Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated

Dubo 2024 ramsey number double star

../

corollary_3: The elementary upper bound on the Ramsey number of the double star with 2m and m leaves, the special case of the paper's Theorem 2.

theorem_2: The paper's main result, an explicit upper bound on the two-color Ramsey number of the double star S(m1,m2) for all positive integers m1, m2 with (sqrt5+1)/2 m2 < m1 < 3m2.


F. Flores Dubó and M. Stein, On the Ramsey number of the double star, Discrete Math. 348 (2025), no. 1, article 114227; DOI 10.1016/j.disc.2024.114227; arXiv:2401.01274.

The copy read for this card is the arXiv preprint 2401.01274v2 (20 April 2024; v1 is of 2 January 2024), seven pages numbered 1--7; the published version was not compared, so the locators below are preprint pages. Source: https://arxiv.org/abs/2401.01274. The arXiv record names arXiv's non-exclusive distribution license (arXiv:2401.01274), every other right reserved.

Read status: claims checked for Theorem 2, Corollary 3 and Lemma 4 (read clause by clause on the page image of p. 3); the proof of Theorem 2 (pp. 4--6) was read for structure; the introduction was read on the page images of pp. 1--3.

Contents

  • Introduction (pp. 1--3): S(m1,m2)S(m_1,m_2) is the double star with mim_i leaves in class ii; RB(T)=max⁡{2t1,t1+2t2}−1R_B(T)=\max\{2t_1,t_1+2t_2\}-1 is the lower bound from the canonical colorings for a tree with bipartition classes t1≥t2≥2t_1\ge t_2\ge2; Burr's belief that R(T)=RB(T)R(T)=R_B(T) unless TT is an odd star, confirmed asymptotically by Haxell, Łuczak and Tingley, who show R(T)≤(1+η)RB(T)R(T)\le(1+\eta)R_B(T) for trees with Δ(T)≤δt1\Delta(T)\le\delta t_1 and t1>t0t_1>t_0, where δ\delta and t0t_0 depend on η>0\eta>0; Grossman, Harary and Klawe's double stars with R>RBR>R_B and their conjecture R≤RB+1R\le R_B+1 for double stars, known for m1≥3m2m_1\ge3m_2 [5] and for m1≤1.699(m2+1)m_1\le1.699(m_2+1) [8], that is, inequality (1) outside the range (2); Norin, Sun and Zhao's lower bounds, which give R(S(2m,m))≥4.2m+o(m)R(S(2m,m))\ge4.2m+o(m) while RB(S(2m,m))=4m+2R_B(S(2m,m))=4m+2, and their Question 1 (is R(S(2m,m))=4.2m+o(m)R(S(2m,m))=4.2m+o(m)?); the remark (p. 3) that Theorem 4.5 of [8] with the fifth invalid pair of its table gives lim⁡m→∞R(S(2m,m))/m≤4.21526\lim_{m\to\infty}R(S(2m,m))/m\le4.21526.
  • Theorem 2 (p. 3): for m1,m2∈N+m_1,m_2\in\mathbb N^+ with 5+12m2<m1<3m2\frac{\sqrt5+1}2m_2<m_1<3m_2, R(S(m1,m2))≤⌈2m12+(m1+m2/2)2+m2/2⌉+1R(S(m_1,m_2))\le\lceil\sqrt{2m_1^2+(m_1+m_2/2)^2}+m_2/2\rceil+1. Corollary 3: R(S(2m,m))≤⌈4.27492m⌉+1R(S(2m,m))\le\lceil4.27492m\rceil+1 for all m∈N+m\in\mathbb N^+ (the abstract rounds the constant to 4.2754.275).
  • Section 2 (pp. 3--4): Lemma 4 (an edge vwvw with d(v)>m1d(v)>m_1, d(w)>m2d(w)>m_2 and ∣N(v)∪N(w)∣≥m1+m2+2|N(v)\cup N(w)|\ge m_1+m_2+2 spans an S(m1,m2)S(m_1,m_2)), Lemma 5 (degree counting in a graph without S(m1,m2)S(m_1,m_2)), Lemma 6 (Lemma 2.3 of [8]: for n≥max⁡{2m1,m1+2m2}+2n\ge\max\{2m_1,m_1+2m_2\}+2 and a two-coloring of KnK_n with no monochromatic S(m1,m2)S(m_1,m_2), one color has all degrees at most m1m_1).
  • Section 3 (pp. 4--6): the proof of Theorem 2 on n=m1+m2+m3+1n=m_1+m_2+m_3+1 vertices with m3=⌈2m12+(m1+m2/2)2−(m1+m2/2)⌉m_3=\lceil\sqrt{2m_1^2+(m_1+m_2/2)^2}-(m_1+m_2/2)\rceil; the acknowledgment records a missing ceiling corrected from an earlier version.

Compiled scope

Pages 1--7 were read on the page images. No proof was checked beyond structure and nothing here is independently reviewed.

Bears on. #549: Corollary 3 is the site's elementary upper bound; Theorem 2 covers the problem's tree S(2k−1,k−1)S(2k-1,k-1) for k≥3k\ge3 and gives R(S(2k−1,k−1))≤(1+572+o(1))kR(S(2k-1,k-1))\le(\frac{1+\sqrt{57}}2+o(1))k, the same constant asymptotically (a specialization made on the Theorem 2 page, not in the paper), an upper bound that neither proves nor refutes the problem's equality; the introduction restates the disproof and the 4.215264.21526 bound and quotes Question 1 of Norin, Sun and Zhao. The paper does not concern the Burr--Erdős conjecture R(T)≤R(K1,n−1)R(T)\le R(K_{1,n-1}) beyond recalling it.

No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.