Finite sampling on the original matrix kernel uses at most its dimension
ProvedWeilDefect.MarkerStability.finite_sampling_on_matrix_kernelactual-zero-samplingconnes-weilfinite-dimensional
For ANY finite complex matrix and any supplied row family , there exists ONE finite set of original row indices, with , such that for EVERY satisfying , vanishing of the rows in is equivalent to vanishing of ALL supplied rows. Restrict the original rows to the ORIGINAL matrix kernel; no matrix or row normalization changes and no injectivity is assumed.
Preamble
import Mathlib.Data.Complex.Basic import Mathlib.LinearAlgebra.Dual.Lemmas import Mathlib.LinearAlgebra.Matrix.ToLin import Theorems.Thm_WeilDefect_MarkerStability_finite_functionals_determine_all set_option autoImplicit false open Matrix open scoped Classical noncomputable section open WeilDefect.MarkerStability
Formal statement
theorem WeilDefect.MarkerStability.finite_sampling_on_matrix_kernel
{I Ω : Type*} [Fintype I] (K : Matrix I I ℂ) (r : Ω → (I → ℂ)) :
∃ G : Finset Ω, G.card ≤ Module.finrank ℂ (LinearMap.ker K.mulVecLin) ∧
∀ c : I → ℂ, K *ᵥ c = 0 →
((∀ ρ ∈ G, r ρ ⬝ᵥ c = 0) ↔ ∀ ρ : Ω, r ρ ⬝ᵥ c = 0) := by sorry
Source
monocap-tech/weil, compiling source ae5ede91582c6434b1676f98d416b23316942daa, exact declaration WeilDefect.MarkerStability.finite_sampling_on_matrix_kernel