Physlib

Physlib.QuantumMechanics.Operators.Unbounded

Unbounded operators

i. Overview

The appropriate mathematical objects for discussing operators in non-relativistic quantum mechanics are partially-defined linear map (`LinearPMap`) between complex Hilbert spaces, `H →ₗ.[ℂ] H'`. An import class of operators in NRQM are those which are both densely defined and closable, which we refer to as _unbounded_. When `H = H'` operators may also be symmetric, self-adjoint or essentially self-adjoint (closure is self-adjoint).

In this module we collect results on how the properties `HasDenseDomain`, `IsUnbounded`, `IsSymmetric`, `IsSelfAdjoint` and `IsEssentiallySelfAdjoint` interact with the basic algebraic operations, closure, adjoints, unitary conjugation and each other.

Notes

- Naming convention : Definitions of `LinearPMap`s for quantum mechanical unbounded operators should have a name of the form `[…]Operator` and notation should use calligraphic capital letters, e.g. `mulOperator f` (`𝓜 f`) for the multiplication operator associated with the function `f`.

- Implementation : Although operators encountered in quantum mechanics are almost always unbounded, we opt to implement unbounded operators via the property `IsUnbounded` on `LinearPMap` rather than as a structure `UnboundedOperator` extending `LinearPMap`. The basic reason for this is that addition/subtraction and composition of unbounded operators in general does not result in another unbounded operator. This means, for example, that any attempt to define addition of `UnboundedOperator`s would inevitably require introducing junk values that spoil associativity.

ii. Key results

Definitions - `HasDenseDomain` : An operator `U : H →ₗ.[ℂ] H'` has dense domain if `U.domain` is dense in `H`. - `IsUnbounded` : An operator is unbounded if it is both densely defined and closable. - `IsSymmetric` : An operator `T : H →ₗ.[ℂ] H` is symmetric if `⟪T x, y⟫_ℂ = ⟪x, T y⟫_ℂ` holds for all `x y : T.domain`. - `IsEssentiallySelfAdjoint` : An operator `T : H →ₗ.[ℂ] H` is essentially self-adjoint if its closure is self-adjoint.

Results - `adjoint_add_le_add_adjoint` : The inequality `U₁† + U₂† ≤ (U₁ + U₂)†` when `U₁ + U₂` has dense domain. - `unitaryConj` : The conjugation `u A u⁻¹ : H' →ₗ.[ℂ] H'` of `A` by a unitary `u`, with domain `u (A.domain)`. - `IsFormalAdjoint.unitaryConj` : Unitary conjugation preserves formal-adjoint pairs. - `HasDenseDomain.unitaryConj_dense_domain` : If `A` has dense domain, then so does `u A u⁻¹`. - `unitaryConj_sub_smul_surjective` : If `A - z` is surjective for a scalar `z : ℂ`, then so is `u A u⁻¹ - z`. - `adjoint_compRestricted_le_compRestricted_adjoint` : The inequality `U† ∘ᵣ V† ≤ (V ∘ᵣ U)†` when `V` and `V ∘ᵣ U` have dense domain. - `IsEssentiallySelfAdjoint.unique_self_adjoint_extension` : The closure of an essentially self-adjoint unbounded operator is its unique self-adjoint extension. - `IsUnbounded.adjoint` : The adjoint of an unbounded operator is also unbounded. - `IsUnbounded.adjoint_closure_eq_adjoint` : An unbounded operator and its closure have the same adjoint. - `IsUnbounded.adjoint_adjoint_eq_closure` : An unbounded operator `U` satisfies `U†† = U.closure`.

iii. Table of contents

- A. Definitions - B. Basic properties - B.1. Dense domain - B.2. Closability - B.3. Adjoints - B.4. Continuity / boundedness - B.5. Unitary conjugation - C. Classes of operators - C.1. Symmetric operators - C.2. Self-adjoint operators - C.3. Essentially self-adjoint operators - C.4. Unbounded operators

iv. References

  • [Reed and Simon, *Methods of Modern Mathematical Physics, Vol. I: Functional Analysis*][Reed1972]
  • [Konrad Schmüdgen, *Unbounded Self-Adjoint Operators on Hilbert Space*][Schmudgen2012]

A. Definitions

See `LinearPMap.instStar` and `LinearPMap.isSelfAdjoint_def` for the definition of `IsSelfAdjoint` for `LinearPMap`s.

B. Basic properties

B.1. Dense domain

B.2. Closability

B.3. Adjoints

B.4. Continuity / boundedness

B.5. Unitary conjugation

C. Classes of operators

C.1. Symmetric operators

C.2. Self-adjoint operators

C.3. Essentially self-adjoint operators

C.4. Unbounded operators

71 declarations

definition

dom(T)\text{dom}(T) is dense in HH

Let HH and HH' be Hilbert spaces over C\mathbb{C}. For a linear partial map TT from HH to HH', this property holds if the domain of TT, denoted as dom(T)\text{dom}(T), is a dense subspace of HH.

theorem

U.HasDenseDomain    Dense(dom(U))U.\text{HasDenseDomain} \iff \text{Dense}(\text{dom}(U))

Let HH and HH' be Hilbert spaces over C\mathbb{C}. For any linear partial map U:HHU: H \to H', the property U.HasDenseDomainU.\text{HasDenseDomain} holds if and only if the domain of UU is a dense subset of HH.

definition

TT is a densely defined and closable operator

Let HH and HH' be complex Hilbert spaces. For a partially defined linear map TT from HH to HH', the predicate `LinearPMap.IsUnbounded` holds if the domain of TT is dense in HH and TT is a closable operator.

theorem

An operator is "unbounded" if and only if it is densely defined and closable

Let HH and HH' be complex Hilbert spaces. For a partially defined linear operator UU from HH to HH', UU is defined as an "unbounded operator" if and only if UU is densely defined (i.e., its domain is dense in HH) and UU is closable.

definition

Symmetry of a partially defined linear operator TT

Let H\mathcal{H} be a complex Hilbert space. A partially defined linear operator TT on H\mathcal{H} (represented by a `LinearPMap`) is said to be **symmetric** if for every pair of vectors x,yx, y in the domain of TT, the inner product satisfies the condition: Tx,y=x,Ty\langle Tx, y \rangle = \langle x, Ty \rangle where ,\langle \cdot, \cdot \rangle denotes the inner product on H\mathcal{H}.

theorem

TT is symmetric     \iff TT is formally adjoint to TT

Let HH be a complex inner product space. For any partially defined linear operator TT on HH (represented as a `LinearPMap`), TT is symmetric if and only if it is formally adjoint to itself. In this context, TT is symmetric if Tx,y=x,Ty\langle Tx, y \rangle = \langle x, Ty \rangle for all x,ydom(T)x, y \in \text{dom}(T), and TT is formally adjoint to SS if Tx,y=x,Sy\langle Tx, y \rangle = \langle x, Sy \rangle for all xdom(T)x \in \text{dom}(T) and ydom(S)y \in \text{dom}(S).

definition

TT is essentially self-adjoint

Let HH be a complex Hilbert space. A partially defined linear map T:D(T)HHT: D(T) \subseteq H \to H (of type `LinearPMap`) is **essentially self-adjoint** if it is densely defined and its closure T\overline{T} is a self-adjoint operator, such that T=(T)\overline{T} = (\overline{T})^*.

theorem

TT is essentially self-adjoint     \iff Tˉ\bar{T} is self-adjoint

Let HH be a complex Hilbert space. For a partially defined linear operator TT on HH, TT is essentially self-adjoint if and only if its closure Tˉ\bar{T} is self-adjoint.

theorem

TT is star-normal     TT=TT\iff T^* T = T T^*

Let HH be a complex Hilbert space and TT be a partial linear map on HH. Then TT is star-normal if and only if TT=TTT^* T = T T^*, where TT^* denotes the adjoint of TT and the product refers to the restricted composition of partial linear maps. The restricted composition grfg \circ_r f is defined on the domain {xdom(f)f(x)dom(g)}\{x \in \text{dom}(f) \mid f(x) \in \text{dom}(g)\}.

theorem

A densely defined operator UU is unbounded     \iff UU is closable

Let HH and HH' be complex Hilbert spaces. Let UU be a linear operator from a subspace of HH to HH' (a `LinearPMap`). If the domain of UU is dense in HH, then UU satisfies the definition of an unbounded operator if and only if UU is closable. Here, an "unbounded operator" is defined as a linear operator that is both densely defined and closable.

theorem

The closure of a densely defined operator is densely defined

Let HH and HH' be complex Hilbert spaces. Let U:D(U)HHU: \mathcal{D}(U) \subseteq H \to H' be a partially defined linear operator. If UU is densely defined (i.e., its domain D(U)\mathcal{D}(U) is dense in HH), then its closure U\overline{U} is also densely defined.

theorem

D(U)D(U)D(\overline{U}) \subseteq \overline{D(U)}

Let HH and HH' be complex Hilbert spaces. For any partially defined linear operator UU from HH to HH' with domain D(U)D(U), let U\overline{U} denote the closure of the operator UU. Then the domain of the closure is a subset of the closure of the domain, i.e., D(U)D(U)D(\overline{U}) \subseteq \overline{D(U)}.

theorem

dom(U)\text{dom}(U) is dense     dom(U)\iff \text{dom}(\overline{U}) is dense

Let HH and HH' be Hilbert spaces. For any partially defined linear operator U:HHU: H \to H', the domain of UU is dense in HH if and only if the domain of its closure U\overline{U} is dense in HH.

theorem

Negation Preserves Density of Domain for Partial Linear Maps

Let HH and HH' be complex Hilbert spaces. For any partially defined linear map UU from HH to HH', if UU has a dense domain, then its negation U-U also has a dense domain.

theorem

Scalar Multiplication Preserves Dense Domains of Partial Linear Maps

Let HH and HH' be complex Hilbert spaces. For any partially defined linear map U:HHU: H \to H', if the domain of UU is dense in HH, then for any scalar cCc \in \mathbb{C}, the domain of the scalar multiple cUc U is also dense in HH.

theorem

Density of the domain of U1+U2U_1 + U_2 when dom(U1)dom(U2)\text{dom}(U_1) \subseteq \text{dom}(U_2)

Let HH and HH' be complex Hilbert spaces. Let U1U_1 and U2U_2 be linear partially defined maps from HH to HH'. If U1U_1 has a dense domain and the domain of U1U_1 is contained within the domain of U2U_2 (i.e., dom(U1)dom(U2)\text{dom}(U_1) \subseteq \text{dom}(U_2)), then the sum operator U1+U2U_1 + U_2 also has a dense domain.

theorem

U1U2U_1 - U_2 is densely defined if D(U1)\mathcal{D}(U_1) is dense and D(U1)D(U2)\mathcal{D}(U_1) \subseteq \mathcal{D}(U_2)

Let HH and HH' be complex Hilbert spaces. Let U1U_1 and U2U_2 be linear partial maps from HH to HH'. If U1U_1 has a dense domain D(U1)\mathcal{D}(U_1) and D(U1)D(U2)\mathcal{D}(U_1) \subseteq \mathcal{D}(U_2), then the domain of the difference U1U2U_1 - U_2 is also dense in HH.

theorem

Common Dense Submodule implies Dense Domain of a Finite Sum of Partial Linear Maps

Let HH and HH' be Hilbert spaces over C\mathbb{C}, and let {Wa}aα\{W_a\}_{a \in \alpha} be a finite family of partial linear maps from HH to HH'. If there exists a submodule EE of HH that is dense in HH and is contained in the domain of each WaW_a (i.e., Edom(Wa)E \subseteq \text{dom}(W_a) for all aαa \in \alpha), then the sum of these partial linear maps aαWa\sum_{a \in \alpha} W_a has a dense domain.

theorem

TnT^n is densely defined if TT is densely defined and T(D(T))D(T)T(D(T)) \subseteq D(T)

Let HH be a Hilbert space over C\mathbb{C} and TT be a partial linear map on HH with domain D(T)HD(T) \subseteq H. If TT is densely defined (i.e., D(T)D(T) is dense in HH) and the domain is invariant under TT (i.e., for all xD(T)x \in D(T), T(x)D(T)T(x) \in D(T)), then for every natural number nn, the nn-th power of TT is also densely defined.

theorem

If TnT^n is densely defined, then TkT^k is densely defined for knk \leq n

Let HH be a complex Hilbert space and TT be a linear partial map from HH to itself. For any natural number nn, if the domain of the nn-th power TnT^n (defined via restricted composition) is dense in HH, then for any natural number kk such that knk \leq n, the domain of TkT^k is also dense in HH.

theorem

(ran U)=kerU(\text{ran } U)^\perp = \ker U^*

Let HH and HH' be Hilbert spaces over C\mathbb{C}, where HH is a complete space. Let U:D(U)HU: \mathcal{D}(U) \to H' be a linear operator with a dense domain D(U)H\mathcal{D}(U) \subseteq H. Then the orthogonal complement of the range of UU is equal to the kernel of its adjoint operator UU^*: (ran U)=kerU(\text{ran } U)^\perp = \ker U^* Here, (ran U)(\text{ran } U)^\perp denotes the set of vectors in HH' orthogonal to the image of UU, and kerU\ker U^* denotes the subspace of vectors yD(U)Hy \in \mathcal{D}(U^*) \subseteq H' such that Uy=0U^* y = 0.

theorem

(kerU)=ran U(\ker U^*)^\perp = \overline{\text{ran } U}

Let HH and HH' be complex Hilbert spaces. For any densely defined linear operator U:D(U)HHU: D(U) \subseteq H \to H', the orthogonal complement of the kernel of its adjoint operator UU^* is equal to the closure of the range of UU. That is, (kerU)=ran U,(\ker U^*)^\perp = \overline{\text{ran } U}, where kerU\ker U^* denotes the kernel of the adjoint operator, ()(\cdot)^\perp denotes the orthogonal complement, and ran U\overline{\text{ran } U} denotes the closure of the range of UU.

theorem

If UU is closed, then closure(U)=U\text{closure}(U) = U

Let HH and HH' be complex Hilbert spaces. For any partially defined linear map UU from HH to HH', if UU is a closed operator, then its closure is equal to UU.

theorem

A closable operator UU is closed     \iff U=UU = \overline{U}

Let HH and HH' be Hilbert spaces over C\mathbb{C}. For any partially defined linear operator U:HHU: H \to H', if UU is closable, then UU is closed if and only if U=UU = \overline{U}, where U\overline{U} denotes the closure of UU.

theorem

Existence of a densely defined formal adjoint implies UU is closable

Let HH and HH' be complex Hilbert spaces. Let UU be a linear operator defined on a dense subspace dom(U)H\text{dom}(U) \subseteq H taking values in HH'. If there exists a densely defined linear operator UU' from dom(U)H\text{dom}(U') \subseteq H' to HH that is a formal adjoint of UU (i.e., Uϕ,ψH=ϕ,UψH\langle U' \phi, \psi \rangle_H = \langle \phi, U \psi \rangle_{H'} for all ϕdom(U)\phi \in \text{dom}(U') and ψdom(U)\psi \in \text{dom}(U)), then UU is closable.

theorem

The zero operator is closable

Let H\mathcal{H} and H\mathcal{H}' be Hilbert spaces over C\mathbb{C}. For any partially defined linear operator U:D(U)HHU: D(U) \subseteq \mathcal{H} \to \mathcal{H}' (represented as a `LinearPMap`), if UU is the zero operator on its domain, then UU is closable.

theorem

UU is closable     cU\implies cU is closable

Let HH and HH' be complex Hilbert spaces, and let UU be a partially defined linear operator from HH to HH'. If UU is closable, then for any complex scalar cCc \in \mathbb{C}, the scalar multiple cUc U is also closable.

theorem

cUc \cdot U is closable iff UU is closable for c0c \neq 0

Let HH and HH' be complex Hilbert spaces, and let UU be a linear map defined on a subspace of HH mapping into HH' (a `LinearPMap`). For any non-zero complex scalar cCc \in \mathbb{C} (c0c \neq 0), the operator cUcU is closable if and only if UU is closable.

theorem

U=1U-U = -1 \cdot U for linear partial maps

Let HH and HH' be inner product spaces over C\mathbb{C}. For any linear partial map UU from HH to HH', the negation of the operator is equal to the scalar multiplication of the operator by 1-1, that is, U=1U-U = -1 \cdot U.

theorem

If UU is closable, then U-U is closable

Let HH and HH' be complex Hilbert spaces. If a partially defined linear operator U:D(U)HHU: \mathcal{D}(U) \subseteq H \to H' is closable, then its negation U-U is also closable.

theorem

(cU)closure=cUclosure(c \cdot U)^{\text{closure}} = c \cdot U^{\text{closure}} for c0c \neq 0

Let HH and HH' be Hilbert spaces over the complex numbers C\mathbb{C}. For any partially defined linear operator UU from HH to HH' and any non-zero scalar cCc \in \mathbb{C}, the closure of the scalar multiple cUc \cdot U is equal to the scalar multiple of the closure of UU: (cU)closure=cUclosure(c \cdot U)^{\text{closure}} = c \cdot U^{\text{closure}}

theorem

The Adjoint of the Identity Operator is the Identity (1=11^\dagger = 1)

Let HH be a complex Hilbert space. Let 11 denote the identity operator on HH defined on the entire space (the identity element of the monoid of partial linear maps). Then the adjoint of this identity operator is equal to itself, i.e., 1=11^\dagger = 1.

theorem

U=0    U=0U = 0 \implies U^\dagger = 0

Let HH be a Hilbert space and HH' be an inner product space over C\mathbb{C}. For any partially defined linear map UU from HH to HH', if UU is the zero operator, then its adjoint operator UU^\dagger is also the zero operator.

theorem

0=00^\dagger = 0

Let HH be a complex Hilbert space and HH' be a complex inner product space. The adjoint of the zero operator 0:HH0: H \to H' is the zero operator 0:HH0: H' \to H.

theorem

(cU)=cˉU(cU)^* = \bar{c}U^* for c0c \neq 0

Let HH and HH' be complex Hilbert spaces, and let U:HHU: H \to H' be a partially defined linear operator. For any non-zero complex scalar cCc \in \mathbb{C}, the adjoint of the scalar multiple cUcU is given by the product of the complex conjugate cˉ\bar{c} and the adjoint UU^*, i.e., (cU)=cˉU(cU)^* = \bar{c}U^*.

theorem

U1+U2(U1+U2)U_1^\dagger + U_2^\dagger \le (U_1 + U_2)^\dagger

Let HH and HH' be complex Hilbert spaces, with HH being a complete space. Let U1U_1 and U2U_2 be partially defined linear operators (linear pre-maps) from HH to HH'. If the sum U1+U2U_1 + U_2 has a dense domain in HH, then the sum of the adjoints of U1U_1 and U2U_2 is a restriction of the adjoint of their sum, denoted as U1+U2(U1+U2)U_1^\dagger + U_2^\dagger \leq (U_1 + U_2)^\dagger.

theorem

U1U2(U1U2)U_1^\dagger - U_2^\dagger \leq (U_1 - U_2)^\dagger

Let HH be a Hilbert space and HH' be an inner product space. Let U1,U2:HHU_1, U_2: H \to H' be partially defined linear operators (linear PMaps). If the operator U1U2U_1 - U_2 has a dense domain, then the difference of their adjoints is extended by the adjoint of their difference: U1U2(U1U2)U_1^\dagger - U_2^\dagger \leq (U_1 - U_2)^\dagger where UU^\dagger denotes the adjoint of UU and \leq denotes the extension relation for partially defined operators.

theorem

UrV(VrU)U^\dagger \circ_r V^\dagger \subseteq (V \circ_r U)^\dagger for partial linear maps

Let HH, HH', and HH'' be complex Hilbert spaces. Let U:HHU: H \to H' and V:HHV: H' \to H'' be partial linear maps. If VV is densely defined and the restricted composition VrUV \circ_r U is also densely defined, then the restricted composition of the adjoints is extended by the adjoint of the restricted composition: UrV(VrU)U^\dagger \circ_r V^\dagger \subseteq (V \circ_r U)^\dagger where \dagger denotes the operator adjoint and ABA \subseteq B means that the partial linear map BB extends AA (i.e., dom(A)dom(B)\text{dom}(A) \subseteq \text{dom}(B) and A(x)=B(x)A(x) = B(x) for all xdom(A)x \in \text{dom}(A)).

theorem

If TnT^n is densely defined, then (T)n(Tn)(T^*)^n \le (T^n)^*

Let HH be a complex Hilbert space and TT be a partial linear operator on HH. For any natural number nn, if the nn-th power of the operator TnT^n (defined via restricted composition) has a dense domain, then the nn-th power of the adjoint of TT is a restriction of the adjoint of the nn-th power of TT. That is, (T)n(Tn) (T^*)^n \le (T^n)^* where TT^* denotes the adjoint of TT, and the relation ABA \le B indicates that the operator BB is an extension of the operator AA.

theorem

Continuity of a linear map is equivalent to its boundedness

Let k\mathbb{k} be a nontrivially normed field, and let EE and FF be seminormed spaces over k\mathbb{k}. A linear map f:EFf: E \to F is continuous if and only if it is bounded, i.e., there exists a constant M>0M > 0 such that for all xEx \in E, f(x)Mx\|f(x)\| \le M \|x\|.

theorem

Continuous Partially Defined Linear Maps are Closable

Let HH and HH' be complex inner product spaces. For any partially defined linear map U:D(U)HHU: D(U) \subseteq H \to H', if UU is continuous on its domain D(U)D(U), then UU is a closable operator.

theorem

D(U)=D(U)D(\overline{U}) = \overline{D(U)} for continuous operators UU

Let HH and HH' be complex Hilbert spaces, and let UU be a partially defined linear operator from HH to HH' with domain D(U)HD(U) \subseteq H. If UU is continuous on its domain, then the domain of its operator closure U\overline{U} is equal to the topological closure of its domain, denoted as D(U)=D(U)D(\overline{U}) = \overline{D(U)}.

theorem

A continuous linear operator is closed     \iff its domain is closed

Let HH and HH' be complex inner product spaces, and assume HH' is a complete space (a Hilbert space). Let U:D(U)HU: \mathcal{D}(U) \to H' be a linear operator with domain D(U)H\mathcal{D}(U) \subseteq H. If UU is continuous on its domain, then UU is a closed operator if and only if its domain D(U)\mathcal{D}(U) is a closed subspace of HH.

theorem

The graph of a closed operator is closed

Let HH and HH' be Hilbert spaces over C\mathbb{C}, and let UU be a partially defined linear operator from HH to HH'. If UU is a closed operator, then its graph, {(x,Ux)xdom(U)}\{(x, Ux) \mid x \in \operatorname{dom}(U)\}, is a closed subset of H×HH \times H'.

theorem

A Closed Operator with a Closed Domain is Continuous

Let HH and HH' be complex Hilbert spaces. Let U:D(U)HHU: D(U) \subseteq H \to H' be a partially defined linear operator. If UU is a closed operator and its domain D(U)D(U) is a closed subspace of HH, then UU is continuous on its domain.

theorem

If U1U_1 is closable and U2U_2 is continuous, then U1+U2U_1 + U_2 is closable

Let HH and HH' be Hilbert spaces over C\mathbb{C}. Let U1U_1 and U2U_2 be partially defined linear operators from HH to HH'. If U1U_1 is closable, U2U_2 is continuous, and the domain of U1U_1 is contained within the domain of U2U_2 (dom(U1)dom(U2)\text{dom}(U_1) \subseteq \text{dom}(U_2)), then the sum operator U1+U2U_1 + U_2 is closable.

theorem

Closability of U1U2U_1 - U_2 for continuous U2U_2

Let HH and HH' be complex Hilbert spaces. Let U1U_1 and U2U_2 be partially defined linear operators from HH to HH' with domains D(U1)D(U_1) and D(U2)D(U_2), respectively. If U1U_1 is closable, U2U_2 is continuous, and D(U1)D(U2)D(U_1) \subseteq D(U_2), then the operator U1U2U_1 - U_2 (defined on D(U1)D(U_1)) is also closable.

theorem

The sum of a closed operator and a continuous operator is closed

Let HH and HH' be complex inner product spaces, where HH' is a complete space (a Hilbert space). Let U1U_1 and U2U_2 be partially defined linear maps from HH to HH'. If U1U_1 is a closed operator, U2U_2 is continuous on its domain, and the domain of U1U_1 is contained within the domain of U2U_2 (dom(U1)dom(U2)\text{dom}(U_1) \subseteq \text{dom}(U_2)), then the sum U1+U2U_1 + U_2 is also a closed operator.

theorem

If U1U_1 is closed and U2U_2 is continuous, then U1U2U_1 - U_2 is closed

Let HH and HH' be complex Hilbert spaces, with HH' being a complete space. Let U1U_1 and U2U_2 be partially defined linear operators from HH to HH'. If U1U_1 is a closed operator, U2U_2 is a continuous operator, and the domain of U1U_1 is contained in the domain of U2U_2 (dom(U1)dom(U2)\operatorname{dom}(U_1) \subseteq \operatorname{dom}(U_2)), then the operator U1U2U_1 - U_2 is also a closed operator.

definition

Unitary conjugation uAu1u A u^{-1} of an operator AA

Let HH and HH' be complex Hilbert spaces. Given a partially defined linear operator AA on HH and a unitary operator u:HHu : H \to H', the unitary conjugation uAu1u A u^{-1} is a partially defined linear operator on HH' with domain dom(uAu1)=u(dom(A))={yHu1ydom(A)} \text{dom}(u A u^{-1}) = u(\text{dom}(A)) = \{ y \in H' \mid u^{-1} y \in \text{dom}(A) \} and the action defined by (uAu1)y=u(A(u1y)) (u A u^{-1}) y = u(A(u^{-1} y)) for all ydom(uAu1)y \in \text{dom}(u A u^{-1}). Since uu and u1u^{-1} are C\mathbb{C}-linear, the resulting operator is also C\mathbb{C}-linear.

theorem

xdom(uAu1)    u1xdom(A)x \in \operatorname{dom}(u A u^{-1}) \iff u^{-1} x \in \operatorname{dom}(A)

Let HH and HH' be complex Hilbert spaces. Let AA be a partially defined linear operator on HH and u:HHu : H \to H' be a unitary operator. For any vector xHx \in H', xx is in the domain of the unitarily conjugated operator uAu1u A u^{-1} if and only if u1xu^{-1} x is in the domain of AA. That is, xdom(uAu1)    u1xdom(A)x \in \operatorname{dom}(u A u^{-1}) \iff u^{-1} x \in \operatorname{dom}(A).

theorem

Formula for the action of unitarily conjugated operator (uAu1)x=u(A(u1x))(u A u^{-1}) x = u(A(u^{-1} x))

Let HH and HH' be complex Hilbert spaces. Let AA be a partially defined linear operator on HH and u:HHu : H \to H' be a unitary operator. For any vector xx in the domain of the unitarily conjugated operator uAu1u A u^{-1}, the value of the operator at xx is given by (uAu1)x=u(A(u1x)). (u A u^{-1}) x = u(A(u^{-1} x)). In this expression, the vector u1xu^{-1} x is an element of the domain of AA.

theorem

uu maps dom(A)\text{dom}(A) into dom(uAu1)\text{dom}(u A u^{-1})

Let HH and HH' be complex Hilbert spaces. Let AA be a partially defined linear operator on HH and u:HHu: H \to H' be a unitary operator. For any vector yy in the domain of AA, its image uyu y is contained in the domain of the unitary conjugation uAu1u A u^{-1}.

theorem

Action of Unitary Conjugation (uAu1)(uy)=u(Ay)(u A u^{-1})(u y) = u(A y)

Let HH and HH' be complex Hilbert spaces. Let AA be a partially defined linear operator on HH with domain dom(A)\text{dom}(A), and let u:HHu: H \to H' be a unitary operator. For any vector ydom(A)y \in \text{dom}(A), the action of the unitary conjugation uAu1u A u^{-1} on the image vector uyu y is given by (uAu1)(uy)=u(Ay). (u A u^{-1})(u y) = u(A y).

theorem

Unitary conjugation preserves formal adjointness

Let HH and HH' be complex Hilbert spaces and u:HHu : H \to H' be a unitary operator. Let AA and BB be partially defined linear operators on HH (of type Hl.[C]HH \to_l.[{\mathbb{C}}] H). If AA is a formal adjoint of BB, then the unitary conjugation uAu1u A u^{-1} is a formal adjoint of uBu1u B u^{-1} as partially defined linear operators on HH'.

theorem

Unitary conjugation preserves dense domain

Let HH and HH' be complex Hilbert spaces, and let AA be a partially defined linear operator on HH. If the domain of AA, denoted by dom(A)\text{dom}(A), is dense in HH, then for any unitary operator u:HHu : H \to H', the domain of the unitary conjugation uAu1u A u^{-1} is dense in HH'.

theorem

Unitary conjugation preserves surjectivity of AzIA - zI

Let HH and HH' be complex Hilbert spaces, AA be a partially defined linear operator on HH, and u:HHu : H \to H' be a unitary operator. For any complex scalar zCz \in \mathbb{C}, if the operator AzIA - zI (where II is the identity operator) is surjective, then the operator uAu1zIuAu^{-1} - zI is also surjective on HH'.

theorem

Polarization identity for the inner product Ty,x\langle Ty, x \rangle

Let HH be a complex Hilbert space and TT be a partially defined linear map (a `LinearPMap`) on HH. For any elements xx and yy in the domain of TT, the inner product of TyTy and xx satisfies the polarization identity: Ty,x=14(T(x+y),x+yT(xy),xy+iT(x+iy),x+iyiT(xiy),xiy)\langle Ty, x \rangle = \frac{1}{4} \left( \langle T(x+y), x+y \rangle - \langle T(x-y), x-y \rangle + i \langle T(x+iy), x+iy \rangle - i \langle T(x-iy), x-iy \rangle \right) where ,\langle \cdot, \cdot \rangle denotes the inner product on HH and ii is the imaginary unit.

theorem

Polarization identity for Tx,y\langle Tx, y \rangle

Let HH be a complex Hilbert space and let TT be a partially defined linear operator on HH with domain D(T)D(T). For any x,yD(T)x, y \in D(T), the inner product Tx,y\langle Tx, y \rangle satisfies the polarization identity: Tx,y=14(T(x+y),x+yT(xy),xyiT(x+iy),x+iy+iT(xiy),xiy)\langle Tx, y \rangle = \frac{1}{4} \left( \langle T(x+y), x+y \rangle - \langle T(x-y), x-y \rangle - i \langle T(x+iy), x+iy \rangle + i \langle T(x-iy), x-iy \rangle \right) where ii is the imaginary unit and the inner product ,\langle \cdot, \cdot \rangle is linear in the second argument and conjugate-linear in the first.

theorem

TT is symmetric     xD(T),Tx,xR\iff \forall x \in D(T), \langle Tx, x \rangle \in \mathbb{R}

Let HH be a complex inner product space and TT be a partially defined linear operator on HH. Then TT is symmetric if and only if for every xx in the domain of TT, the inner product of TxTx and xx is a real number, i.e., Tx,x=Tx,x.\langle Tx, x \rangle = \overline{\langle Tx, x \rangle}.

theorem

Symmetric densely defined operators are closable

Let HH be a complex Hilbert space. Let TT be a linear operator with domain D(T)HD(T) \subseteq H (represented as a `LinearPMap`). If TT is symmetric (i.e., Tx,y=x,Ty\langle Tx, y \rangle = \langle x, Ty \rangle for all x,yD(T)x, y \in D(T)) and has a dense domain in HH, then TT is closable.

theorem

A symmetric operator is densely defined and closable iff its domain is dense

Let HH be a complex Hilbert space and TT be a linear operator on HH with a partial domain. If TT is symmetric (i.e., Tx,y=x,Ty\langle Tx, y \rangle = \langle x, Ty \rangle for all x,yx, y in the domain of TT), then TT is a densely defined and closable operator if and only if the domain of TT is dense in HH.

theorem

TT is symmetric     TT\iff T \le T^\dagger

Let HH be a complex Hilbert space. For a densely defined linear operator TT on HH (represented as a `LinearPMap`), TT is symmetric if and only if TTT \le T^\dagger, where TT^\dagger denotes the adjoint operator and the relation TTT \le T^\dagger signifies that TT^\dagger is an extension of TT.

theorem

A symmetric operator TT is self-adjoint     dom(T)=dom(T)\iff \text{dom}(T^*) = \text{dom}(T)

Let HH be a complex Hilbert space and TT be a densely defined symmetric linear operator on HH. Then TT is self-adjoint if and only if the domain of its adjoint TT^* is equal to the domain of TT.

theorem

T+TT + T^\dagger is symmetric

Let HH be a complex Hilbert space and TT be a linear operator on HH with a dense domain. Then the operator T+TT + T^\dagger, representing the sum of TT and its adjoint TT^\dagger, is symmetric.

theorem

If TT is symmetric, then TnT^n is symmetric

Let HH be a complex inner product space and TT be a partially defined linear map on HH. If TT is symmetric, meaning Tx,y=x,Ty\langle Tx, y \rangle = \langle x, Ty \rangle for all x,ydom(T)x, y \in \text{dom}(T), then for any natural number nn, the nn-th power TnT^n (defined via restricted composition) is also symmetric.

theorem

If TT is symmetric, then T-T is symmetric

Let HH be a complex Hilbert space and let TT be a linear partial map (a linear operator defined on a subspace of HH). If TT is symmetric, then its negation T-T is also symmetric.

theorem

The closure of a densely defined symmetric operator is symmetric

Let HH be a complex Hilbert space and TT be a partially defined linear operator on HH with a dense domain. If TT is symmetric, such that Tx,y=x,Ty\langle Tx, y \rangle = \langle x, Ty \rangle for all x,ydom(T)x, y \in \text{dom}(T), then its closure T\overline{T} is also symmetric.

theorem

TT\overline{T} \subseteq T^* for symmetric operators

Let HH be a complex Hilbert space. If TT is a densely defined symmetric linear operator on HH, then the closure of TT, denoted by T\overline{T}, is a restriction of the adjoint of TT, denoted by TT^*. That is, TT\overline{T} \subseteq T^*.

theorem

TT is essentially self-adjoint     dom(T)=dom(T)\iff \text{dom}(T^*) = \text{dom}(\overline{T})

Let HH be a complete complex Hilbert space and let TT be a densely defined symmetric linear operator on HH. Then TT is essentially self-adjoint if and only if the domain of its adjoint TT^* is equal to the domain of its closure T\overline{T}, denoted by dom(T)=dom(T)\text{dom}(T^*) = \text{dom}(\overline{T}).

theorem

Surjectivity of T±iIT \pm iI implies Self-Adjointness for Symmetric Operators

Let HH be a complex Hilbert space and TT be a densely defined symmetric operator on HH. If the operators T+iIT + iI and TiIT - iI (where ii is the imaginary unit and II is the identity operator) are both surjective—that is, their ranges are equal to the entire space HH—then TT is self-adjoint.