Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Certified numerical bounds for the twin-prime constant, Euler's constant and the prime logarithmic sum

Proved
PrimePairSieve.sieve_constants_certificate

by andreaskapfer · Sep 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

number-theoryverified-computation

Statement

Let C=S2=2∏p>2(1−1(p−1)2)C=\mathfrak S_2 = 2\prod_{p>2}(1-\tfrac1{(p-1)^2})C=S2​=2∏p>2​(1−(p−1)21​) be the twin-prime singular constant, γ\gammaγ Euler's constant, and

P=∑plog⁡pp(p−1).P=\sum_{p}\frac{\log p}{p(p-1)} .P=p∑​p(p−1)logp​.

The theorem states the certified numerical bounds

1.25 ≤ C ≤ 1.3217,γ ≥ 0.5722,P ≥ 0.74.1.25\ \le\ C\ \le\ 1.3217,\qquad \gamma\ \ge\ 0.5722,\qquad P\ \ge\ 0.74 .1.25 ≤ C ≤ 1.3217,γ ≥ 0.5722,P ≥ 0.74.

Meaning

These are exactly the numerical inputs consumed by the threshold-domination child PrimePairSieve.reciprocal_base_denominator_dominates_threshold: the sharp reciprocal expansion PrimePairSieve.reciprocal_base_denominator_quadratic_expansion_sharp gives

B(z)≥L2+(4γ+4P−log⁡2)L−322C−103z−1/3,L=log⁡z,B(z)\ge \frac{L^2+(4\gamma+4P-\log2)L-\tfrac32}{2C}-\frac{10}{3}z^{-1/3},\qquad L=\log z,B(z)≥2CL2+(4γ+4P−log2)L−23​​−310​z−1/3,L=logz,

and combining the three bounds above with log⁡2≤0.6931471808\log 2\le 0.6931471808log2≤0.6931471808 yields 4γ+4P−log⁡2≥4.55564\gamma+4P-\log2\ge 4.55564γ+4P−log2≥4.5556, which drives the elementary comparison for all z≥40z\ge40z≥40.

Route

All four bounds are certified finite computations, in the style used throughout this project.

  • C≥1.25C\ge1.25C≥1.25. The partial products P(Q)=∏2<p≤Q(1−1(p−1)2)P(Q)=\prod_{2<p\le Q}(1-\tfrac1{(p-1)^2})P(Q)=∏2<p≤Q​(1−(p−1)21​) decrease in QQQ and C=2inf⁡QP(Q)C=2\inf_Q P(Q)C=2infQ​P(Q). For Q≤80Q\le80Q≤80, P(Q)≥P(80)P(Q)\ge P(80)P(Q)≥P(80), an exact rational with P(80)≈0.66163P(80)\approx0.66163P(80)≈0.66163; for Q≥80Q\ge80Q≥80, the primes ppp with 80<p≤Q80<p\le Q80<p≤Q contribute factors indexed by k=p−1∈[81,Q−1]k=p-1\in[81,Q-1]k=p−1∈[81,Q−1], so
P(Q) ≥ P(80)∏k=81Q−1(1−1k2) ≥ P(80)⋅8081 ≥ 0.653.P(Q)\ \ge\ P(80)\prod_{k=81}^{Q-1}\Bigl(1-\frac1{k^2}\Bigr)\ \ge\ P(80)\cdot\frac{80}{81}\ \ge\ 0.653.P(Q) ≥ P(80)k=81∏Q−1​(1−k21​) ≥ P(80)⋅8180​ ≥ 0.653.
  • C≤1.3217C\le1.3217C≤1.3217. C/2=inf⁡QP(Q)≤P(200)C/2=\inf_Q P(Q)\le P(200)C/2=infQ​P(Q)≤P(200), and the exact rational P(200)P(200)P(200) satisfies 2P(200)≤1.321322P(200)\le1.321322P(200)≤1.32132 (the product has only 464646 prime factors).
  • γ≥0.5722\gamma\ge0.5722γ≥0.5722. Mathlib's monotone sequence γn=Hn−log⁡(n+1)\gamma_n=H_n-\log(n+1)γn​=Hn​−log(n+1) satisfies γn<γ\gamma_n<\gammaγn​<γ, so it suffices that H100−log⁡101≥0.5722H_{100}-\log101\ge0.5722H100​−log101≥0.5722. Now H100H_{100}H100​ is an exact rational and the cubic Taylor bound log⁡(1+u)≤u−u22+u33\log(1+u)\le u-\tfrac{u^2}2+\tfrac{u^3}3log(1+u)≤u−2u2​+3u3​ gives log⁡101=log⁡100+log⁡(1+1100)≤4.6151213\log 101=\log100+\log(1+\tfrac1{100})\le4.6151213log101=log100+log(1+1001​)≤4.6151213, while H100≥5.1873775H_{100}\ge5.1873775H100​≥5.1873775; hence γ≥0.572257\gamma\ge0.572257γ≥0.572257.
  • P≥0.74P\ge0.74P≥0.74. The summand is positive, so the tsum is at least any finite partial sum; the sum is summable by comparison with ∑p−3/2\sum p^{-3/2}∑p−3/2. For primes p≤1000p\le1000p≤1000 one bounds log⁡p=klog⁡2+log⁡(1+u)\log p=k\log2+\log(1+u)logp=klog2+log(1+u) with u=p/2k−1≤12u=p/2^k-1\le\tfrac12u=p/2k−1≤21​ and the alternating Taylor expansion log⁡(1+u)≥u−u22+u33−u44\log(1+u)\ge u-\tfrac{u^2}2+\tfrac{u^3}3-\tfrac{u^4}4log(1+u)≥u−2u2​+3u3​−4u4​; the resulting rational lower bound for the partial sum over the 168168168 primes up to 100010001000 exceeds 0.7460.7460.746, while the tail beyond 100010001000 is positive.

Reference

Numerical certificate feeding the endgame comparison for TaoFivePrimes.siebert_prime_pair_bound; constants as in H. Riesel and R. C. Vaughan, On sums of primes, Ark. Mat. 21 (1983), 46--74.

Preamble
import Definitions.Def_TaoFivePrimes_SingularSeries
import Mathlib.Data.Nat.Prime.Basic
import Mathlib.NumberTheory.Harmonic.EulerMascheroni
import Mathlib.Topology.Algebra.InfiniteSum.Real

open scoped BigOperators
Formal statement
namespace PrimePairSieve

theorem sieve_constants_certificate :
    (1.25 : ℝ) ≤ TaoFivePrimes.twinPrimeConstant ∧
      TaoFivePrimes.twinPrimeConstant ≤ 1.3217 ∧
      (0.5722 : ℝ) ≤ Real.eulerMascheroniConstant ∧
      (0.74 : ℝ) ≤
        ∑' p : Nat.Primes,
          Real.log (p : ℝ) / ((p : ℝ) * ((p : ℝ) - 1)) := by sorry

end PrimePairSieve
Source
Numerical certificate for the endgame constants of TaoFivePrimes.siebert_prime_pair_bound.

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