A clean constant bound for the p-series beyond exponent two #
∑' m : ℕ, m ^ (-t) ≤ 2 for every real t ≥ 2. Mathlib supplies the exact value at the endpoint,
ζ (2) = π ^ 2 / 6, and summability throughout t > 1, but no inequality valid across a range of
exponents; that is what this file adds.
The bound is deliberately lossy. The supremum over t ≥ 2 is ζ (2) = 1.6449…, so 2 gives away
about 18%. A round constant is the useful thing to expose: consumers carry it through chains of
inequalities and none of them wants π in the goal.
Nothing here is specific to any application, and the file contains no number theory. The m = 0
term is 0, by the junk value of 0 ^ (-t).
Main results #
TauCeti.tsum_nat_rpow_neg_le_two—∑' m : ℕ, m ^ (-t) ≤ 2for2 ≤ t.