Wiki
Wiki

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

Updated


Statement

Setting. B(x)B(x) and D(x)D(x) are as in Theorem I and Theorem II, and c1c_1 is an absolute positive constant.

Theorem III (p. 258). For all sufficiently large values of xx,

D(x)−B(x)>c1log⁡log⁡log⁡x.D(x)-B(x)>c_1\log\log\log x.

The paper also notes (p. 258 and its footnote) that infinitely many B-numbers are not D-numbers and infinitely many D-numbers are not B-numbers.

Source. P. Erdős and L. Mirsky, The distribution of values of the divisor function d(n)d(n), Proc. London Math. Soc. (3) 2 (1952), 257--271; Theorem III on p. 258, its proof in §7, pp. 264--265. The copy read is identified on the source card.

Read depth. Claims checked: the statement was read clause by clause on the page images and the proof was read; its estimates were not re-derived. Nothing here is independently reviewed.

Proof pointer

§7, pp. 264--265. Each value of dd has one D-number mm and one B-number m∗≥mm^*\ge m (p. 259), so D(x)−B(x)D(x)-B(x) counts the D-numbers m≤xm\le x with m∗>xm^*>x. Take tt with p1⋯pt≤x<p1⋯pt+1p_1\cdots p_t\le x<p_1\cdots p_{t+1} and rr with 22r−1<pt<22r+1−12^{2^r-1}<p_t<2^{2^{r+1}-1}. For 3≤ν≤r3\le\nu\le r the integer 22ν−1p2⋯pt−12^{2^\nu-1}p_2\cdots p_{t-1} is below xx and has 2t+ν−22^{t+\nu-2} divisors, so the D-number mνm_\nu with that divisor count is at most xx, while the B-number with that count is p1⋯pt+ν−2>xp_1\cdots p_{t+\nu-2}>x. This gives at least r−2r-2 such D-numbers, and r−2>c1log⁡log⁡log⁡xr-2>c_1\log\log\log x.

Dependencies

The correspondence between D-numbers and B-numbers set up on p. 259.

Bears on

No Erdős problem in the corpus; the theorem compares the two counts of Theorem IV.