKn)=(n1)!χach(ΛKn). If possible, assume that there exists a complete coloring c:V(ΛKn){c1,c2,,c(n1)!+1} of ΛKn. This implies that all ((n1)!+12) pairs of colors appear on at least one pair of adjacent vertices.

Let c(vπ1)=c(n1)!+1, for some vπ1V(ΛKn). Based on the adjacency pattern of vertices in the graph ΛKn, any vertex in Vi1; 1in, is a part of exactly one clique of order (n1)!, which is induced by the corresponding Vi and these vertices are adjacent to exactly (n1)!1 vertices of that clique. Without loss of generality, select V1 for further analysis, as the choice of any Vi does not affect the adjacency pattern of vertices in the maximal cliques induced. As all (n1)! vertices in V1 are distinctly colored, vπ1V1. Also, any vertex in Vi1 is adjacent to only (n1)!1 vertices in ΛKn and hence vπ1Vi1; for any 1in.

On assigning (n1)! colors to the vertices of V1, the number of vertices in ViV1, for any 2in, that are to be colored, is less than (n1)!. Also, the cardinality of the sets Vij=1i1Vj, for 1in, in order, decrease from (n1)! to ρ(n1), as the Vi’s are non-disjoint and hence a vertex corresponding to a permutation having the least possible cardinality of fix is adjacent to the maximum possible vertices in Vij=1i1Vi, for any 1in. Therefore, any vertex v(t1)(t2);2t1t2n, can be assigned the color c(n1)!+1, as they have the same degree and the vertex v(t1)(t2) is adjacent to all other vertices of ΛKn that corresponds to permutations that fix either t1 or t2. Therefore, without loss of generality, assume that the vertex v(2)(3) is assigned the color c(n1)!+1. Since v(2)(3) is not adjacent to all (n1)! vertices of V1, the colors assigned to the vertices of V1 to which v(2)(3) is not adjacent to must be assigned to the vertices of either V2V1 or V3(V1V2). Since |V2V1|+|V3(V1V2)|<|V1|, even if all these vertices in V2V1 or V3(V1V2) are given distinct colors, there shall not exist any edge having the end vertices colored with the pair c(n1)!+1, ct; for some 1t(n1)!, yielding a contradiction. Hence, χach(ΛKn)=(n1)!. The same coloring can be extended to the graph ΛKn, as the other components apart from ΛKn are trivial. ◻

Theorem 2.2. The achromatic number of Λ¯Kn is n+ρ(n).

Proof. As the graph Λ¯Kn has ω(Λ¯Kn)=χ(Λ¯Kn)=ρ(n)+n, χach(Λ¯Kn)ρ(n)+n. Conversely, assume that χach(Λ¯Kn)=ρ(n)+n+1 and let c be such a complete coloring of Λ¯Kn with ρ(n)+n+1 colors, say c1,c2,,cρ(n)+n+1. As every complete coloring is proper, the clique of order ρ(n)+n induced by the ρ(n) universal vertices along with each set of n vertices with distinct fix in Vi1;1in, vertices must be colored with distinct colors.

Therefore, assign the color ci;1in, to the vertices of Vi1, for the corresponding 1in, and the colors ci;n+1iρ(n)+n, to the ρ(n) universal vertices. For c to be a complete coloring of Λ¯Kn, the color cρ(n)+n+1 has to be assigned to any vertex vπVik; 1in, and 2kn2, as vπ0 is adjacent to only the ρ(n) universal vertices and the subgraph of Λ¯Kn induced by i=1nVi1 is a complete n-partite graph, with each partite having ρ(n1) vertices.

Let c(vπ)=cρ(n)+n+1, for some π with 2|fix(π)|n2. Therefore, vπVt1Vt2, for some 1t1<t2n. As vπ is not adjacent to the vertices of Vt1Vt2, some vertex of Vi(Vt1Vt2); 1it1t2n, must be assigned the colors ct1 and ct2 to satisfy the complete coloring protocol. This assignment cannot be done because all the vertices of Vi(Vt1Vt2); 1it1t2n, are adjacent to all the vertices of Vt11 and Vt21, which are colored using the colors ct1 and ct2, respectively. Hence, χach(Λ¯Kn)<n+ρ(n)+1; completing the proof. ◻

A proper vertex coloring of a graph G is said to be an exact coloring of G if every pair of colors appear on exactly one pair of adjacent vertices and the maximum number of colors used to obtain an exact coloring of G is the exact chromatic number of G (see [5]).

Proposition 2.3. The graphs ΛKn, ΛKn and Λ¯Kn do not admit exact coloring.

Proof. As vπ0 is a universal vertex of the graph ΛKn, we need n!ρ(n) colors in order to obtain an exact coloring with respect to the color assigned to vπ0. For ΛKn to have exactly one edge having end vertices with all (n!ρ(n)12) pairs of colors, ΛKn must be a complete graph, which cannot happen for any n1. Therefore, the graphs ΛKn and ΛKn do not admit an exact coloring. Using the similar argument based on the existence of ρ(n) universal vertices in Λ¯Kn, it can be proved that Λ¯Kn also does not admit an exact coloring. ◻

A proper coloring in which every pair of colors appear on at most one pair of adjacent vertices of a graph G is called a harmonious coloring of G and the minimum number of colors used in a harmonious coloring of G is the harmonic chromatic number of G, denoted by χh(G) (c.f. [5]).

Theorem 2.4. For nN,

  1. χh(ΛKn)=χh(ΛKn)=n!ρ(n).

  2. χh(Λ¯Kn)=n!.

Proof. The graph ΛKn contains a universal vertex vπ0, and hence every vertex in ΛKn must be given a unique color to obtain a harmonious coloring of ΛKn. If there exists two vertices given the same color, it violates the definition of a harmonious coloring as two edges will have the same pair of colors on its end vertices. Hence, χh(ΛKn)=n!ρ(n). Also, any harmonic coloring of ΛKn can be extended as the harmonic coloring of ΛKn, by assigning any of the n!ρ(n) colors used to color the vertices of ΛKn to the ρ(n) isolated vertices of ΛKn, as the isolated vertices do not contribute to any edges in the graph. Therefore, χh(ΛKn)=χh(ΛKn)=n!ρ(n). The proof of (ii) follows immediately based on the arguments of (i), as there are ρ(n) universal vertices in Λ¯Kn◻

Assigning colors to the vertices of a graph G such that no vertex vV(G) is adjacent to two vertices of the same color class is called an injective coloring of G and the minimum number of colors used to obtain such a coloring is called the injective chromatic number of G, denoted by χi(G) (ref. [15]). Note that an injective coloring of a graph is not usually not a proper coloring, but in the following result, we can observe that any injective coloring of the graphs ΛKn and Λ¯Kn turns out to be a proper coloring.

Proposition 2.5. For nN,

  1. χi(ΛKn)=χi(ΛKn)=n!ρ(n).

  2. χi(Λ¯Kn)=n!.

Proof. According to the injective coloring protocol, if a vertex of ΛKn and its neighbour has to be given the same color, it can only be vπ0 and one of its neighbour, say vπ, as vπ0 is a universal vertex in the graph ΛKn. Any neighbour vπ of vπ is a neigbour of vπ0, and hence this assignment will conflict the injective coloring protocol as two vertices in the neighbourhood vπ will have the same color. Therefore, any injective coloring of the graph ΛKn is a proper coloring of ΛKn, which demands the assignment of unique colors to all its vertices. Any injective coloring of ΛKn is also an injective coloring of ΛKn, as the ρ(n) trivial components do not have neighbours to alter the coloring protocol. Hence, χi(ΛKn)=χi(ΛKn)=n!ρ(n). On the same lines, (ii) is also proved. ◻

In this section, we study the vertex colorings of the graphs ΛKn, ΛKn and Λ¯Kn, whose coloring protocols are given based on the neighbourhood of the vertices in the graph. Firstly, we investigate the Grundy coloring of these graphs, where a Grundy coloring of a graph G is a proper vertex coloring such that every vertex is colored by the smallest integer which has not appeared in any of its previously colored neighbours. The maximum number of colors that are used to obtain a Grundy coloring of G is the Grundy chromatic number of G, χgr(G) (ref. [14]).

Theorem 2.6. For the graphs ΛKn, ΛKn and Λ¯Kn,

  1. χgr(ΛKn)=χgr(ΛKn)=(n1)!.

  2. χgr(Λ¯Kn)=n+ρ(n).

Proof.

  1. Let c:V(ΛKn){c1,c2,,c(n1)!} be a proper vertex coloring of ΛKn given as follows. First, assign the colors c1,c2,,c(n1)! to the vertices vπV1, as it induces a clique of order (n1)! in ΛKn. Following this, the vertices vπVij=1i1Vj, for each 2in2, in order, are colored based on its neighbours in j=1iVj that are previously colored.

    If vπVi1;2in, then we assign distinct colors to vπ such that c(vπ)=c(vπ), where vπV11, as these vertices are not adjacent in ΛKn. If vπVjk; k>1, there exists at least one permutation πSn such that fix(π)fix(π)=, as |fix(π)|n2, for any non-identity πSn. If |fix(π)|=n2, the existence of such a permutation is unique; otherwise, there exists a one-to-one correspondence between these permutations, as ρ(nk) is a constant. Therefore, we assign the color c(vπ)=c(vπ), where vπV1 and corresponds to vπVi;2in. Continuing the assignment for each 2in, in order, we obtain a Grundy coloring of ΛKn with (n1)! colors; establishing χgr(ΛKn)(n1)!. Coloring the ρ(n) isolated vertices with the least value among the colors assigned to the vertices of ΛKn, the Grundy coloring of ΛKn can be extended to a Grundy coloring of ΛKn with the same Grundy chromatic number. As we know that χgr(G)χach(G), for any graph G (ref. [8]), the result follows from Theorem 2.1.

  2. Consider the coloring c:V(Λ¯Kn){c1,c2,,cρ(n)+n} such that the vertices of Vij=1i1Vi, for 1in, are assigned the colors ci; 1in, for the corresponding i values and the colors ci;n+1in+ρ(n), are assigned to the ρ(n) universal vertices. This is a Grundy coloring of Λ¯Kn, as every vertex is colored by the smallest cj such that it has not occurred in its colored neighbors and χgr(Λ¯Kn)n+ρ(n). As the other inequality follows from Theorem 2.2, χgr(Λ¯Kn)=n+ρ(n).

 ◻

A vertex v of a graph G is said to be colorful or is said to have a rainbow neighbourhood if v is adjacent to at least one vertex from every color class. The vertex coloring of G such that every vertex is colorful or possesses a rainbow neighbourhood is called a fall coloring or J-coloring. The minimum and maximum number of colors used to obtain a fall coloring of G are called the fall chromatic numbers of G, denoted by χf(G) and ψf(G), respectively (see [8]). The maximum number of colors that can be used to obtain a J-coloring of G is called the J-chromatic number of G, denoted by χj(G) (For details, refer to [16]). Note that the notions of fall and J-coloring of a graph G coincide, and hence χj(G)=ψf(G).

Theorem 2.7. For any n3, χf(ΛKn)=ψf(ΛKn)=(n1)!.

Proof. Every vertex in ΛKn is a part of a clique of order (n1)! and also each vertex of ΛKn is adjacent to at least (n1)!1 vertices of a clique Vi, for some 1in. Hence, all the vertices of ΛKn possess a rainbow neighbourhood or equivalently, are colorful. Hence, χf(ΛKn)=(n1)! and ψf(ΛKn)(n1)!. The vertex connectivity of ΛKn is (n1)!1, as the vertices in Vi1; 1in, are adjacent only to the vertices of the corresponding Vi’s. Hence, ψf(ΛKn) cannot be greater than (n1)!; thereby proving the result. ◻

Proposition 2.8. The graphs ΛKn and Λ¯Kn do not admit a fall or J-coloring.

Proof. The graph ΛKn cannot admit a fall coloring or J-coloring as it is disconnected and the isolated vertices cannot be colorful or have a rainbow neighbourhood. In the graph Λ¯Kn, the vertex vπ0 cannot be a colorful vertex because vπ0 is adjacent only to the ρ(n) universal vertices of Λ¯Kn; whereas, ω(Λ¯Kn)=χ(Λ¯Kn)=n+ρ(n). Hence the result. ◻

The number of vertices of a graph G having rainbow neighbourhood is called the rainbow neighbourhood number of G, denoted by rχ(G) (c.f. [16]). As a consequence of Theorem 2.7 and Proposition 2.8, the following corollary is obtained.

Corollary 2.9. For n3,

  1. rχ(ΛKn)=rχ(ΛKn)=n!ρ(n).

  2. rχ(Λ¯Kn)=nρ(n1)+ρ(n).

Proof. As every vertex of ΛKn, except its isolated vertices, is colorful in ΛKn, rχ(ΛKn)=rχ(ΛKn)=n!ρ(n). In any proper coloring of Λ¯Kn, every vertex vπ1Vik, for some 1kn, is adjacent to at least the ρ(n) universal vertices and the vertices vπ2Vj1, where ifix(π2). Since the vertices of Vi1;1in, induce a complete n-partite graph, it can be deduced that in the rainbow coloring of Λ¯Kn, each vπV(Λ¯Kn) is adjacent to the vertices of ρ(n)+(nk) color classes, where k=|fix(π)|. Hence the result. ◻

The minimum number of colors required to properly color the vertices of a graph G such that each vertex of G is adjacent to vertices of at least t1 different color classes is called the t-colorful chromatic number of G, denoted by χtcol(G), and such a coloring of G is called the t-colorful coloring of G (see [18]).

Corollary 2.10. For n3 and 1tδ, χtcol(ΛKn)=(n1)! and χtcol(Λ¯Kn)=ρ(n)+n.

A vertex coloring of a graph G is said to be a b-coloring if at least one vertex of each color class is colorful. The minimum number of colors used in a b-coloring of G is called the b-chromatic number of G, denoted by χb(G) (ref. [8]).

Theorem 2.11. For n3,

  1. χb(ΛKn)=χb(ΛKn)=χ(ΛKn).

  2. χb(Λ¯Kn)=χ(Λ¯Kn).

Proof. As a consequence of Theorem 2.7, we deduce that every vertex of ΛKn is colorful. Hence, the proper vertex coloring of ΛKn by itself is a b-coloring of ΛKn. This coloring when extended to the isolated vertices by assigning one of the (n1)! colors used in any proper coloring of ΛKn does not alter the b-chromatic number of ΛKn, as any b-coloring requires only one vertex of a color class to be colorful. Similarly, it can be observed that the chromatic coloring of Λ¯Kn is a b-coloring of Λ¯Kn, as the ρ(n) universal vertices and the vertices of Vi1; 1in, are adjacent to each vertex of all color classes, except itself. Hence the result. ◻

The coloring protocols in which the assignment of colors between two vertices u,v in a graph G is based on the distance d(u,v)1 are called distance related vertex colorings. In these colorings where the protocols involve only the assignment of distinct colors to vertices u,v at some distance d(u,v), and the parameter associated is the number of colors required, reduce to either the chromatic or the harmonious coloring of ΛKn and Λ¯Kn, as these graphs have diameter 2. Hence, we study the distance related coloring schemes for these graphs, which does not coincide with any other proper coloring.

An L(2,1)-coloring of a graph G is an assignment of non-negative integers as colors, to the vertices of G such that the colors assigned to adjacent vertices differ by at least 2, and the colors assigned to vertices at distance 2 differs by at least 1. For an L(2,1)-coloring c of G, the c-cap of G, λc(G)=max{c(v):vV(G)} and L-cap of G, λL(G)=min{λc(G)}, where the minimum is taken over all possible L(2,1)-colorings c of G (c.f. [2]).

Theorem 2.12. For n3, λL(ΛKn)=λL(ΛKn)=n!ρ(n).

Proof. As the diameter of the graph ΛKn is 2, any L(2,1)-coloring c of ΛKn assigns distinct colors to all the vertices. Hence, the number of colors required to color the vertices is n!ρ(n). As 0 is also a color that can be assigned in an L(2,1) coloring, and there exists a universal vertex vπ0 in ΛKn, λL(ΛKn)n!ρ(n). To prove that λL(ΛKn)=n!ρ(n), we obtain an L(2,1)-coloring of ΛKn which assigns all colors 2,3,,n!ρ(n) and 0, as follows.

First, assign the color 0 to vπ0. Next, assign colors to the vertices of Vik; n2kn2, and Vj1, for 1ijn, alternatively. As the vertices of i=1n[k=n2+1n2Vik] induce a clique, these vertices cannot be assigned consecutive colors. Since every vertex of Vik; n2kn2, corresponding to a k-element fix is not adjacent to (nk)ρ(n1) vertices of Vi1;1in, we assign consecutive colors from 2 to k=n2n2(nk)ρ(nk)+1 to the vertices of Vik; n2kn2, and Vi1;1in, by alternating between them. As |i=1n[k=n2n2Vik]|=k=n2n2(nk)ρ(nk)<nρ(n1), for all n3, there exists nρ(n1)k=n2n2(nk)ρ(nk) vertices of Vi1;1in, left uncolored, at this stage.

The fixes of permutations in R1 are either same or disjoint. Hence, the vertices of Vi1; 1in, corresponding to these permutations with disjoint fixes are non-adjacent and we assign consecutive integers to these vertices by not assigning colors to two vertices of the same Vi1, for any 1in, back to back. Hence, the value of the color given to the last vertex of ΛR1;1in, is nρ(n1)+k=n2n2(nk)ρ(nk)+1.

Following this, we color the vertices of i=1nVik; 2kn21, for which, first consider the (nk) distinct k-element fixes, for each 2kn21, and partition them into subsets of order nk, such that each of them is a maximal subset of pairwise disjoint k-element fixes. Order the partite sets and the vertices in these partite sets, such that the first and the last element of a partite set has disjoint k-element fix with the last and the first element of the preceding and succeeding partite sets. Note that this ordering does not affect the nature of the partite set as the nk fixes in a partite set are mutually disjoint. Since each Rk;2kn21, consists of ρ(nk) permutations with the same k-element fix, we extend the same partitioning and the ordering to all the ρ(nk) sets of (nk) distinct k-element fixes; thereby to the vertices corresponding to them, as well.

Assign colors to all vertices corresponding to the ρ(nk) sets of each of the partite set, from the first to the k(nk)nth, in order. In this assignment, all vertices can be assigned consecutive integers because each distinct k-element fix has (nkk) disjoint k-element fixes, for all 2kn21, and (nkk)>nk+1. Also, every vertex in Vik1;1in, is not adjacent to least one vertex of Vjk;1ijn, for all 1kn21. Hence, all consecutive colors starting from nρ(n1)+k=n2n2(nk)ρ(nk)+2, until nρ(n1)+k=n2n2(nk)ρ(nk)+1+k=2n21(nk)ρ(nk) are used to color the vertices; which on simplification yields λL(ΛKn)=n!ρ(n). Any of the colors used in this L(2,1)-coloring of ΛKn can be given to the ρ(n) isolated vertices of ΛKn to obtain an L(2,1)-coloring of ΛKn, and hence yielding λL(ΛKn)=n!ρ(n)◻

Theorem 2.13. For n3, λL(Λ¯Kn)=n!+ρ(n).

Proof. The graph Λ¯Kn has ρ(n) universal vertices to which we assign the colors 0,2,4,, 2(ρ(n)1). Following this, assign the color 2ρ(n) to the vertex vπ0. Next, we begin by assigning colors to the vertices of Vij=1i1;1in, such that the first vertex and the last vertex assigned color belongs to Vi1 and ViVi+1 (unless i=n) of the respective Vi’s. This assignment ensures that consecutive colors starting from 2ρ(n)+1 to n!+ρ(n) are given to these n!ρ(n)1 vertices, as each Vi is an independent set and also, the first and last vertex in the sequence as per the requirement exists as Vi’s are not disjoint. As all possible consecutive labels are assigned to the non-universal vertices, and the universal vertices must be given non-consecutive colors in any L(2,1)-coloring of the graph Λ¯Kn, we get λL(Λ¯Kn)=n!+ρ(n)◻

A proper vertex coloring of a graph G such that two colors i and j are assigned to u,vV(G) when d(u,v)+|ij|1+r, for some fixed rN; not greater than the diameter of G is called r-radio coloring of G. The value of a radio coloring c of G, rcr(c)=max{c(v):vV(G)} and the r-radio chromatic number of G, rcr(G)=min{rcr(c)}, where the minimum is taken over all possible r-radio colorings c of G (c.f. [2]). When r is the diameter of G and r is one less than the diameter of G, the colorings are called the radio and antipodal colorings of G, and the corresponding chromatic numbers are called the radio and antipodal numbers of G, denoted by rn(G) and an(G), respectively. Since the diameter of ΛKn and Λ¯Kn is 2, we obtain the following result.

Proposition 2.14. For the graphs ΛKn and Λ¯Kn,

  1. rn(ΛKn)=λL(ΛKn).

  2. an(ΛKn)=χ(ΛKn).

  3. rn(Λ¯Kn)=λL(Λ¯Kn).

  4. an(Λ¯Kn)=χ(Λ¯Kn).

2.4. Equitable coloring of the n-inordinate invariant intersection graphs

A proper vertex coloring of a graph G in which the cardinality of the color classes differ by at most 1 is called an equitable coloring of G. The minimum number of colors used to obtain an equitable coloring of G is the equitable chromatic number, denoted by χe(G) (ref. [14]). In this section, we obtain the equitable coloring pattern and determine the equitable chromatic number of the graphs ΛKn, ΛKn and Λ¯Kn.

Theorem 2.15. The equitable chromatic number of ΛKn is n!ρ(n)12+1.

Proof. The graph ΛKn has a universal vertex vπ0 and hence in any equitable coloring of ΛKn, each color class must be of cardinality either 1 or 2. Therefore, χe(ΛKn)n!ρ(n)12+1. To prove the converse, we partition the n!ρ(n)1 non-universal vertices of ΛKn into independent sets of size 2 and claim that there can exist at most one singleton in such a partition.

For each 1kn2, consider the graph ΛRkRnk. As |Rk|=(nk)ρ(nk) and |Rnk|=(nk)ρ(k), |Rk||Rnk|, for any 1kn2. Also, for each 1kn2, every vertex of ΛRnk is not adjacent to exactly kρ(nk) vertices of ΛRk and hence every vertex of ΛRnk can be paired with a vertex of ΛRk such that they are not adjacent. Pairing the vertices of ΛRkRnk; 1kn2, in this manner can be viewed as an injective mapping f:V(ΛRnk)V(ΛRk), that gives (nk)ρ(k) independent sets {v,f(v)} of cardinality 2. Therefore, at this stage, (nk)[ρ(nk)ρ(k)] vertices in each ΛRk; 1kn2, are unpaired, implying that all the uncolored vertices are in ΛRk; 1kn2.

Consider the (nk)[ρ(nk)ρ(k)] vertices in each ΛRk; 1kn2. As each ΛRk; 1kn2, is not a clique, there exists at least one vertex of ΛRk to which a vertex of ΛRk is not adjacent to or equivalently, for every k-element fix, there exists at least one other k-element fix, such that they are disjoint. Hence, these vertices corresponding to these disjoint fixes of each ΛRk can be paired.

If (nk) is even, then the vertices of that ΛRk can be paired within themselves. If (nk) is odd, for some 1kn2, then after pairing the first set of (nk)1 vertices that correspond to distinct k-element fixes, we pair the left out vertex to a vertex in the second set among the ρ(nk)ρ(k) sets of (nk) vertices that correspond to distinct k-element fixes. As there exists vertices of at least 2 distinct k-element fixes to which a vertex that corresponds to a permutation fixing k elements is not adjacent to, for any 1k<n2, such pairing is possible in any ΛRk; 1k<n2. If ρ(nk) is even and (nk) is odd, at this point, all vertices are paired. Otherwise, when both ρ(nk) and (nk) are odd, there will exist exactly one unpaired vertex from each ΛRk; 1k<n2, of this kind. If exactly one vertex is left unpaired among all ΛRk’s, we are done.

If there exists more than one unpaired vertex at this stage, we pair these unpaired vertices based on their non-adjacency, which in turn leaves us with at most one singleton, yielding the required number of pairs. The pairing of these unpaired vertices can be done because |fix(π)|n2 for all non-identity permutations of Sn and the unpaired vertices of the graph ΛRk can be permuted with the vertices of ΛRk in such a way that the graph induced by these left out vertices from each ΛRk of this kind has at least one unique vertex to which every unpaired vertex is not adjacent to, owing to the structure of Sn. Hence, χe(ΛKn)=n!ρ(n)12+1◻

Theorem 2.16. For the graph Λ¯Kn, χe(Λ¯Kn)=n!ρ(n)2+ρ(n).

Proof. The graph Λ¯Kn has ρ(n) universal vertices. Therefore, in any equitable coloring of Λ¯Kn, the colors are assigned to the vertices of Λ¯Kn such that each color class is of cardinality either 1 or 2, implying that χe(Λ¯Kn)n!ρ(n)2+ρ(n). To prove that χe(Λ¯Kn)=n!ρ(n)2+ρ(n), we partition the n!ρ(n) vertices of Λ¯Kn into independent sets of order 1 or 2, by pairing them, and claim that there can exist at most one singleton in such a partition.

In Λ¯Kn, the graph k=n2+1nΛ¯Rk is an empty graph. Hence, these vertices in k=n2+1nΛ¯Rk can be paired in any manner. If k=n2+1n(nk)ρ(nk) is even, we obtain no unpaired vertices; otherwise, pair the left out vertex with vπ0. Note that vπ0 can be paired with any of the non-universal vertices of Λ¯Kn. Based on the number of vertices in each Λ¯Rk; 1kn2, we pair them as follows.

Consider Λ¯Rk, for 1kn2. Every vertex of Λ¯Rk is not adjacent to at least ρ(nk)1 vertices in Λ¯Rk, as they correspond to permutations having the same k-element fix. If both ρ(nk) and (nk) are even, then two vertices corresponding to the same k-element fix can be paired and we obtain ρ(nk)2(nk) pairs of vertices in each of the corresponding Λ¯Rk’s. If ρ(nk) is odd and (nk) is even, then two vertices corresponding to the same k-element fix can be paired until we obtain ρ(nk)12(nk) pairs of vertices of in the corresponding Λ¯Rk and one set among the ρ(nk) sets of (nk) vertices are paired within themselves. For any 2kn2, this pairing is possible because there exists at least two k-element fixes that are not disjoint and hence the vertices corresponding to them can be paired.

If both ρ(nk) and (nk) are odd, pair the ρ(nk)((nk)1) vertices as mentioned above. Hence, we obtain exactly one unpaired vertex here. If this is the only unpaired vertex among all Λ¯Rk’s, we either pair it with vπ0 or it becomes the only singleton; proving the result. If there is one such unpaired vertex from more than one Λ¯Rk;1kn2, we pair the vertices in that set of unpaired vertices if each vertex of this set has at least one unique vertex to which it is not adjacent to. Otherwise, the unpaired vertex of Λ¯Rk which does not have such a unique vertex to which it is not adjacent to, is permuted with the other paired vertices of Λ¯Rk in such a way that the unpaired vertex obtained after permuting has a unique vertex in the set of other unpaired vertices, to which it is not adjacent to. This swapping of vertices among the vertices of Λ¯Rk, for each k is possible owing to the structure of Sn. Therefore, we obtain at most one unpaired vertex at the end of the process, yielding the required value of the equitable chromatic number. ◻

Note that, unlike the other coloring schemes, the equitable coloring of ΛKn cannot be extended to the graph ΛKn because ΛKn has ρ(n) isolated vertices, to which any color can be assigned. As ω(ΛKn)=ω(ΛKn)=(n1)!, we know that χe(ΛKn)(n1)!. Also, α(ΛKn)=n. This implies that ideally, the n! vertices of ΛKn must be distributed equally to the (n1)! color classes; that is, n per color class, which will be an equitable coloring with the minimum number of colors. With respect to the assignment of colors to the vertices of ΛKn in a chromatic coloring, the color class of vπ0 is of cardinality 1 and only for the ρ(n1) color classes assigned to the vertices of Vi1, we can ensure cardinality n. Therefore, for this assignment of colors to the vertices of ΛKn to be an equitable assignment of colors to the vertices of ΛKn, the ρ(n) isolated vertices are distributed to each of the (n1)!ρ(n1) color classes, such that all the color classes have the same cardinality n or differ at most by 1.

The number of isolated vertices to be added to each of the color class which has less than n vertices cannot be precisely mentioned as it depends on the assignment. Also, as the value of ρ(n) is computed based on the integer partitioning of n and the permutations corresponding to these partitioning that does not fix (nk) elements, it grows rapidly and at this stage of the study, a pattern for ρ(n) values also does not exists, owing to its dependency on the theory of integer partitioning, which by itself is an unsolved problem (For details, refer to [1]). Hence, the possibility of assignment, though ideally must hold, (holds till n=10), could not be proved with the expected rigour. Therefore, we conclude the section by posing the following conjecture.

Conjecture 2.17. The equitable chromatic number of the graph ΛKn is (n1)!

3. Conclusion

In this article, we examined various proper vertex coloring schemes on the n-inordinate invariant intersection graphs and their complements. It is interesting to see that almost all the chromatic numbers associated with various coloring patterns are equal to the chromatic number or the order of the graphs. This increases the interest to realise some coloring in the literature for which the assignment is different, prompting a wide scope to investigate other coloring schemes like improper vertex colorings and domination related vertex colorings, edge colorings and so on of the n-inordinate invariant intersection graphs and the n-inordinate invariant non-intersection graphs.

References:

  1. G. E. Andrews and K. Eriksson. Integer Partitions. Cambridge University Press, 2004.
  2. G. Chartrand and P. Zhang. Chromatic Graph Theory, 2019.
  3. I. Dejter. Totally multicolored subgraphs of complete Cayley graphs. Congress Numerantium, 70:53–64, 1990.
  4. I. J. Dejter and R. E. Giudici. On unitary Cayley graphs. Journal of Combinatorial Mathematics and Combinatorial Computing, 18:121–124, 1995.
  5. K. Edwards. Detachments of complete graphs. Combinatorics, Probability and Computing, 14(3):275–310, 2005.
  6. C. Godsil and G. F. Royle. Algebraic Graph Theory, 2001.
  7. I. N. Herstein. Topics in Algebra, 2006.
  8. M. Kalpana and D. Vijayalakshmi. Fall coloring and b-coloring of graphs. Journal of Physics: Conference Series, 1139(1):012045, 2018.
  9. C. L. Liu. Introduction to Combinatorial Mathematics.
  10. S. Madhumitha and S. Naduvath. Invariant intersection graph of a graph. Journal of Discrete Mathematical Sciences and Cryptography, To Appear.
  11. S. Madhumitha and S. Naduvath. Graphs defined on rings: A review. Mathematics, 11(17):3643, 2023.
  12. S. Madhumitha and S. Naduvath. Graphs on groups in terms of the order of elements: A review. Discrete Mathematics, Algorithms & Applications, 16(3), 2024.
  13. S. Madhumitha, S. Naduvath, and V. S. Prebhath. n-inordinate invariant intersection graphs and their derived graphs. Communicated, 2024.
  14. V. M. Mathew, S. Naduvath, and T. J. Varghese. New results on orthogonal component graphs of vector spaces over Zp. Communications in Combinatorics and Optimization, 2023.
  15. S. Naduvath, K. Chithra, S. Satheesh, and J. Kok. A study on the injective coloring parameters of certain graphs. Discrete Mathematics, Algorithms & Applications, 8(03):1650052, 2016.
  16. S. Naduvath and J. Kok. Some new graph coloring problems. Recent Advancements in Graph Theory, pages 289–314, 2020.
  17. D. B. West. Introduction to Graph Theory, 2001.
  18. Z. Zhang and J. Guo. Colorful graph coloring. International Workshop on Frontiers in Algorithmics, pages 141–161, 2022.