betaFeedback

WorksMathematics

On an Erdős–Kac-type conjecture of Elliott

Abstract

Elliott and Halberstam proved that ∑p<n2ω(n−p)\sum_{p<n} 2^{ω(n-p)} is asymptotic to ϕ(n)ϕ(n). In analogy to the Erdős–Kac Theorem, Elliott conjectured that if one restricts the summation to primes pp such that ω(n−p)≤2log⁡log⁡n+λ(2log⁡log⁡n)1/2ω(n-p)\le 2 \log \log n+λ(2\log \log n)^{1/2} then the sum will be asymptotic to ϕ(n)∫−∞λe−t2/2dt/2πϕ(n)\int_{-\infty}^λ e^{-t^2/2}dt/\sqrt{2π}. We show that this conjecture follows from the Bombieri–Vinogradov Theorem. We further prove a related result involving Poisson–Dirichlet distribution, employing deeper lying level of distribution results of the primes.

Introduction and motivation

Let ω(m)=∑p∣m1\omega(m)=\sum_{p\mid m}1 be the prime divisor function. Elliott and Halberstam proved that [9]

∑p<n2ω(n−p)=ϕ(n)+O(nlog⁡n(log⁡log⁡n)2)(1)\sum_{p<n}2^{\omega(n-p)}=\phi(n)+O\left(\frac{n}{\log n}(\log\log n)^2\right) \tag*{(1)}

holds for n≥3n\ge3. Here and later pp denotes a prime and ϕ\phi is Euler’s totient function. Elliott gave a talk at the meeting held in Urbana–Champaign, June 5–7, 2014, in memory of Paul and Felice Bateman, and Heini Halberstam, where he revisited his works with Halberstam. A telegraphic representation of the talk was published, which contains the following conjecture [10 Conj. B]: for each real λ\lambda,

∑p<nω(n−p)≤2log⁡log⁡n+λ(2log⁡log⁡n)1/22ω(n−p)→ϕ(n)2π∫−∞λe−u2/2 du(2)\sum_{\substack{p<n\\ \omega(n-p)\le2\log\log n+\lambda(2\log\log n)^{1/2}}}2^{\omega(n-p)} \to\frac{\phi(n)}{\sqrt{2\pi}}\int_{-\infty}^{\lambda}e^{-u^2/2}\,du \tag*{(2)}

as n→∞n\to\infty. In this note we establish the conjecture.

Theorem 1.1. The asymptotic relation (1.2) holds for each real λ\lambda.

Theorem 1.1 can be restated as follows. Let XnX_n be a random variable taking the value m∈{1,2,…,n}m\in\{1,2,\ldots,n\} with probability proportional to 2ω(m)1n−m a prime2^{\omega(m)}\mathbf{1}_{n-m\text{ a prime}}. Then

ω(Xn)−2log⁡log⁡n(2log⁡log⁡n)1/2→dN(0,1),n→∞,(3)\frac{\omega(X_n)-2\log\log n}{(2\log\log n)^{1/2}}\xrightarrow{d}N(0,1),\qquad n\to\infty, \tag*{(3)}

where N(0,1)N(0,1) is the standard gaussian distribution and the arrow indicates convergence in distribution. The proof of Theorem 1.1 will follow from

Proposition 1.2. Fix k,B≥1k,B\ge1. For any 1≤R≤n1/1001\le R\le n^{1/100} and S⊆NS\subseteq\mathbb{N} we have

∑a≤R, a∈Sω(a)≤k(a,n)=1μ2(a)∣P(a divides Xn)−2ω(a)a∣≪k,B(log⁡log⁡n)3+log⁡Rlog⁡n∑a≤R, a∈Sω(a)≤k(a,n)=1μ2(a)a+1log⁡Bn.\sum_{\substack{a\le R,\ a\in S\\ \omega(a)\le k\\ (a,n)=1}}\mu^2(a)\left|\mathbb{P}(a\text{ divides }X_n)-\frac{2^{\omega(a)}}{a}\right| \ll_{k,B}\frac{(\log\log n)^3+\log R}{\log n}\sum_{\substack{a\le R,\ a\in S\\ \omega(a)\le k\\ (a,n)=1}}\frac{\mu^2(a)}{a}+\frac{1}{\log^B n}.

The reduction of Theorem 1.1 to Proposition 1.2 is explained in §3. The proof of Proposition 1.2 is given in §4 and uses the classical tools utilized in [9] (the Brun–Titchmarsh and Bombieri–Vinogradov Theorems). Proposition 1.2 has little content when log⁡R≍log⁡n\log R\asymp\log n. This is remedied by the following proposition, proved using recent work on primes in arithmetic progressions [6].

This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 851318).

Proposition 1.3. There is an absolute constant η>0\eta> 0 with the following property. For every k≥1k \ge1,

∑a≤nω(a)≤k(a,n)=1μ2(a)∣P(a divides Xn)−2ω(a)a+2Daϕ(a)log⁡n∣≪k(log⁡log⁡n)Cklog⁡n(4)\sum_{\substack{a \le n\\ \omega(a) \le k\\ (a,n)=1}} \mu^2(a)\left|\mathbb{P}(a\text{ divides }X_n)-\frac{2^{\omega(a)}}{a}+\frac{2D_a}{\phi(a)\log n}\right| \ll_k \frac{(\log\log n)^{C_k}}{\log n} \tag*{(4)}

where CkC_k is a positive constant depending only on kk,

Da:=∑d∣a∞g(d)log⁡d,g(d):=∏pk∥d(−1)k−1(p−2)(p−1)−k.(5)D_a := \sum_{d\mid a^\infty} g(d)\log d,\qquad g(d):=\prod_{p^k\parallel d}(-1)^{k-1}(p-2)(p-1)^{-k}. \tag*{(5)}

Since ∑d∣a∞∣g(d)∣=2ω(a)−12ω(a)\sum_{d\mid a^\infty}|g(d)|=2^{\omega(a)-1}2^{\omega(a)}, we have

∣Da∣≤∑p∣a∑i≥1log⁡(pi)∑pi∣∣d∣a∞∣g(d)∣≤2ω(a)∑p∣a∑i≥1log⁡(pi)∣g(pi)∣≪2ω(a)log⁡a.(6)|D_a|\le\sum_{p\mid a}\sum_{i\ge1}\log(p^i)\sum_{p^i\mid\mid d\mid a^\infty}|g(d)| \le2^{\omega(a)}\sum_{p\mid a}\sum_{i\ge1}\log(p^i)|g(p^i)|\ll2^{\omega(a)}\log a. \tag*{(6)}

Proposition 1.3 is proved in §5. The lower order term in (4) leads to the following application. Given an integer m>1m>1 we write it as m=∏i=1Ω(m)pi(m)m=\prod_{i=1}^{\Omega(m)}p_i(m) where Ω(m)\Omega(m) is the number of prime factors of mm (multiplicity counted) and p1(m)≥⋯≥pΩ(m)(m)p_1(m)\ge\cdots\ge p_{\Omega(m)}(m). We define

V(m):=(log⁡pi(m)log⁡m)i≥1(7)V(m):=\left(\frac{\log p_i(m)}{\log m}\right)_{i\ge1} \tag*{(7)}

where pi(m):=1p_i(m):=1 for i>Ω(m)i>\Omega(m). This vector sums to 1 and its entries are nonincreasing. Let PD⁡(θ)\operatorname{PD}(\theta) be the Poisson–Dirichlet distribution with parameter θ\theta (see Chapter 1 of [5] for its definition).

Corollary 1.4. If V(Xn)V(X_n) tends in distribution to PD⁡(θ)\operatorname{PD}(\theta) for some θ>0\theta>0 then θ=2\theta=2.

Let X~n\widetilde{X}_n be the random variable taking the value m∈{1,2,…,n}m\in\{1,2,\ldots,n\} with probability proportional to 2ω(m)2^{\omega(m)}. As observed by Elliott [10],

ω(X~n)−2log⁡log⁡n(2log⁡log⁡n)1/2→dN(0,1),n→∞.(8)\frac{\omega(\widetilde{X}_n)-2\log\log n}{(2\log\log n)^{1/2}}\xrightarrow{d}N(0,1),\qquad n\to\infty. \tag*{(8)}

See §6 for a self-contained proof of (8). As shown in [8], V(X~n)V(\widetilde{X}_n) tends to PD⁡(2)\operatorname{PD}(2). We expect Theorem 1.1 and Corollary 1.4 to hold because XnX_n is distributed like X~n\widetilde{X}_n, only with support restricted to shifted primes which are believed to behave in most ways like random integers.

The structure of the paper is as follows. In the next short section we show how Corollary 1.4 follows from Proposition 1.3. Afterwards in §3 we reduce the conjecture of Elliott we prove, Theorem 1.1 to Proposition 1.2. In §4 we show how Proposition 1.2 follows from the classical Bombieri–Vinogradov and Brun–Titchmarsh Theorems. At this point the proof of Theorem 1.1 is complete. In the penultimate §5 we go beyond the Bombieri–Vinogradov range by employing a result in [6], based on spectral theory of automorphic forms, to prove Proposition 1.3. As mentioned in the last paragraph, we conclude the paper with a short proof of (8).

Proof of Corollary 1.4

We now show how Corollary 1.4 follows from Proposition 1.3. Suppose V(Xn)→dPD⁡(θ)V(X_n)\xrightarrow{d}\operatorname{PD}(\theta), and in particular V(Xn)i→dPD⁡(θ)iV(X_n)_i\xrightarrow{d}\operatorname{PD}(\theta)_i for all ii. Since 2ω(n−p)=no(1)2^{\omega(n-p)}=n^{o(1)}, we have P(Xn≤T)≪T/n1+o(1)\mathbb{P}(X_n\le T)\ll T/n^{1+o(1)}, so that log⁡Xn/log⁡n\log X_n/\log n tends to 1 in probability. This implies V~(Xn)i→dPD⁡(θ)i\widetilde{V}(X_n)_i\xrightarrow{d}\operatorname{PD}(\theta)_i for V~(Xn)i:=log⁡pi(Xn)/log⁡n\widetilde{V}(X_n)_i:=\log p_i(X_n)/\log n. By [12 Thm. 2.1],

lim⁡n→∞E∑i=1∞1V~(Xn)i∈[a,b]=θ∫ab(1−t)θ−1t dt\lim_{n\to\infty}\mathbb{E}\sum_{i=1}^{\infty}\mathbf{1}_{\widetilde{V}(X_n)_i\in[a,b]}=\theta\int_a^b\frac{(1-t)^{\theta-1}}{t}\,dt

for all 0<a<b<10<a<b<1 (the sum appears infinite but can be truncated at i≤⌊1/a⌋i\le\lfloor1/a\rfloor since ∑i≥1V~(Xn)i≤1\sum_{i\ge1}\widetilde{V}(X_n)_i\le1). This is simplified as

References

  1. [1]R. Arratia, F. Kochman, and V. S. Miller. Extensions of Billingsley’s Theorem via Multi-Intensities. arXiv preprint arXiv:1401.1555, 2014.arxiv.org/abs/1401.1555
  2. [2]Edgar Sising, Valentin Blomer, and Junxian Li. Uniform Titchmarsh divisor problems. Adv. Math., 393:51, 2021. Id/No 108076.
  3. [3]A. Bharadwaj and B. Rodgers. Large prime factors of well-distributed sequences. arXiv preprint arXiv:2402.11884, 2024.
  4. [4]Patrick Billingsley. On the central limit theorem for the prime divisor function. Am. Math. Mon., 76:132–139, 1969.
  5. [5]Patrick Billingsley. Convergence of probability measures. Wiley Ser. Probab. Stat. Chichester: Wiley, 2nd ed. edition, 1999.
  6. [6]V. Blomer, L. Grimmelt, J. Li, and S. L. Rydin Myerson. Additive problems with almost prime squares. Geom. Funct. Anal., 33(5):1173–1242, 2023.arxiv.org/abs/2111.01601
  7. [7]Rick Durrett. Probability. Theory and examples, volume 49 of Camb. Ser. Stat. Probab. Math. Cambridge: Cambridge University Press, 5th edition edition, 2019.
  8. [8]Dor Elboim and Ofir Gorodetsky. Multiplicative arithmetic functions and the generalized Ewens measure. arXiv preprint arXiv:1909.00601, 2019. To appear in Israel Journal of Mathematics.DOI
  9. [9]P. D. T. A. Elliott and H. Halberstam. Some applications of Bombieri’s theorem. Mathematika, 13:196–203, 1966.DOI
  10. [10]P. D. T. A. Elliott. Additive and multiplicative functions from Nottingham to Urbana-Champaign. Int. J. Number Theory, 11(5):1357–1366, 2015.DOI
  11. [11]Andrew Granville and K. Soundararajan. Sieving and the Erdős-Kac theorem. In Equidistribution in number theory, an introduction, volume 237 of NATO Sci. Ser. II Math. Phys. Chem., pages 15–27. Springer, Dordrecht, 2007.DOI
  12. [12]Kenji Handa. The two-parameter Poisson-Dirichlet point process. Bernoulli, 15(4):1082–1116, 2009.DOI
  13. [13]Hugh L. Montgomery and Robert C. Vaughan. Multiplicative number theory. I. Classical theory, volume 97 of Cambridge Studies in Advanced Mathematics. Cambridge University Press, Cambridge, 2007.

Paper details

vlnc:W_jscc7fx5
Version
v2 · latest
Published
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents