Wiki
Wiki

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 n!n! 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 mm, which extends the source's lemma.

Definitions

For an integer N≥1N\ge1 and an integer 1≤R≤N1\le R\le N that does not divide NN, the bracketing divisors of RR are the consecutive divisors d<R<bd<R<b of NN: dd is the largest divisor of NN below RR and bb the smallest above it. Both exist because 1∣N1\mid N and N∣NN\mid N.

The greedy expansion of an integer 1≤m≤N1\le m\le N is the sequence R0=m>R1>R2>⋯R_0=m>R_1>R_2>\cdots defined as follows: if RiR_i divides NN the expansion terminates and RiR_i is its last term; otherwise Ri+1=Ri−diR_{i+1}=R_i-d_i, where di<Ri<bid_i<R_i<b_i are the bracketing divisors of RiR_i, and did_i is the divisor chosen at step ii. A step at which RiR_i does not divide NN is nonterminal.

Statement

Let NN be an integer whose consecutive divisors have ratio at most 22, and let 1≤R≤N1\le R\le N. If RR divides NN (in particular if R=NR=N), the greedy expansion terminates at this step. Otherwise let d<R<bd<R<b be the bracketing divisors of RR. Here log⁡\log is the natural logarithm, as the source fixes on p. 1. Then

R−d<d,R−d≤2Rlog⁡bd.R-d<d,\qquad R-d\le2R\log\frac bd .

Consequently the divisors chosen at successive nonterminal steps of the greedy expansion strictly decrease, hence are distinct.

Consequence (compilation-supplied). For 1≤m≤N1\le m\le N the greedy expansion of mm terminates after finitely many steps at some RkR_k dividing NN, and m=d0+⋯+dk−1+Rkm=d_0+\cdots+d_{k-1}+R_k is a sum of distinct divisors of NN. 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 mm is used there without being stated.

Proof

Since d<bd<b are consecutive divisors of NN, the hypothesis gives b≤2db\le2d, and R<bR<b. Hence

R−d<b−d≤2d−d=d.R-d<b-d\le2d-d=d .

For the second inequality, R−d<b−d=d (b/d−1)≤R (b/d−1)R-d<b-d=d\,(b/d-1)\le R\,(b/d-1), because d<Rd<R and b/d−1≥0b/d-1\ge0. On the interval [1,2][1,2] the function x↦x−1−2log⁡xx\mapsto x-1-2\log x vanishes at x=1x=1 and has derivative 1−2/x≤01-2/x\le0, so x−1≤2log⁡xx-1\le2\log x there. As 1<b/d≤21<b/d\le2, this gives R (b/d−1)≤2Rlog⁡(b/d)R\,(b/d-1)\le2R\log(b/d).

For the consequence (the source's proof gives only that the next chosen divisor is below dd; the terminal-divisor case and the rest are supplied here): at a nonterminal step ii the chosen divisor is did_i, and the next remainder satisfies Ri+1=Ri−di<diR_{i+1}=R_i-d_i<d_i by the first inequality. The divisor used at the next step, whether the chosen divisor di+1<Ri+1d_{i+1}<R_{i+1} or the terminal divisor Ri+1R_{i+1} itself, is therefore below did_i. So the divisors used form a strictly decreasing sequence of positive integers. The remainders are positive integers that strictly decrease, so some RiR_i divides NN (at the latest Ri=1R_i=1), and then m=d0+⋯+di−1+Rim=d_0+\cdots+d_{i-1}+R_i is a sum of distinct divisors of NN.

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 n!n! have ratio at most 22. 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 n≥2n\ge2, let dd be a divisor of n!n! with d<n!d<n!, and put q=n!/d>1q=n!/d>1. It suffices to find a divisor d′d' of n!n! with d<d′≤2dd<d'\le2d. If qq is even, d′=2dd'=2d divides n!n!. If qq is odd, let pp be a prime factor of qq; then pp is odd, p≤np\le n, and vp(d)<vp(n!)v_p(d)<v_p(n!). Let 2c2^c be the largest power of 22 below pp, so that 2c<p<2c+12^c<p<2^{c+1}. Since 2c<p≤n2^c<p\le n, 2c2^c divides n!n!; and since qq is odd, v2(d)=v2(n!)≥cv_2(d)=v_2(n!)\ge c. Put d′=dp/2cd'=dp/2^c. It is an integer, vp(d′)=vp(d)+1≤vp(n!)v_p(d')=v_p(d)+1\le v_p(n!), v2(d′)=v2(d)−c≥0v_2(d')=v_2(d)-c\ge0, and every other prime has the same exponent in d′d' as in dd; so d′d' divides n!n!. Finally d<d′<2dd<d'<2d because 1<p/2c<21<p/2^c<2.