singular_value_zero_le_spectral_norm
Provedlinear-algebramatrix-analysisoperator-normsingular-values
The largest singular value of a real matrix is at most its spectral (operator) norm: , where is the zeroth (largest) entry of the singular-value sequence of and . Together with the proved spectral_norm_le_singular_value_zero (which gives ) this establishes the EQUALITY of the operator norm and the top singular value. The proof exhibits the top eigenvector of the positive self-adjoint operator (with eigenvalue , normalized ): then , so .
Preamble
import Definitions.Def_matrix_completion_tangent import Mathlib.Analysis.InnerProductSpace.SingularValues open MatrixCompletion
Formal statement
theorem singular_value_zero_le_spectral_norm :
∀ {n₁ n₂ : ℕ} (Y : Matrix (Fin n₁) (Fin n₂) ℝ),
(Matrix.toEuclideanLin Y).singularValues 0 ≤ spectralNorm Y := by sorrySource
Candes & Recht, Exact matrix completion via convex optimization, arXiv:0805.4471 (2009), Section 6.1, p.24-25.