Equi Neighbor Polynomial of Some Binary Graph Operations

Dhanya P.1, Anil Kumar V.2
1Department of Mathematics, CKGM Govt. College, Perambra P. O. Kozhikode, Kerala, India 673 525
2Department of Mathematics, University of Calicut, Malappuram, Kerala, India 673 635

Abstract

Let G(V,E) be a simple graph of order n with vertex set V and edge set E. Let (u,v) denote an unordered vertex pair of distinct vertices of G. For a vertex uG, let N(u) be the set of all vertices of G which are adjacent to u in G. Then for 0in1, the i-equi neighbor set of G is defined as: Ne(G,i)={(u,v):u,vV,uv and |N(u)|=|N(v)|=i}. The equi-neighbor polynomial Ne[G;x] of G is defined as Ne[G;x]=i=0(n1)|Ne(G,i)|xi. In this paper we discuss the equi-neighbor polynomial of graphs obtained by some binary graph operations.

Keywords: iEqui neighbor set, Equi neighbor polynomial

1. Introduction

Let G(V,E) be a simple graph of order n with vertex set V and edge set E. Let (u,v) denote an unordered vertex pair of distinct vertices of G. For a vertex uG, let N(u) be the set of all vertices of G which are adjacent to u in G. Then for 0in1, the i-equi neighbor set of G is defined as: Ne(G,i)={(u,v):u,vV,uv and |N(u)|=|N(v)|=i}. The equi-neighbor polynomial Ne[G;x] of G is defined as Ne[G;x]=i=0(n1)|Ne(G,i)|xi [1].

Binary graph operations are used to model the action between two network systems. Binary graph operations are usually known as graph products in which two initial graphs are acted together according to some specific rules to produce a new graph.

In this paper we discuss the equi-neighbor polynomial of graphs obtained by some binary graph operations.

2. Main Results

In this section, we discuss the equi-neighbor polynomial of graphs obtained by some binary graph operations.

If H and K are two graphs, then the join, HK is the graph with vertex set V(H)V(K) and the edge set E(H)E(K){uv:uV(H),vV(K)}.

Theorem 1. If H and K are any two graphs with h and k vertices respectively, then the equi neighbor polynomial of the join of H and K is given by Ne[HK;x]=xkNe[H;x]+xhNe[K;x]+i=0h1(H,i)n(K,i+kh)xi+k, where n(H,i) denotes the number of vertices of H with i neighbors and n(K,i+kh) denotes the number of vertices of K with i+kh neighbors.

Proof. Observe that from the definition of HK, each vertex of K has k more than the number of neighbors as in H and each vertex K has h more than the number of neighbors as in K. Therefore it is enough to consider the following three cases:

Case (i) Let u,vV(H); |N(u)|=|N(v)|=i in H.

Then u has (i+k) neighbors in HK, which are the i neighbors in H and the k vertices in K. The same is true for v. Hence the pair (u,v) has (i+k)-equi neighbors in HK and there are |Ne(H,i)| such pairs.

Case (ii) Let u,vV(K); |N(u)|=|N(v)|=i in K.

Then u has (i+h) neighbors in HK, which are the i neighbors in K and the h vertices in H. The same is true for v. Hence the pair (u,v) has (i+h)-equi neighbors in HK and there are |Ne(K,i)| such pairs.

Case (iii) Let uV(H),vV(K); |N(u)|=i in H, |N(v)|=i+kh in H.

Then u and v have (i+h) neighbors in HK. Hence the pair (u,v) has (i+k)-equi neighbors in HK and there are n(H,i)n(K,i+kh) such pairs.
Thus we have,

Ne[HK;x]=i=0h1|Ne(H,i)|xi+k+i=0k1|Ne(K,i)|xi+h+i=0h1n(H,i)n(K,i+kh)xi+k=xkNe[H;x]+xhNe[K;x]+i=0h1n(H,i)n(K,i+hk)xi+k. This completes the proof. ◻

Consider the graph G(V,E) and let wV. Then the graph G+w is a graph obtained from G including the vertex w in G and joining it to all other vertices of G.

Corollary 1. Ne[G+w;x]=xNe[G;x]+n(G,n1)xn, where n(G,n1) represents the number of vertices of G with (n1) neighbors.

Proof. The result follows from the fact that G+w is isomorphic to GK1, where K1 is the complete graph on a single vertex with V(K1)={w}. ◻

A Wheel graph Wn,n>3 is obtained by taking the join of the cycle graph Cn1 and the complete graph K1.

Lemma 1.

For a cycle graph Cn, Ne[Cn;x]=(n2)x2,n3.

Corollary 2. If Wn,n>3 is a wheel graph with n vertices, then Ne[Wn;x]={(n12)x3,if  n4,6x3,if  n=4.

Proof. Observe that Wn is isomorphic to Cn1K1; where Cn1 is the cycle graph on n vertices and K1 is the complete graph on a single vertex. Therefore, using lemma 1 and Theorem 1, it follows that

Case (i) Let n4. Then, Ne[Cn1K1;x]=x(n12)x2+x×0=(n12)x3.

Case (ii) Let n=4. Then, Ne[Cn1K1;x]=x×3x2+x3×0+3×1×x3=6x3.

This completes the proof. ◻

A Shell graph Sn, where n3 is obtained from the cycle graph Cn by adding the edges corresponding to the (n3) concurrent chords of the cycle.

Lemma 2. If Pn is a path graph of length n; n2, we have Ne[Pn;x]= x+(n22)x2;n2.

Corollary 3. For a shell graph Sn; n3, we have Ne[Sn;x]={(n32)x3+x2,if n3,4,3x2,if n=3,x3+x2,if n=4.

Proof. Note that Sn=Pn1K1. Then the result follows from Lemma 2 and Theorem 1◻

The corona of two graphs K and H is formed from one copy of K and |V(K)| copies of H where the ith vertex of K is adjacent to every vertex in the ith copy of H [2]. It is denoted by KH.

Theorem 2. If K is a graph having k vertices and H is a graph having h vertices, then the equi neighbor polynomial of corona of K and H is given by Ne[KH;x]=xhNe[K;x]+kxNe[H;x]+(k2)i=0h1[n(H,i)]2xi+1+n(K,0)n(H,h1)xh, where n(H,i) represents the number of vertices of H with i neighbors and n(K,0) represents the number of isolated vertices of K.

Proof. Observe that, from the definition of KH,

  1. each vertex of K has h more than the number of neighbors as in K,

  2. each vertex in a copy of H has one more than the number of neighbors as in H and there are k copies of H.

Therefore it is enough to consider the following four cases;

Case (i) Let u,vV(K); |N(u)|=|N(v)|=i in K.

Then u has (i+h) neighbors in KH, which are the i neighbors in K and the h vertices in a copy of H corresponding to the vertex u. v also has (i+h) neighbors in KH, which are the i neighbors in K and the h vertices in the copy of H corresponding to the vertex v. Hence the pair (u,v) has (i+h)-equi neighbors in KH and there are |Ne(K,i)| such pairs.

Case (ii) Let u,vV(Hj); where Hj denotes the jth copy of H;j=1,2,k; |N(u)|=|N(v)|=i in Hj.

Then u has (i+1) neighbors in KH, which are the i neighbors in H and the vertex in K corresponding to Hj. The same is true for v. Hence the pair (u,v) has (i+1)-equi neighbors in KH and there are |Ne(H,i)| such pairs corresponding to the copy Hj of H. Note that there are k such copies of H.

Case (iii) Let uV(Hj), vV(Hk), where Hj anf Hk denote the jth and kth copy of H respectively; j,k{1,2,,k}; jk; |N(u)|=|N(v)|=i in H.

Then the pair (u,v) has (i+1)-equi neighbors in KH and there are (k2)[n(H,i)]2such pairs.

Case (iv) Let uV(K) and vV(Hj), where Hj denote the jth copy of H; j{1,2,,k}; |N(u)|=0 in K; |N(v)|=h1 in H.

Then u and v have h neighbors in KH. Hence the pair (u,v) has h-equi neighbors in KH and there are n(K,0)n(H,h1) such pairs. Thus,

Ne[KH;x]=i=0k1|Ne(K,i)|xi+h+ki=0h1|Ne(H,i)|xi+1+(k2)i=0h1[n(H,i)]2xi+1+n(K,0)n(H,h1)xh=xhNe[K;x]+kxNe[H;x]+(k2)i=0h1[n(H,i)]2xi+1+n(K,0)n(H,h1)xh. This completes the proof. ◻

The graph Q(m,n) ;m1,n>1 is obtained by identifying each vertex of the complete graph Km with a vertex of a unique Kn where there are m copies of Kn [3].

Corollary 4. For m1,n>1, we have Ne[Q(m,n);x]={(m2)(n1)2+m(n12)}xn1+(m2)xm+n2.

Proof. The result follows from the fact that Q(m,n) and KmKn1 are isomorphic. ◻

The Cartesian product of two graphs K and H is the graph K◻H with vertex set V(K)×V(H) and the vertices (u,v) and (x,y) are adjacent if and only if u=x and vyE(H) or uxE(G) and v=y.

Theorem 3. If K is a graph with vertex set V(K)={u1,u2,,uk} and H is a graph with vertex set V(H)={v1,v2,,vh}, then the equi neighbor polynomial of cartesian product of K and H is given by, Ne[K◻H;x]=Ne[K;x]r=1hxdr1+Ne[H;x]r=1kxdr+2(i,j)I×I|Ne(K,i)||Ne(H,j)|xi+j+m=1k+h2{i,jI;ijn(K,i)n(K,j)n(H,mi)n(H,mj)xm}, where dr=|N(ur)|,r=1,2,,k, dr1=|N(vr)|,r=1,2,,h, I={0,1,2,,k1}, n(K,i) and n(H,i) represent the number of vertices in K and H respectively with i neighbors.

Proof. Let the vertices of K◻H be denoted by uivj where i{1,2,,k} and j{1,2,,h}. Observe that from the definition of K◻H the number of neighbors of a vertex uivj in K◻H is equal to the sum of the number of neighbors of ui in K and the number of neighbors of vj in H; i{1,2,,k}, j{1,2,,h}. Therefore it is enough to consider the following four cases;

Case (i) Consider the vertex pairs of K◻H of the form (urvs,urvt), r{1,2,,k}, s,t{1,2,,h}, st; with |N(ur)|=dr in K and |N(vs)|=|N(vt)|=i in H. Then the pair (urvs,urvt) has (i+dr)-equi neighbors in K◻H and there are |Ne(H,i)| such pairs, for each r{1,2,,k}.

Case (ii) Consider the vertex pairs of K◻H of the form (usvr,utvr), where s,t{1,2,,k}, r{1,2,,h}, st, with |N(us)|=|N(ut)|=i in K and |N(vr)|=dr1 in H. Then the pair (usvr,utvr) has (i+dr1)-equi neighbors in K◻H and there are |Ne(K,i)| such pairs, for each r{1,2,,h}.

Case (iii) Consider the vertex pairs of K◻H of the form (usvr,utvl), where s,t{1,2,,k}, r,l{1,2,,h}, st, rl, with |N(us)|=|N(ut)|=i in K and |N(vr)|=|N(vl)|=j in H. Then the pair (usvr,utvl) has (i+j)-equi neighbors in K◻H and there are 2|Ne(K,i)||Ne(H,j)| such pairs.

Case (iv) Consider the vertex pairs of K◻H of the form (usvr,utvl), s,t{1,2,,k}, r,l{1,2,,h} with |N(us)|=i, |N(ut)|=j in K, where ij. Then the pair (usvr,utvl) has m-equi neighbors in K◻H and there are n(K,i)n(K,j)n(H,mi)n(H,mj) such pairs for each m{1,2,,h+k2}. Thus,

Ne[K◻H;x]=i=0h1|Ne(H,i)|xi×r=1kxdr+i=0k1|Ne(K,i)|xi×r=1hxdr1+(i,j)I×I2|Ne(K,i)||Ne(H,j)|xi+j+m=1k+h2{i,jI;ijn(K,i)n(K,j)n(H,mi)n(H,mj)xm}=Ne[K;x]r=1hxdr1+Ne[H;x]r=1kxdr+2(i,j)I×I|Ne(H,i)||Ne(H,j)|xi+j+m=1k+h2{i,jI;ijn(K,i)n(K,j)n(H,mi)n(H,mj)xm}. This completes the proof. ◻

A Ladder graph Ln;n2 is obtained as the cartesian product of two paths one of which has only one edge.

Corollary 5. For a Ladder graph Ln, we have Ne[Ln;x]=(n2)(2n5)x3+6x2;n2.

Proof. Since Ln is isomorphic to Pn◻P2, N[Ln;x]=N[Pn◻P2;x]. Note that Ne[Pn;x]=(n22)x2+x. So we obtain, Ne[Ln;x]=x[2x+(n2)x2]+[x+(n22)x2]2x+2[x2+(n22)x3]=(n2)(2n5)x3+6x2. This completes the proof. ◻

A Circular ladder graph CLn;n3 is obtained as the cartesian product of the cycle graph Cn and the path P2.

Corollary 6. For a Circular ladder graph CLn, we have Ne[CLn;x]=n(2n1)x3;n3.

Proof. Using the Lemma 1, Theorem 3 and using the fact that CLn is isomorphic to Cn◻P2, it follows that, Ne[CLn;x]=x×nx2+(n2)x2×2x+2(n2)×1×x3=n(2n1)x3. This completes the proof. ◻

A mbook graph is obtained as the cartesian product of the star graph K1,m and the path graph P2.

Corollary 7. If Bm is a mbook graph, then we have Ne[Bm;x]={xm+1+m(2m1)x2,if  m>1,6x2,if  m=1.

Proof. Since Bm is isomorphic to K1,m◻P2, Ne[Bm;x]=Ne[K1,m◻P2;x]. Note that Ne[K1,m;x]={(m2)x, m1,x, m=1.. Thus we obtain,

Case (i) Let m>1. Then, Ne[K1,m◻P2;x]=x[xm+mx]+(m2)x×2x+2(m2)×1×x2=xm+1+m(2m1)x2.

Case (ii) Let m=1. Then, Ne[K1,m◻P2;x]=x×2x+x×2x+2×1×1×x2=6x2.

This completes the proof. ◻

A rooted graph is a graph in which one vertex is distinguished as a root. The Rooted product of a graph G and a rooted graph H is obtained as follows;

Take |V(G)| copies of H and for each vertex vi of G, identify vi with the root vertex of the ith copy of H [4].

Theorem 4. Let G be a graph with n vertices. Let H be a rooted graph with m vertices, having a root vertex v1 with degree d. If G is the rooted product of G and H, then,

Ne[G;x]=xdNe[G;x]+n2Ne[H;x]+(n2)i=0,idm1n(H,i)xi+i=0n1n(G,i)n(H,i+d)xi(n+1)2n(H,d)+n(G,0)n(H,d)+(n+1)2n(G,0)nxd, where n(G,i) and n(H,i) represent the number of vertices of G and H respectively with i neighbors.

Proof. Let V(G)={u1,u2,,un} and let V(H)={v1,v2,,vm} where v1 is the root vertex. Then a vertex of G can be represented by urvs where r{1,2,,n} and s{1,2,,m}. Observe that, from the definition of the rooted product of G and H,

  1. the number of neighbors of vertices of G of the form urv1, r{1,2,,n} is equal to the sum of the number of neighbors of ur in G and the number of neighbors of v1 in H.

  2. the number of neighbors of vertices of G of the form urvs, r{1,2,,n}; s{2,3,,m} is equal to the number of neighbors of vs in H.

Therefore it is enough to consider the following seven cases;

Case (i) Consider the vertex pairs of G of the form (urv1,usv1), where r,s{1,2,,n}; rs; with |N(ur)|=|N(us)|=i in G. Then urv1 has (i+d) neighbors in G. The same is true for usv1. Hence the pair (urv1,usv1) has (i+d)-equi neighbors in G and there are |Ne(G,i)| such pairs.

Case (ii) Consider the vertex pairs of G of the form (urvs,urvt), where r{1,2,,n}, s,t{2,3,,m}, st, with |N(vs)|=|N(vt)|=id in H. Then the pair (urvs,urvt) has i-equi neighbors in G and there are |Ne(H,i)| such pairs, corresponding to each r{1,2,,n}.

Case (iii) Consider the vertex pairs of G of the form (urvs,urvt), where r{1,2,,n}, s,t{2,3,,m}, st, with |N(vs)|=|N(vt)|=d in H. Then the pair (urvs,urvt) has d-equi neighbors in G and there are |Ne(H,d)|n(H,d)+1 such pairs, corresponding to each r{1,2,,n}.

Case (iv) Consider the vertex pairs of G of the form (urv1,urvs), where r{1,2,,n}, s{2,3,,m}, with |N(ur)|=i in G and |N(vs)|=i+d in H.

Subcase (i) Let i=0. Then the pair (urv1,urvs) has d-equi neighbors in G and there are n(G,0){n(H,d)1} such pairs.

Subcase (ii) Let i0. Then the pair (urv1,urvs) has (i+d)-equi neighbors in G and there are n(G,i)n(H,i+d) such pairs.

Case (v) Consider the vertex pairs of G of the form (urv1,usvt), where r,s{1,2,,n}, t{2,3,,m}, rs, with |N(ur)|=i in G and |N(vt)|=i+d in H.

Subcase (i) Let i=0. Then the pair (urv1,usvt) has d-equi neighbors in G and there are (n1)n(G,0){n(H,d)1} such pairs.

Subcase (ii) Let i0. Then the pair (urv1,usvt) has (i+d)-equi neighbors in G and there are (n1)n(G,i)n(H,i+d) such pairs.

Case (vi) Consider the vertex pairs of G of the form (urvs,utvl), where r,t{1,2,,n}, s,l{2,3,,m}, rt, with |N(vs)|=|N(vl)|=id in H. Then the pair (urvs,utvl) has i-equi neighbors in G and there are (n2)[2|Ne(H,i)|+n(H,i)] such pairs.

Case (vii) Consider the vertex pairs of G of the form (urvs,utvl), where r,t{1,2,,n}, s,l{2,3,,m}, rt, with |N(vs)|=|N(vl)|=d in H. Then the pair (urvs,utvl) has d-equi neighbors in G and there are (n2)[2|Ne(H,d)|n(H,d)+1] such pairs. Thus,

Ne[G;x]=xdNe[G;x]+ni=0;idm1|Ne(H,i)|xi+n{|Ne(H,d)|n(H,d)+1}xd+n(G,0){n(H,d)1}xd+i=1n1n(G,i)n(H,i+d)xi+d+(n1)n(G,0){n(H,d)1}xd+(n1)i=1n1n(G,i)n(H,i+d)xi+d+(n2)i=0;idm1{2|Ne(H,i)|+n(H,i)}xi+(n2){2|Ne(H,d)|n(H,d)+1}xd=xdNe[G;x]+n2Ne[H;x]+(n2)i=0,idm1n(H,i)xi+i=1n1n(G,i)n(H,i+d)xi(n+1)2n(H,d)+n(G,0)n(H,d)+(n+1)2n(G,0)}nxd. This completes the proof. ◻

The Tensor product of two graphs K and H is the graph K×H with vertex set V(K)×V(H) and the vertices (u,v) and (x,y) are adjacent if and only if uxE(K) and vyE(H).

Theorem 5. If K is a graph with vertex set V(K)={u1,u2,,uk} and H is a graph with vertex set V(H)={v1,v2,,vh}, then the equi neighbor polynomial of Tensor product of K and H is given by Ne[K×H;x]=r=1kNe[K;xdr1]+r=1hNe[H;xdr]+2(i,j)I1×I2|Ne(H,i)||Ne(H,j)|xij+i,jI1;ij{(m,n)I2×I2;im=jnn(K,i)n(K,j)n(H,m)n(H,n)xim}+n(K,0)m,nI2;mnn(H,m)n(H,n), where I1={0,1,2,,k1}, I2={0,1,2,,h1}, dr=|N(ur)|,r{1,2,,k}, dr1=|N(vr)|, r{1,2,,h}, n(K,i) and n(H,i) represent the number of vertices in K and H with i neighbors respectively.

Proof. Let the vertices of K×H be denoted by uivj, where i{1,2,,k} and j{1,2,,h}. Observe that from the definition of K×H the number of neighbors of a vertex uivj in K×H is equal to the product of the number of neighbors of ui in K and the number of neighbors of vj in H; i{1,2,,k}, j{1,2,,h}. Therefore it is enough to consider the following five cases;

Case (i) Consider the vertex pairs of K×H of the form (urvs,urvt), where r{1,2,,k}, s,t{1,2,,h}, st, with |N(ur)|=dr in K and |N(vs)|=|N(vt)|=i in H. Then the pair (urvs,urvt) has (i×dr)– equi neighbors in K×H and there are |Ne(H,i)| such pairs, corresponding to each r{1,2,,k}.

Case (ii) Consider the vertex pairs of K×H of the form (usvr,utvr) where s,t{1,2,,k}, r{1,2,,h}, st, with |N(us)|=|N(ut)|=i in K and |N(vr)|=dr1 in H. Then the pair (usvr,utvr) has (i×dr1)– equi neighbors in K×H and there are |Ne(K,i)| such pairs, corresponding to each r{1,2,,h}.

Case (iii) Consider the vertex pairs of K×H of the form (usvr,utvl), where s,t{1,2,,k}, r,l{1,2,,h}, st, rl, with |N(us)|=|N(ut)|=i in K and |N(vr)|=|N(vl)|=j in H. Then the pair (usvr,utvl) has (i×j)– equi neighbors in K×H and there are 2|Ne(K,i)||Ne(H,j)| such pairs.

Case (iv) Consider the vertex pairs of K×H of the form (usvr,utvl), where s,t{1,2,,h}, r,l{1,2,,k} with |N(us)|=i, |N(ut)|=j in K, where ij, and |N(vr)|=m, |N(vl)|=n satisfying im=jn. Then for any set of values for i and j satisfying ij, we get (m,n)I2×I2;im=jnn(K,i)n(K,j)n(H,m)n(H,n) pairs of vertices with im-equi neighbors.

Case (v) Consider the vertex pairs of K×H of the form (urvs,urvt), where r{1,2,,k}, s,t{1,2,,h}, with |N(ur)|=0, |N(vs)|=m, |N(vt)|=n, mn. Then the pair (urvs,urvt) has zero-equi neighbors and there are n(K,0)m,nI2;mnn(H,m)n(H,n) such pairs.

This completes the proof. ◻

3. Conclusion

In this paper, we discussed the equi neighbor polynomial of graphs obtained by some graph binary operations. The paper has provided explicit formulae to find the equi neighbor polynomial of some well-known graph products such as join, corona, cartesian product, rooted product, and tensor product of two graphs in terms of the equi neighbor polynomial of the parent graphs.

Acknowledgments

We thank the anonymous referees for their helpful comments.

References:

  1. Dhanya, P., and Kumar, A. V., 2004. The number of equi neighbor sets graphs. Advances and Applications in Discrete Mathematics
  2. Frucht, R., and Harary, F., 1970. On the corona of two graphs. Aequationes Mathematicae, 4, pp.322-325.
  3. Shyama, M. P., and Anilkumar, V., 2016. On the roots of Hosoya polynomials. Journal of Discrete Mathematical Sciences and Cryptography, 19, pp.199-219.
  4. Godsil, C. D., and McKay, B. D., 1978. A new graph product and its spectrum. Bulletin of Australian Mathematical Society, 18, pp.21-28.