Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Statement
Lemma 2 (p. 2). Let be a positive integer and a finite set of positive integers, and put
(the paper's (1)). Then
(the paper's (2)) and
(the paper's (3)).
The lemma is stated for every finite set ; it is not restricted to representations, where .
Source. Max A. Alekseyev, On partitions into squares of distinct integers whose reciprocals sum to 1, in The Mathematics of Various Entertaining Subjects, Volume 3 (2019), pp. 213--221, read in the arXiv version identified on the source card: the lemma and its proof on p. 2, in Section 1 (pp. 2--3).
Read depth. Claims checked: the statement and its hypotheses were read clause by clause on the page image; the short proof was read and its steps followed.
Proof pointer
P. 2. With , the harmonic mean is at most the -th power mean , which rearranges to (2). The lower bound in (3) comes from , and the two upper bounds from combined with (2), and from .
Use in the paper
With the bounds (3), applied to the remaining reciprocal sum and the remaining sum of squares after each chosen element, give the range of the next element in a backtracking search (Algorithm 1, p. 3); the bound (2) makes the search terminate. Run on with , the search finds no representation, which is the paper's Lemma 3 (p. 3), the sharpness half of Theorem 1. The search was not rerun here.
Bears on
- Problem 283: no case of the problem on its own; the lemma bounds the computation behind the paper's Lemma 3, that is not a sum of squares of distinct positive integers whose reciprocals sum to , which makes the threshold of Theorem 1 for exact.