arXiv++ Combinatorics

We study finite extremal problems for palettes, which arise from the palette framework for the uniform Turán densities of $3$-uniform hypergraphs. Recent work has developed reductions from palette colorability questions to extremal problems for digraphs. In this paper we prove an exact degree-square refinement of these reductions for a natural family of left and right tournament palettes. For a tournament $T$, let $P_T^L$ and $P_T^R$ denote the left and right palettes generated by $T$. We prove that if $T$ is self-converse and has at least two vertices, then for every $m\ge 1$ the maximum number of admissible triples in an $m$-color palette avoiding both $P_T^L$ and $P_T^R$ is \[ \operatorname{ex}_2^+(m,T) = \max\left\{ \sum_{v\in V(D)} d_D^+(v)^2: |V(D)|=m,\; D\text{ is }T\text{-free} \right\}. \] The proof attaches two auxiliary digraphs to each palette and converts the palette optimization into a degree-square Turán problem. We also prove a general majorization principle for convex out-degree moments in $F$-free digraphs. Whenever an ordinary Turán extremal construction has extremal initial segments, the same construction maximizes every nondecreasing convex function of the out-degree sequence. Applying this to the Brown--Harary and Zhou--Li extremal digraphs for directed cycles gives exact formulas for all convex out-degree moments in $\overrightarrow{C}_{\ell}$-free digraphs. In particular, $\operatorname{ex}_2^+(m,\overrightarrow{C}_{3}) =\frac{m(m^2-1)}{3}.$ Consequently, for $m$ color the sharp density avoiding the two cyclic-triangle palettes is $\frac13-\frac1{3m^2}.$ Combining this exact finite endpoint with the palette classification theorem, we obtain finite $3$-graphs $H_m$ satisfying \[ \frac13-\frac1{3m^2} \le π_{\mathrm u}(H_m) \le \frac13. \] Thus the densities of these finite hypergraphs converge to $\frac13$.

Erdős and Simonovits asked the following question: For an integer $c\geq 2$ and a family of non-bipartite graphs $\mathcal{F}$, what is the infimum of $α$ such that any $\mathcal{F}$-free $n$-vertex graph with $n$ large enough and minimum degree at least $αn$ has chromatic number at most $c$? Denote the infimum as $δ_χ(\mathcal{F}, c)$. A fundamental result of Erdős, Stone and Simonovits implies that if $3\le r+1=χ(\mathcal{F})=\min\{χ(F): F\in \mathcal{F}\}$, then for any $c\le r-1$, $δ_χ(\mathcal{F}, c)=1-{1 \over r}$. So the remaining challenge is to determine $δ_χ(\mathcal{F}, c)$ for $c\ge χ(\mathcal{F})-1$. Most previous known results are under the condition that $c= χ(\mathcal{F})-1$. When $c\ge χ(\mathcal{F})$, the only known exact results are $δ_χ(K_3, 3)$ by Häggkvist and Jin, and $δ_χ(K_3, c)$ for every $c\ge4$ by Brandt and Thomassé, $δ_χ(K_r, r)$ and $δ_χ(K_r, r+1)$ by Goddard and Lyle, and Nikiforov. Combining results of Thomassen and Ma, $Ω\bigg((c+1)^{-8(k+1)}\bigg)=δ_χ(C_{2k+1}, c)=O(\frac{k}{c})$ for $c\ge 3$. In this paper, we determine $δ_χ(C_{2k+1}, c)$ for all $c\ge 2$ and $k\ge 3c+4$. We also obtain the following corollary. If $G$ is a graph on $n$ vertices with $c\ge 3$, $χ(G)>c$ and $δ(G)> {n \over 2c+2}$, then $C_{2k+1} \subset G$ for all $k\in [3c+4, {n \over 108(c+1)^c}]$. Methods to obtain all previous known results related to odd cycles cannot be applied to solve for $δ_χ(C_{2k+1}, c)$ for $c\ge 3$.The innovation of our proof is to give the concept of a `strong $2k$-core'. We think that this concept grasps the essence of the problem and it makes our proof concise and elementary (we do not need to borrow any other tools). How to define a proper `core' might be a key to this type of questions.

Füredi and Gunderson showed that $ex(n, C_{2k+1})$ is achieved only on $K_{\lfloor\frac{n}{2}\rfloor, \lceil\frac{n}{2}\rceil}$ if $n\ge 4k-2$. It is natural to study how far a $ C_{2k+1}$-free graph is from being bipartite.Let $T^*(r, n)$ be obtained by adding a suspension $K_{r}$ with $1$ suspension point to $K_{\lfloor\frac{n-r+1}{2}\rfloor, \lceil\frac{n-r+1}{2}\rceil}$. We show that for integers $r, k$ with $3\le r\le 2k-4$ and $n\ge 20(r+2)^2k$, if $G$ is a $C_{2k+1}$-free $n$-vertex graph with $e(G)\ge e(T^*(r, n))$, then $G$ is obtained by adding suspensions to a bipartite graph one by one and the total number of vertices in all suspensions minus intersection points is no more than $r-1$. In other words, $G=B\bigcup\limits_{i=1}^p G_i$, where $B$ is a bipartite graph, $G_1$ is a suspension to $B$, $G_j$ is a suspension to $B\bigcup\limits_{i=1}^{j-1} G_i$ for $2\le j\le p$ and $\sum\limits_{i=1}^p \vert V(G_i)-V(G_i)\cap V(B\bigcup\limits_{i=1}^{j-1} G_i) \vert\le r-1$. Furthermore, $\sum\limits_{i=1}^p \vert V(G_i)-V(G_i)\cap V(B\bigcup\limits_{i=1}^{j-1} G_i) \vert= r-1$ if and only if $G=T^*(r, n)$. Let $d_2(G)=\min\{|T|: T\subseteq V(G), G-T \ \text{is bipartite}\}$ and $γ_2(G)=\min\{|E|: E\subseteq E(G), G-E \ \text{is bipartite}\}$. Our structural stability result implies that $d_2(G)\le r-1$ and $γ_2(G)\le {\lceil\frac{r}{2}\rceil \choose 2}+{\lfloor\frac{r}{2}\rfloor \choose 2}$ under the same condition, which is a recent result of Ren-Wang-Wang-Yang [SIAM J. Discrete Math. 38 (2024)]. They proved $d_2(G)\le r-1$ and $γ_2(G)\le {\lceil\frac{r}{2}\rceil \choose 2}+{\lfloor\frac{r}{2}\rfloor \choose 2}$ separately. We introduce a new concept strong-$2k$-core which is the key that we can give a stronger structural stability result but a simpler proof.

The Lagrangian density of an $r$-uniform hypergraph $H$ is $r!$ multiplying the supremum of the Lagrangians of all $H$-free $r$-uniform hypergraphs. For an $r$-uniform graph $H$ with $t$ vertices, it is clear that $π_λ(H)\ge r!λ{(K_{t-1}^r)}$. We say that an $r$-uniform hypergraph $H$ with $t$ vertices is $λ$-perfect if $π_λ(H)= r!λ{(K_{t-1}^r)}$. A theorem of Motzkin and Straus implies that all $2$-uniform graphs are $λ$-perfect. It is interesting to understand what kind of hypergraphs are $λ$-perfect. The property `$λ$-perfect' is monotone in the sense that an $r$-graph obtained by removing an edge from a $λ$-perfect $r$-graph (keep the same vertex set) is $λ$-perfect. It's interesting to understand the relation between the number of edges in a hypergraph and the `$λ$-perfect' property. We propose that the number of edges in a hypergraph no more than the number of edges in a linear hyperpath would guarantee the `$λ$-perfect' property. We show some partial result to support this conjecture. We also give some partial result to support the conjecture that the disjoint union of two $λ$-perfect $r$-uniform hypergraph is $λ$-perfect. We show that the disjoint union of a $λ$-perfect $3$-graph and $S_{2,t}=\{123,124,125,126,...,12(t+2)\}$ is perfect. This result implies the earlier result of Heftz and Keevash, Jiang, Peng and Wu, and several other earlier results.

The {\em Turán number} of an $r$-uniform graph $F$, denoted by $ex(n,F)$, is the maximum number of edges in an $F$-free $r$-uniform graph on $n$ vertices. The {\em Turán density} of $F$ is defined as $π(F)=\underset{n\rightarrow\infty}{\lim}{ex(n,F) \over {n \choose r }}.$ For graphs, Erdős-Stone-Simonovits (\cite{ESi}, \cite{ES}) showed that $Π_{\infty}^{(2)}=Π_{fin}^{(2)}=Π_{1}^{(2)}=\{0, {1 \over 2}, {2 \over 3}, \ldots,{l-1 \over l}, ...\}.$ We know quite few about the Turán density of an $r$-uniform graph for $r\ge 3$. Baber and Talbot \cite{BT}, and Pikhurko \cite{Pikhurko2} showed that there is an irrational number in $Π_{3}^{(3)}$ and $Π_{fin}^{(3)}$ respectively, disproving a conjecture of Chung and Graham \cite{FG}. Baber and Talbot \cite{BT} asked whether $Π_{1}^{(r)}$ contains an irrational number. In this paper, we show that the Lagrangian density of $F=\{123, 124, 134, 234, 567\}$ (the disjoint union of $K_4^3$ and an edge) is ${\sqrt 3\over 3}$, consequently, the Turán density of the extension of $F$ is an irrational number, answering the question of Baber and Talbot.

A real number $α\in [0, 1)$ is a jump for an integer $r\ge 2$ if there exists $c>0$ such that no number in $(α, α+ c)$ can be the Turán density of a family of $r$-uniform graphs. A classical result of Erd\H os and Stone \cite{ES} implies that that every number in $[0, 1)$ is a jump for $r=2$. Erd\H os \cite{E64} also showed that every number in $[0, r!/r^r)$ is a jump for $r\ge 3$ and asked whether every number in $[0, 1)$ is a jump for $r\ge 3$. Frankl and Rödl \cite{FR84} gave a negative answer by showing a sequence of non-jumps for every $r\ge 3$. After this, Erd\H os modified the question to be whether $\frac{r!}{r^r}$ is a jump for $r\ge 3$? What's the smallest non-jump? Frankl, Peng, Rödl and Talbot \cite{FPRT} showed that ${5r!\over 2r^r}$ is a non-jump for $r\ge 3$. Baber and Talbot \cite{BT0} showed that every $α\in[0.2299, 0.2316)\cup [0.2871, \frac{8}{27})$ is a jump for $r=3$. Pikhurko \cite{Pikhurko2} showed that the set of all possible Turán densities of $r$-uniform graphs has cardinality of the continuum for $r\ge 3$. However, whether $\frac{r!}{r^r}$ is a jump for $r\ge 3$ remains open, and $\frac{5r!}{2r^r}$ has remained the known smallest non-jump for $r\ge 3$. In this paper, we give a smaller non-jump by showing that ${54r!\over 25r^r}$ is a non-jump for $r\ge 3$. Furthermore, we give infinitely many irrational non-jumps for every $r\ge 3$.

We say that a graph $F$ can be embedded into a graph $G$ if $G$ contains an isomorphic copy of $F$ as a subgraph. Guo and Volkmann \cite{GV} conjectured that if $G$ is a connected graph with at least $n$ vertices and minimum degree at least $n-3$, then any tree with $n$ vertices and maximum degree at most $n-4$ can be embedded into $G$. In this paper, we give a result slightly stronger than this conjecture and obtain a sufficient and necessary condition that a tree with $n$ vertices and maximum degree at most $n-3$ can be embedded into a connected graph G with at least $n$ vertices and minimum degree at least $n-3$. Our result implies that the conjecture of Guo and Volkmann is true with one exception. We also give an application to the Ramsey number of a tree versus a star.

Given bipartite graphs $H_1$, \dots , $H_k$, the bipartite Ramsey number $br(H_1,\dots, H_k)$ is the minimum integer $N$ such that any $k$-edge-coloring of complete bipartite graph $K_{N, N}$ contains a monochromatic $H_i$ in color $i$ for $1\le i\le k$. There are considerable results on asymptotic values of bipartite Ramsey numbers of cycles. For exact value, Zhang-Sun \cite{Zhangs} determined $br(C_4, C_{2n})$, Zhang-Sun-Wu \cite{Zhangsw} determined $br(C_6, C_{2n})$, and Gholami-Rowshan \cite{GR} determined $br(C_8, C_{2n})$. In this paper, we solve all remaining cases and give the exact values of $br(C_{2n}, C_{2m})$ for all $n\ge m\ge 5$, this answers a question concerned by Bucić-Letzter-Sudakov \cite{BLS}, Gholami-Rowshan \cite{GR}, Zhang-Sun \cite{Zhangs}, and Zhang-Sun-Wu \cite{Zhangsw}.

The Lagrangian density of an $r$-uniform hypergraph $F$ is $r!$ multiplying the supremum of the Lagrangians of all $F$-free $r$-uniform hypergraphs. For an $r$-graph $H$ with $t$ vertices, it is clear that $π_λ(H)\ge r!λ{(K_{t-1}^r)}$. We say that an $r$-unform hypergraph $H$ with $t$ vertices is perfect if $π_λ(H)= r!λ{(K_{t-1}^r)}$. A theorem of Motzkin-Straus implies that all $2$-uniform graphs are perfect. It is interesting to explore what kind of hypergraphs are perfect. A hypergraph is linear if any 2 edges have at most 1 vertex in common. We propose the following conjecture: (1) For $r\ge 3$, there exists $n$ such that a linear $r$-unofrm hypergraph with at least $n$ vertices is perfect. (2) For $r\ge 3$, there exists $n$ such that if $G, H$ are perfect $r$-uniform hypergraphs with at least $n$ vertices, then $G\bigsqcup H$ is perfect. Regarding this conjecture, we obtain a partial result: Let $S_{2,t}=\{123,124,125,126,...,12(t+2)\}$. (An earlier result of Sidorenko states that $S_{2,t}$ is perfect \cite{Sidorenko-89}.) Let $H$ be a perfect $3$-graph with $s$ vertices. Then $F=S_{2,t}\bigsqcup H$ is perfect if $s\geq 3$ and $t\geq 3$.