Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Source. Scott D. Hughes, Sums of distinct divisors of factorials, arXiv:2609.10902v1, Lemma 4 (greedy step) with its proof, physical p. 2 of the five-page PDF held by its library source card, Hughes (2026), whose result page is Lemma 4. The statement and proof were read in the canonical conversion beside the PDF and checked against the page image. The lemma is consumed by the Theorem 1 reconstruction.
Standing. Author-recorded reconstruction; not an independent review; it changes no status and assigns no tier. The ratio property of the divisors of is cited by the source from Tenenbaum–Yokota and Yokota and not proved there; the proof given at the end of this page is supplied by the compilation and labeled as such, as is the consequence on termination and the representation of , which extends the source's lemma.
Definitions
For an integer and an integer that does not divide , the bracketing divisors of are the consecutive divisors of : is the largest divisor of below and the smallest above it. Both exist because and .
The greedy expansion of an integer is the sequence defined as follows: if divides the expansion terminates and is its last term; otherwise , where are the bracketing divisors of , and is the divisor chosen at step . A step at which does not divide is nonterminal.
Statement
Let be an integer whose consecutive divisors have ratio at most , and let . If divides (in particular if ), the greedy expansion terminates at this step. Otherwise let be the bracketing divisors of . Here is the natural logarithm, as the source fixes on p. 1. Then
Consequently the divisors chosen at successive nonterminal steps of the greedy expansion strictly decrease, hence are distinct.
Consequence (compilation-supplied). For the greedy expansion of terminates after finitely many steps at some dividing , and is a sum of distinct divisors of . The source's lemma ends at "strictly decreasing, hence distinct"; the termination rule and the counting of the final divisor are stated in the opening of its Section 3 (p. 2), and the representation of is used there without being stated.
Proof
Since are consecutive divisors of , the hypothesis gives , and . Hence
For the second inequality, , because and . On the interval the function vanishes at and has derivative , so there. As , this gives .
For the consequence (the source's proof gives only that the next chosen divisor is below ; the terminal-divisor case and the rest are supplied here): at a nonterminal step the chosen divisor is , and the next remainder satisfies by the first inequality. The divisor used at the next step, whether the chosen divisor or the terminal divisor itself, is therefore below . So the divisors used form a strictly decreasing sequence of positive integers. The remainders are positive integers that strictly decrease, so some divides (at the latest ), and then is a sum of distinct divisors of .
The ratio property for factorials (compilation-supplied proof)
The source recalls from Tenenbaum–Yokota (J. Number Theory 35 (1990), 150–156, proof of Lemma 4) and Yokota (Res. Bull. Hiroshima Inst. Tech. 29 (1995), 25–28, Lemma 2) that consecutive divisors of have ratio at most . Neither paper is held here. The following proof is supplied by the compilation so that the reconstruction does not rest on an unread citation.
Let , let be a divisor of with , and put . It suffices to find a divisor of with . If is even, divides . If is odd, let be a prime factor of ; then is odd, , and . Let be the largest power of below , so that . Since , divides ; and since is odd, . Put . It is an integer, , , and every other prime has the same exponent in as in ; so divides . Finally because .