Wiki
Wiki

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

Updated


Statement

Theorem (unnumbered, printed p. 8). "If the edges of KnK_n colored [sic] with two colors then for each ll there exist ll paths, each monochromatic in the same color, such that they cover at least n(l+1)l+2\frac{n(l+1)}{l+2} vertices of KnK_n."

The paragraph after it (p. 8) calls the bound essentially best possible: color red the edges inside a set of p=⌊n(l+1)l+2⌋p=\lfloor\frac{n(l+1)}{l+2}\rfloor vertices and every other edge blue. The same kind of coloring with p=⌊2n3⌋p=\lfloor\frac{2n}3\rfloor lets ll vertex-disjoint paths of one color cover only ⌊2n3⌋+l−1\lfloor\frac{2n}3\rfloor+l-1 vertices, and the authors suggest that the theorem may still hold for edge-disjoint paths. Problem 1 (p. 8): "Is the theorem above true for edge disjoint paths?" Corollary 1 (p. 9), "Diagonal case of the path-path Ramsey number established in [2]": "In a coloring of KnK_n there is a monochromatic path of at least ⌊2n3⌋+1\lfloor\frac{2n}3\rfloor+1 vertices."

Source. P. Erdős and A. Gyárfás, Vertex covering with monochromatic paths, Math. Pannon. 6 (1995), no. 1, 7--10 (received October 1994); the copy read is the journal's own PDF of the four printed pages with no usable text layer, printed p. nn being PDF p. n−6n-6. The Theorem, the sharpness paragraph and Problem 1 on printed p. 8 (PDF p. 2), Corollary 1 on printed p. 9 (PDF p. 3), read on the rendered page images. The paper's [2] is Gerencsér and Gyárfás, Ann. Univ. Sci. Budapest. Eötvös Sect. Math. 10 (1967), 167--170.

Read depth. Claims checked: the Theorem, the sharpness paragraph, Problem 1, the Lemma (p. 8) and Corollary 1 were read clause by clause on the page images. The proof (pp. 8--10: the Lemma, Corollary 1, and an induction on ll) was read for its structure and not checked; nothing here is independently reviewed.

Proof pointer

A cut coloring (p. 8) is "a coloring where the endpoints of a maximum monochromatic (say red) path are connected by a red edge". The Lemma (p. 8): if the coloring is not a cut coloring, AA is the vertex set of a maximum monochromatic path, say red, and BB is any vertex set disjoint from AA with ∣B∣<⌈∣A∣/2⌉|B|<\lceil|A|/2\rceil, then some blue path contains all of BB and ∣B∣+2|B|+2 vertices of AA; and when ∣A∣|A| is even and ∣B∣=∣A∣/2|B|=|A|/2, some blue path on 2∣B∣+12|B|+1 vertices contains ∣B∣+1|B|+1 vertices of AA. Corollary 1 follows from the Lemma, and the Theorem is proved by induction on ll starting from Corollary 1 (pp. 9--10): a maximum monochromatic path, say red, with vertex set AA leaves a small complement BB, the Lemma gives a blue path meeting AA in ∣B∣+1|B|+1 vertices, and that set is deleted before the induction hypothesis is applied.

Dependencies

None outside the paper. The induction starts from Corollary 1, the diagonal path Ramsey number of Gerencsér and Gyárfás (1967), reproved here from the Lemma; the l=1l=1 case of the Theorem, a monochromatic path on at least 2n/32n/3 vertices, falls one vertex short of it when 3∣n3\mid n.

Bears on