Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Littlewood: the Riemann hypothesis implies M(N)=O(N1/2+ε)M(N)=O(N^{1/2+\varepsilon})M(N)=O(N1/2+ε)

Proved
moebius_summatory_rpow_bound_of_riemannHypothesis

by marozols · Nov 20, 2026 · Mathlib 0df444a (Lean v4.33.1)

analytic-number-theorymertens-functionmoebius-functionriemann-hypothesis

Let μ\muμ be the Möbius function and let

M(N)=∑n=1Nμ(n)(N∈N)M(N)=\sum_{n=1}^{N}\mu(n)\qquad(N\in\mathbb N)M(N)=n=1∑N​μ(n)(N∈N)

be the Mertens function. Assume the Riemann hypothesis in Mathlib's form: every zero sss of ζ\zetaζ with s≠1s\neq 1s=1 that is not a trivial zero −2,−4,−6,…-2,-4,-6,\dots−2,−4,−6,… satisfies Re⁡s=12\operatorname{Re}s=\tfrac12Res=21​. Then for every real ε>0\varepsilon>0ε>0

M(N)=Oε ⁣(N1/2+ε)(N→∞).M(N)=O_\varepsilon\!\left(N^{1/2+\varepsilon}\right)\qquad(N\to\infty).M(N)=Oε​(N1/2+ε)(N→∞).

That is, for each ε>0\varepsilon>0ε>0 there is a constant CεC_\varepsilonCε​, independent of NNN, with ∣M(N)∣≤CεN1/2+ε|M(N)|\le C_\varepsilon N^{1/2+\varepsilon}∣M(N)∣≤Cε​N1/2+ε for all sufficiently large NNN.

This is the classical, unconditionally provable implication "RH ⇒\Rightarrow⇒ Mertens bound" due to Littlewood (1912). It is the forward direction of the criterion in Titchmarsh–Heath-Brown, Theorem 14.25(C). It is a theorem, not a conjecture: the Riemann hypothesis is a hypothesis of the statement.

Classical proof outline. Under RH one has log⁡ζ(s)=O((log⁡t)2−2σ+δ)\log\zeta(s)=O\big((\log t)^{2-2\sigma+\delta}\big)logζ(s)=O((logt)2−2σ+δ) for 12<σ≤1\tfrac12<\sigma\le121​<σ≤1 (Titchmarsh–Heath-Brown §14.2). Consequently

1ζ(σ+it)=Oδ,ε ⁣(∣t∣ε)\frac{1}{\zeta(\sigma+it)}=O_{\delta,\varepsilon}\!\left(|t|^{\varepsilon}\right)ζ(σ+it)1​=Oδ,ε​(∣t∣ε)

uniformly for σ≥12+δ\sigma\ge\tfrac12+\deltaσ≥21​+δ and ∣t∣≥1|t|\ge1∣t∣≥1. Apply the truncated Perron formula to 1/ζ(w)=∑nμ(n)n−w1/\zeta(w)=\sum_n\mu(n)n^{-w}1/ζ(w)=∑n​μ(n)n−w on the line Re⁡w=1+1/log⁡x\operatorname{Re}w=1+1/\log xRew=1+1/logx with height TTT. Then move the contour to Re⁡w=12+ε\operatorname{Re}w=\tfrac12+\varepsilonRew=21​+ε. No poles are crossed, because 1/ζ1/\zeta1/ζ is holomorphic for Re⁡w>12\operatorname{Re}w>\tfrac12Rew>21​ under RH. Bound the shifted integrals using the estimate above and choose T=xT=xT=x. The result is M(x)=O(x1/2+2ε)M(x)=O(x^{1/2+2\varepsilon})M(x)=O(x1/2+2ε).

Formalization Note. The hypothesis is Mathlib's RiemannHypothesis. The conclusion is verbatim the statement of the platform theorem moebius_summatory_rpow_bound: the integer-valued partial sums are cast to C\mathbb CC, the comparison function is the real power N1/2+εN^{1/2+\varepsilon}N1/2+ε, and Asymptotics.IsBigO Filter.atTop expresses the OOO-estimate along N→∞N\to\inftyN→∞ in N\mathbb NN.

Preamble
import Mathlib.NumberTheory.LSeries.RiemannZeta
import Mathlib.NumberTheory.ArithmeticFunction.Moebius
Formal statement
theorem moebius_summatory_rpow_bound_of_riemannHypothesis (hRH : RiemannHypothesis) :
    ∀ ε : ℝ, 0 < ε →
      Asymptotics.IsBigO Filter.atTop
        (fun N : ℕ => ∑ n ∈ Finset.Icc 1 N, (ArithmeticFunction.moebius n : ℂ))
        (fun N : ℕ => (N : ℝ) ^ (1 / 2 + ε)) := by sorry
Source
J. E. Littlewood, Quelques conséquences de l'hypothèse que la fonction ζ(s) de Riemann n'a pas de zéros dans le demi-plan Re(s) > 1/2, C. R. Acad. Sci. Paris 154 (1912), 263–266. E. C. Titchmarsh, The Theory of the Riemann Zeta-function, 2nd ed., revised by D. R. Heath-Brown, Oxford University Press, 1986, §14.25, Theorem 14.25(C), equation (14.25.2), p. 370 (forward direction: RH implies (14.25.2)), using the conditional bounds for log ζ and 1/ζ of §14.2. https://sites.math.rutgers.edu/~zeilberg/EM18/TitchmarshZeta.pdf

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me