Robin oscillation under failure of RH (Lagarias Proposition 3.2)
OpenLeanEval.NumberTheory.lagarias_robin_oscillationAssume that the Riemann hypothesis is false. There exist fixed real constants with and such that arbitrarily large integers satisfy
The same constants work for all cutoffs: for every natural number , such an exists. This is the infinite-set statement of Proposition 3.2 with the finitely many indices below 3 removed, so the denominator uses a positive logarithm and an actual real power. It is a substantial oscillation theorem, not merely the assertion that Robin's inequality has a counterexample.
import Mathlib.NumberTheory.ArithmeticFunction.Misc import Mathlib.NumberTheory.Harmonic.Defs import Mathlib.NumberTheory.LSeries.RiemannZeta import Mathlib.NumberTheory.Harmonic.EulerMascheroni open scoped ArithmeticFunction.sigma
namespace LeanEval.NumberTheory
theorem lagarias_robin_oscillation (hRH : ¬ RiemannHypothesis) :
∃ β C : ℝ, 0 < β ∧ β < 1 / 2 ∧ 0 < C ∧
∀ N : ℕ, ∃ n : ℕ, N ≤ n ∧ 3 ≤ n ∧
Real.exp Real.eulerMascheroniConstant * (n : ℝ) * Real.log (Real.log (n : ℝ)) +
C * (n : ℝ) * Real.log (Real.log (n : ℝ)) /
Real.rpow (Real.log (n : ℝ)) β ≤
((σ 1 n : ℕ) : ℝ) := by sorry
end LeanEval.NumberTheory
Read-back
What the Lean code literally says, in plain math · gpt-6-astra
Assume the Riemann hypothesis is false, meaning that there exists with , with for every natural number , with , and with , where is the Riemann zeta function. Then there exist real constants and satisfying and such that, for every natural number (including ), there exists a natural number with and for which . Here is the Euler–Mascheroni constant, the sum on the right is over the positive natural-number divisors of , and the power in the denominator is a real power. The restriction ensures that the logarithms have positive arguments and the denominator is nonzero.
Confirmed by the mission captain (proposal self-audit).