Approximate Eigenvectors at the Bottom of the Operator Spectrum
ProvedFunctionalInequalities.Analysis.Semigroup.Generated.exists_approx_eigenvector_of_gap_readablefunctional-inequalitieskls-conjecturekls-modularkls-readablelog-concave
This is an approximate-eigenvector construction for .
Let be complete real Hilbert spaces, let be closed with dense domain, and let with . Assume:
- For every orthogonal to , .
- is orthogonal to .
- .
- .
Then there exists , orthogonal to , such that
The theorem does not assume in its type and does not assert the existence of an exact eigenvector.
Original Lean declaration header
Within namespace FunctionalInequalities.Analysis.Semigroup.Generated. Shown in its original module context. Proof omitted.
theorem exists_approx_eigenvector_of_gap (T : H₀ →ₗ.[ℝ] H₁) (hT : T.IsClosed)
(hd : Dense (T.domain : Set H₀)) {lam η : ℝ} (hlam : 0 < lam)
(hgap : ∀ y : T.domain, (∀ k : T.domain, T k = 0 → ⟪(y : H₀), (k : H₀)⟫ = 0) →
lam * ‖(y : H₀)‖ ^ 2 ≤ ‖T y‖ ^ 2)
(y : T.domain) (hy : ∀ k : T.domain, T k = 0 → ⟪(y : H₀), (k : H₀)⟫ = 0)
(hy1 : ‖(y : H₀)‖ = 1) (hTy : ‖T y‖ ^ 2 ≤ lam * (1 + η)) :
∃ z : (adjointComp T).domain,
(∀ k : T.domain, T k = 0 → ⟪(z : H₀), (k : H₀)⟫ = 0) ∧
‖(y : H₀) - (z : H₀)‖ ≤ Real.sqrt (η / 8) ∧
‖adjointComp T z - lam • (z : H₀)‖ ≤ lam * Real.sqrt (η / 2)
Module navigation
Module: Analysis/Semigroup/Generated/ApproximateEigenvector.
Definition modules: Analysis/Semigroup/AdjointComp.
Source excerpt, lines 55–64 · Frozen source 408915f.
Preamble
import Definitions.Def_KLSModular_FunctionalInequalities_Analysis_Semigroup_Generated_ApproximateEigenvector
open scoped InnerProductSpace RealInnerProductSpace
variable {H₀ H₁ : Type*} [NormedAddCommGroup H₀] [InnerProductSpace ℝ H₀] [CompleteSpace H₀]
[NormedAddCommGroup H₁] [InnerProductSpace ℝ H₁] [CompleteSpace H₁]
Formal statement
theorem FunctionalInequalities.Analysis.Semigroup.Generated.exists_approx_eigenvector_of_gap_readable (T : H₀ →ₗ.[ℝ] H₁) (hT : T.IsClosed)
(hd : Dense (T.domain : Set H₀)) {lam η : ℝ} (hlam : 0 < lam)
(hgap : ∀ y : T.domain, (∀ k : T.domain, T k = 0 → ⟪(y : H₀), (k : H₀)⟫ = 0) →
lam * ‖(y : H₀)‖ ^ 2 ≤ ‖T y‖ ^ 2)
(y : T.domain) (hy : ∀ k : T.domain, T k = 0 → ⟪(y : H₀), (k : H₀)⟫ = 0)
(hy1 : ‖(y : H₀)‖ = 1) (hTy : ‖T y‖ ^ 2 ≤ lam * (1 + η)) :
∃ z : (adjointComp T).domain,
(∀ k : T.domain, T k = 0 → ⟪(z : H₀), (k : H₀)⟫ = 0) ∧
‖(y : H₀) - (z : H₀)‖ ≤ Real.sqrt (η / 8) ∧
‖adjointComp T z - lam • (z : H₀)‖ ≤ lam * Real.sqrt (η / 2) := by sorry
Source
Lean source: https://github.com/human-ai-mathematics/kls-lean/blob/408915f/formalization/FunctionalInequalities/Analysis/Semigroup/Generated/ApproximateEigenvector.lean#L49-L124 (repository human-ai-mathematics/kls-lean, commit 408915f, module `FunctionalInequalities.Analysis.Semigroup.Generated.ApproximateEigenvector`, declaration `FunctionalInequalities.Analysis.Semigroup.Generated.exists_approx_eigenvector_of_gap`; the repository is private at upload time).