Wiki
Wiki

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

Updated


Source. Lemma 2, p. 69, proved on pp. 69--70, of Paul Erdős and Aleksandar Ivić, The distribution of values of a certain class of arithmetic functions at consecutive integers, Number Theory (Budapest, 1987), Colloq. Math. Soc. János Bolyai 51, North-Holland, Amsterdam (1990), 45--91, as identified on the source card. The paper attributes the lemma to A. Schinzel and thanks him for an unpublished result (p. 51).

Statement

Notation (pp. 45--46): P(k)P(k) is the number of unrestricted partitions of kk; ω(n)\omega(n) is the number of distinct prime factors of nn.

Lemma 2 (A. Schinzel; p. 69, quoted). "If ω(n)\omega(n) is the number of distinct prime factors of nn, then"

lim⁡n→∞ω(∏j=1nP(j))=∞.(4.8)\lim_{n\to\infty}\omega\Bigl(\prod_{j=1}^{n}P(j)\Bigr)=\infty. \qquad(4.8)

Since ω(∏j≤nP(j))\omega\bigl(\prod_{j\le n}P(j)\bigr) is nondecreasing in nn, the lemma says that infinitely many primes divide some value P(j)P(j). It gives no rate of growth.

Proof pointer

Pp. 69--70. The proof uses an asymptotic formula for P(n)P(n), displayed as (4.9) and cited to M. Knopp, Modular functions in analytic number theory (1970), p. 90, with main term 143 eaλn(n−1/24)−1(1−1aλn)\frac{1}{4\sqrt3}\,e^{a\lambda_n}(n-1/24)^{-1}\bigl(1-\frac{1}{a\lambda_n}\bigr), where λn=(n−1/24)1/2\lambda_n=(n-1/24)^{1/2} and a=π(2/3)1/2a=\pi(2/3)^{1/2}. Supposing that finitely many primes q1,…,qrq_1,\ldots,q_r account for the prime factors of every P(n)P(n), n≥2n\ge2, it invokes R. Tijdeman's theorem (reference [30], On integers with many small prime factors, Compositio Math. 26 (1973), 319--330): there is a constant C<0C<0 such that two integers AA, BB composed only of q1,…,qrq_1,\ldots,q_r with ∣A−B∣≤Alog⁡CA|A-B|\le A\log^CA are equal. It takes two sets of positive integers a1,…,aka_1,\ldots,a_k and b1,…,bkb_1,\ldots,b_k whose mmth power sums agree for m=1,…,[c/2]m=1,\ldots,[c/2] but not for m=[c/2]+1m=[c/2]+1, and from (4.9) and Taylor expansion obtains that A=∏jP(n+aj)A=\prod_jP(n+a_j) and B=∏jP(n+bj)B=\prod_jP(n+b_j) have ratio 1+O(n−1/2−[c/2])1+O(n^{-1/2-[c/2]}) (4.10) and also 1+Ω(n−1/2−[c/2])1+\Omega(n^{-1/2-[c/2]}) (4.11). The first gives A−B≪A(log⁡A)−2[c/2]−1A-B\ll A(\log A)^{-2[c/2]-1}, hence A=BA=B by Tijdeman's theorem, against the second. The print names Tijdeman's constant CC and then works with [c/2][c/2] without defining cc separately; the existence of the two sets of integers is asserted without proof or reference (both observations of this page).

Dependencies

The asymptotic formula (4.9) for P(n)P(n) and Tijdeman's theorem, both cited. The lemma is used for Theorem 4. Read depth: claims checked; the statement was read clause by clause on p. 69 and the proof for its structure on pp. 69--70.

Bears on

  • Problem 1106: with the problem's p(n)p(n) written P(n)P(n) here, the lemma states that F(n)F(n), the number of distinct prime factors of p(1)p(2)⋯p(n)p(1)p(2)\cdots p(n), tends to infinity, which is the problem's first question. It gives no rate and does not address the second question, whether F(n)>nF(n)>n for all large nn.