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

Structure and coloring of some (P7P_{7}, C4C_{4})-free graphs

Ran Chen1,111Email: [email protected],   Di Wu1,222Email: [email protected],   Baogang Xu1,333Email: [email protected] OR [email protected]. Supported by NSFC 11931006

1Institute of Mathematics, School of Mathematical Sciences
Nanjing Normal University, 1 Wenyuan Road, Nanjing, 210023, China
Abstract

Let GG be a graph. We use PtP_{t} and CtC_{t} to denote a path and a cycle on tt vertices, respectively. A diamond is a graph obtained from two triangles that share exactly one edge. A kite is a graph consists of a diamond and another vertex adjacent to a vertex of degree 2 of the diamond. A gem is a graph that consists of a P4P_{4} plus a vertex adjacent to all vertices of the P4P_{4}. In this paper, we prove some structural properties to (P7,C4,(P_{7},C_{4}, diamond)-free graphs, (P7,C4,(P_{7},C_{4}, kite)-free graphs and (P7,C4,(P_{7},C_{4}, gem)-free graphs. As their corollaries, we show that (i) χ(G)max{3,ω(G)}\chi(G)\leq\max\{3,\omega(G)\} if GG is (P7,C4,(P_{7},C_{4}, diamond)-free, (ii) χ(G)ω(G)+1\chi(G)\leq\omega(G)+1 if GG is (P7,C4,(P_{7},C_{4}, kite)-free and (iii) χ(G)2ω(G)1\chi(G)\leq 2\omega(G)-1 if GG is (P7,C4,(P_{7},C_{4}, gem)-free. These conclusions generalize some results of Choudum et al [4] and Lan et al [14].

Key words and phrases: P7P_{7}-free graphs, chromatic number, clique number

AMS 2000 Subject Classifications: 05C15, 05C75

1 Introduction

All graphs considered in this paper are finite and simple. We follow [1] for undefined notations and terminology. We use PtP_{t} and CtC_{t} to denote a path and a cycle on tt vertices, respectively. Let GG be a graph, let vV(G)v\in V(G), and let XX and YY be two subsets of V(G)V(G). We say that vv is complete to XX if vv is adjacent to all vertices of XX, and say that vv is anticomplete to XX if vv is not adjacent to any vertex of XX. We say that XX is complete (resp. anticomplete) to YY if each vertex of XX is complete (resp. anticomplete) to YY.

A hole of GG is an induced cycle of length at least 4, and a kk-hole is a hole of length kk. Given a graph HH, we say that GG is HH-free if it dose not contain an induced subgraph isomorphic to HH. Given a set {H1,H2,}\{H_{1},H_{2},\cdots\} of graphs, we say that GG is (H1,H2,)(H_{1},H_{2},\cdots)-free if GG is HiH_{i}-free for each ii.

A clique blowup of a graph HH with V(H)={v1,v2,,vn}V(H)=\{v_{1},v_{2},\cdots,v_{n}\} is any graph GG such that V(G)V(G) can be partitioned into nn nonempty cliques AiA_{i}, viV(H)v_{i}\in V(H) such that AiA_{i} is complete to AjA_{j} if vivjv_{i}\sim v_{j}, and AiA_{i} is anticomplete to AjA_{j} if vi≁vjv_{i}\not\sim v_{j}. Particularly, if |Ai|=t|A_{i}|=t for all i{1,,n}i\in\{1,\cdots,n\}, then we call GG a tt-size clique blowup of HH.

For vV(G)v\in V(G), let NG(v)N_{G}(v) be the set of vertices adjacent to vv, dG(v)=|NG(v)|d_{G}(v)=|N_{G}(v)|, NG[v]=NG(v){v}N_{G}[v]=N_{G}(v)\cup\{v\}, and MG(v)=V(G)NG[v]M_{G}(v)=V(G)\setminus N_{G}[v]. For XV(G)X\subseteq V(G), let NG(X)={uV(G)X|uN_{G}(X)=\{u\in V(G)\setminus X\;|\;u has a neighbor in X}X\} and MG(X)=V(G)(XNG(X))M_{G}(X)=V(G)\setminus(X\cup N_{G}(X)). Moreover, G[X]G[X] denotes the subgraph of GG induced by XX. If it does not cause any confusion, we usually omit the subscript GG and simply write N(v)N(v), d(v)d(v), N[v]N[v], M(v)M(v), N(X)N(X) and M(X)M(X). Let δ(G)\delta(G) denote the minimum degree of GG. For uu, vV(G)v\in V(G), we simply write uvu\sim v if uvE(G)uv\in E(G), and write u≁vu\not\sim v if uvE(G)uv\not\in E(G).

A clique (stable set) of GG is a set of mutually adjacent (non-adjacent) vertices in G.G. The clique number of GG, denoted by ω(G),\omega(G), is the maximum size of a clique in GG. For positive integer kk, a kk-coloring of GG is a function c:V(G){1,,k}c:V(G)\rightarrow\{1,\cdots,k\}, such that for each edge uvuv of GG, c(u)c(v)c(u)\not=c(v). The chromatic number of GG, denoted by χ(G)\chi(G), is the minimum number kk for which there exists a kk-coloring of GG. A graph is perfect if all its induced subgraphs HH satisfy χ(H)=ω(H)\chi(H)=\omega(H). A family 𝒢{\cal G} of graphs is said to be χ\chi-bounded [11] if there exists a function ff such that for every graph G𝒢G\in{\cal G}, χ(G)f(w(G))\chi(G)\leq f(w(G)). If such a function ff does exist to 𝒢{\cal G}, then ff is called a χ\chi-binding function of 𝒢{\cal G}.

A clique cutset of GG is a clique KK in GG such that GKG-K has more components than GG. A vertex uu is a cut vertex of GG if it is a clique cutset of size 1.

A diamond is a graph obtained from two triangles that share exactly one edge. A kite is a graph consists of a diamond and another vertex adjacent to a vertex of degree 2 of the diamond. A gem is a graph that consists of a P4P_{4} plus a vertex adjacent to all vertices of the P4P_{4}. A bull is a graph consisting of a triangle with two disjoint pendant edges. Let C=v1v2v3v4v5v6v7v1C=v_{1}v_{2}v_{3}v_{4}v_{5}v_{6}v_{7}v_{1} be a 7-hole. We use FF to denote a graph obtained from CC by adding a stable set {y1,y2,y3}\{y_{1},y_{2},y_{3}\} such that N(y1)V(C)={v1,v4,v5}N(y_{1})\cap V(C)=\{v_{1},v_{4},v_{5}\}, N(y2)V(C)={v2,v5,v6}N(y_{2})\cap V(C)=\{v_{2},v_{5},v_{6}\} and N(y3)V(C)={v3,v6,v7}N(y_{3})\cap V(C)=\{v_{3},v_{6},v_{7}\}. (See Figure 1).

Refer to caption
Figure 1: Illustration of diamond, kite, gem, bull, Petersen graph and FF.

Gyárfás [11] showed that χ(G)(t1)ω(G)1\chi(G)\leq(t-1)^{\omega(G)-1} for all PtP_{t}-free graphs. This upper bound was improved to χ(G)(t2)ω(G)1\chi(G)\leq(t-2)^{\omega(G)-1} in [10]. However, this χ\chi-binding function is exponential in ω(G)\omega(G). It is natural to ask that is it possible to improve the exponential bound for PtP_{t}-free graphs to a polynomial bound [18]. Interested readers are referred to [17, 19, 21] for more results and still open problems on this topic.

It is known that P3P_{3}-free graphs are complete graphs and P4P_{4}-free graphs are perfect [22]. But up to now, no polynomial binding function for PtP_{t}-free graphs has been found, when t5t\geq 5. We refer the interested readers to [9] for results of P5P_{5}-free graphs, and list here some conclusions on binding functions of some P6P_{6}-free or P7P_{7}-free graphs.

Choudum, Karthick and Shalu [5] proved that every (P6P_{6}, gem)-free graph GG satisfies χ(G)8ω(G)\chi(G)\leq 8\omega(G). Cameron, Huang and Merkel [2] proved that every (P6P_{6}, diamond)-free graph GG satisfies χ(G)ω(G)+3\chi(G)\leq\omega(G)+3. Gasper and Huang [8] proved that every (P6,C4)(P_{6},C_{4})-free graph GG satisfies χ(G)32ω(G)\chi(G)\leq\lfloor\frac{3}{2}\omega(G)\rfloor. This bound was recently improved by Karthick and Maffray [13] to χ(G)54ω(G)\chi(G)\leq\lceil\frac{5}{4}\omega(G)\rceil. Mishra [15] proved that every (P6P_{6}, diamond, bull)-free graphs GG satisfies χ(G)4\chi(G)\leq 4 when ω(G)=2\omega(G)=2 and χ(G)=ω(G)\chi(G)=\omega(G) when ω(G)2\omega(G)\neq 2, and every (P7P_{7}, diamond, bull)-free graph GG satisfies χ(G)max{7,ω(G)}\chi(G)\leq\max\{7,\omega(G)\}. Cameron, Huang, Penev and Sivaraman [3] showed that every (P7,C4,C5)(P_{7},C_{4},C_{5})-free graph satisfies χ(G)32ω(G)\chi(G)\leq\frac{3}{2}\omega(G). This upper bound was recently improved by Huang [12] to χ(G)119ω(G)\chi(G)\leq\lceil\frac{11}{9}\omega(G)\rceil.

In 2021, Choudum, Karthick and Belavadi [4] proved that χ(G)2ω(G)1\chi(G)\leq 2\omega(G)-1 if GG is (P7,C7,C4(P_{7},C_{7},C_{4}, gem)-free, and χ(G)max{3,ω(G)}\chi(G)\leq\max\{3,\omega(G)\} if GG is (P7,C7,C4(P_{7},C_{7},C_{4}, diamond)-free. In 2022, Lan, Zhou and Liu [14] proved that χ(G)max{3,ω(G)}\chi(G)\leq\max\{3,\omega(G)\} if GG is (P6,C4(P_{6},C_{4}, diamond)-free.

In this paper, we study (P7,C4,H)P_{7},C_{4},H)-free graphs for HH\in {diamond, kite, gem}, and get some structural properties of these graphs. For graphs GG and HH, we use G+HG+H to denote a graph with vertex set V(G)V(H)V(G)\cup V(H) and edge set E(G)E(H){uv|uV(G),vV(H)}E(G)\cup E(H)\cup\{uv\;|\;u\in V(G),v\in V(H)\}. A bisimplicial vertex is one whose neighborhoods is the union of two cliques.

Theorem 1.1

Let GG be a connected (P7,C4(P_{7},C_{4}, diamond)-free graph without clique cutsets. If GG is isomorphic to neither FF nor the Petersen graph, then δ(G)max{2,ω(G)1}\delta(G)\leq\max\{2,\omega(G)-1\}.

Theorem 1.2

Let GG be a connected (P7,C4(P_{7},C_{4}, kite)-free graph with δ(G)ω(G)+1\delta(G)\geq\omega(G)+1. If GG has no clique cutsets, then there is an integer 0\ell\geq 0 such that G=K+HG=K_{\ell}+H, where HH is the Petersen graph or FF.

Theorem 1.3

Let GG be a connected (P7,C4(P_{7},C_{4}, gem)-free graph without clique cutsets. If GG is not the clique blowup of the Petersen graph, then GG has a bisimplicial vertex.

There exist graphs showing that all the requirements in our theorems are necessary. We will give these examples separately after completing their proofs. As immediate consequences of these theorems, we have the following corollaries that generalize some results of Choudum, Karthick and Belavadi [4], and of Lan, Zhou and Liu [14].

Corollary 1.1

Every (P7,C4(P_{7},C_{4}, diamond)-free graph GG satisfies χ(G)max{3,ω(G)}\chi(G)\leq\max\{3,\omega(G)\}.

Corollary 1.2

Every (P7,C4(P_{7},C_{4}, kite)-free graph GG satisfies χ(G)ω(G)+1\chi(G)\leq\omega(G)+1.

Corollary 1.3

Every (P7,C4(P_{7},C_{4}, gem)-free graph GG satisfies χ(G)2ω(G)1\chi(G)\leq 2\omega(G)-1.

We will prove Theorem 1.1 in Section 2, prove Theorem 1.2 in Section 3, and prove Theorem 1.3 in Section 4.

2 The class of (P7,C4,(P_{7},C_{4}, diamond)-free graphs

In this section, we focus on (P7,C4(P_{7},C_{4}, diamond)-free graphs, and prove Theorem 1.1 and Corollary 1.1. Theorem 1.1 generalizes the following result from [4].

Lemma 2.1

[4] Let GG be a connected (P7,C7,C4(P_{7},C_{7},C_{4}, diamond)-free graph. Then GG has a clique cutset or δ(G)max{2,ω(G)1}\delta(G)\leq max\{2,\omega(G)-1\} or GG is the Petersen graph.

Proof of Theorem 1.1: Let GG be a connected (P7,C4(P_{7},C_{4}, diamond)-free graph. If GG is C7C_{7}-free, then the statement follows directly from Lemma 2.1. So, we suppose that GG has 7-holes, and let v1v2v3v4v5v6v7v1v_{1}v_{2}v_{3}v_{4}v_{5}v_{6}v_{7}v_{1} be a 7-hole of GG. Let A={v1,v2,,v7}A=\{v_{1},v_{2},\cdots,v_{7}\}, and let R=M(A)R=M(A). During the proof of Theorem 1.1, every subscript is understood to be modulo 7. For each ii\in{1, \cdots, 7}, let

Xi\displaystyle X_{i} =\displaystyle= {xN(A)|N(x)A={vi,vi+3}},and\displaystyle\{x\in N(A)|~{}N(x)\cap A=\{v_{i},v_{i+3}\}\},~{}\mbox{and}
Yi\displaystyle Y_{i} =\displaystyle= {xN(A)|N(x)A={vi,vi+3,vi+4}}.\displaystyle\{x\in N(A)|~{}N(x)\cap A=\{v_{i},v_{i+3},v_{i+4}\}\}.

Let X=X1X7X=X_{1}\cup\cdots\cup X_{7} and Y=Y1Y7Y=Y_{1}\cup\cdots\cup Y_{7}. Next, we prove some useful properties of N(A)N(A).

N(A)=XYN(A)=X\cup Y, and V(G)=AXYRV(G)=A\cup X\cup Y\cup R. (1)

It is certain that XYN(A)X\cup Y\subseteq N(A). Since GG is diamond-free, we have that no vertex of N(A)N(A) may have three consecutive neighbors in AA. Let xN(A)x\in N(A). By symmetry, we may assume xv1x\sim v_{1}. If N(x)A={v1}N(x)\cap A=\{v_{1}\}, then G[{x,v1,v2,v3,v4,v5,v6}]=P7G[\{x,v_{1},v_{2},v_{3},v_{4},v_{5},v_{6}\}]=P_{7}, a contradiction. Therefore |N(x)A|2|N(x)\cap A|\geq 2.

Suppose xv2x\sim v_{2}. Then, xx has neighbors in {v3,,v7}\{v_{3},\cdots,v_{7}\} as otherwise G[{x,v2,v3,v4,v5,v6,v7}]=P7G[\{x,v_{2},v_{3},v_{4},v_{5},v_{6},v_{7}\}]=P_{7}, and x≁v3x\not\sim v_{3} and x≁v7x\not\sim v_{7} as otherwise xx has three consecutive neighbors in AA. To forbid a 4-hole on {x,v2,v3,v4}\{x,v_{2},v_{3},v_{4}\} or {x,v1,v6,v7}\{x,v_{1},v_{6},v_{7}\}, we have that x≁x4x\not\sim x_{4} and x≁x6x\not\sim x_{6}. Therefore xv5x\sim v_{5}, which implies that xYx\in Y. By symmetry, we have xYx\in Y when xv7x\sim v_{7}. So, we way assume that x≁v2x\not\sim v_{2} and x≁v7x\not\sim v_{7}, and thus x≁v3x\not\sim v_{3} and x≁v6x\not\sim v_{6} to forbid a 4-hole on {x,v1,v2,v3}\{x,v_{1},v_{2},v_{3}\} or {x,v1,v6,v7}\{x,v_{1},v_{6},v_{7}\}.

Now, xXx\in X if xx has exactly one neighbor in {v4,v5}\{v_{4},v_{5}\}, and xYx\in Y if xx is complete to {v4,v5}\{v_{4},v_{5}\}. This proves (1).

For each i{1,,7}i\in\{1,\cdots,7\}, |Xi|1|X_{i}|\leq 1 and |Yi|1|Y_{i}|\leq 1. (2)

Suppose to its contrary, we may assume by the symmetry that X1X_{1} or Y1Y_{1} has two distinct vertices xx and xx^{\prime}. Then G[{x,x,v1,v4}]G[\{x,x^{\prime},v_{1},v_{4}\}] is a diamond if xxx\sim x^{\prime}, and xv1xv4xxv_{1}x^{\prime}v_{4}x is a 4-hole if x≁xx\not\sim x^{\prime}. Both are contradictions. Therefore, (2) holds.

For i{1,,7}i\in\{1,\cdots,7\}, let Xi={xi}X_{i}=\{x_{i}\} if XiØX_{i}\neq\mbox{{\rm\O}}, and Yi={yi}Y_{i}=\{y_{i}\} if YiØY_{i}\neq\mbox{{\rm\O}}.

N(A)N(A) is a stable set. (3)

Suppose to its contrary that N(A)N(A) has two adjacent vertices uu and vv. First we assume that {u,v}XØ.\{u,v\}\cap X\neq\mbox{{\rm\O}}. Without loss of generality, suppose that uX1.u\in X_{1}. Then v≁vjv\not\sim v_{j} for jj\in{2, 7} as otherwise G[{u,v,v1,vj}]G[\{u,v,v_{1},v_{j}\}] is a diamond if vv1v\sim v_{1}, and uvvjv1uuvv_{j}v_{1}u is a 4-hole if v≁v1v\not\sim v_{1}. Moreover, v≁vkv\not\sim v_{k} for k{3,5}k\in\{3,5\} as otherwise G[{u,v,v4,vk}]G[\{u,v,v_{4},v_{k}\}] is a diamond if vv4v\sim v_{4}, and uvvkv4uuvv_{k}v_{4}u is a 4-hole if v≁v4v\not\sim v_{4}. Therefore N(v)A{v1,v4,v6}N(v)\cap A\subseteq\{v_{1},v_{4},v_{6}\}. Since |N(v)A|2|N(v)\cap A|\geq 2 and |X1|1|X_{1}|\leq 1, we have that N(v)A{{v1,v6},{v4,v6},{v1,v4,v6}}N(v)\cap A\in\{\{v_{1},v_{6}\},\{v_{4},v_{6}\},\{v_{1},v_{4},v_{6}\}\}. Then vv6v7v1vvv_{6}v_{7}v_{1}v is a 4-hole if N(v)A{{v1,v6},{v1,v4,v6}}N(v)\cap A\in\{\{v_{1},v_{6}\},\{v_{1},v_{4},v_{6}\}\}, and vv4v5v6vvv_{4}v_{5}v_{6}v is a 4-hole if N(v)A={v4,v6}N(v)\cap A=\{v_{4},v_{6}\}, both are contradictions.

So, we may assume that uYu\in Y and vYv\in Y. Without loss of generality, suppose that uY1u\in Y_{1}. Then v≁vjv\not\sim v_{j} for j{2,7}j\in\{2,7\} as otherwise G[{u,v,v1,vj}]G[\{u,v,v_{1},v_{j}\}] is a diamond if vv1v\sim v_{1}, and uvvjv1uuvv_{j}v_{1}u is a 4-hole if v≁v1v\not\sim v_{1}. Moreover, v≁v3v\not\sim v_{3} to forbid a diamond on {u,v,v3,v4}\{u,v,v_{3},v_{4}\} if vv4v\sim v_{4} or a 4-hole on {u,v,v3,v4}\{u,v,v_{3},v_{4}\} if v≁v4v\not\sim v_{4}. If vv5v\sim v_{5}, then v≁v4v\not\sim v_{4} as otherwise vY1v\in Y_{1} and thus |Y1|2|Y_{1}|\geq 2, a contradiction. But now G[{u,v,v4,v5}]G[\{u,v,v_{4},v_{5}\}] is a diamond. Therefore, N(v)A={v1,v4,v6}N(v)\cap A=\{v_{1},v_{4},v_{6}\} since |N(v)A|=3|N(v)\cap A|=3. However, vv4v5v6vvv_{4}v_{5}v_{6}v is a 4-hole, a contradiction. This proves (3).

For each i{1,,7}i\in\{1,\cdots,7\}, ethier Xi=ØX_{i}=\mbox{{\rm\O}} or Xi+2Xi+5=ØX_{i+2}\cup X_{i+5}=\mbox{{\rm\O}}. (4)

If this is not true, we may assume by symmetry that X1ØX_{1}\neq\mbox{{\rm\O}} and X3ØX_{3}\neq\mbox{{\rm\O}}, then x1≁x3x_{1}\not\sim x_{3} by (3), which implies that G[{x1,x3,v1,v2,v4,v5,v6}]=P7G[\{x_{1},x_{3},v_{1},v_{2},v_{4},v_{5},v_{6}\}]=P_{7}, a contradiction. This proves (4).

For each i{1,,7}i\in\{1,\cdots,7\}, ethier Yi=ØY_{i}=\mbox{{\rm\O}} or Yi+3Yi+4=ØY_{i+3}\cup Y_{i+4}=\mbox{{\rm\O}}. (5)

If (5) does not hold, we may assume by symmetry that Y1ØY_{1}\neq\mbox{{\rm\O}} and Y4ØY_{4}\neq\mbox{{\rm\O}}, then y1≁y4y_{1}\not\sim y_{4} by (3), which implies that y1v4y4v1y1y_{1}v_{4}y_{4}v_{1}y_{1} is a 4-hole, a contradiction. This proves (5).

For each i{1,,7}i\in\{1,\cdots,7\}, ethier Xi=ØX_{i}=\mbox{{\rm\O}} or YiYi+1Yi+2Yi+3=ØY_{i}\cup Y_{i+1}\cup Y_{i+2}\cup Y_{i+3}=\mbox{{\rm\O}}. (6)

If it is not the case, we may assume by symmetry that X1ØX_{1}\neq\mbox{{\rm\O}} and Y1Y2Y3Y4ØY_{1}\cup Y_{2}\cup Y_{3}\cup Y_{4}\neq\mbox{{\rm\O}}. Let yY1Y2Y3Y4y\in Y_{1}\cup Y_{2}\cup Y_{3}\cup Y_{4}. Then x1≁yx_{1}\not\sim y by (3), which implies that x1v4yv1x1x_{1}v_{4}yv_{1}x_{1} is a 4-hole if yY1Y4y\in Y_{1}\cup Y_{4}, G[{x1,y,v2,v3,v4,v6,v7}]=P7G[\{x_{1},y,v_{2},v_{3},v_{4},v_{6},v_{7}\}]=P_{7} if yY2y\in Y_{2}, and G[{x1,y,v1,v2,v4,v5,v6}]=P7G[\{x_{1},y,v_{1},v_{2},v_{4},v_{5},v_{6}\}]=P_{7} if yY3y\in Y_{3}, all of which are contradictions. This proves (6).

Suppose that Y=Ø.Y=\mbox{{\rm\O}}. By (4), we have that {X1,X2,,X7}\{X_{1},X_{2},\cdots,X_{7}\} has at most three nonempty elements, and thus d(vi)=2d(v_{i})=2 for some i{1,,7}i\in\{1,\cdots,7\}. If ω(G)=2\omega(G)=2, then Y=ØY=\mbox{{\rm\O}}, and thus δ(G)2\delta(G)\leq 2.

Suppose that ω(G)4\omega(G)\geq 4 and suppose by symmetry that Y1ØY_{1}\neq\mbox{{\rm\O}}. Then, Y4Y5X1X7X6X5=ØY_{4}\cup Y_{5}\cup X_{1}\cup X_{7}\cup X_{6}\cup X_{5}=\mbox{{\rm\O}} by (5) and (6), and one can verify easily that d(v1)=3ω(G)1max{2,ω(G)1}d(v_{1})=3\leq\omega(G)-1\leq\max\{2,\omega(G)-1\}.

So, we suppose that

  • ω(G)=3\omega(G)=3 and δ(G)3\delta(G)\geq 3,

  • GG is not isomorphic to FF and has no clique cutsets, and

  • Yi0={yi0}ØY_{i_{0}}=\{y_{i_{0}}\}\neq\mbox{{\rm\O}} for some integer i0i_{0}.

By (5) and (6), we have that Yi0+3Yi0+4Xi0Xi01Xi02Xi03=ØY_{i_{0}+3}\cup Y_{i_{0}+4}\cup X_{i_{0}}\cup X_{i_{0}-1}\cup X_{i_{0}-2}\cup X_{i_{0}-3}=\mbox{{\rm\O}}, and so

X=Xi0+1Xi0+2Xi0+3 and Y=Yi0Yi0+1Yi0+2Yi02Yi01.X=X_{i_{0}+1}\cup X_{i_{0}+2}\cup X_{i_{0}+3}\mbox{ and }Y=Y_{i_{0}}\cup Y_{i_{0}+1}\cup Y_{i_{0}+2}\cup Y_{i_{0}-2}\cup Y_{i_{0}-1}. (7)

We will show that

Xi0+1=Ø if Yi0Ø.\mbox{$X_{i_{0}+1}=\mbox{{\rm\O}}$ if $Y_{i_{0}}\neq\mbox{{\rm\O}}$}. (8)

For simplicity, we may take i0=1i_{0}=1 by symmetry to illustrate the procedure. Before proving (8), we firstly prove that

Xi0+2=Ø if Xi0+1Ø.\mbox{$X_{i_{0}+2}=\mbox{{\rm\O}}$ if $X_{i_{0}+1}\neq\mbox{{\rm\O}}$}. (9)

Notice that we take i0=1i_{0}=1. Suppose X2ØX_{2}\neq\mbox{{\rm\O}} and X3ØX_{3}\neq\mbox{{\rm\O}}. By (4) and (6), we have that X4X7Y2Y3Y4Y5Y6=ØX_{4}\cup X_{7}\cup Y_{2}\cup Y_{3}\cup Y_{4}\cup Y_{5}\cup Y_{6}=\mbox{{\rm\O}}, and so X=X2X3X=X_{2}\cup X_{3} and Y=Y1Y7Y=Y_{1}\cup Y_{7}. Since Y7=ØY_{7}=\mbox{{\rm\O}} implies that d(v7)=2d(v_{7})=2, we have that Y7={y7}ØY_{7}=\{y_{7}\}\neq\mbox{{\rm\O}}.

Since δ(G)3\delta(G)\geq 3 and N(A)N(A) is a stable set, both x2x_{2} and x3x_{3} have neighbors in RR. Let uu be a neighbor of x2x_{2} in RR and vv be a neighbor of x3x_{3} in RR, and let SS and TT be the components of G[R]G[R] which contains uu and vv, respectively. Since G[{x2,v1,v2,v6,v7}]=P5G[\{x_{2},v_{1},v_{2},v_{6},v_{7}\}]=P_{5} and GG is P7P_{7}-free, we have that x2x_{2} must be complete to V(S)V(S), and thus ω(S)2\omega(S)\leq 2 as ω(G)=3\omega(G)=3. Moreover, SS is P3P_{3}-free as GG is diamond-free, and thus SS is a K1K_{1} or a K2K_{2}. Similarly, x3x_{3} must be complete to V(T)V(T), and TT is a K1K_{1} or a K2K_{2}.

Suppose that S=TS=T. Then |V(S)|=|V(T)|=1|V(S)|=|V(T)|=1 as otherwise G[{x,y,x2,x3}]G[\{x,y,x_{2},x_{3}\}] is a diamond for any edge xyxy of SS, and so V(S)=V(T)={u}={v}V(S)=V(T)=\{u\}=\{v\}. If uy7u\sim y_{7}, then v3x3uy7v3v_{3}x_{3}uy_{7}v_{3} is a 4-hole, a contradiction. So, u≁y7u\not\sim y_{7}. By symmetry, u≁y1u\not\sim y_{1}. But now d(u)=2d(u)=2, a contradiction. Therefore, STS\neq T.

If V(S)={u}V(S)=\{u\}, then we have that u≁y1u\not\sim y_{1} to forbid a 4-hole on {u,y1,x2,v5}\{u,y_{1},x_{2},v_{5}\}, and thus ux3u\sim x_{3} and uy7u\sim y_{7} since δ(G)3\delta(G)\geq 3, which implies that there is a 4-hole on {u,x3,v3,y7}\{u,x_{3},v_{3},y_{7}\}, a contradiction. So, SS is a K2K_{2}. Similarly, TT is a K2K_{2}. Let V(S)={u,u}V(S)=\{u,u^{\prime}\} and V(T)={v,v}V(T)=\{v,v^{\prime}\}.

To forbid a 4-hole on {y1,x2,v5,u}\{y_{1},x_{2},v_{5},u\} or {y1,x2,v5,u}\{y_{1},x_{2},v_{5},u^{\prime}\}, y1y_{1} must be anticomplete to V(S)V(S). Similarly, y7y_{7} must be anticomplete to V(T)V(T). Since δ(G)3\delta(G)\geq 3, we have that each vertex of SS has neighbors in {x3,y7}\{x_{3},y_{7}\}. Since GG is diamond-free, we may assume, without loss of generality, that ux3u\sim x_{3} and uy7u^{\prime}\sim y_{7}. Similarly, we may suppose that vx2v\sim x_{2} and vy1v^{\prime}\sim y_{1}. But now, ux2vx3uux_{2}vx_{3}u is a 4-hole, a contradiction. This proves (9).

We can now turn to prove (8). Recall that we take i0=1i_{0}=1 by symmetry.

Suppose to its contrary that X2ØX_{2}\neq\mbox{{\rm\O}}. Then X3=ØX_{3}=\mbox{{\rm\O}} by (9). By (4) and (6), we have that X4X7Y2Y3Y4Y5=ØX_{4}\cup X_{7}\cup Y_{2}\cup Y_{3}\cup Y_{4}\cup Y_{5}=\mbox{{\rm\O}}, and thus X=X2X=X_{2} and Y=Y1Y6Y7Y=Y_{1}\cup Y_{6}\cup Y_{7}. Since, for i{6,7}i\in\{6,7\}, Yi=ØY_{i}=\mbox{{\rm\O}} implies that d(vi)=2d(v_{i})=2, we have that Y6ØY_{6}\neq\mbox{{\rm\O}} and Y7ØY_{7}\neq\mbox{{\rm\O}}. Since δ(G)3\delta(G)\geq 3 and N(A)N(A) is a stable set, we have that x2x_{2} must have a neighbor, say uu, in RR. Let BB be the component of G[R]G[R] that contains uu.

Since GG is P7P_{7}-free and diamond-free, we have that x2x_{2} must be complete to V(B)V(B) and BB is a K1K_{1} or a K2K_{2}. Suppose that V(B)={u}V(B)=\{u\}. Then, uu has at least two neighbors in N(A){x2}N(A)\setminus\{x_{2}\} as δ(G)3\delta(G)\geq 3. Since uy1u\sim y_{1} implies that there is a 4-hole on {u,y1,x2,v5}\{u,y_{1},x_{2},v_{5}\}, we have that u≁y1u\not\sim y_{1}, and thus uy6u\sim y_{6} and uy7u\sim y_{7}. But then, uy6v3y7uuy_{6}v_{3}y_{7}u is a 4-hole, a contradiction. So BB is a K2K_{2}. Let V(B)={u,v}V(B)=\{u,v\}. To forbid a 4-hole on {a,y1,x2,v5}\{a,y_{1},x_{2},v_{5}\} or {a,x2,v2,y6}\{a,x_{2},v_{2},y_{6}\}, where aV(B)a\in V(B), we have that V(B)V(B) is anticomplete to {y1y_{1}, y6y_{6}}. Since δ(G)3\delta(G)\geq 3, each vertex in BB must have neighbor in {y7}\{y_{7}\}, which implies a diamond G[{u,v,x2,y7}]G[\{u,v,x_{2},y_{7}\}], a contradiction. This proves (8).

Next, we prove that

X=Ø and Y=Yi0Yi0+1Yi0+2Yi01Yi02.X=\mbox{{\rm\O}}\mbox{ and }Y=Y_{i_{0}}\cup Y_{i_{0}+1}\cup Y_{i_{0}+2}\cup Y_{i_{0}-1}\cup Y_{i_{0}-2}. (10)

We still take i0=1i_{0}=1 for simplicity. By (7) and (8), we have that X=X3X4X=X_{3}\cup X_{4} and Y=Y1Y2Y3Y6Y7Y=Y_{1}\cup Y_{2}\cup Y_{3}\cup Y_{6}\cup Y_{7}.

Suppose X3ØX_{3}\neq\mbox{{\rm\O}}. By (6), we have that Y3Y4Y5Y6=ØY_{3}\cup Y_{4}\cup Y_{5}\cup Y_{6}=\mbox{{\rm\O}}, and so X=X3X4X=X_{3}\cup X_{4} and Y=Y1Y2Y7Y=Y_{1}\cup Y_{2}\cup Y_{7}. Since d(v2)3d(v_{2})\geq 3, we have that Y2ØY_{2}\neq\mbox{{\rm\O}}, and consequently X3=ØX_{3}=\mbox{{\rm\O}} by taking i0=2i_{0}=2 in (8), a contradiction. Therefore, X3=ØX_{3}=\mbox{{\rm\O}}, and X=X4X=X_{4} and Y=Y1Y2Y3Y6Y7Y=Y_{1}\cup Y_{2}\cup Y_{3}\cup Y_{6}\cup Y_{7}.

Suppose that X4ØX_{4}\neq\mbox{{\rm\O}}. Then, Y4Y5Y6Y7=ØY_{4}\cup Y_{5}\cup Y_{6}\cup Y_{7}=\mbox{{\rm\O}} by (6), and so Y=Y1Y2Y3Y=Y_{1}\cup Y_{2}\cup Y_{3}. Moreover, Y2ØY_{2}\neq\mbox{{\rm\O}} and Y3ØY_{3}\neq\mbox{{\rm\O}} as d(v2)3d(v_{2})\geq 3 and d(v3)3d(v_{3})\geq 3. Consequently, we have a contradiction as X4=ØX_{4}=\mbox{{\rm\O}} by taking i0=3i_{0}=3 in (8). Therefore, X4=ØX_{4}=\mbox{{\rm\O}}, and thus (10) holds.

Recall that Yi0ØY_{i_{0}}\neq\mbox{{\rm\O}} by our assumption. By symmetry, we may set i0=1i_{0}=1 in (10). Then, we have that

X=ØX=\mbox{{\rm\O}} and Y=Y1Y2Y3Y6Y7Y=Y_{1}\cup Y_{2}\cup Y_{3}\cup Y_{6}\cup Y_{7}. (11)

We prove further that

R=Ø if Y2Ø and Y3Ø.\mbox{$R=\mbox{{\rm\O}}$ if $Y_{2}\neq\mbox{{\rm\O}}$ and $Y_{3}\neq\mbox{{\rm\O}}$}. (12)

Suppose that Y2ØY_{2}\neq\mbox{{\rm\O}}, Y3ØY_{3}\neq\mbox{{\rm\O}}, and RØR\neq\mbox{{\rm\O}}. By (5), we have that Y5Y6Y7=ØY_{5}\cup Y_{6}\cup Y_{7}=\mbox{{\rm\O}}, and thus Y=Y1Y2Y3={y1,y2,y3}Y=Y_{1}\cup Y_{2}\cup Y_{3}=\{y_{1},y_{2},y_{3}\}. Let QQ be a component of G[R]G[R]. Since GG is connected, there must exist a path PP between QQ and YY. If we can show that PP must contain y2y_{2}, then y2y_{2} is a cutvertex of GG, which leads to a contradiction. Suppose that V(P)Y={y}V(P)\cap Y=\{y\}. To prove (12), it suffices to prove the following Claim 2.1.

Claim 2.1

yy3y\neq y_{3} and yy1.y\neq y_{1}.

Proof. By symmetry, we only need to prove yy3y\neq y_{3}. Suppose to its contrary that y=y3y=y_{3}. Since G[{v1,v2,v3,y3}]=P4G[\{v_{1},v_{2},v_{3},y_{3}\}]=P_{4} and GG is P7P_{7}-free, we have that d(u,y3)2d(u,y_{3})\leq 2 for each vertex uu of QQ. Let Ni={uV(Q)|d(u,y3)=i}N_{i}=\{u\in V(Q)|~{}d(u,y_{3})=i\} for ii\in{1, 2}. Since GG is diamond-free and y3y_{3} is complete to N1N_{1}, and since ω(G)=3\omega(G)=3, we have that each component of G[N1]G[N_{1}] is a K1K_{1} or a K2K_{2}. We show that

N2 is a stable set.\mbox{$N_{2}$ is a stable set}. (13)

Suppose to the contrary that N2N_{2} has two adjacent vertices xx^{\prime} and yy^{\prime}. Let SS^{\prime} be the component of G[N2]G[N_{2}] which contains xx^{\prime} and yy^{\prime}. Let x′′x^{\prime\prime} be a neighbor of xx^{\prime} in N1N_{1}. Since d(u,y3)2d(u,y_{3})\leq 2 for each vertex uu of QQ, we have that x′′x^{\prime\prime} is complete to V(S)V(S^{\prime}), which implies that ω(S)=2\omega(S^{\prime})=2 as ω(G)=3\omega(G)=3. Moreover, SS^{\prime} is P3P_{3}-free as GG is diamond-free. Then, S=xyS^{\prime}=x^{\prime}y^{\prime}.

Let S′′S^{\prime\prime} be the component of G[N1]G[N_{1}] which contains x′′x^{\prime\prime}. Note that S′′S^{\prime\prime} is a K1K_{1} or a K2K_{2}. For each vertex t1N1{x′′}t_{1}\in N_{1}\setminus\{x^{\prime\prime}\} and t2V(S)t_{2}\in V(S^{\prime}), to avoid a 4-hole on {x′′,y3,t1,t2}\{x^{\prime\prime},y_{3},t_{1},t_{2}\} if t1≁x′′t_{1}\not\sim x^{\prime\prime} or a diamond on {y3,x′′,t1,t2}\{y_{3},x^{\prime\prime},t_{1},t_{2}\} if t1x′′t_{1}\sim x^{\prime\prime}, {x,y}\{x^{\prime},y^{\prime}\} must be anticomplete to N1{x′′}N_{1}\setminus\{x^{\prime\prime}\}, which implies that xx^{\prime} and yy^{\prime} both have a neighbor in {y1,y2}\{y_{1},y_{2}\} as δ(G)3\delta(G)\geq 3.

Suppose xy1x^{\prime}\sim y_{1} and yy1y^{\prime}\sim y_{1}. Then x′′y1x^{\prime\prime}\sim y_{1} to forbid a diamond on {x,y,x′′,y1}\{x^{\prime},y^{\prime},x^{\prime\prime},y_{1}\}. If y2y_{2} has a neighbor in SS^{\prime}, say xx^{\prime} by symmetry, then y2yy_{2}\sim y^{\prime} to forbid an induced P7P_{7} on {x,y,y2,y3,v1,v2,v7}\{x^{\prime},y^{\prime},y_{2},y_{3},v_{1},v_{2},v_{7}\}, and thus G[{x,y,y1,y2}]G[\{x^{\prime},y^{\prime},y_{1},y_{2}\}] is a diamond, a contradiction. So, y2y_{2} has no neighbors in SS^{\prime}, and thus {x′′,y1}\{x^{\prime\prime},y_{1}\} is a clique cutset, which leads to a contradiction. This proves that x≁y1x^{\prime}\not\sim y_{1} or y≁y1y^{\prime}\not\sim y_{1}.

If xy2x^{\prime}\sim y_{2} and yy2y^{\prime}\sim y_{2}, then x′′≁y2x^{\prime\prime}\not\sim y_{2} to forbid a 4-hole on {x′′,y2,y3,v6}\{x^{\prime\prime},y_{2},y_{3},v_{6}\}, which forces G[{x,y,x′′,y2}]G[\{x^{\prime},y^{\prime},x^{\prime\prime},y_{2}\}] to be a diamond, a contradiction. Therefore, x≁y2x^{\prime}\not\sim y_{2} or y≁y2y^{\prime}\not\sim y_{2}.

By symmetry, we may assume that xy1x^{\prime}\sim y_{1}, x≁y2x^{\prime}\not\sim y_{2}, yy2y^{\prime}\sim y_{2}, and y≁y1y^{\prime}\not\sim y_{1}. Then, GG has an induced P7P_{7} on {x,y,y1,y3,v1,v2,v3}\{x^{\prime},y^{\prime},y_{1},y_{3},v_{1},v_{2},v_{3}\}, a contradiction. This proves (13).

Suppose that N2ØN_{2}\not=\mbox{{\rm\O}}. Let wN2w\in N_{2} and ww^{\prime} be a neighbor of ww in N1N_{1}. To avoid a diamond on {w,w,w′′,y3}\{w,w^{\prime},w^{\prime\prime},y_{3}\} if w′′ww^{\prime\prime}\sim w^{\prime} or a 4-hole on {w,w,w′′,y3}\{w,w^{\prime},w^{\prime\prime},y_{3}\} if w′′≁ww^{\prime\prime}\not\sim w^{\prime} for each vertex w′′N1{w}w^{\prime\prime}\in N_{1}\setminus\{w^{\prime}\}, ww must be anticomplete to V(Q){w,w}V(Q)\setminus\{w^{\prime},w\}, and so ww is complete to {y1,y2}\{y_{1},y_{2}\} because δ(G)3\delta(G)\geq 3. But then, wy1v5y2wwy_{1}v_{5}y_{2}w is a 4-hole, which is a contradiction. This shows that N2=ØN_{2}=\mbox{{\rm\O}}, and thus Q=N1Q=N_{1}.

Suppose that N1ØN_{1}\neq\mbox{{\rm\O}}. Let MM be a component of G[N1]G[N_{1}], and let aV(M)a\in V(M). To avoid a 4-hole ay3y6y2aay_{3}y_{6}y_{2}a, we have that a≁y2a\not\sim y_{2}. Since δ(G)3\delta(G)\geq 3, we have that MM is a K2K_{2} of which both vertices are adjacent to y1y_{1}, which forces a diamond G[V(M){y1,y3}]G[V(M)\cup\{y_{1},y_{3}\}] and leads to a contradiction. Therefore, N1=ØN_{1}=\mbox{{\rm\O}}, and so V(Q)=ØV(Q)=\mbox{{\rm\O}}. This completes the proof of Claim 2.1, and proves (12).  

If Y2ØY_{2}\neq\mbox{{\rm\O}} and Y3ØY_{3}\neq\mbox{{\rm\O}}, then GG is isomorphic to FF by (11) and (12).

If Y2=ØY_{2}=\mbox{{\rm\O}} and Y3ØY_{3}\not=\mbox{{\rm\O}}, then Y6Y7=ØY_{6}\cup Y_{7}=\mbox{{\rm\O}} by (5), which implies that Y=Y1Y3Y=Y_{1}\cup Y_{3} and d(v2)=2d(v_{2})=2, a contradiction.

So, we suppose that Y3=ØY_{3}=\mbox{{\rm\O}} and Y2ØY_{2}\neq\mbox{{\rm\O}}. Then, Y5=Y6=ØY_{5}=Y_{6}=\mbox{{\rm\O}} by (5). Since Y7=ØY_{7}=\mbox{{\rm\O}} implies that d(v7)=2d(v_{7})=2, we have that Y7ØY_{7}\not=\mbox{{\rm\O}}, and so Y=Y1Y2Y7={y1,y2,y7}Y=Y_{1}\cup Y_{2}\cup Y_{7}=\{y_{1},y_{2},y_{7}\}. By setting i0=7i_{0}=7 and replacing the tuple (Y1,Y2,Y3)(Y_{1},Y_{2},Y_{3}) with (Y7,Y1,Y2)(Y_{7},Y_{1},Y_{2}), we can deduce, with the same argument as that used in proving (12), that R=ØR=\mbox{{\rm\O}} and thus GG is isomorphic to FF. This totally completes the proof of Theorem 1.1.  

The following three classes of graphs show that the requirements P7P_{7}-free, C4C_{4}-free, and diamond-free are all necessary.

Let G1G_{1} be a tt-size clique blowup of a 7-hole with t2t\geq 2. It is certain that G1G_{1} is (P7,C4(P_{7},C_{4})-free. Moreover, we can easily verify that G1G_{1} does not satisfy the conclusion of Theorem 1.1.

Let G2G_{2} be a graph whose vertex-set is partitioned into seven stable sets S1S_{1}, S2S_{2}, \cdots, S7S_{7} such that for each i{1,2,,7}i\in\{1,2,\ldots,7\}, |Si|2|S_{i}|\geq 2, and SiS_{i} is complete to Si+1Si1S_{i+1}\cup S_{i-1} and anticomplete to Si+2Si2Si+3Si3S_{i+2}\cup S_{i-2}\cup S_{i+3}\cup S_{i-3}. It is certain that G2G_{2} is (P7P_{7}, diamond)-free and does not satisfy the conclusion of Theorem 1.1.

Let A1=x1x2x3x4x5x6x7x1A_{1}=x_{1}x_{2}x_{3}x_{4}x_{5}x_{6}x_{7}x_{1} be a 7-hole. We use G3G_{3} to denote the graph obtained from A1A_{1} by adding a tree with vertex set {a,b,c,d,g1,g2}\{a,b,c,d,g_{1},g_{2}\} and edge set {ag2,bg1,cg2,dg1,g1g2}\{ag_{2},bg_{1},cg_{2},dg_{1},g_{1}g_{2}\} such that N(a)V(A1)={x2,x6}N(a)\cap V(A_{1})=\{x_{2},x_{6}\}, N(b)V(A1)={x3,x7}N(b)\cap V(A_{1})=\{x_{3},x_{7}\}, N(c)V(A1)={x1,x4}N(c)\cap V(A_{1})=\{x_{1},x_{4}\}, N(d)V(A1)={x1,x5}N(d)\cap V(A_{1})=\{x_{1},x_{5}\} and N({g1,g2})V(A1)=ØN(\{g_{1},g_{2}\})\cap V(A_{1})=\mbox{{\rm\O}}. See Figure 2. It is certain that G3G_{3} is (C4C_{4}, diamond)-free, and does not satisfy the conclusion of Thoerem 1.1.

Refer to caption
Figure 2: Illustration of G3G_{3}.

Now, we can prove Corollary 1.1 by induction on |V(G)||V(G)|. Let GG be a (P7,C4(P_{7},C_{4}, diamond)-free graph. We may assume that GG is connected and has no clique cutsets. By Theorem 1.1, δ(G)max{2,ω(G)1}\delta(G)\leq\max\{2,\omega(G)-1\}, or GG is the Petersen graph or the graph FF. It follows immediately that χ(G)max{3,ω(G)}\chi(G)\leq\max\{3,\omega(G)\}.

3 The class of (P7,C4,(P_{7},C_{4}, kite)-free graphs.

In this section, we consider (P7,C4(P_{7},C_{4}, kite)-free graphs, and prove Theorem 1.2 and Corollary 1.2. We will need the following Lemma 3.1 from [7].

Lemma 3.1

[7] Let GG be a connected (C4C_{4}, kite)-free graph. If GG has no clique cutsets, then G=K+GG=K_{\ell}+G^{\prime} for some diamond-free graph GG^{\prime}, where \ell is a nonnegative integer.

Proof of Theorem 1.2: Let GG be a connected (P7P_{7}, C4C_{4}, kite)-free graph such that GG has no clique cutsets and δ(G)ω(G)+1\delta(G)\geq\omega(G)+1.

By Lemma 3.1, there exist an integer 0\ell\geq 0 and a diamond-free graph GG^{\prime} such that G=K+GG=K_{\ell}+G^{\prime}. We have that GG^{\prime} has no clique cutsets; Otherwise if GG^{\prime} has a clique cutset QQ, then QV(K)Q\cup V(K_{\ell}) is a clique cutset of GG, a contradiction. Therefore, GG^{\prime} is a (P7,C4(P_{7},C_{4}, diamond)-free graph without clique cutsets. By Theorem 1.1, we have that GG^{\prime} is isomorphic to FF or the Petersen graph, or δ(G)max{2,ω(G)1}\delta(G^{\prime})\leq\max\{2,\omega(G^{\prime})-1\}.

If GG^{\prime} is isomorphic to FF or the Petersen graph, then we are done. So, we suppose that GG^{\prime} is isomorphic to neither FF nor the Petersen graph. Thus, δ(G)max{2,ω(G)1}\delta(G^{\prime})\leq\max\{2,\omega(G^{\prime})-1\}. If GG^{\prime} is a single vertex, then G=K+1G=K_{\ell+1} and δ(G)==ω(G)1\delta(G)=\ell=\omega(G)-1. If ω(G)=2\omega(G^{\prime})=2, then δ(G)2\delta(G^{\prime})\leq 2, and so δ(G)δ(G)++2=ω(G)\delta(G)\leq\delta(G^{\prime})+\ell\leq\ell+2=\omega(G). If ω(G)3\omega(G^{\prime})\geq 3, then δ(G)ω(G)1\delta(G^{\prime})\leq\omega(G^{\prime})-1, and so δ(G)δ(G)+ω(G)+1ω(G)1\delta(G)\leq\delta(G^{\prime})+\ell\leq\omega(G^{\prime})+\ell-1\leq\omega(G)-1. All contradict our assumption that δ(G)ω(G)+1\delta(G)\geq\omega(G)+1. This proves Theorem 1.2.  

Notice that the graph G1G_{1} defined in Section 2 is (P7,C4(P_{7},C_{4})-free, and the graph G2G_{2} defined in Section 2 is (P7P_{7}, kite)-free. Moreover, one can easily check that both G1G_{1} and G2G_{2} do not satisfy the conclusion of Theorem 1.2.

Let A2A_{2} be the graph obtained from a 8-cycle y1y2y3y4y5y6y7y8y1y_{1}y_{2}y_{3}y_{4}y_{5}y_{6}y_{7}y_{8}y_{1} by adding a 6-cycle u1u2u3u4u5u6u1u_{1}u_{2}u_{3}u_{4}u_{5}u_{6}u_{1} such that the edge set between them is {y1u1,y4u2,y5u4,y6u6}\{y_{1}u_{1},y_{4}u_{2},y_{5}u_{4},y_{6}u_{6}\}. We use G4G_{4} to denote the graph obtained from A2A_{2} by adding a stable set {t1,t2}\{t_{1},t_{2}\} such that N(t1)V(A2)={y2,y7,u5}N(t_{1})\cap V(A_{2})=\{y_{2},y_{7},u_{5}\} and N(t2)V(A2)={y3,y8,u3}N(t_{2})\cap V(A_{2})=\{y_{3},y_{8},u_{3}\} (see Figure 3). Obviously, G4G_{4} is (C4C_{4}, kite)-free and it does not satisfy the conclusion of Theorem 1.2.

These graphs show that the requirements P7P_{7}-free, C4C_{4}-free, and kite-free in Theorem 1.2 are all necessary.

Refer to caption
Figure 3: Illustration of G4G_{4}.

Now, we can prove Corollary 1.2 by induction on |V(G)||V(G)|. Let GG be a (P7,C4(P_{7},C_{4}, kite)-free graph. We may assume that GG is connected and has no clique cutsets. By Theorem 1.2, we have that δ(G)ω(G)\delta(G)\leq\omega(G) or there is an integer 0\ell\geq 0 such that G=K+HG=K_{\ell}+H, where HH is the Petersen graph or FF. It follows immediately that χ(G)ω(G)+1\chi(G)\leq\omega(G)+1.

4 The class of (P7,C4,(P_{7},C_{4}, gem)-free graphs.

In this section, we turn our attention to (P7,C4(P_{7},C_{4}, gem)-free graphs, and prove Theorem 1.3 and Corollary 1.3. Theorem 1.3 generalizes the following Lemma 4.1 from [4]. In our proof, we will also use a conclusion from [13]. Recall that a bisimplicial vertex is a vertex whose neighborhoods is the union of two cliques.

Lemma 4.1

[4] Let GG be a connected (P7,C7,C4(P_{7},C_{7},C_{4}, gem)-free graph. If GG has no clique cutsets, then GG has a bisimplicial vertex, or GG is the clique blowup of the Petersen graph.

Lemma 4.2

[13] If GG is any clique blowup of the Petersen graph, then χ(G)54ω(G)\chi(G)\leq\lceil\frac{5}{4}\omega(G)\rceil.

Proof of Theorem 1.3: Let GG be a connected (P7,C4(P_{7},C_{4}, gem)-free graph. If GG is C7C_{7}-free, then the statement follows directly from Lemma 4.1. So, we suppose that GG has 7-holes, and let v1v2v3v4v5v6v7v1v_{1}v_{2}v_{3}v_{4}v_{5}v_{6}v_{7}v_{1} be a 7-hole of GG. Let A={v1,,v7}A=\{v_{1},\cdots,v_{7}\} and let R=M(A)R=M(A). During the proof of Theorem 1.3, every subscript is understood to be modulo 7. Since GG is gem-free, it is certain that

no vertex of N(A)N(A) may have four consecutive neighbors in AA. (14)

Let ii be an arbitrary integer in {1,,7}\{1,\cdots,7\}, let

Xi\displaystyle X_{i} =\displaystyle= {xN(A)|N(x)A={vi,vi+3}},\displaystyle\{x\in N(A)|~{}N(x)\cap A=\{v_{i},v_{i+3}\}\},
Yi\displaystyle Y_{i} =\displaystyle= {xN(A)|N(x)A={vi,vi+1,vi+2}},\displaystyle\{x\in N(A)|~{}N(x)\cap A=\{v_{i},v_{i+1},v_{i+2}\}\},
Zi\displaystyle Z_{i} =\displaystyle= {xN(A)|N(x)A={vi,vi+3,vi+4}},\displaystyle\{x\in N(A)|~{}N(x)\cap A=\{v_{i},v_{i+3},v_{i+4}\}\},

and let X=i=17XiX=\cup_{i=1}^{7}X_{i}, Y=i=17YiY=\cup_{i=1}^{7}Y_{i} and Z=i=17ZiZ=\cup_{i=1}^{7}Z_{i}. We will firstly prove some useful properties about XiX_{i}, YiY_{i} and ZiZ_{i}.

(M1) N(A)=XYZN(A)=X\cup Y\cup Z, and V(G)=AXYZRV(G)=A\cup X\cup Y\cup Z\cup R.

It suffices to verify that N(A)XYZN(A)\subseteq X\cup Y\cup Z. Let xN(A)x\in N(A). Without loss of generality, we suppose that xv1x\sim v_{1}. If N(x)A={v1}N(x)\cap A=\{v_{1}\}, then G[{x,v1,v2,v3,v4,v5,v6}]=P7G[\{x,v_{1},v_{2},v_{3},v_{4},v_{5},v_{6}\}]=P_{7}, a contradiction. Therefore, |N(x)A|2|N(x)\cap A|\geq 2.

Suppose xv2x\sim v_{2}. To avoid an induced P7=xv2v3v4v5v6v7P_{7}=xv_{2}v_{3}v_{4}v_{5}v_{6}v_{7}, xx must have a neighbor in {v3,,v7}\{v_{3},\cdots,v_{7}\}. If xv3x\sim v_{3}, then x≁v4x\not\sim v_{4} and x≁v7x\not\sim v_{7} by (14), and thus x≁v5x\not\sim v_{5} and x≁v6x\not\sim v_{6} to forbid a 4-hole xv3v4v5xxv_{3}v_{4}v_{5}x or xv1v7v6xxv_{1}v_{7}v_{6}x, which leads to xYx\in Y as N(x)A={v1,v2,v3}N(x)\cap A=\{v_{1},v_{2},v_{3}\}. By symmetry, we have that xYx\in Y if xv7x\sim v_{7}. So, we may assume that x≁v3x\not\sim v_{3} and x≁v7x\not\sim v_{7}. Consequently, x≁v4x\not\sim v_{4} and x≁v6x\not\sim v_{6} to forbid a 4-hole xv2v3v4xxv_{2}v_{3}v_{4}x or xv1v7v6xxv_{1}v_{7}v_{6}x. Now, xv5x\sim v_{5} and hence xZx\in Z as N(x)A={v1,v2,v5}N(x)\cap A=\{v_{1},v_{2},v_{5}\}.

Similarly, xYx\in Y or ZZ if xv7x\sim v_{7}.

So, we suppose that x≁v2x\not\sim v_{2} and x≁v7x\not\sim v_{7}. Then, x≁v3x\not\sim v_{3} and x≁v6x\not\sim v_{6} to forbid a 4-hole on xv1v2v3xxv_{1}v_{2}v_{3}x or xv1v7v6xxv_{1}v_{7}v_{6}x. Since |N(x)A|2|N(x)\cap A|\geq 2, we have that xv4x\sim v_{4} or xv5x\sim v_{5}. If xv4x\sim v_{4} and xv5x\sim v_{5}, then xZx\in Z. Otherwise, xXx\in X. Therefore, (M1) holds.

(M2) Both XiZiX_{i}\cup Z_{i} and YiY_{i} are cliques.

Let xx, xXiZix^{\prime}\in X_{i}\cup Z_{i}. If x≁xx\not\sim x^{\prime}, then xvixvi+3xxv_{i}x^{\prime}v_{i+3}x is a 4-hole. Let yy, yYiy^{\prime}\in Y_{i}. If y≁yy\not\sim y^{\prime}, then yviyvi+2yyv_{i}y^{\prime}v_{i+2}y is a 4-hole. This proves (M2).

(M3) YiY_{i} is complete to Yi+1Yi+6Y_{i+1}\cup Y_{i+6}.

Suppose to its contrary, and let xYix\in Y_{i} and yYi+1Yi+6y\in Y_{i+1}\cup Y_{i+6}. If x≁yx\not\sim y, then G[{x,y,vi,vi+1,vi+2}]G[\{x,y,v_{i},v_{i+1},v_{i+2}\}] is a gem, which leads to a contradiction. Therefore, (M3) holds.

(M4) YiY_{i} is anticomplete to Yi+2Yi+3Yi+4Yi+5Y_{i+2}\cup Y_{i+3}\cup Y_{i+4}\cup Y_{i+5}.

Suppose that (M4) does not hold. By symmetry, let xY1x\in Y_{1} and yY3Y4Y5Y6y\in Y_{3}\cup Y_{4}\cup Y_{5}\cup Y_{6} such that xyx\sim y. If yY3y\in Y_{3}, then G[{x,y,v2,v3,v4}]G[\{x,y,v_{2},v_{3},v_{4}\}] is a gem. If yY6y\in Y_{6}, then G[{x,y,v1,v2,v7}]G[\{x,y,v_{1},v_{2},v_{7}\}] is a gem. If yY4y\in Y_{4}, then xv3v4yxxv_{3}v_{4}yx is a 4-hole. If yY5y\in Y_{5}, then xv1v7yxxv_{1}v_{7}yx is a 4-hole. All are contradictions. This proves (M4).

(M5) Either Xi=ØX_{i}=\mbox{{\rm\O}} or Xi+2Xi+5=ØX_{i+2}\cup X_{i+5}=\mbox{{\rm\O}}.

If it is not the case, we may choose, without loss of generality, that uX1u\in X_{1} and vX3v\in X_{3} then uvv3v4uuvv_{3}v_{4}u is a 4-hole if uvu\sim v, and G[{v,u,v1,v2,v4,v5,v6}]=P7G[\{v,u,v_{1},v_{2},v_{4},v_{5},v_{6}\}]=P_{7} otherwise, which leads to a contradiction. This proves (M5).

(M6) XiX_{i} is anticomplete to Xi+1Xi+3Xi+4Xi+6X_{i+1}\cup X_{i+3}\cup X_{i+4}\cup X_{i+6}.

Suppose that (M6) does not hold. By symmetry, let xX1x\in X_{1} and yX2X4X5X7y\in X_{2}\cup X_{4}\cup X_{5}\cup X_{7} such that xyx\sim y. If yX2y\in X_{2}, then xyv2v1xxyv_{2}v_{1}x is a 4-hole. If yX4y\in X_{4}, then xyv7v1xxyv_{7}v_{1}x is a 4-hole. If yX5y\in X_{5}, then xyv5v4xxyv_{5}v_{4}x is a 4-hole. If yX7y\in X_{7}, then xyv3v4xxyv_{3}v_{4}x is a 4-hole. All are contradictions. This proves (M6).

(M7) XiX_{i} is complete to Yi+2Yi+6Y_{i+2}\cup Y_{i+6}.

If (M7) does not hold, we may choose by symmetry that xXix\in X_{i} and yYi+2Yi+6y\in Y_{i+2}\cup Y_{i+6} such that x≁yx\not\sim y, then G[{x,y,vi,vi+1,vi+2,vi+4,vi+5}]=P7G[\{x,y,v_{i},v_{i+1},v_{i+2},v_{i+4},v_{i+5}\}]=P_{7} whenever yYi+2y\in Y_{i+2}, and G[{x,y,vi+1,vi+3,vi+4,vi+5,vi+6}]=P7G[\{x,y,v_{i+1},v_{i+3},v_{i+4},v_{i+5},v_{i+6}\}]=P_{7} whenever yYi+6y\in Y_{i+6}, which leads to a contradiction. Therefore, (M7) hlds.

(M8) XiX_{i} is anticomplete to YiYi+1Yi+3Yi+4Yi+5Y_{i}\cup Y_{i+1}\cup Y_{i+3}\cup Y_{i+4}\cup Y_{i+5}.

Suppose that (M8) does not hold. Without loss of generality, let xX1x\in X_{1} and yY1Y2Y4Y5Y6y\in Y_{1}\cup Y_{2}\cup Y_{4}\cup Y_{5}\cup Y_{6} such that xyx\sim y. If yY1y\in Y_{1}, then G[{x,y,v1,v2,v3}]G[\{x,y,v_{1},v_{2},v_{3}\}] is a gem. If yY2y\in Y_{2}, then xyv2v1xxyv_{2}v_{1}x is a 4-hole. If yY4y\in Y_{4}, then G[{x,y,v4,v5,v6}]G[\{x,y,v_{4},v_{5},v_{6}\}] is a gem. If yY5y\in Y_{5}, then xyv5v4xxyv_{5}v_{4}x is a 4-hole. If yY6y\in Y_{6}, then G[{x,y,v1,v6,v7}]G[\{x,y,v_{1},v_{6},v_{7}\}] is a gem. All are contradictions. This proves (M8).

(M9) Either Zi=ØZ_{i}=\mbox{{\rm\O}} or Zi+3Zi+4=ØZ_{i+3}\cup Z_{i+4}=\mbox{{\rm\O}}.

If this is not true, we may choose by symmetry that z1Z1z_{1}\in Z_{1} and z4Z4z_{4}\in Z_{4}, then z1v4z4v1z1z_{1}v_{4}z_{4}v_{1}z_{1} is a 4-hole whenever z1≁z4z_{1}\not\sim z_{4}, and G[{z1,z4,v1,v4,v5}]G[\{z_{1},z_{4},v_{1},v_{4},v_{5}\}] is a gem whenever z1z4z_{1}\sim z_{4}. This proves (M9).

(M10) ZiZ_{i} is anticomplete to Zi+1Zi+2Zi+5Zi+6Z_{i+1}\cup Z_{i+2}\cup Z_{i+5}\cup Z_{i+6}.

Suppose that (M10) does not hold. By symmetry, let xZ1x\in Z_{1} and yZ2Z3Z6Z7y\in Z_{2}\cup Z_{3}\cup Z_{6}\cup Z_{7} such that xyx\sim y. If yZ2Z6y\in Z_{2}\cup Z_{6}, then xyv2v1xxyv_{2}v_{1}x is a 4-hole. If yZ3Z7y\in Z_{3}\cup Z_{7}, then xyv7v1xxyv_{7}v_{1}x is a 4-hole. Therefore, (M10) is true.

(M11) Either Zi=ØZ_{i}=\mbox{{\rm\O}} or XiXi+4Xi+6=ØX_{i}\cup X_{i+4}\cup X_{i+6}=\mbox{{\rm\O}}.

Suppose that (M11) does not hold. Without loss of generality, we choose uZ1u\in Z_{1} and vX1X5X7v\in X_{1}\cup X_{5}\cup X_{7}. If u≁vu\not\sim v, then uv1vv4uuv_{1}vv_{4}u is a 4-hole whenever vX1v\in X_{1}, uv1vv5uuv_{1}vv_{5}u is a 4-hole whenever vX5v\in X_{5}, and G[{u,v,v1,v2,v3,v5,v6}]=P7G[\{u,v,v_{1},v_{2},v_{3},v_{5},v_{6}\}]=P_{7} whenever vX7v\in X_{7}. Therefore, uvu\sim v, which leads to either a gem G[{u,v,v1,v4,v5}]G[\{u,v,v_{1},v_{4},v_{5}\}] if vX1X5v\in X_{1}\cup X_{5}, or a 4-hole uvv3v4uuvv_{3}v_{4}u if vX7v\in X_{7}. This proves (M11).

(M12) ZiZ_{i} is anticomplete to Xi+1Xi+2Xi+3Xi+5X_{i+1}\cup X_{i+2}\cup X_{i+3}\cup X_{i+5}.

Suppose that (M12) does not hold. By symmetry, choose xZ1x\in Z_{1} and yX2X3X4X6y\in X_{2}\cup X_{3}\cup X_{4}\cup X_{6} such that xyx\sim y. If yX2y\in X_{2}, then xyv2v1xxyv_{2}v_{1}x is a 4-hole. If yX3y\in X_{3}, then xyv3v4xxyv_{3}v_{4}x is a 4-hole. If yX4y\in X_{4}, then xyv7v1xxyv_{7}v_{1}x is a 4-hole. If yX6y\in X_{6}, then xyv6v5xxyv_{6}v_{5}x is a 4-hole. All are contradictions. This proves (M12).

(M13) ZiZ_{i} is complete to Yi+2Yi+3Yi+6Y_{i+2}\cup Y_{i+3}\cup Y_{i+6}.

Without loss of generality, let zZ1z\in Z_{1} and yY3Y4Y7y\in Y_{3}\cup Y_{4}\cup Y_{7}. If z≁yz\not\sim y, then G[{z,y,v3,v4,v5}]G[\{z,y,v_{3},v_{4},v_{5}\}] is a gem whenever yY3y\in Y_{3}, G[{z,y,v4,v5,v6}]G[\{z,y,v_{4},v_{5},v_{6}\}] is a gem whenever yY4y\in Y_{4}, and G[{z,y,v2,v3,v5,v6,v7}]=P7G[\{z,y,v_{2},v_{3},v_{5},v_{6},v_{7}\}]=P_{7} whenever yY7y\in Y_{7}. Therefore, zyz\sim y, and thus (M13) holds.

(M14) ZiZ_{i} is anticomplete to YiYi+1Yi+4Yi+5Y_{i}\cup Y_{i+1}\cup Y_{i+4}\cup Y_{i+5}.

By symmetry, choose zZ1z\in Z_{1} and yY1Y2Y5Y6y\in Y_{1}\cup Y_{2}\cup Y_{5}\cup Y_{6}. If zyz\sim y, then G[{z,y,v1,v2,v3}]G[\{z,y,v_{1},v_{2},v_{3}\}] is a gem whenever yY1y\in Y_{1}, zyv2v1zzyv_{2}v_{1}z is a 4-hole whenever yY2y\in Y_{2}, G[{z,y,v4,v5,v6}]G[\{z,y,v_{4},v_{5},v_{6}\}] is a gem whenever yY5y\in Y_{5}, and G[{z,y,v1,v6,v7}]G[\{z,y,v_{1},v_{6},v_{7}\}] is a gem whenever yY6y\in Y_{6}. Therefore, z≁yz\not\sim y, and thus (M14) holds.

Let Yi=Yi1{vi}Y^{\prime}_{i}=Y_{i-1}\cup\{v_{i}\} for each i{1,,7}i\in\{1,\cdots,7\} and Y=Y1Y7Y^{\prime}=Y^{\prime}_{1}\cup\cdots\cup Y^{\prime}_{7}. By (M2), (M3) and (M4), we have that G[Y]G[Y] is a clique blowup of C7C_{7} and hence G[Y]G[Y^{\prime}] is a nonempty clique blowup of C7C_{7}. Follow from (M1) to (M14), G[N(A)A]G[N(A)\cup A] is a clique blowup of some graph like the one as shown in Figure 4.

Refer to caption
Figure 4: One situation of G[N(A)A]G[N(A)\cup A].

Now, we can prove Theorem 1.3, by considering whether XiX_{i} is empty or not, to find a bisimplicial vertex of GG in AA.

By (M5), we have that {X1,,X7}\{X_{1},\cdots,X_{7}\} has at most three nonempty elements. So, there must exist an integer i{1,,7}i\in\{1,\cdots,7\} such that viv_{i} is anticomplete to XX.

Suppose that Z=ØZ=\mbox{{\rm\O}}. Then, N(vi)=(Yi2{vi1}Yi1)(Yi{vi+1})N(v_{i})=(Y_{i-2}\cup\{v_{i-1}\}\cup Y_{i-1})\cup(Y_{i}\cup\{v_{i+1}\}) by (M1). Notice that vi1v_{i-1} is complete to Yi2Yi1Y_{i-2}\cup Y_{i-1}, and vi+1v_{i+1} is complete to YiY_{i}, and Yi2Y_{i-2} is complete to Yi1Y_{i-1} by (M3). We have that Yi2{vi1}Yi1Y_{i-2}\cup\{v_{i-1}\}\cup Y_{i-1} and Yi{vi+1}Y_{i}\cup\{v_{i+1}\} are two nonempty cliques by (M2), and thus viv_{i} is a bisimplicial vertex of GG.

Therefore, we suppose that ZØZ\neq\mbox{{\rm\O}}. Without loss of generality, suppose Z1ØZ_{1}\neq\mbox{{\rm\O}}. By (M9) and (M11), Z4Z5X1X5X7=ØZ_{4}\cup Z_{5}\cup X_{1}\cup X_{5}\cup X_{7}=\mbox{{\rm\O}}, and so

X=X2X3X4X6, and Z=Z1Z2Z3Z6Z7.X=X_{2}\cup X_{3}\cup X_{4}\cup X_{6},\mbox{ and }Z=Z_{1}\cup Z_{2}\cup Z_{3}\cup Z_{6}\cup Z_{7}.

Firstly, we prove that

if X2=ØX_{2}=\mbox{{\rm\O}}, then v5v_{5} is a bisimplicial vertex of GG. (15)

Suppose that X2=ØX_{2}=\mbox{{\rm\O}}. By (M1), we can deduce that N(v5)=(Y4Y3Z1{v4})(Y5Z2{v6})N(v_{5})=(Y_{4}\cup Y_{3}\cup Z_{1}\cup\{v_{4}\})\cup(Y_{5}\cup Z_{2}\cup\{v_{6}\}). By (M3) and (M13), Y3Y_{3} is complete to Y4Y_{4}, Z1Z_{1} is complete to Y3Y4Y_{3}\cup Y_{4}, and Y5Y_{5} is complete to Z2Z_{2}. By the definition of ZiZ_{i} and YiY_{i}, v4v_{4} is complete to Y4Y3Z1Y_{4}\cup Y_{3}\cup Z_{1}, and v6v_{6} is complete to Y5Z2Y_{5}\cup Z_{2}. It follows from (M2) that Y4Y3Z1{v4}Y_{4}\cup Y_{3}\cup Z_{1}\cup\{v_{4}\} and Y5Z2{v6}Y_{5}\cup Z_{2}\cup\{v_{6}\} are two nonempty cliques. Therefore, v5v_{5} is a bisimplicial vertex of GG. This proves (15).

Next, we prove that

if X4=ØX_{4}=\mbox{{\rm\O}}, then v4v_{4} is a bisimplicial vertex of GG. (16)

Suppose that X4=ØX_{4}=\mbox{{\rm\O}}. Then, N(v4)=(Y3Y2Z7{v3})(Y4Z1{v5})N(v_{4})=(Y_{3}\cup Y_{2}\cup Z_{7}\cup\{v_{3}\})\cup(Y_{4}\cup Z_{1}\cup\{v_{5}\}) by (M1). By (M3) and (M13), Y2Y_{2} is complete to Y3Y_{3}, Z7Z_{7} is complete to Y2Y3Y_{2}\cup Y_{3}, and Z1Z_{1} is complete to Y4Y_{4}. By (M2) and by the definition of ZiZ_{i} and YiY_{i}, we have that Z1Z_{1}, Z7Z_{7}, Y2Y_{2}, Y3Y_{3} and Y4Y_{4} are all cliques, and v3v_{3} is complete to Y3Y2Z7Y_{3}\cup Y_{2}\cup Z_{7}, and v5v_{5} is complete to Y4Z1Y_{4}\cup Z_{1}. Therefore, Y3Y2Z7{v3}Y_{3}\cup Y_{2}\cup Z_{7}\cup\{v_{3}\} and Y4Z1{v5}Y_{4}\cup Z_{1}\cup\{v_{5}\} are two cliques, and so v4v_{4} is a bisimplicial vertex of GG. This proves (16).

By (M5), we have that one of X2X_{2} and X4X_{4} is empty. It follows immediately from (15) and (16) that GG must have a bisimplicial vertex. This completes the proof of Theorem 1.3.  

It is certain that G2G_{2} defined in Section 2 is (P7P_{7}, gem)-free and contains a 7-hole, and it is easy to check that G2G_{2} does not satisfy the conclusion of Theorem 1.3.

Let A3=z1z2z3z4z5z6z7z1A_{3}=z_{1}z_{2}z_{3}z_{4}z_{5}z_{6}z_{7}z_{1} be a 7-hole. We use G5G_{5} to denote the graph obtained from A3A_{3} by adding seven vertices a1a_{1}, a2a_{2}, \cdots, a7a_{7} such that N(ai)(V(A3){a1,a2,,a7})={zi,zi+3,zi+4,ai+3,ai+4}N(a_{i})\cap(V(A_{3})\cup\{a_{1},a_{2},\cdots,a_{7}\})=\{z_{i},z_{i+3},z_{i+4},a_{i+3},a_{i+4}\} for each i{1,,7}i\in\{1,\cdots,7\} (see Figure 5). Obviously, G5G_{5} is (P7,C4(P_{7},C_{4})-free and contains a 7-hole and thus the clique blowup of G5G_{5} is (P7,C4(P_{7},C_{4})-free and contains a 7-hole. Moreover, one can easily check that the clique blowup of G5G_{5} does not satisfy the conclusion of Theorem 1.3.

Refer to caption
Figure 5: Illustration of G5G_{5}.

We use G6G_{6} to denote the tt-size clique blowup of G3G_{3}, defined in Section 2, with t1t\geq 1. It is certain that G6G_{6} is (C4C_{4}, gem)-free and contains a 7-hole, and we can check that G6G_{6} does not satisfy the conclusion of Theorem 1.3.

These graphs show that the requirements P7P_{7}-free, C4C_{4}-free, and gem-free in Theorem 1.3 are all necessary.

Now, we can prove Corollary 1.3 by induction on |V(G)||V(G)|. Let GG be a (P7,C4(P_{7},C_{4}, gem)-free graph. We may assume that GG is connected and has no clique cutsets. By Theorem 1.3, we have that GG has a bisimplicial vertex or GG is the clique blowup of the Petersen graph. By Lemma 4.2, it follows immediately that χ(G)2ω(G)1\chi(G)\leq 2\omega(G)-1.


Remarks. Recall that we use PkP_{k} to denote a path on kk vertices, and use diamond to denote the graph obtained from a K4K_{4} by removing an edge. In 1998, Randerath [16] proved that χ(G)ω(G)+1\chi(G)\leq\omega(G)+1 for every (P5P_{5}, diamond)-free graph. Cameron, Huang and Merkel [2] proved that χ(G)ω(G)+3\chi(G)\leq\omega(G)+3. for every (P6P_{6}, diamond)-free graph. Schiermeyer [20] asked a question: Is it true that there exists a constant cc such that χ(G)ω(G)+c\chi(G)\leq\omega(G)+c for every (P7(P_{7}, diamond)-free graph?? This is still open.


Acknowledgement: We thank T. Karthick for informing us reference [20], and thank I. Schiermeyer for helpful suggestions.

References

  • [1] J. A. Bondy, U. S. R. Murty, Graph Theory, Springer, New York, 2008.
  • [2] K. Cameron, S. Huang, O. Merkel, An optimal χ\chi-bound for (P6P_{6}, diamond)-free graphs, J. of Graph Theory, 97 (2021) 451-465.
  • [3] K. Cameron, S. Huang, I. Penev, V. Sivaraman, The class of (P7,C4,C5)(P_{7},C_{4},C_{5})-free graphs: Decomposition, algorithms, and χ\chi-boundedness, J. of Graph Theory, 93 (2020) 503-552.
  • [4] S. A. Choudum, T. Karthick, M. M. Belavadi, Structural domination and coloring of some (P7,C7)(P_{7},C_{7})-free graphs, Disc. Math., 344 (2021) 112244.
  • [5] S. A. Choudum, T. Karthick, M. A. Shalu, Perfect coloring and linearly χ\chi-bound P6P_{6}-free graphs, J. of Graph Theory, 54 (2007) 293-306.
  • [6] M. Chudnovsky, V. Sivaraman, Perfect divisibility and 2-divisibility, J. of Graph Theory, 90 (2019) 54-60.
  • [7] D. J. Fraser, A. M. Hamel, C. T. Hoáng, On the structure of (even-hole, kite)-free graphs, Graphs and Combinatorics, 34 (2018) 989-999.
  • [8] S. Gaspers, S. Huang, Linearly χ\chi-Bounding (P6,C4)(P_{6},C_{4}) -Free Graphs, International Workshop on Graph-Theoretic Concepts in Computer Science, Springer, Cham, 2017.
  • [9] M. Geißer, Colourings of P5P_{5}-free graphs, PhD thesis, 2022.
  • [10] S. Gravier, C. T. Hoáng, F. Maffray, Coloring the hypergraph of maximal cliques of a graph with no long path, Disc. Math., 272 (2003) 285-290.
  • [11] A. Gyárfás, On Ramsey covering-numbers, Infinite and Finite Sets, 2 (1975) 801-816.
  • [12] S. Huang, The optimal χ\chi-bound for (P7,C4,C5)(P_{7},C_{4},C_{5})-free graphs, arXiv:2212.05239, 2022.
  • [13] T. Karthick, F. Maffray, Square-free graphs with no six-vertex induced path, SIAM J. on Disc. Math., 33 (2019) 874-909.
  • [14] K. Lan, Y. Zhou, F. Liu, The chromatic number of (P6,C4(P_{6},C_{4}, diamond)-free graphs, Bulletin of the Australian Mathematical Society, (2022) 1-10.
  • [15] S. Mishra, On graphs with no induced bull and no induced diamond, arXiv:2107.03750, 2021.
  • [16] B. Randerath, The Vizing bound for the chromatic number based on forbidden Pairs (Ph. D. Thesis), Shaker Verlag, Aachen, (1998).
  • [17] B. Randerath, I. Schiermeyer, Vertex colouring and forbidden subgraphs-A survey, Graphs and Combinatorics, 20 (2004) 1-40.
  • [18] I. Schiermeyer, Chromatic number of P5P_{5}-free graphs: Reed’s conjecture, Disc. Math., 339 (2016) 1940-1943.
  • [19] I. Schiermeyer, B. Randerath, Polynomial χ\chi-binding functions and forbidden induced subgraphs: A survey, Graphs and Combinatorics, 35 (2019) 1-31.
  • [20] I. Schiermeyer, Polynomial χ\chi-binding functions for P5P_{5}-free graphs, personal communication.
  • [21] A. Scott, P. Seymour, A survey of χ\chi-boundedness, J. of Graph Theory, 95 (2020) 473-504.
  • [22] D. Seinsche, On a property of the class of nn-colorable graphs, J. of Combinatorial Theory, Ser. B, 16 (1974) 191-193.