Abel transform of the mu-squared-over-phi log-power sum
Provedgpy_mertensMuSqSum_abelnumber-theory
For a natural number and a bound , let the one-variable Mertens-type sum be , where when is squarefree and otherwise (the square of the Möbius value is the squarefree indicator). Then summation by parts gives
This is the exact entry point for Mertens-type estimates: they bound the partial sums , and Abel summation converts those bounds into bounds on the log-power sums. Formalization Note: the weight and the sum are unfolded into explicit range sums with the squarefree condition inline.
Preamble
import Mathlib open Classical
Formal statement
theorem gpy_mertensMuSqSum_abel (m : ℕ) (T : ℕ) :
(∑ 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)
= (∑ D ∈ Finset.range (T + 1),
(if Squarefree D then (ArithmeticFunction.moebius D : ℝ) ^ 2
/ ((Nat.totient D : ℕ) : ℝ) else 0)) * (Real.log (T : ℝ)) ^ m
- ∑ i ∈ Finset.range T,
(∑ D ∈ Finset.range (i + 1),
(if Squarefree D then (ArithmeticFunction.moebius D : ℝ) ^ 2
/ ((Nat.totient D : ℕ) : ℝ) else 0))
* ((Real.log ((i + 1 : ℕ) : ℝ)) ^ m - (Real.log (i : ℝ)) ^ m) := by sorrySource
GPY conditional monomial evaluation: the Abel transform of the one-variable mu-squared-over-phi log-power sum, directly from summation by parts