Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Statement
Here is the function of the Theorem (p. 307) and is the quantity of display (3.2) (p. 310), logarithms to base 2.
Remark (p. 313, unnumbered, closing the paper). The authors state that the lower-bound process of Section 3 can be iterated to give , and after iterations , which in their words "grows up as high as ". They do not carry this out, citing messy details. The remark ends: "It is conceivable that for every and ."
The final sentence is an expectation, not a proved statement, and the iterated bounds are announced without proof. Since and for the of (3.2), the announced bounds are all at most , so the in "as high as " tends to with (an observation made here). The last sentence, for every , is equivalent to .
Source. S. L. G. Choi, J. Komlós and E. Szemerédi, On sum-free subsequences, Trans. Amer. Math. Soc. 212 (1975), 307--313, DOI 10.1090/S0002-9947-1975-0376594-1; the remark on printed p. 313, with (3.2) and footnote 2 on p. 310, read in the journal's printing. The paper and its edition are recorded on the source card.
Read depth. Claims checked: the remark was read clause by clause on the page image. It carries no proof.
Proof pointer
None: the paper gives no proof of the iterated bounds and none is claimed for the last sentence.
Dependencies
The lower-bound argument of the Theorem (Section 3, pp. 310--313), which the remark proposes to iterate.
Bears on
- Problem 790: the last sentence is what the site's commentary renders as the authors' "conjecture that ", the paper's being the problem's . If true it would answer the problem's second displayed question, whether for some , in the negative. The paper proves nothing toward it beyond the Theorem's lower bound.