Wiki
Wiki

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

Updated


Claim. For all sufficiently large nn, every nn-uniform hypergraph with at most 0.7n/ln⁡n 2n0.7\sqrt{n/\ln n}\,2^n edges is 2-colorable (Theorems 2.1 and 3.1, which also give fast randomized algorithms that find such a coloring with high probability, and derandomized parallel versions). In the notation of Problem 901, m(n)>0.7n/ln⁡n 2nm(n)>0.7\sqrt{n/\ln n}\,2^n for large nn, that is, m(n)≫n/log⁡n 2nm(n)\gg\sqrt{n/\log n}\,2^n. The proof refines Beck's recoloring: after a random 2-coloring, the vertices are processed in a random order, and a vertex lying in an edge that was monochromatic under the initial coloring is flipped, with probability p, only if such an edge is still monochromatic when the vertex is reached. Theorem 4.2 gives a local version through the Lovász Local Lemma: for large nn, an nn-uniform hypergraph in which every edge meets at most 0.17n/ln⁡n 2n0.17\sqrt{n/\ln n}\,2^n other edges is 2-colorable. The source card records the theorems.

Covers. The lower bound m(n)>0.7n/ln⁡n 2nm(n)>0.7\sqrt{n/\ln n}\,2^n for large nn, the best lower bound on record, which improved Beck's [[problems/set_systems/E0901/claims/1978_01_01_beck|n1/3−o(1)2nn^{1/3-o(1)}2^n]]. The order of m(n)m(n) stays open: the upper bound is Erdős's [[problems/set_systems/E0901/claims/1964_09_01_erdos|n22n+1n^22^{n+1}]], and the paper's Section 5 shows that Erdős's random construction still gives O(n22n)O(n^22^n) edges for a restricted family of hypergraphs with small pairwise intersections.

Depends on. No page of this wiki.

Acceptance. J. Radhakrishnan and A. Srinivasan, Improved bounds and algorithms for hypergraph 2-coloring, Random Structures Algorithms 16 (2000), no. 1, 4--32, a refereed journal (refereed); a preliminary version appeared in the Proceedings of the 39th Annual Symposium on Foundations of Computer Science (1998), 684--693, the second link. The issue is dated January 2000 without a day, so the page is dated to the first day of that month. The curator of erdosproblems.com, Thomas Bloom, credits the bound to this paper in the problem's commentary, but the site labels the problem OPEN, so that credit is not acceptance of this partial claim.