Physlib

PhyslibAlpha.ProbabilisticTheory.HilbertSpace.Unbounded.Cayley.Certificate

Spectral data of bounded normal and unitary operators

i. Overview

`BoundedNormalSpectralData` is a spectral measure on `ℂ` whose integral of the identity is a given bounded normal operator. `BoundedUnitarySpectralData` is the same for a unitary whose spectral measure is supported on the unit circle away from `1`. Pulling such a measure back along the Cayley transform gives a spectral measure on `ℝ`, which is determined by its Cayley pushforward.

ii. Key results

- `BoundedNormalSpectralData` : a spectral measure reconstructing a bounded normal operator. - `BoundedUnitarySpectralData` : a spectral measure reconstructing a unitary, supported away from `1`. - `BoundedUnitarySpectralData.realSpectralMeasure` : the pulled-back spectral measure on `ℝ`. - `BoundedUnitarySpectralData.realSpectralMeasure_eq_of_cayleyMap_eq` : a real spectral measure is determined by its Cayley pushforward.

iii. Table of contents

  • A. Bounded normal spectral data
  • B. Bounded unitary spectral data

A. Bounded normal spectral data

B. Bounded unitary spectral data

7 declarations

theorem

Equality of Bounded Normal Spectral Data via Spectral Measures

Let DD and EE be bounded normal spectral data associated with an operator UU. If the spectral measures of DD and EE are equal, then DD and EE are equal.

theorem

Extensionality of Bounded Normal Spectral Data via Spectral Integrals

Let DD and EE be bounded normal spectral data associated with a bounded normal operator UU. If for every bounded measurable function f:C→Cf: \mathbb{C} \to \mathbb{C}, the spectral integrals of ff with respect to the spectral measures of DD and EE are equal as bounded linear operators (i.e., ∫Cf dμD=∫Cf dμE\int_{\mathbb{C}} f \, d\mu_D = \int_{\mathbb{C}} f \, d\mu_E), then D=ED = E.

theorem

Extensionality of Bounded Unitary Spectral Data: D.spectralMeasure=E.spectralMeasure  ⟹  D=ED.spectralMeasure = E.spectralMeasure \implies D = E

For any two instances of bounded unitary spectral data DD and EE (which represent spectral measures on C\mathbb{C} that reconstruct a unitary operator), if their associated spectral measures are equal, then D=ED = E.

theorem

Equality of bounded integrals implies D=ED = E for bounded unitary spectral data

Let HH be a Hilbert space and let DD and EE be bounded unitary spectral data on HH. If for every bounded measurable function f:C→Cf: \mathbb{C} \to \mathbb{C}, the operator-valued integrals with respect to the spectral measures μD\mu_D and μE\mu_E associated with DD and EE are equal, i.e., ∫Cf dμD=∫Cf dμE,\int_{\mathbb{C}} f \, d\mu_D = \int_{\mathbb{C}} f \, d\mu_E, then the spectral data are equal (D=ED = E).

definition

Real spectral measure of bounded unitary spectral data

Given a bounded unitary spectral data DD on a Hilbert space H\mathcal{H} (where the spectral measure is supported on the unit circle T⊂C\mathbb{T} \subset \mathbb{C} excluding 11), this definition constructs the associated spectral measure on the real line R\mathbb{R}. This is obtained by pulling back the spectral measure of DD along the Cayley transform c:R→T∖{1}c: \mathbb{R} \to \mathbb{T} \setminus \{1\}, defined by c(x)=x−ix+ic(x) = \frac{x - i}{x + i}. The resulting measure is a spectral measure on R\mathbb{R} taking values in the operators on H\mathcal{H}.

theorem

cayleyMap(D.realSpectralMeasure)=D.spectralMeasure\text{cayleyMap}(D.\text{realSpectralMeasure}) = D.\text{spectralMeasure}

Let DD be a bounded unitary spectral data on a Hilbert space H\mathcal{H}. Let μD\mu_D be the spectral measure associated with DD (which is supported on the unit circle T∖{1}⊂C\mathbb{T} \setminus \{1\} \subset \mathbb{C}), and let νD\nu_D be the corresponding real spectral measure on R\mathbb{R} obtained by pulling back μD\mu_D along the Cayley transform c:R→T∖{1}c: \mathbb{R} \to \mathbb{T} \setminus \{1\}, c(x)=x−ix+ic(x) = \frac{x - i}{x + i}. Then, the push-forward of νD\nu_D along the Cayley transform (denoted as `cayleyMap`) recovers the original spectral measure μD\mu_D: cayleyMap(νD)=μD\text{cayleyMap}(\nu_D) = \mu_D

theorem

Uniqueness of the real spectral measure determined by its Cayley pushforward

Let H\mathcal{H} be a Hilbert space and let DD be an object of bounded unitary spectral data, which includes a spectral measure EDE_D on C\mathbb{C} supported on the unit circle excluding 1 (T∖{1}\mathbb{T} \setminus \{1\}). Let ED,RE_{D, \mathbb{R}} be the associated real spectral measure on R\mathbb{R} constructed by pulling back EDE_D along the Cayley transform c:R→T∖{1}c: \mathbb{R} \to \mathbb{T} \setminus \{1\} defined by c(x)=x−ix+ic(x) = \frac{x - i}{x + i}. For any spectral measure μS\mu_S on R\mathbb{R}, if the pushforward of μS\mu_S along the Cayley transform (denoted as cayleyMap(μS)\text{cayleyMap}(\mu_S)) is equal to EDE_D, then μS\mu_S must be equal to ED,RE_{D, \mathbb{R}}.