Wiki
Wiki

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

Updated


Statement

Theorem 2 (p. 4, quoted). "For integers m≥2m\ge2, define

γm=lim inf⁡x→∞inf⁡ℓ1x∑n≤xn≡am (mod ℓ)1,andγm′=lim inf⁡x→∞inf⁡ℓ1log⁡x∑n≤xn≡am (mod ℓ)1n.\gamma_m=\liminf_{x\to\infty}\inf_\ell\frac1x \sum_{\substack{n\le x\\ n\equiv a^m\ (\mathrm{mod}\ \ell)}}1, \qquad\text{and}\qquad \gamma_m'=\liminf_{x\to\infty}\inf_\ell\frac1{\log x} \sum_{\substack{n\le x\\ n\equiv a^m\ (\mathrm{mod}\ \ell)}}\frac1n.

Then γ2=δ0\gamma_2=\delta_0, γ2′=1/2\gamma_2'=1/2, and for m≥3m\ge3,

0<γm≤ρ(m)(=1mm+o(m))<12m−1≤γm′≤min⁡β≥01eβ∑k=0∞βkm(km)!(∼1em/e).0<\gamma_m\le\rho(m)\left(=\frac1{m^{m+o(m)}}\right)<\frac1{2^{m-1}} \le\gamma_m'\le\min_{\beta\ge0}\frac1{e^\beta}\sum_{k=0}^\infty \frac{\beta^{km}}{(km)!}\left(\sim\frac1{e^{m/e}}\right).

Here ρ(u)\rho(u) is the Dickman-de Bruijn function, defined by ρ(u)=1\rho(u)=1 for 0≤u≤10\le u\le1, and uρ′(u)=−ρ(u−1)u\rho'(u)=-\rho(u-1) for all u≥1u\ge1."

The sums count the integers n≤xn\le x that are mm-th power residues mod ℓ\ell; the sentence introducing the theorem (p. 4) speaks of a prime modulus ℓ\ell. The constant δ0=0.171500…\delta_0=0.171500\ldots is the quadratic residue constant of Corollary 1. The paper adds (p. 4) that the exact values of γm\gamma_m and γm′\gamma_m' are unknown for every m≥3m\ge3, and reports the numerical bounds γ3′≤0.3245\gamma_3'\le0.3245, γ4′≤0.2187\gamma_4'\le0.2187, γ5′≤0.14792\gamma_5'\le0.14792 and γ6′≤0.1003\gamma_6'\le0.1003 from minimizing over β\beta. In words: for each m≥2m\ge2 there is πm>0\pi_m>0 such that, for xx sufficiently large and all primes pp, more than πm%\pi_m\% of the integers up to xx are mm-th power residues mod pp.

Source. Andrew Granville and K. Soundararajan, The spectrum of multiplicative functions, Ann. of Math. (2) 153 (2001), no. 2, 407--470; read as arXiv:math/9909190v1 (8 September 1999), printed page == PDF page: Theorem 2 and the remarks after it on p. 4, Section 2 on pp. 13--18. The published pagination differs and was not compared. The edition read is identified on the source card.

Read depth. Claims checked: the statement and the remarks after it were read clause by clause on the page image. The proof was not checked.

Proof pointer

Section 2 (pp. 13--18). The paper says (p. 13) that γ2=δ0\gamma_2=\delta_0 is clear from the introduction, where it follows from Corollary 1. The upper bound γm≤ρ(m)\gamma_m\le\rho(m) takes, by the Chebotarev density theorem, a prime ℓ≡1(modm)\ell\equiv1\pmod m with a character of order mm equal to 11 on primes up to x1/mx^{1/m} and to e2πi/me^{2\pi i/m} on the primes from there to xx, so that the mm-th power residues up to xx are the x1/mx^{1/m}-smooth integers (p. 13). Positivity of γm\gamma_m iterates Proposition 2.2 (p. 14), which rests on Hall's Lemma 1′' and Hildebrand's Lemma 2.1 (p. 13), and is completed on p. 15. Section 2b (pp. 15--18) treats the logarithmic proportions, the lower bound coming from the zero-sum count of Lemma 2.3 (p. 16) through Corollary 2.5 (p. 17).

Dependencies

Corollary 1 for γ2=δ0\gamma_2=\delta_0; Lemma 1′' (Hall, p. 5), Lemma 2.1 (Hildebrand, p. 13), Proposition 2.2 (p. 14), Lemmas 2.3 and 2.4 and Corollary 2.5 (pp. 16--17) of the same paper.

Bears on

No Erdős problem page of the corpus cites this theorem.