§8, Proof, claim A, p. 150 — linear independence of columns survives sufficiently small modification
ProvedExecCompLP.Consistency.claimA_linearIndependent_perturblinear-algebralinear-independencep2o-batch-pfp2ap2o-gran-per-chapterp2o-plan-paperp2o-v1perturbation
Let be a real matrix and a set of column indices such that the columns are linearly independent in . Then there is such that for every real matrix with
the columns are again linearly independent.
This is claim A of the proof of consistency in §8: linear independence of a set of columns is an open condition on the entries of the matrix. It guarantees that a basis of the true program remains a basis of the sample program once the sampling errors are small.
Formalization Note Closeness of matrices is entrywise, since the paper names no norm; all norms on give the same statement.
Preamble
import Mathlib import Definitions.Def_ExecCompLP_Consistency_Setting
Formal statement
namespace ExecCompLP.Consistency
theorem claimA_linearIndependent_perturb {m n : ℕ} (ξ : Matrix (Fin m) (Fin n) ℝ)
(S : Finset (Fin n)) (hS : LinearIndependent ℝ (fun j : S => fun i => ξ i j)) :
∃ δ > 0, ∀ X : Matrix (Fin m) (Fin n) ℝ, (∀ i j, |X i j - ξ i j| < δ) →
LinearIndependent ℝ (fun j : S => fun i => X i j) := by sorry
end ExecCompLP.Consistency
Source
Charnes, Cooper & Ferguson, Optimal estimation of executive compensation by linear programming, Management Science 1(2) (1955), p. 150, §8, Proof, claim A
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.