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

The maximum number of cliques in graphs with bounded odd circumference

Zequn Lv Alfréd Rényi Institute of Mathematics, Hungarian Academy of Sciences. Department of Mathematical Sciences, Tsinghua University. Ervin Győri Alfréd Rényi Institute of Mathematics, Hungarian Academy of Sciences. Zhen He Alfréd Rényi Institute of Mathematics, Hungarian Academy of Sciences. Department of Mathematical Sciences, Tsinghua University. Nika Salia Alfréd Rényi Institute of Mathematics, Hungarian Academy of Sciences. Chuanqi Xiao Alfréd Rényi Institute of Mathematics, Hungarian Academy of Sciences. Department of Computer Science and Information Theory, Budapest University of Technology and Economics. Xiutao Zhu Alfréd Rényi Institute of Mathematics, Hungarian Academy of Sciences. Department of Mathematics, Nanjing University.
Abstract

In this work, we give the sharp upper bound for the number of cliques in graphs with bounded odd circumferences. This generalized Turán-type result is an extension of the celebrated Erdős and Gallai theorem and a strengthening of Luo’s recent result. The same bound for graphs with bounded even circumferences is a trivial application of the theorem of Li and Ning.

1 Introduction

A central topic of extremal combinatorics is to investigate sufficient conditions for the appearance of given substructures. In particular, for a given graph HH and a set of graph \mathcal{F}, the generalized Turán number ex(n,H,){\rm ex}(n,H,\mathcal{F}) denotes the maximum number of copies of HH in a graph on nn vertices containing no FF as a subgraph, for every FF\in\mathcal{F}. In 1959, Erdős and Gallai [1] determined the maximum number of edges in a graph with a small circumference. For every integers nn and kk such that nk3n\geq k\geq 3, they proved

ex(n,K2,𝒞k)(k1)(n1)2,{\rm ex}(n,K_{2},\mathcal{C}_{\geq k})\leq\frac{(k-1)(n-1)}{2},

where KkK_{k} denotes the complete graph with kk vertices and 𝒞k\mathcal{C}_{\geq k} denotes the family of cycles of length at least kk. The bound is sharp for every nn congruent to one modulo k2k-2. The equality is attained by graphs with n1k2\frac{n-1}{k-2} maximal 22-connected blocks each isomorphic to Kk1K_{k-1}. Recently Li and Ning [2] proved that in order to obtain the same upper-bound it is enough to forbid only long even cycles

ex(n,K2,𝒞2keven)(2k1)(n1)2,{\rm ex}(n,K_{2},\mathcal{C}^{even}_{\geq 2k})\leq\frac{(2k-1)(n-1)}{2}, (1)

where 𝒞2keven\mathcal{C}^{even}_{\geq 2k} denotes the family of even cycles of length at least 2k2k, that is {C2k,C2k+2,}\{C_{2k},C_{2k+2},\dots\}. We also denote the family of odd cycles of length at least 2k+12k+1 by 𝒞2k+1odd:={C2k+1,C2k+3,}\mathcal{C}^{odd}_{\geq 2k+1}:=\{C_{2k+1},C_{2k+3},\dots\}.

Note that by considering the complete balanced bipartite graph, one can not achieve a linear bound for edges in graphs with a small odd circumference. On the other hand, Voss and Zuluga [5] proved that every 22-connected graph GG with minimum degree at least k3k\geq 3, with at least 2k+12k+1 vertices, contains an even cycle of length at least 2k2k. Even more, if GG is not bipartite then it contains an odd cycle of length at least 2k12k-1.

The celebrated Erdős and Gallai theorem is well studied and well understood. There are numerous papers strengthening, generalizing, and extending for different classes of graphs. Recently Luo [3] proved

ex(n,Kr,𝒞k)(n1)k2(k1r),{\rm ex}(n,K_{r},\mathcal{C}_{\geq k})\leq\frac{(n-1)}{k-2}\binom{k-1}{r}, (2)

for all nk4n\geq k\geq 4. This bound is a great tool for obtaining results in hypergraph theory. In order to highlight its significance, that bound was consequently reproved with different methods multiple times [4, 6]. In this paper, we strengthen Luo’s theorem. In particular, we obtain the tight same bounds for graphs with bounded odd circumferences. On the other hand, the result for graphs with bounded even circumference is trivial after applying Equation 1 and the following easy corollary of Equation 2,

ex(n,Kr,Pk+1)nk(kr),{\rm ex}(n,K_{r},P_{k+1})\leq\frac{n}{k}\binom{k}{r},

where Pk+1P_{k+1} denotes the path of length kk. Since the graph GG does not contain a cycle of length 2k2k, for each vertex the neighborhood contains no path of length 2k22k-2. In particular, for all r3r\geq 3 we have

ex(n,Kr,𝒞2keven)1rvd(v)(2k2)(2k2r1)1rex(n,K2,𝒞2keven)(k1)(2k2r1)n12k2(2k1r).{\rm ex}(n,K_{r},\mathcal{C}^{even}_{\geq 2k})\leq\frac{1}{r}\sum_{v}\frac{d(v)}{(2k-2)}\binom{2k-2}{r-1}\leq\frac{1}{r}\frac{ex(n,K_{2},\mathcal{C}^{even}_{\geq 2k})}{(k-1)}\binom{2k-2}{r-1}\leq\frac{n-1}{2k-2}\binom{2k-1}{r}.

For graphs with small odd circumferences, we have the following theorem.

Theorem 1.

For integers n,k,rn,k,r satisfying n2krn\geq 2k\geq r and k6k\geq 6,

ex(n,Kr,𝒞2k+1odd)n12k1(2kr).{\rm ex}(n,K_{r},\mathcal{C}^{odd}_{\geq 2k+1})\leq\frac{n-1}{2k-1}\binom{2k}{r}.

The equality holds if and only if n1n-1 is divisible by 2k12k-1 for a connected nn-vertex graph which consists of n12k1\frac{n-1}{2k-1} maximal 22-connected blocks isomorphic to K2kK_{2k}.

It is interesting to ask for which integers and for which congruence classes the same phenomenon still holds. In this direction, we would like to rise a modest conjecture.

Conjecture 1.

For an integer k2k\geq 2 and a sufficiently large nn. Let GG be an nn vertex C3+1C_{3\ell+1}-free graph for every integer k\ell\geq k. Then for every r2r\geq 2, the number of cliques of size rr in GG is at most

n13k1(3kr).\frac{n-1}{3k-1}\binom{3k}{r}.

The equality holds if and only if n1n-1 is divisible by 3k13k-1 for a connected nn-vertex graph which consists of n13k1\frac{n-1}{3k-1} maximal 22-connected blocks isomorphic to K3kK_{3k}.

Moreover, we would like to propose the following conjecture. Let pp be a prime number and 𝒞pprime\mathcal{C}_{\geq p}^{prime} be the family of all cycles of length at least pp.

Conjecture 2.

For integers nn, rr and a prime pp satisfying r<pr<p, we have

ex(n,Kr,𝒞pprime)n1p2(p1r).ex(n,K_{r},\mathcal{C}_{\geq p}^{prime})\leq\frac{n-1}{p-2}\binom{p-1}{r}.

The equality holds if and only if n1n-1 is divisible by p2p-2, for a connected nn-vertex graph which consists of n1p2\frac{n-1}{p-2} maximal 22-connected blocks isomorphic to Kp1K_{p-1}.

2 Proof of the main result

Proof.

We prove Theorem 1 by induction on the number of vertices of the graph. The base cases for n2kn\leq 2k are trivial. Let GG be a graph on nn vertices where n>2kn>2k. We assume that every 𝒞2k+1odd\mathcal{C}^{odd}_{\geq 2k+1}-free graph on mm vertices, for some m<nm<n, contains at most m12k1(2kr)\frac{m-1}{2k-1}\binom{2k}{r} copies of KrK_{r} and the equality is achieved for the class of graphs described in the statement of the theorem.

If δ(G)\delta(G), the minimum degree of GG, is at most k+2k+2 then we are done by the induction hypothesis

Kr(G)Kr(G[V(G){v}])+(k+2r1)n22k1(2kr)+(k+2r1)<n12k1(2kr),K_{r}(G)\leq K_{r}(G[V(G)\setminus\{v\}])+\binom{k+2}{r-1}\leq\frac{n-2}{2k-1}\binom{2k}{r}+\binom{k+2}{r-1}<\frac{n-1}{2k-1}\binom{2k}{r},

since k6k\geq 6. From here, we may assume GG is a graph with δ(G)>k+2\delta(G)>k+2, and each edge of GG is in a copy of KrK_{r}.

Let v1v2v3vmv_{1}v_{2}v_{3}\dots v_{m} be a longest path of GG such that v1v_{1} is adjacent to vtv_{t} and tt is the maximum possible among the longest paths.

Claim 2.

If t2kt\leq 2k, then vtv_{t} is a cut vertex.

Proof.

Consider a family 𝒫\mathcal{P} contains all longest paths of GG on the vertex set {v1,v2,v3,,vm}\{v_{1},v_{2},v_{3},\dots,v_{m}\} such that vtvt+1vmv_{t}v_{t+1}\dots v_{m} is a sub-path of them. Let T1T_{1} be the set of terminal vertices, excluding vmv_{m}, of paths from 𝒫\mathcal{P}. If T1={v1,v2,,vt1}T_{1}=\{v_{1},v_{2},\dots,v_{t-1}\}, then by the maximality of tt, the vertex vtv_{t} is a cut vertex. Let the path u1u2ut1utvt+1vmu_{1}u_{2}\cdots u_{t-1}u_{t}v_{t+1}\cdots v_{m} be a path from 𝒫\mathcal{P} such that urT1u_{r}\in T_{1} and ur+1T1u_{r+1}\notin T_{1} minimizing rr. Note that ut=vtu_{t}=v_{t} and {v1,v2,,vt}={u1,u2,,ut}\{v_{1},v_{2},\dots,v_{t}\}=\{u_{1},u_{2},\dots,u_{t}\}.

By the minimality of rr, the vertex u1u_{1} is not adjacent to two consecutive vertices from {ur+1,ur+2,,vt}\{u_{r+1},u_{r+2},\dots,v_{t}\}. Since if v1v_{1} is adjacent to uiu_{i} and ui+1u_{i+1} for r+1itr+1\leq i\leq t, then

u2u3urur+1uiu1ui+1utvt+1vmu_{2}u_{3}\cdots u_{r}u_{r+1}\cdots u_{i}u_{1}u_{i+1}\cdots u_{t}v_{t+1}\cdots v_{m}

forms a path from 𝒫\mathcal{P}, contradicting the minimality of rr.

Note that u1u_{1} is not adjacent to vertices ur+2,ur+3,,u2r+1u_{r+2},u_{r+3},\dots,u_{2r+1}. Otherwise, suppose u1u_{1} is adjacent to uiu_{i} for r+2i2r+1r+2\leq i\leq 2r+1, then we get a path from 𝒫\mathcal{P}

ui1ui2ur+1uru1uiui+1utvt+1vm,u_{i-1}u_{i-2}\dots u_{r+1}u_{r}\dots u_{1}u_{i}u_{i+1}\dots u_{t}v_{t+1}\dots v_{m},

contradicting the minimality of rr since ur+1T1u_{r+1}\notin T_{1}.

Finally we have d(u1)(r1)+t(2r+1)+12=t22<kd(u_{1})\leq(r-1)+\frac{t-(2r+1)+1}{2}=\frac{t-2}{2}<k, a contradiction. ∎

Claim 3.

We have t2kt\leq 2k.

Proof.

Suppose otherwise, t>2kt>2k. Let CC denote the cycle v1v2vtv1v_{1}v_{2}\dots v_{t}v_{1}. Since GG is 𝒞2k+1odd\mathcal{C}^{odd}_{\geq 2k+1}-free, tt is even and vertices vt3v_{t-3} and vt2v_{t-2} have no common neighbors in G[V(G)V(C)]G[V(G)\setminus V(C)]. On the other hand, since every edge is in a KrK_{r} there is a vertex vlv_{l} incident with both vt3v_{t-3} and vt2v_{t-2}. Let us denote c:=t2kc:=t-2k. Since GG is 𝒞2k+1odd\mathcal{C}^{odd}_{\geq 2k+1}-free, c1l2k4c-1\leq l\leq 2k-4 holds. We take indices modulo tt in this claim.

There is no ii such that viv_{i} is adjacent with vt1v_{t-1} and vi+1v_{i+1} is adjacent with v1v_{1}. Since otherwise, the following cycle is odd with a length greater than 2k2k,

v1v2vivt1vt2vi+1v1.v_{1}v_{2}\dots v_{i}v_{t-1}v_{t-2}\dots v_{i+1}v_{1}.

Note that, since vt1,v1T1v_{t-1},v_{1}\in T_{1} and from maximality of tt, we have N(v1),N(vt1)V(C)N(v_{1}),N(v_{t-1})\subseteq V(C). Therefore we have N(v1)N+(vt1)=N(v_{1})\cap N^{+}(v_{t-1})=\emptyset. Where N+(vt1)N^{+}(v_{t-1}) denotes the following set {vi+1:viN+(vt1)}\{v_{i+1}:v_{i}\in N^{+}(v_{t-1})\}.

There is no such ii such that both l<i<l+c2l<i<l+c-2 and v1viv_{1}v_{i} is an edge of GG hold. Since otherwise, one of the following cycles is an odd cycle longer than 2k2k,

v1v2vlvt2vt3viv1 or v1v2vlvt3vt4viv1.v_{1}v_{2}\dots v_{l}v_{t-2}v_{t-3}\dots v_{i}v_{1}\mbox{~{}or~{}}v_{1}v_{2}\dots v_{l}v_{t-3}v_{t-4}\dots v_{i}v_{1}.

Similarly, there is no ii such that l<i<l+cl<i<l+c and vt1viv_{t-1}v_{i} is an edge of GG. Finally, we have

N(v1){vl+2,,vl+c3}=.N(v_{1})\cap\{v_{l+2},\dots,v_{l+c-3}\}=\emptyset.

Also, we have

N(v1)N+(vt1)=N(v_{1})\cap N^{+}(v_{t-1})=\emptyset

and

N+(vt1){vl+2,,vl+c3}=.N^{+}(v_{t-1})\cap\{v_{l+2},\dots,v_{l+c-3}\}=\emptyset.

Hence, we get

|N(v1)||{v1,v2,,v2k+c}N+(vt1){vl+2,,vl+c3}|2k+c(k+3)(c5)<δ(G),\left|{N(v_{1})}\right|\leq\left|{\{v_{1},v_{2},\dots,v_{2k+c}\}\setminus N^{+}(v_{t-1})\setminus\{v_{l+2},\dots,v_{l+c-3}\}}\right|\leq 2k+c-(k+3)-(c-5)<\delta(G),

a contradiction. ∎

From Claims 2 and 3, GG contains a 22-connected block of size at most 2k2k. By the induction hypothesis, we see that graph GG contains at most n12k1(2kr)\frac{n-1}{2k-1}\binom{2k}{r} edges. The equality is achieved if and only if every maximal 22-connected component of GG is isomorphic to K2kK_{2k} and the number of the maximal 22-connected components is n12k1\frac{n-1}{2k-1}. ∎

3 Acknowledgements

The research of Győri is partially supported by the National Research, Development and Innovation Office – NKFIH, grant K116769, K132696 and SNN117879. The research of Salia was supported by the Institute for Basic Science (IBS-R029-C4). The research of Zhu was supported by program B for an outstanding Ph.D. candidate at Nanjing University.

References

  • [1] Paul Erdős and Tibor Gallai. On maximal paths and circuits of graphs. Acta Mathematica Academiae Scientiarum Hungarica, 10(3-4):337–356, 1959.
  • [2] Binlong Li and Bo Ning. Eigenvalues and cycles of consecutive lengths. arXiv preprint arXiv:2110.05670, 2021.
  • [3] Ruth Luo. The maximum number of cliques in graphs without long cycles. Journal of Combinatorial Theory, Series B, 128:219–226, 2018.
  • [4] Bo Ning and Xing Peng. Extensions of the Erdős–Gallai theorem and Luo’s theorem. Combinatorics, Probability and Computing, 29(1):128–136, 2020.
  • [5] HJ Voss and C. Zuluga. Maximale gerade und ungerade kreise in graphen i. Wiss. Z. Tech. Hochschule Ilmenau, 4:57–70, 1977.
  • [6] Xiutao Zhu, Ervin Győri, Zhen He, Zequn Lv, Nika Salia, and Chuanqi Xiao. Stability version of dirac’s theorem and its applications for generalized tur\\backslash’an problems. arXiv preprint arXiv:2207.12465, 2022.

E-mail addresses:
E. Győri: [email protected]
Z. He: [email protected]
J. Lv: [email protected]
N. Salia: [email protected]
C. Xiao: [email protected]
X. Zhu: [email protected]