Wiki
Wiki

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

Updated


Statement

Notation (p. 38): ω(n)\omega(n) is the number of distinct prime factors of nn and P(n)P(n) its greatest prime factor.

Theorem 1 (p. 39). Let 0<ε<10<\varepsilon<1, and let f:R>1→Rf:\mathbb R_{>1}\to\mathbb R be a function with f(x)→∞f(x)\to\infty as x→∞x\to\infty and f(x)/log⁡xf(x)/\log x monotone and non-increasing. Let kk and ll be positive integers with 2≤l≤(log⁡k)/f(k)2\le l\le(\log k)/f(k), where kk exceeds an effectively computable number depending only on ε\varepsilon and ff. Then there are distinct positive integers a1,…,aka_1,\ldots,a_k and distinct non-negative integers b1,…,blb_1,\ldots,b_l with

P(∏i=1k∏j=1l(ai+bj))<((1+ε)log⁡kllog⁡(log⁡kl))l.P\Bigl(\prod_{i=1}^k\prod_{j=1}^l(a_i+b_j)\Bigr) <\Bigl((1+\varepsilon)\frac{\log k}{l}\log\Bigl(\frac{\log k}{l}\Bigr)\Bigr)^l .

Consequence stated by the authors (pp. 38--39). For sets A,BA,B of positive integers with k=∣A∣≥l=∣B∣≥2k=|A|\ge l=|B|\ge2, Győry, Stewart and Tijdeman's bound is ω(∏a∈A,b∈B(a+b))>C2log⁡k\omega\bigl(\prod_{a\in A,b\in B}(a+b)\bigr)>C_2\log k, display (1), and combining it with the prime number theorem the paper obtains P(a+b)>C3log⁡klog⁡log⁡kP(a+b)>C_3\log k\log\log k for some a∈Aa\in A, b∈Bb\in B, display (2), with effectively computable positive constants C2,C3C_2,C_3. The authors state that Theorem 1 shows that for l=2l=2 the right-hand sides of (1) and (2) cannot be replaced by ((1/8)+ε)(log⁡k)2log⁡log⁡k((1/8)+\varepsilon)(\log k)^2\log\log k and ((1/4)+ε)(log⁡klog⁡log⁡k)2((1/4)+\varepsilon)(\log k\log\log k)^2 respectively, for any ε>0\varepsilon>0, and that they cannot be replaced by (log⁡k)l(\log k)^l when l>2log⁡log⁡kl>2\log\log k (p. 39).

Proof pointer

The theorem follows from Lemma 3 (p. 40), which combines Lemma 1 with the Canfield--Erdős--Pomerance lower bound for the count ψ(x,y)\psi(x,y) of yy-smooth integers up to xx (Lemma 2, p. 40). Lemma 3 takes WW to be the integers up to NN whose prime factors are at most t=⌊c((log⁡N)/l)l⌋t=\lfloor c((\log N)/l)^l\rfloor and applies Lemma 1 to obtain many aa with every a+bja+b_j in WW (proof pp. 41--42). The proof of Theorem 1 (p. 42) applies Lemma 3 with c=1c=1, δ=ε/5\delta=\varepsilon/5 and N=⌊exp⁡((1+ε)(log⁡k)log⁡w)⌋N=\lfloor\exp((1+\varepsilon)(\log k)\log w)\rfloor, w=(log⁡k)/lw=(\log k)/l.

Read depth

Claims checked: the statement and the authors' consequences for l=2l=2 were read clause by clause on the page images of the print. The proof was followed for the outline above and is not independently verified.

Dependencies

  • Lemma 1 (p. 39), through Lemma 3 (p. 40).

Source. P. Erdős, C. L. Stewart and R. Tijdeman, Some diophantine equations with many solutions, Compositio Mathematica 66 (1988), 37--56; the edition read is named on the source card.

Bears on

  • Problem 126: the problem concerns the product of a+ba+b over distinct elements a,ba,b of a single set AA, while Theorem 1 concerns the product of ai+bja_i+b_j over two different sets, one of only ll elements. The paper derives no bound for the problem's product from it. The paper does record (p. 38) the Erdős--Turán lower bound ω(∏a,a′∈A(a+a′))>C1log⁡k\omega\bigl(\prod_{a,a'\in A}(a+a')\bigr)>C_1\log k for ∣A∣=k≥2|A|=k\ge2, with the product as printed over all pairs a,a′∈Aa,a'\in A; the problem asks whether the number of prime factors of the product over distinct elements grows faster than log⁡k\log k.