hermitian_kth_eigenvalue_dual_witness
Provedsensitivity_theorem
Dual min-max witness (<=) for the i-th descending eigenvalue: there exists a subspace W of (V -> R) of dimension exactly (card V) - i on which the Rayleigh quotient is at most lambda_i for every vector. Naturally satisfied by the eigenvector tail span.
Preamble
import Mathlib.Analysis.Matrix.Spectrum import Mathlib.LinearAlgebra.Matrix.Hermitian import Mathlib.LinearAlgebra.FiniteDimensional.Defs import Mathlib.Data.Matrix.Mul import Mathlib.Data.Fintype.Basic import Mathlib.Data.Real.Star open Matrix /-! # Generalized min-max characterization (upper-bound dual witness form) For any Hermitian real matrix `A : Matrix V V ℝ` and descending-sorted index `i`, there exists a subspace `W ⊆ (V → ℝ)` of dimension `(card V) − i` on which the Rayleigh quotient is bounded above by the `i`-th descending eigenvalue. This is the existential, witness-form *dual* min-max characterization (the other direction of Courant–Fischer): on a `(n − i)`-dim subspace (natural witness: the span of the bottom `(n − i)` eigenvectors), every vector's Rayleigh quotient is at most `λ_i`. Reusable spectral-theory result; not currently in Mathlib in this form. -/ /-- **Dual min-max witness for the `i`-th descending eigenvalue.** For any real Hermitian `A : Matrix V V ℝ` and any descending index `i`, there exists a subspace `W ⊆ (V → ℝ)` of dimension `(card V) − i` such that every `u ∈ W` has `(A *ᵥ u) ⬝ᵥ u ≤ hA.eigenvalues₀ i * (u ⬝ᵥ u)`. -/
Formal statement
theorem hermitian_kth_eigenvalue_dual_witness
{V : Type*} [Fintype V] [DecidableEq V] [Nonempty V]
{A : Matrix V V ℝ} (hA : A.IsHermitian) (i : Fin (Fintype.card V)) :
∃ W : Submodule ℝ (V → ℝ),
Module.finrank ℝ W = Fintype.card V - (i : ℕ) ∧
∀ u ∈ W, A *ᵥ u ⬝ᵥ u ≤ hA.eigenvalues₀ i * (u ⬝ᵥ u) := by sorrySource
Huang, Hao. "Induced subgraphs of hypercubes and a proof of the Sensitivity Conjecture." Annals of Mathematics 190.3 (2019): 949-955.