Error propagation through an approximate bilinear map (Lemma 2.9)
ProvedIntMul.HvdH.bilinear_error_propagationinteger-multiplication
Let be a continuous -bilinear map with and let be any function . Let , , , and . Then
This is Lemma 2.9 of Harvey–van der Hoeven.
Preamble
import Mathlib
Formal statement
namespace IntMul.HvdH
theorem bilinear_error_propagation {U V W : Type*} [NormedAddCommGroup U] [NormedSpace ℂ U]
[NormedAddCommGroup V] [NormedSpace ℂ V] [NormedAddCommGroup W] [NormedSpace ℂ W]
(A : U →L[ℂ] V →L[ℂ] W) (hA : ‖A‖ ≤ 1) (Aa : U → V → W)
(u ut : U) (v vt : V) (eA eu ev : ℝ) (hu1 : ‖u‖ ≤ 1) (hvt1 : ‖vt‖ ≤ 1)
(hAa : ‖Aa ut vt - A ut vt‖ ≤ eA) (hu : ‖ut - u‖ ≤ eu) (hv : ‖vt - v‖ ≤ ev) :
‖Aa ut vt - A u v‖ ≤ eA + eu + ev := by sorry
end IntMul.HvdHSource
Harvey, van der Hoeven, Integer multiplication in time O(n log n), Annals of Math. 193 (2021), https://hal.science/hal-02070778, Section 2.6