Wiki
Wiki

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

Updated


Statement

The paper writes G(n)G^{(n)} for a graph on nn vertices and Gl(n)(k)G^{(n)}_l(k) for one with nn vertices and ll edges every vertex of which has valency ≥k\ge k (p. 227). An open Hamilton line is a path through all nn vertices. The paper introduces the result by saying that the argument of Pósa's paper gives it (p. 228).

Theorem (p. 228, quoted). "Let G(n)G^{(n)} be a graph and assume that for every 1≤k<(n−1)/21\le k<(n-1)/2 G(n)G^{(n)} has at most kk vertices of valency ≤k\le k. Then G(n)G^{(n)} has an open Hamilton line. The theorem is best possible."

Edge-count form (p. 228). Using this theorem and the argument of the Theorem of p. 227, the paper obtains that every Gμk(n)(k)G^{(n)}_{\mu_k}(k) has an open Hamilton line, where

μk=1+max⁡k≤t<n−12[(n−t−12)+t(t+1)],\mu_k=1+\max_{k\le t<\frac{n-1}2}\Bigl[\binom{n-t-1}2+t(t+1)\Bigr],

and it states that this result is best possible. The paper does not restate the range of kk for this form, and it gives no extremal graph for either statement.

Source. P. Erdős, Remarks on a paper of Pósa, Magyar Tud. Akad. Mat. Kutató Int. Közl. 7 (1962), 227--229 (received August 2, 1962); both statements on printed p. 228. The edition read is identified in the source digest.

Read depth. Claims checked: the theorem and the edge-count form, with the definition of μk\mu_k, were read clause by clause on the page image of p. 228. The paper gives no proof of either (see below). Nothing here is independently reviewed.

Proof pointer

None in the paper: the theorem's proof "can be left to the reader of Pósa's paper" (p. 228), and the edge-count form is said to follow by the same argument as the Theorem of p. 227, which counts the edges at and away from tt vertices of small valency.

Dependencies

Pósa's theorem (the paper's [3]: L. Pósa, A theorem concerning Hamilton lines, Publications of the Math. Inst. 7 (1962) A. 225--226, as the paper cites it) and the argument of the Theorem of p. 227.

Bears on

No problem page of this corpus. Problem 1012 concerns long cycles, not Hamilton paths, and its page uses only the Theorem of p. 227.