Wiki
Wiki

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

Updated


Claim. There are positive constants cc and α\alpha such that for every sufficiently large nn there is a graph GG on nn vertices with maximum degree 33 and

r^(G)≥cn(log⁡2n)α,\hat r(G)\ge cn(\log_2n)^\alpha,

where r^(G)\hat r(G) is the least number of edges of a graph HH every 22-coloring of whose edges contains a monochromatic copy of GG (the problem's R^(G)\hat R(G)). The proof fixes c=110c=\frac1{10} and α=160\alpha=\frac1{60}. Since (log⁡2n)α→∞(\log_2n)^\alpha\to\infty, no constant c(3)c(3) satisfies r^(G)≤c(3) n\hat r(G)\le c(3)\,n along this family, which is the negation of the statement at d=3d=3 (adding a disjoint star K1,dK_{1,d}, which cannot lower r^\hat r, gives the failure for every d≥3d\ge3 read as an exact maximum degree: an elementary remark of this corpus, not a statement of the paper). The paper poses the question as Beck's and answers it in the negative. The graph is a disjoint union of pairwise nonisomorphic graphs, each a binary tree on 2m2m leaves closed by a cycle through the leaves, with mm of order log⁡2n/log⁡2log⁡2n\log_2n/\log_2\log_2n; the lower bound comes from the fact that no graph with nlnl edges is Ramsey for GG, where l=110n1/(15m)≥110(log⁡2n)1/60l=\frac1{10}n^{1/(15m)}\ge\frac1{10}(\log_2n)^{1/60}. The theorem is paged at Theorem 1 of the library's source card. The same question is settled again, with a larger lower bound, on the page Tikhomirov 2022.

Acceptance. Reviewed: the site's curator, T. F. Bloom, labels the problem DISPROVED and credits Rödl and Szemerédi with the disproof for d=3d=3 in the problem's commentary (page last edited 18 January 2026, accessed 2026-09-17); the curator is independent of the authors. Refereed: On size Ramsey numbers of graphs with bounded degree, Combinatorica 20 (2000), no. 2, 257--262, received 21 December 1998 and published in the February 2000 issue (the Crossref record), the date this page is named by. Tikhomirov (2022), Conlon, Nenadov and Trujić (2022) and Draganić and Petrova (2022) each cite the paper as the negative answer to Beck's question. Formalization: the Lean 4 file pinned above, in Boris Alexeev's repository, declares itself a formalization of a solution to this problem, names Rödl and Szemerédi as the informal authors and Codex and GPT-5.6 Sol as the formal authors, and proves the negative answer at degree three by a deterministic finite version of the construction; its first commit is dated 17 August 2026, and the formal-conjectures statement file ErdosProblems/559.lean (added 20 September 2026) points to it as the formal proof of erdos_559 and of the degree-three variant. It is a third party's formalization of this claim, so it is a link on this page and not a page of its own; this corpus has not built it or audited its statement, so formalized is not listed, and the claim is accepted on the curator's credit and the refereed publication alone.

Read depth. Claims checked: Theorem 1, the constants fixed in its proof and the Fact of p. 259 were read (pp. 258--259); the proof (pp. 259--261) was read for its structure and its estimates were not checked. The theorem states ∣V∣=n|V|=n while its construction gives a graph on at most nn vertices, a discrepancy noted on the result page; it does not affect the disproof. Of the Lean file, the header and the statement of its main theorem were read, not the proof. Nothing is independently reviewed in this corpus.

Depends on. Nothing in this wiki; the result is the paper's own theorem.