Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Conlon 2008 hypergraph ramsey numbers
section_6_2: The paper's restatement of Erdős's density threshold function and its two-sided logarithmic bound in the graph case, stated without proof, as printed.
theorem_1_1: The three-color Ramsey number of the complete 3-uniform hypergraph is at least 2^{n^{c log n}}, improving the 2^{cn^2 log^2 n} of Erdős and Hajnal by a stepping-up construction.
theorem_1_2: For fixed s ≥ 4 the off-diagonal 3-uniform Ramsey number r_3(s, n) has logarithm at most ((s−3)/(s−2)! + o(1)) n^{s−2} log n, improving the exponent of the Erdős–Rado bound by a factor n^{s−2}/polylog n.
theorem_1_3: A superexponential lower bound for the off-diagonal 3-uniform Ramsey number, which gives log r_3(4, n)/n → ∞ as Erdős and Hajnal suggested in 1972.
theorem_2_4: The diagonal two-color 3-uniform Ramsey number satisfies log_2 log_2 r_3(k, k) ≤ (2 + o(1))k, improving the Erdős–Rado bound r_3(k, k) ≤ 2^{2^{4k}}; stated without a written proof.
theorem_6_2: Every r-coloring of the k-tuples of an N-set has a subset of size more than (log N)^β with more than a (1 − η) share of its k-sets in one color, so Erdős's F^{(k)}(N, α) is at least a power of log N for every fixed α > 0.
D. Conlon, J. Fox and B. Sudakov, Hypergraph Ramsey numbers. arXiv:0808.3760v1 (27 August 2008), 20 pages; published as J. Amer. Math. Soc. 23 (2010), no. 1, 247--266, DOI 10.1090/S0894-0347-09-00645-6 (published online 18 August 2009; Crossref record read).
Edition read. The copy read for this card is the arXiv v1 of 27 August 2008 (the only arXiv version; abstract page https://arxiv.org/abs/0808.3760 read), with a complete text layer; the folder's year follows that version. Provenance: 259,083 bytes. The published version was not compared; page locators below are the preprint's. The arXiv record names arXiv's non-exclusive distribution license (arXiv:0808.3760), every other right reserved.
Read status: claims checked for the Section 6.2 passage consumed by Problem 563 (the definition of and the display , p. 16, read clause by clause on the page image) and for the statements of Theorems 1.1, 1.2, 1.3, 2.4 and 6.2, each read clause by clause on the page images; the abstract, pp. 2--12 and the rest of Section 6.2 (pp. 16--18) were read on the page images for the statements and proof outlines listed below; no proof was checked.
Results. Theorem 1.1, p. 3 (three-color lower bound ); Theorem 1.2, p. 3 (off-diagonal upper bound on for fixed ); Theorem 1.3, p. 3 (off-diagonal lower bound for ); Theorem 2.4, p. 8 (diagonal upper bound ); Theorem 6.2, p. 16 (almost monochromatic subsets of size , with the consequence for on p. 17); Section 6.2, p. 16 (the definition of and the graph bound). Section 5 (on the Erdős--Hajnal function ), Proposition 2.5 and Section 6.1 are not recorded here.
Contents
- Abstract (p. 1): is the least such that every red-blue coloring of the -tuples of an -set has a red -set or a blue -set. Main results as stated there: for fixed , improving Erdős and Rado (1952) by a factor of in the exponent; for , the first superexponential lower bound for fixed , answering a question of Erdős and Hajnal (1972); , improving Erdős and Hajnal.
- Section 6.2, Discrepancy in hypergraphs (pp. 16--18): the Erdős--Hajnal fact that every two-coloring of the triples of an -set has a set of size with at least triples in one color, and Erdős's remark about ; Theorem 6.2 (p. 16): for and positive integers there is such that every -coloring of the -tuples of an -set has a subset of size with more than -sets in one color. Then the restatement in terms of Erdős's function (page section_6_2): , "introduced by Erdős in [11]" (the paper's [11] is the 1990 chapter, its p. 21), the remark that is essentially the inverse of , and "It is easy to show that for , " (p. 16, no proof). P. 17: the hypergraph bounds for with sufficiently small, the bounds that the conjecture of Erdős, Hajnal and Rado would imply, Erdős's prize question whether changes continuously or in jumps (the site's Problem 161), and the consequence of Theorem 6.2 that for every fixed ; Theorem 6.3 (p. 17): for all positive integers there is such that the -color Ramsey number of the -uniform blow-up , with parts of size , satisfies (the exponent as printed; the proof's first line takes ); Question 6.4 (p. 18), Erdős's question on pairs with and all triples of meeting both in one class, open for two classes and answered negatively for four.
Compiled scope
Read: p. 1 (abstract), pp. 2--12 (Sections 1--4 and the start of Section 5) and pp. 16--20 (Section 6.2 and the references) on the page images. Statements, with proof outlines on the result pages; no proof was checked. The rest of Section 5 and Section 6.1 were not compiled. The paper deduces Theorem 6.2 from Theorem 6.3 (p. 17) only through the remark that the blow-up's edge density tends to .
Bears on.
- #563 (Section 6.2, p. 16: the definition of and the two-sided logarithmic bound for , both stated without proof; the paper's wording of the definition differs from Erdős's and the site's in one word, recorded on the result page; the asymptotic the problem asks for is not addressed).
- #162 (Section 6.2, p. 16: the same display; the site's wording of #162 prints "largest", as the paper does, and corrected it asks #563's question).
- #161 (Section 6.2, p. 17, read on the page image: "Erdős [4] asked (and offered a $500 cash reward) if the change in occurs continuously, or there are jumps? He suspected the only jump occurs at ", with the paper's contribution that for bounded away from Theorem 6.2 gives , a power of ; the paper's reference [4], cited on p. 16 as "the book [4]", is F. Chung and R. Graham, Erdős on Graphs. His Legacy of Unsolved Problems (A K Peters, 1998), as the reference list on p. 18 gives it; the bound, recorded on the Theorem 6.2 page, does not decide whether a jump occurs).
- #564 (context only: Theorem 2.4, p. 8, is an upper bound for the two-color number the problem asks to bound below, and Theorem 1.1, p. 3, is a three-color lower bound; neither gives the lower bound the problem asks for).
No file of this source is held: no license on record permits its redistribution, and the card cites the edition it names above.