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

Infinite families of kk-vertex-critical (P5P_{5}, C5C_{5})-free graphs

Ben Cameron
   Chính T. Hoàng
Abstract

A graph is kk-vertex-critical if χ(G)=k\chi(G)=k but χ(Gv)<k\chi(G-v)<k for all vV(G)v\in V(G). We construct a new infinite families of kk-vertex-critical (P5,C5)(P_{5},C_{5})-free graphs for all k6k\geq 6. Our construction generalizes known constructions for 44-vertex-critical P7P_{7}-free graphs and 55-vertex-critical P5P_{5}-free graphs and is in contrast to the fact that there are only finitely many 55-vertex-critical (P5,C5)(P_{5},C_{5})-free graphs. In fact, our construction is actually even more well-structured, being (2P2,K3+P1,C5)(2P_{2},K_{3}+P_{1},C_{5})-free.

1 Introduction

For a given integer kk, the kk-Colouring problem is to determine whether a given graph is kk-colourable. This problem is known to be NP-complete for all k3k\geq 3 [23] which led to tremendous interest in restricting the structure of input graphs such that polynomial-time kk-Colouring algorithms can be developed. One of the most popular structural restrictions is to forbid an induced subgraph. Yet, for every graph HH that contains an induced cycle [21] or claw [24, 17], it remains NP-complete to solve kk-Colouring on HH-free graphs for all k3k\geq 3. Thus, assuming P\neqNP, if kk-Colouring can be solved in polynomial time for HH-free graphs for k3k\geq 3, HH must be a linear forest, that is, a disjoint union of paths.

To this end, it was shown that kk-Colouring can be solved in polynomial-time for P5P_{5}-free graphs for all kk [15]. Further, 44-Colouring P6P_{6}-free graphs [11, 12, 13] and 33-Colouring P7P_{7}-free graphs [2] can also be solved in polynomial-time. However, solving kk-Colouring of PtP_{t}-free graphs remains NP-complete if t7t\geq 7 and k4k\geq 4 or t=6t=6 and k5k\geq 5 [18]. The complexity remains an open question for PtP_{t}-free when t8t\geq 8.

The fact that P5P_{5} is the largest connected subgraph that can be forbidden such that kk-Colouring can be solved in polynomial-time for all kk (again assuming P\neqNP) has generated special interest in further properties of P5P_{5}-free graphs. One such property is determining which subfamilies of P5P_{5}-free graphs admit polynomial-time certifying kk-Colouring algorithms. An algorithm is certifying if it returns a simple and easily verifiable witness with each output. The kk-Colouring algorithms for P5P_{5}-free graphs developed in [15] return kk-colourings of the input graph in the case that the graph is kk-colourable, but there is no certificate when the graph is not kk-colourable. In general, a kk-Colouring algorithm can return a (k+1)(k+1)-vertex-critical induced subgraph to certify negative outputs. In fact, when the set of (k+1)(k+1)-vertex-critical graphs in a given family of graphs is finite, then there exists a polynomial-time certifying kk-Coloring algorithm for that family (see, for example, [5]).

For kk-vertex-critical P5P_{5}-free graphs, it is known that there are only finitely many for k=4k=4 [16] and this finite list was used to develop a linear-time 33-Colouring algorithm for P5P_{5}-free graphs [26]. For k5k\geq 5, however, there are infinitely many kk-vertex-critical P5P_{5}-free graphs [16]. Thus, the search for subfamilies of P5P_{5}-free graphs admitting polynomial-time certifying kk-Coloruing algorithms for k4k\geq 4 has turned to graphs that are P5P_{5}-free and HH-free for other graphs HH. We call these graphs (P5,H)(P_{5},H)-free. The most comprehensive result to date on kk-vertex-critical (P5,H)(P_{5},H)-free graphs is the dichotomy theorem from [8] that there are only finitely many for all kk if and only if HH is not 2P22P_{2} or K3+P1K_{3}+P_{1}. It is further known that there are only finitely many 55-vertex-critical (P5,H)(P_{5},H)-free graphs when HH is any of:

  • chair [20];

  • bull [19];

  • C5C_{5} [16].

For the above graphs, the cardinality of kk-vertex-critical graphs remains unknown for all k6k\geq 6.

On the infinite side, there are only two known constructions of HH-free graphs where HH is a linear forest:

  • GrG_{r} is 44-vertex-critical P7P_{7}-free for all r1r\geq 1 [9], and

  • GpG_{p} is 55-vertex-critical P5P_{5}-free for all p1p\geq 1 (in fact, (2P2,K3+P1)(2P_{2},K_{3}+P_{1})-free) [16].

These two families turn out to be special cases of the new infinite families of kk-vertex-critical (P5,C5)(P_{5},C_{5})-free graphs for each k6k\geq 6 that we will construct in this paper. Our families are the first defined by forbidden induced subgraphs where the boundary between only finitely many and infinitely many kk-vertex-critical graphs is k=6k=6 (all others happen at k5k\leq 5). In light of the Strong Perfect Graph Theorem [10], our result is somewhat surprising as every non-complete P5P_{5}-free kk-vertex-critical graph must contain C5C_{5} or C2k+1¯\overline{C_{2k+1}} for some k3k\geq 3. In addition, our result makes progress toward an open problem posed in [8] to find a dichotomy theorem for the number of kk-vertex-critical (P5,H)(P_{5},H)-free graphs for all kk when HH is of order 55. To see the state-of-the-art on this problem, see Table LABEL:tab:order5suvery.

We provide our infinite families of vertex-critical graphs in Section 2. Following that we provide a large table summarizing whether there are only finitely many or infinitely many kk-vertex-critical (P5,H)(P_{5},H)-free graphs for all kk in Section 3. From this, we find that there are only 1212 such graphs HH remaining to get the complete dichotomy to solve the open problem from [8]. We end this section with a brief section on notations and definitions.

1.1 Notation and Definitions

Let χ(G)\chi(G) denote the chromatic number of GG. A graph GG is kk-vertex-critical if χ(G)=k\chi(G)=k and χ(Gv)<k\chi(G-v)<k for all vV(G)v\in V(G). For SV(G)S\subseteq V(G), let G[S]G[S] denote the subgraph of GG induced by SS. For vertices u,vV(G)u,v\in V(G), we write uvu\sim v if uu and vv are adjacent and uvu\nsim v is uu and vv are non-adjacent. For S,TV(G)S,T\subseteq V(G), we say SS is complete (anti-complete) to TT if sts\sim t (sts\nsim t) for all sSs\in S and tTt\in T. For a vV(G)v\in V(G), N(v)N(v) is the open neighbourhood and denotes the set {uV(G):uv}\{u\in V(G):u\sim v\} and N[v]N[v] is the closed neighbourhood, denoting the set N(v){v}N(v)\cup\{v\}. For SV(G)S\subseteq V(G), we say SS is a stable set if uvu\nsim v for all u,vSu,v\in S. For graphs GG and HH, G+HG+H denotes the disjoint union of GG and HH where V(G+H)=V(G)V(H)V(G+H)=V(G)\cup V(H) and E(G+H)=E(G)E(H)E(G+H)=E(G)\cup E(H). A graph GG is HH-free if it does not contain any induced subgraph isomorphic to HH.

2 New infinite families of kk-vertex-critical graphs

We are now ready to give the constructions for our infinite families of vertex-critical graphs.

Definition 2.1.

Fix q1q\geq 1 and k3k\geq 3. Let G(q,k)G(q,k) be a graph on vertex set {v0,v1,,vkq}\{v_{0},v_{1},...,v_{kq}\} where, with each integer taken modulo kq+1kq+1, the neighbourhood of vertex viv_{i} is

{vi1,vi+1}{vi+kj+m:m=2,3,k1 and j=0,1,,q1}.\{v_{i-1},v_{i+1}\}\cup\{v_{i+kj+m}:m=2,3,...k-1\text{ and }j=0,1,...,q-1\}.

Examples to help visualize this definition can be found in Figure 1

\GraphInit\Vertexv0v_{0}\Vertexv1v_{1}\Vertexv2v_{2}\Vertexv3v_{3}\Vertexv4v_{4}\Vertexv5v_{5}\Vertexv6v_{6}\Vertexv7v_{7}\Vertexv8v_{8}\Vertexv9v_{9}\Vertexv10v_{10}\Vertexv11v_{11}\Vertexv12v_{12}\Vertexv13v_{13}\Vertexv14v_{14}\Vertexv15v_{15}\Vertexv16v_{16}\Vertexv17v_{17}\Vertexv18v_{18}\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge
\GraphInit\Vertexv0v_{0}\Vertexv1v_{1}\Vertexv2v_{2}\Vertexv3v_{3}\Vertexv4v_{4}\Vertexv5v_{5}\Vertexv6v_{6}\Vertexv7v_{7}\Vertexv8v_{8}\Vertexv9v_{9}\Vertexv10v_{10}\Vertexv11v_{11}\Vertexv12v_{12}\Vertexv13v_{13}\Vertexv14v_{14}\Vertexv15v_{15}\Vertexv16v_{16}\Vertexv17v_{17}\Vertexv18v_{18}\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge
Figure 1: The neighbourhood of v0v_{0} in G(3,6)G(3,6) (left) and the graph G(3,6)G(3,6) (right).

From this definition, it is clear that {G(q,3):q1}={Gr:r1}\{G(q,3):q\geq 1\}=\{G_{r}:r\geq 1\} where GrG_{r} is the infinite family 44-vertex-critical P7P_{7}-free graph in [9] and {G(q,4):q1}={Gp:p1}\{G(q,4):q\geq 1\}=\{G_{p}:p\geq 1\} where GpG_{p} is the infinite family of 55-vertex-critical 2P22P_{2}-free graphs in [16]. Further, our new construction is very natural in the sense that G(1,k)Kk+1G(1,k)\cong K_{k+1} and G(2,k)C2k+1¯G(2,k)\cong\overline{C_{2k+1}} for all kk, which include all prototypical (k+1)(k+1)-vertex-critical P5P_{5}-free graphs except C5C_{5} from the Strong Perfect Graph Theorem [10].

For a given qq and kk and for 0ik10\leq i\leq k-1, let Vi={vt:t=i(modk)}V_{i}=\{v_{t}:t=\ i\pmod{k}\}. It is clear that the ViV_{i}’s partition the vertex set of G(q,k)G(q,k).

Lemma 2.2.

For 1ik1\leq i\leq k, ViV_{i} is a stable set of G(q,k)G(q,k) and the only edge in V0V_{0} is v0vqkv_{0}v_{qk}.

Proof.

Fix i{0,1,2,qk}i\in\{0,1,2\ldots,qk\}. Since it is clear that vivi+1v_{i}\nsim v_{i+1} and vivi1v_{i}\nsim v_{i-1} with the exception of v0vqkv_{0}\sim v_{qk}, we will focus on the subset of N(vi)N(v_{i}) defined by {vi+kj+m:m=2,3,k1 and j=0,1,,q1}\{v_{i+kj+m}:m=2,3,...k-1\text{ and }j=0,1,...,q-1\}. Fix m{2,3,k1}m\in\{2,3,...k-1\} and j={0,1,,q1}j=\{0,1,...,q-1\}. The result now follows by considering two cases to consider.

Case 1: Suppose i+kj+mqki+kj+m\leq qk. Then i+kj+mi+m(modk)i+kj+m\equiv i+m\pmod{k}, and since 2mk12\leq m\leq k-1, ii+m(modk)i\not\equiv i+m\pmod{k}. Therefore, vhN(vi)v_{h}\not\in N(v_{i}) for all hi(modk)h\equiv i\pmod{k}.

Case 2: Suppose i+kj+m>qki+kj+m>qk, then i+kj+mi+kj+m1(modqk+1)i+kj+m\equiv i+kj+m-1\pmod{qk+1}, which is congruent to i+m1(modk)i+m-1\pmod{k}. Since 1m1k21\leq m-1\leq k-2, it follows that ii+m1(modk)i\not\equiv i+m-1\pmod{k}. Therefore, vhN(vi)v_{h}\not\in N(v_{i}) for all hi(modk)h\equiv i\pmod{k}.

Lemma 2.3.

For all k4k\geq 4 and q1q\geq 1, G(q,k)G(q,k) is 2K22K_{2}-free.

Proof.

Let G=G(q,k)G=G(q,k) and suppose GG is not 2K22K_{2}-free. By symmetry, we may suppose without loss of generality that v0v_{0} belongs to an induced 2K22K_{2}. Let v,x,yV(G)v_{\ell},x,y\in V(G) such that {v0,v,x,y}\{v_{0},v_{\ell},x,y\} induces a 2K22K_{2} in GG such that v0vv_{0}\sim v_{\ell} and xyx\sim y. It is clear by the definition of GG that N(v0)={v1,vqk}V2V3Vk1N(v_{0})=\{v_{1},v_{qk}\}\cup V_{2}\cup V_{3}\cup\cdots\cup V_{k-1}. Therefore, if {x,y}(V0{v0,vqk})V1\{x,y\}\not\subseteq(V_{0}\setminus\{v_{0},v_{qk}\})\cup V_{1}, then v0v_{0} will be adjacent to xx or yy, contradicting our assumption. Further, since xyx\sim y, we must have {x,y}(V0{v0,vqk})(V1{v1})\{x,y\}\subseteq(V_{0}\setminus\{v_{0},v_{qk}\})\cup(V_{1}\setminus\{v_{1}\}). Since both V1{v1}V_{1}\setminus\{v_{1}\} and V0{v0,vqk}V_{0}\setminus\{v_{0},v_{qk}\} are stable sets from Lemma 2.2, we may assume without loss of generality that xV0{v0,vqk}x\in V_{0}\setminus\{v_{0},v_{qk}\} and yV1{v1}y\in V_{1}\setminus\{v_{1}\}. There are now only three cases to consider.

Case 1: Suppose vV2V3Vk2v_{\ell}\in V_{2}\cup V_{3}\cup\cdots\cup V_{k-2}. Let =(modk)\ell^{\prime}=\ell\pmod{k}. Since k4k\geq 4, 2k22\leq\ell^{\prime}\leq k-2. So, for vhkV0v_{hk}\in V_{0} where <hkqk\ell<hk\leq qk, we have that hk=+k(h1)+(k)hk=\ell+k(h-1)+(k-\ell^{\prime}). Thus, vhkN(v)v_{hk}\in N(v_{\ell}) by the definition of GG. For vhkV0v_{hk}\in V_{0} where qk<hk<qk<hk<\ell, we have that hk=+k(h1)+(k+1)(modqk+1)hk=\ell^{\prime}+k(h-1)+(k-\ell^{\prime}+1)\pmod{qk+1}, so again vhkN(v)v_{hk}\in N(v_{\ell}). It follows that V0N(v)V_{0}\subseteq N(v_{\ell}) and, thus, vxv_{\ell}\sim x. This contradicts our assumption on the induced 2K22K_{2} in GG.

Case 2: Suppose vVk1v_{\ell}\in V_{k-1}. We show that Vk1N(y)V_{k-1}\subseteq N(y). Let h{0,,qk}h\in\{0,\ldots,qk\} such that y=vhy=v_{h}. We know that h=1(modk)h=1\pmod{k} since y=vhV1y=v_{h}\in V_{1}. Let h{0,,qk}h^{\prime}\in\{0,\ldots,qk\} such that h=k1(modk)h^{\prime}=k-1\pmod{k}. If h<hh<h^{\prime}, then since h+kj+k2k1(modk)h+kj+k-2\equiv k-1\pmod{k}, it follows that vhvhv_{h}\sim v_{h^{\prime}}. If h<hh^{\prime}<h, then since h+kj+k1(modqk+1)k1(modk)h+kj+k-1\pmod{qk+1}\equiv k-1\pmod{k}, it follows that vhvhv_{h}\sim v_{h^{\prime}}. Thus, Vk1N(y)V_{k-1}\subseteq N(y), and yvy\sim v_{\ell} which again contradicts our assumption on the induced 2K22K_{2} in GG.

Case 3: Suppose v{vqk,v1}v_{\ell}\in\{v_{qk},v_{1}\}. If v=v1v_{\ell}=v_{1}, then since 1+jk+m(modqk+1)=1+jk+m1+jk+m\pmod{qk+1}=1+jk+m for all 0jq10\leq j\leq q-1 and 2mk12\leq m\leq k-1, we can obtain all vhv_{h} such that vhV0v_{h}\in V_{0} in the set {v1+(q1)j+(k1):j=0,1,,q1}N(v1)\{v_{1+(q-1)j+(k-1)}:j=0,1,\ldots,q-1\}\subseteq N(v_{1}). Also, we can obtain all vhv_{h} such that vhVk1v_{h}\in V_{k-1} in the set {v1+(q1)j+(k2):j=0,1,,q1}N(v1)\{v_{1+(q-1)j+(k-2)}:j=0,1,\ldots,q-1\}\subseteq N(v_{1}). Thus, xv1=vx\sim v_{1}=v_{\ell} which contradicts our assumption on the induced 2K22K_{2} in GG. Similarly, for v=vqkv_{\ell}=v_{qk}, we get that yvy\sim v_{\ell} which again contradicts our assumption on the induced 2K22K_{2} in GG.

All three cases together show that GG is 2K22K_{2}-free. ∎

Lemma 2.4.

For all k4k\geq 4 and q1q\geq 1, G(q,k)G(q,k) is (K3+P1)(K_{3}+P_{1})-free.

Proof.

Let G=G(q,k)G=G(q,k) and suppose GG has an induced K3+P1K_{3}+P_{1}. By the symmetry of GG, we may assume that the isolated vertex in the K3+P1K_{3}+P_{1} is v0v_{0}. Let S={v0,vi1,vi2,vi3}S=\{v_{0},v_{i_{1}},v_{i_{2}},v_{i_{3}}\} be the subset of V(G)V(G) that induces the K3+P1K_{3}+P_{1} with v0v_{0} the isolated vertex. Since v0v_{0} is anticomplete with {vi1,vi2,vi3}\{v_{i_{1}},v_{i_{2}},v_{i_{3}}\}, it follows that {vi1,vi2,vi3}V0V1{vqk}\{v_{i_{1}},v_{i_{2}},v_{i_{3}}\}\subset V_{0}\cup V_{1}\setminus\{v_{qk}\}. However, since V0V_{0} and V1{vqk}V_{1}\setminus\{v_{qk}\} stable sets by Lemma 2.2, it follows that G[V0V1{vqk]}G[V_{0}\cup V_{1}\setminus\{v_{qk}]\} is bipartite and therefore triangle-free. Therefore, GG is (K3+P1)(K_{3}+P_{1})-free. ∎

Lemma 2.5.

For all k5k\geq 5 and q1q\geq 1, G(q,k)G(q,k) is C5C_{5}-free.

Proof.

Let k5k\geq 5, q1q\geq 1, and G=G(q,k)G=G(q,k). Suppose by way of contradiction that GG contains an induced C5C_{5}. Without loss of generality suppose v0v_{0} is on an induced C5C_{5} in GG and let {v0,vi1,vi2,vi3,vi4}\{v_{0},v_{i_{1}},v_{i_{2}},v_{i_{3}},v_{i_{4}}\} induce a C5C_{5} in GG such that v0vi1v_{0}\sim v_{i_{1}}, v0vi4v_{0}\sim v_{i_{4}}, and vijvij+1v_{i_{j}}\sim v_{i_{j+1}} for j=1,2,3j=1,2,3. By similar arguments to the proof of Lemma 2.3, we must have, without loss of generality, vi2V0{v0,vqk}v_{i_{2}}\in V_{0}\setminus\{v_{0},v_{qk}\} and vi3V1{v1}v_{i_{3}}\in V_{1}\setminus\{v_{1}\}. Further, since N(v0)={v1,vqk}V2V3Vk1N(v_{0})=\{v_{1},v_{qk}\}\cup V_{2}\cup V_{3}\cup\cdots\cup V_{k-1}, we must have {vi4,vi5}V2V3Vk1\{v_{i_{4}},v_{i_{5}}\}\subseteq V_{2}\cup V_{3}\cup\cdots\cup V_{k-1}. By a similar argument to that of Case 1 of the proof of Lemma 2.3 and since v1vqkv_{1}\sim v_{qk}, we must have vi4Vk1v_{i_{4}}\in V_{k-1} (or else vi2vi4v_{i_{2}}\sim v_{i_{4}}) and vi3Vjv_{i_{3}}\in V_{j} for some 2jk22\leq j\leq k-2. Further, j=k2j=k-2, or else vi1vi4v_{i_{1}}\sim v_{i_{4}}. But now since k5k\geq 5, and therefore k23k-2\geq 3, we have vi3vi1v_{i_{3}}\sim v_{i_{1}}. This contradicts our assumption on the induced C5C_{5}. Therefore, GG is C5C_{5}-free. ∎

Note our proof of the following lemma is adapted directly from Theorem 2.4 of [16] where it was shown that G(q,4,G(q,4,) (under the name GpG_{p}) is 44-vertex-critical for all qq.

Lemma 2.6.

For all k3k\geq 3 and q1q\geq 1, G(q,k)G(q,k) is (k+1)(k+1)-vertex-critical.

Proof.

Let G=G(q,k)G=G(q,k). Note that for all 0i(q1)k+10\leq i\leq(q-1)k+1, the set {vi+j:j=0,1,,k1}\{v_{i+j}:j=0,1,\ldots,k-1\} induces a KkK_{k}. Thus, χ(G)k\chi(G)\geq k. Suppose GG is kk-colourable, and without loss of generality, let vertices viv_{i} be assigned colour ii for i=0,1,,k1i=0,1,\ldots,k-1. We now must have vertex vjv_{j} be assigned colour j(modk)j\pmod{k} for all kjqk1k\leq j\leq qk-1. But now N(vqk)N(v_{qk}) contains all kk colours. So GG is not kk-colourable. From Lemma 2.2, however, this partial colouring is valid and we may give vertex vqkv_{qk} colour kk to get that χ(G)=k+1\chi(G)=k+1. Since vqkv_{qk} is the only vertex with colour kk, it follows by the symmetry of GG that GG is (k+1)(k+1)-vertex-critical. ∎

Theorem 2.7.

There are infinitely many kk-vertex-critical (2P2,K3+P1,C5)(2P_{2},K_{3}+P_{1},C_{5})-free graphs for all k6k\geq 6.

Proof.

The proof follows from Lemmas 2.3-2.6

Corollary 2.8.

There are infinitely many kk-vertex-critical (P5,C5)(P_{5},C_{5})-free graphs for all k6k\geq 6.

3 State of (P5,H)(P_{5},H)-free graphs when HH is of order 5

In this section we detail in a table whether there are only finitely many or infinitely many kk-vertex-critical (P5,H)(P_{5},H)-free graphs for all nonisomorphic graphs HH of order 55. This is to make progress on the open problem raised in [8]. For the names of graphs we are mostly following https://www.graphclasses.org/smallgraphs.html#nodes5 with the notable exception that we use ++ to denote disjoint union.

Table 1: The state-of-the-art for HH of order 55.
Graph Graph name Finite/Infinite Reference
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44
K5¯\overline{K_{5}} finite Ramsey’s Theorem [27]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge
P2+3P1P_{2}+3P_{1} finite [7]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge
P3+2P1P_{3}+2P_{1} finite [1]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge
claw+P1+P_{1} unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge
K1,4K_{1,4} finite [22] (see also [25, Theorem 1])
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge
2K2+P12K_{2}+P_{1} infinite Contains 2K22K_{2} [16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge
P4+P1P_{4}+P_{1} unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge
P3+P2P_{3}+P_{2} infinite Contains 2K22K_{2} [16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge
K3+2P1K_{3}+2P_{1} infinite Contains K3+P1K_{3}+P_{1} [8, 16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge
chair finite k5k\leq 5, unknown k6k\geq 6 [20]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge
co-dart infinite Contains K3+P1K_{3}+P_{1} [8, 16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge
diamond+P1¯\overline{\text{diamond}+P_{1}} unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge
C4+P1C_{4}+P_{1} unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge
banner finite [3, Theorem 3(i)]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge
diamond+P1+P_{1} infinite Contains K3+P1K_{3}+P_{1} [8, 16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge
bull finite k=5k=5, unknown k6k\geq 6 [19]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge
dart unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge
K2,3K_{2,3} finite [22]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge\Edge
K3+2P2¯\overline{K_{3}+2P_{2}} unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge
P5P_{5} infinite Contains 2K22K_{2} [16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge
K3+P2K_{3}+P_{2} infinite Contains 2K22K_{2} [16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge
co-banner infinite Contains 2K22K_{2} [16] (see also [3])
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge
butterfly infinite Contains 2K22K_{2} [16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge
C5C_{5} finite k5k\leq 5, infinite k6k\geq 6 [16], Corollary 2.8
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge
P5¯\overline{P_{5}} finite [14]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge\Edge
gem finite [4] (see also [6])
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge
kite infinite Contains K3+P1K_{3}+P_{1} [8, 16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge
K4+P1K_{4}+P_{1} infinite Contains K3+P1K_{3}+P_{1} [8, 16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge\Edge
claw+K1¯\overline{\text{claw}+K_{1}} infinite Contains K3+P1K_{3}+P_{1} [8, 16]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge
P3+2P1¯\overline{P_{3}+2P_{1}} unknown N/A
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge\Edge
paraglider or P3+P2¯\overline{P_{3}+P_{2}} finite [4]
\GraphInit\Vertex0\Vertex11\Vertex22\Vertex33\Vertex44\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge
W4W_{4} unknown N/A
\GraphInit\Vertex011223344\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge
K5eK_{5}-e unknown N/A
\GraphInit011223344\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge\Edge
K5K_{5} infinite k=5k=5, unknown k6k\geq 6 [16]

References

  • [1] T. Abuadas, B. Cameron, C. T. Hoàng, and J. Sawada. Vertex-critical (P3+P1)(P_{3}+\ell P_{1})-free and vertex-critical (gem, co-gem)-free graphs. preprint, arXiv : 2206.03422[math.CO], 2022.
  • [2] F. Bonomo, M. Chudnovsky, P. Maceli, O. Schaudt, M. Steinz, and M. Zhong. Three-coloring and list three-coloring of graphs without induced paths on seven vertices. Combinatorica, 38(4):779–801, 2018.
  • [3] C. Brause, M. Geißer, and I. Schiermeyer. Homogeneous sets, clique-separators, critical graphs, and optimal χ\chi-binding functions. Discrete Appl. Math., 320:211–222, 2022.
  • [4] Q. Cai, J. Goedgebeur, and S. Huang. Some results on kk-critical P5P_{5}-free graphs. Discrete Appl. Math., 334:91–100, 2023.
  • [5] Q. Cai, S. Huang, T. Li, and Y. Shi. Vertex-critical (P5,banner)(P_{5},banner)-free graphs. In Frontiers in Algorithmics - 13th International Workshop, FAW 2019, Sanya, China, April 29 - May 3, 2019, Proceedings, volume 11458, pages 111–120, Sanya, China, 2019. Springer.
  • [6] B. Cameron and C. T. Hoàng. A refinement on the structure of vertex-critical (P5P_{5}, gem)-free graphs. Theoretical Computer Science, 961:113936, 2023.
  • [7] B. Cameron, C. T. Hoàng, and J. Sawada. Dichotomizing kk-vertex-critical HH-free graphs for HH of order four. Discrete Appl. Math., 312:106–115, 2022. Ninth Workshop on Graph Classes, Optimization, and Width Parameters.
  • [8] K. Cameron, J. Goedgebeur, S. Huang, and Y. Shi. kk-critical graphs in P5P_{5}-free graphs. Theoretical Computer Science, 864:80–91, 2021.
  • [9] M. Chudnovsky, J. Goedgebeur, O. Schaudt, and M. Zhong. Obstructions for three-coloring graphs without induced paths on six vertices. J. Combin., Ser. B, 140:45–83, 2020.
  • [10] M. Chudnovsky, N. Robertson, P. Seymour, and R. Thomas. The strong perfect graph theorem. Ann. of Math., 164(1):51–229, 2006.
  • [11] M. Chudnovsky, S. Spirkl, and M. Zhong. Four-coloring P6P_{6}-free graphs. I. Extending an excellent precoloring. preprint, arXiv : 1802.02282[math.CO], 2018.
  • [12] M. Chudnovsky, S. Spirkl, and M. Zhong. Four-coloring P6P_{6}-free graphs. II. Finding an excellent precoloring. preprint, arXiv : 1802.02283[math.CO], 2018.
  • [13] M. Chudnovsky, S. Spirkl, and M. Zhong. Four-coloring P6P_{6}-free graphs. In Proceedings of the Thirtieth Annual ACM-SIAM Symposium on Discrete Algorithms, SODA ’19, page 1239–1256, USA, 2019. Society for Industrial and Applied Mathematics.
  • [14] H. S. Dhaliwal, A. M. Hamel, C. T. Hoàng, F. Maffray, T. J. D. McConnell, and S. A. Panait. On color-critical (P5,(P_{5},co-P5)P_{5})-free graphs. Discrete Appl. Math., 216:142–148, 2017.
  • [15] C. T. Hoàng, M. Kamiński, V. Lozin, J. Sawada, and X. Shu. Deciding kk-colorability of P5P_{5}-free graphs in polynomial time. Algorithmica, 57:74–81, 2010.
  • [16] C. T. Hoàng, B. Moore, D. Recoskie, J. Sawada, and M. Vatshelle. Constructions of kk-critical P5P_{5}-free graphs. Discrete Appl. Math., 182:91–98, 2015.
  • [17] I. Holyer. The NP-completeness of edge-coloring. SIAM J. Comput., 10(4):718–720, 1981.
  • [18] S. Huang. Improved complexity results on kk-coloring PtP_{t}-free graphs. European J. Combin., 51:336–346, 2016.
  • [19] S. Huang, J. Li, and W. Xia. Critical (P5P_{5}, bull)-free graphs. Discrete Applied Mathematics, 334:15–25, 2023.
  • [20] S. Huang and Z. Li. Vertex-Critical (P5,chair)(P_{5},chair)-free Graphs. preprint, arXiv : arXiv:2301.02436 [math.CO], 2023.
  • [21] M. Kamiński and V. Lozin. Coloring edges and vertices of graphs without short or long cycles. Contrib. Discrete Math., 2(1):61–66, 2007.
  • [22] M. Kamiński and A. Pstrucha. Certifying coloring algorithms for graphs without long induced paths. Discrete Appl. Math., 261:258–267, 2019.
  • [23] R. M. Karp. Reducibility among combinatorial problems. In Complexity of computer computations (Proc. Sympos., IBM Thomas J. Watson Res. Center, Yorktown Heights, N.Y., 1972), pages 85–103, 1972.
  • [24] D. Leven and Z. Gail. NP completeness of finding the chromatic index of regular graphs. J. Algorithms, 4:35–44, 1983.
  • [25] V. Lozin and D. Rautenbach. Some results on graphs without long induced paths. Inform. Process. Lett., 88(4):167–171, 2003.
  • [26] F. Maffray and G. Morel. On 33-colorable P5P_{5}-free graphs. SIAM J. Discrete Math., 26(4):1682–1708, 2012.
  • [27] F. P. Ramsey. On a problem of formal logic. Proc. Lond. Math. Soc., (s2-30):264–286, 1928.