Documentation

LeanPool.CCDN.AuxiliaryPolynomial.LargeHeightRatios

In the large-height range, the coefficient saving absorbs the radius and leaves a negative logarithmic-height term to control the bad-prime factor.

theorem CCDN.auxiliary_large_height_bracket_bound {r d M X a k u b l h T : ℝ} (hr : 0 ≤ r) (hd : 1 ≤ d) (hM : 8 * d * (r + 1) ≤ M) (hX : 0 < X) (halow : X * (M - d) / ((r + 1) * M) ≤ a) (haup : a ≤ X * (M + r) / ((r + 1) * M)) (hk : 0 ≤ k) (hklow : 1 / (d * (r + 1)) - 1 / M ≤ k) (hb : 0 ≤ b) (hl : 0 ≤ l) (hT : 0 ≤ T) (hheight : 2 * d * (r + 1) * b ≤ h) (hposition : h ≤ l + T * d * X) (hcomparison : a * u ≤ b - k * l) :
u ≤ T - h / (8 * d * X)

Abstract real-variable form of the large-height step: if a is the normalized rank ratio, k the kernel ratio and the comparison a * u ≤ b - k * l holds, with M ≥ 8 d (r + 1), the height h at least 2 d (r + 1) b and at most l + T d X, then u ≤ T - h / (8 d X).

theorem CCDN.log_one_add_le_linear_add_log {x c : ℝ} (hx : 0 ≤ x) (hc : 1 ≤ c) :
Real.log (1 + x) ≤ x / c + Real.log c

A logarithmic bad-prime loss is absorbed by a linear height saving.