Wiki
Wiki

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

Updated


Claim. There is an absolute constant c>0c>0 such that for G=G(n,1/2)G=G(n,1/2), with high probability,

τ(G)≤n−(2+2c)log⁡2n≤n−(1+c)α(G).\tau(G)\le n-(2+2c)\log_2n\le n-(1+c)\alpha(G).

This is Theorem 1.1 of N. Alon, T. Bohman and H. Huang, More on the bipartite decomposition of random graphs, J. Graph Theory 84 (2017), no. 1, 45--52, first posted as arXiv:1409.6165 on 2014-09-22; the corpus states it on its result page. The second inequality uses α(G)=(2+o(1))log⁡2n\alpha(G)=(2+o(1))\log_2n whp. Since α(G)→∞\alpha(G)\to\infty, the bound puts τ(G)\tau(G) strictly below n−α(G)n-\alpha(G) with high probability for every nn, so the statement of Problem 807 fails with probability tending to 11, not only for the most nn covered by Alon's Theorem 1.1. The paper also gives, with a short proof, the weaker τ(G)≤n−α(G)−Ω(log⁡log⁡n)\tau(G)\le n-\alpha(G)-\Omega(\log\log n) whp (its inequality (1)). The proof of Theorem 1.1 applies the second moment method to the number of induced copies of members of a family of kk-vertex bipartite graphs whose bipartition number is at most 0.01k0.01k, for kk slightly above 2log⁡2n2\log_2n. The paper notes that the method cannot reach n−2α(G)n-2\alpha(G) and asks whether τ(G)=n−O(α(G))\tau(G)=n-O(\alpha(G)) whp; the typical value of τ(G(n,1/2))\tau(G(n,1/2)) is not asked by the problem and remains open.

Depends on. Theorem 1.1 of the paper, the library's result page; the result is otherwise self-contained.

Acceptance. refereed: the Journal of Graph Theory is a refereed journal, and the Crossref record of the DOI gives volume 84 (2017), no. 1, pages 45--52, published online 22 February 2016, the paper link's date. reviewed: the curator of erdosproblems.com, T. F. Bloom, labels the problem DISPROVED and records in its commentary that this paper proves τ(G)≤n−(1+c)α(G)\tau(G)\le n-(1+c)\alpha(G) almost surely for an absolute constant c>0c>0 (the site's page as of 2026-09-18, with an empty thread and an empty proof-claim tab); the site's label is the discussion link. The arXiv version, the only one, is cited, on its source card; Theorem 1.1 and inequality (1) are checked statements (p. 2), the proof (Section 3, pp. 3--6) is not checked, and the journal text is not held.

Formalization. Boris Alexeev's repository plby/lean-proofs holds, at its commit of 15 September 2026, the file src/latest/ErdosProblems/Erdos807.lean and the repository's notes page for the problem. The file's header declares it a formalization of a solution to Problem 807, with Noga Alon, Tom Bohman and Hao Huang as informal authors and Codex and GPT-5.6 Sol as formal authors. Its theorem alon_bohman_huang states Theorem 1.1, both inequalities for some c>0c>0, and its proof takes c=1/1000c=1/1000. Its theorem not_erdos_807, also named erdos_807, is the negation of the statement that τ(G)=n−α(G)\tau(G)=n-\alpha(G) with high probability for G(n,1/2)G(n,1/2), proved from the same bound. No formal-conjectures statement file existed for the problem on 2026-10-07. The corpus has not built or audited the development, so the page lists no formalized evidence.