Wiki
Wiki

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

Updated


Statement

Setting (p. 638). For an odd prime pp,

Rp={0,1,…,12(p−1)},rp=#Rp=12(p+1),θp=log⁡rplog⁡p.R_p=\Big\{0,1,\ldots,\tfrac12(p-1)\Big\},\qquad r_p=\#R_p=\tfrac12(p+1),\qquad \theta_p=\frac{\log r_p}{\log p}.

By Kummer's theorem, p∤(2nn)p\nmid\binom{2n}{n} exactly when every base-pp digit of nn lies in RpR_p.

Lemma 1 (p. 638), quoted: "For each odd prime pp and all real numbers x≥2x\geq2, the number of integers 1≤n≤x1\leq n\leq x with p∤(2nn)p\nmid\binom{2n}{n} is at most pxθppx^{\theta_p}."

For p=2p=2 the paper notes (p. 638) that (2nn)\binom{2n}{n} is always even.

Heuristic that follows (p. 639, unlabeled, not a theorem). Reading Lemma 1 as a probability, at most pxθp−1px^{\theta_p-1} when xx is an integer, that p∤(2nn)p\nmid\binom{2n}{n} for nn random in [1,x][1,x], with θp−1\theta_p-1 greater than −0.37-0.37, −0.32-0.32 and −0.29-0.29 for p=3,5,7p=3,5,7, and assuming the events for different primes independent, the paper expects at least x0.02x^{0.02} integers n≤xn\le x with (2nn)\binom{2n}{n} coprime to 105105. The same heuristic for the four primes 3,5,7,113,5,7,11 suggests only finitely many nn with (2nn)\binom{2n}{n} coprime to 11551155, such as n=3160n=3160; the paper reports that no larger example is known, though the search has gone up to 1010410^{10^4}, citing Mauldin and Ulam.

Context the paper gives (p. 638). The numbers n=1,10,756n=1,10,756 have (2nn)\binom{2n}{n} coprime to 105105; whether there are infinitely many is Graham's problem, with a prize as reported in the paper's references [2, 4]. For m=pqm=pq a product of two odd primes, (2nn)\binom{2n}{n} is coprime to mm for infinitely many nn, a result of Erdős, Graham, Ruzsa and Straus (1975).

Source. Carl Pomerance, Divisors of the middle binomial coefficient, Amer. Math. Monthly 122 (2015), no. 7, 636--644, doi:10.4169/amer.math.monthly.122.7.636: the setting, Lemma 1 and its proof in Section 4 (pp. 638--639), the heuristic on p. 639. The edition read is identified on the source card.

Read depth. Claims checked: the setting, the statement and the proof were read clause by clause on the printed pages. Nothing here is independently reviewed.

Proof pointer

Pages 638--639. With D=⌊1+log⁡x/log⁡p⌋D=\lfloor1+\log x/\log p\rfloor, every n≤xn\le x has at most DD base-pp digits, and restricting each to RpR_p leaves at most rpD<prplog⁡x/log⁡p=pxθpr_p^D<pr_p^{\log x/\log p}=px^{\theta_p} choices.

Dependencies

Kummer's theorem (Section 3, p. 637).

Bears on

  • Problem 376: the problem asks whether infinitely many nn have (2nn)\binom{2n}{n} coprime to 105105. Lemma 1 bounds, for each of p=3,5,7p=3,5,7 separately, how many n≤xn\le x escape divisibility by pp; it gives no lower bound and says nothing about the three primes jointly. The independence heuristic built on it predicts infinitely many such nn but is not a proof, and the paper does not settle the problem.