Wiki
Wiki

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

Updated


Statement

H(k)H(k) is the minimal diameter hk−h1h_k-h_1 of an admissible kk-tuple, a tuple of kk increasing integers avoiding at least one residue class modulo every prime (pp. 1, 9). In the section "Narrow admissible tuples", under "Sieving methods" (p. 78): the sieve of Eratosthenes, which sieves [2,x][2,x] by the class 0(modp)0\pmod p for all p≤kp\le k and takes the survivors pm+1,…,pm+kp_{m+1},\ldots,p_{m+k} with m=π(k)m=\pi(k), "yields the upper bound"

H(k)≤klog⁡k+klog⁡log⁡k−k+o(k)(149)H(k)\le k\log k+k\log\log k-k+o(k) \qquad (149)

by the prime number theorem in the forms pk=klog⁡k+klog⁡log⁡k−k+O(klog⁡log⁡k/log⁡k)p_k=k\log k+k\log\log k-k+O(k\log\log k/\log k) and π(x)=x/log⁡x+O(x/log⁡2x)\pi(x)=x/\log x+O(x/\log^2x). Hensley and Richards [44--46] improved (149) by sieving the symmetric interval [−x/2,x/2][-x/2,x/2] in place of [2,x][2,x], which gives admissible kk-tuples made of −1-1, 11, the primes pm+1,…,pm+⌊(k+1)/2⌋−1p_{m+1},\ldots,p_{m+\lfloor(k+1)/2\rfloor-1} and the negatives of pm+1,…,pm+⌊k/2⌋−1p_{m+1},\ldots,p_{m+\lfloor k/2\rfloor-1}, again with mm as small as possible (the printed display of this tuple omits the minus sign of −pm+1-p_{m+1}); "It follows from Lemma 5 of [45] that one can take m=o(k/log⁡k)m=o(k/\log k), leading to the improved upper bound"

H(k)≤klog⁡k+klog⁡log⁡k−(1+log⁡2)k+o(k).(150)H(k)\le k\log k+k\log\log k-(1+\log2)k+o(k). \qquad (150)

Theorem 17 (vi), (xi) (p. 10) states the Eratosthenes bound as a theorem with an effective o(k)o(k); the paper then describes the shifted Schinzel and shifted greedy sieves (pp. 78--79) as further numerical improvements and (p. 79) conjectures klog⁡k+kk\log k+k as an upper bound for all large kk.

Source. D. H. J. Polymath, Variants of the Selberg sieve, and bounded intervals containing many primes, Res. Math. Sci. 1 (2014), Art. 12; displays (149)--(150) on p. 78 of the journal PDF, read in the text layer; Theorem 17 on p. 10. The arXiv version (1407.4897) was not compared; its section numbering differs from the journal's unnumbered headings.

Read depth. Claims checked: the two displays, the sentences around them and Theorem 17 (vi), (xi) were read clause by clause in the text layer. The displays come with a one-sentence justification each, and the derivation of (150) from Lemma 5 of Hensley and Richards is not carried out in the paper; nothing was checked beyond the statements.

Proof pointer

Page 78: (149) from the prime number theorem applied to pπ(k)+kp_{\pi(k)+k}; (150) from the symmetric sieve of [−x/2,x/2][-x/2,x/2], whose tuple is drawn from ±1\pm1 and the ±pj\pm p_j with j>mj>m, where m=o(k/log⁡k)m=o(k/\log k) suffices, which the paper attributes to Lemma 5 of Hensley and Richards 1974 (that paper's Theorem gives the same gain in the form ρ∗(x)−π(x)≥(log⁡2−ε)x/(log⁡x)2\rho^*(x)-\pi(x)\ge(\log2-\varepsilon)x/(\log x)^2).

Dependencies

The prime number theorem with the stated error terms; Hensley and Richards's Lemma 5 (their Acta Arithmetica paper, held). References [44]--[46] of the paper are Hensley and Richards 1973 (the symposium paper), Hensley and Richards 1974 (Acta Arith. 25) and Richards 1974 (Bull. Amer. Math. Soc. 80).

Bears on

  • Problem 1204: H(k)H(k) is the problem's A(k)A(k), and (150) is the second-order improvement of the upper bound that the site attributes to Hensley and Richards.