Axler theta error, analytic range: for
OpenTaoFivePrimes.axler_theta_log_four_tailThe analytic half of Axler's explicit fourth-power theta estimate.
Let
be the Chebyshev theta function. This node asserts the pointwise estimate
It is the analytic half of Proposition 1 of Axler (2018), which is stated for all . Splitting that proposition at separates the two logically different ingredients: on the estimate is a finite verification, while from on it follows from the explicit relation between and the zeros of the Riemann zeta function. This node is the analytic ingredient.
Role. This is a quantitative prime number theorem statement: since , it says the relative error of is at most , i.e. below from on. Above this estimate is far from tight ( is of size up to logarithmic factors, whereas is larger by a factor at ), so the constant is comfortable in this range; the source needs the analytic estimate in this form because it is then fed into partial summation against the kernel . Together with the companion node TaoFivePrimes.axler_theta_log_four_finite_certificate it yields the target statement TaoFivePrimes.axler_theta_error_log_four on the whole range .
Formalization Note. The statement is unconditional and uniform in ; equivalently, writing , it says for all .
import Mathlib open Finset
theorem TaoFivePrimes.axler_theta_log_four_tail (x : ℝ) (h1 : 10 ^ 8 ≤ x) :
|(∑ p ∈ Nat.primesLE ⌊x⌋₊, Real.log (p : ℝ)) - x| <
100 * x / (Real.log x) ^ 4 := by sorry