Unconditional Mertens bound for mu^2/phi-weighted log-power sums
Provedgpy_mertensMuSqBoundnumber-theory
For every and , the one-variable sum differs from by at most . The constant is exact, and the factor comes from the explicit bound with constant . The argument is Abel summation against the bound plus the elementary harmonic bound; no prime number theorem or Chebyshev estimate is used. This is the unconditional one-variable Mertens input for the GPY monomial evaluation (hypothesis hMono). Formalization Note: the sum is written out in Mathlib terms, with no local definitions.
Preamble
import Mathlib open Classical
Formal statement
theorem gpy_mertensMuSqBound (m T : ℕ) :
abs ((∑ D ∈ Finset.range (T + 1), (if Squarefree D then (ArithmeticFunction.moebius D : ℝ) ^ 2
/ ((Nat.totient D : ℕ) : ℝ) else 0) * (Real.log (D : ℝ)) ^ m)
- (Real.log (T : ℝ)) ^ (m + 1) / ((m : ℝ) + 1))
≤ (6 + 2 * 500) * (Real.log (T : ℝ)) ^ m := by sorrySource
GPY S0 bound: unconditional one-variable Mertens bound for hMono