Physlib

PhyslibAlpha.AlgebraicFramework.HilbertSpace.TraceClass.Polar

Polar decomposition for bounded operators

Ported from `unbounded-alpha-public`'s `TraceClass/Polar.lean`, restated against this repo's `H →L[ℂ] H` (no bundled trace-class subtype involved at all here — this file only uses `CFC.abs`).

For a bounded operator `T`, its absolute value `S = |T|` satisfies `‖T x‖ = ‖S x‖`. The resulting isometry from `range S` to `H` is extended to the closure of that range, producing a genuine partial isometry `polarFactor T` with `polarFactor T * |T| = T` and, crucially, `star (polarFactor T) * T = |T|` for *every* bounded `T` — the fact that finally lets trace-class theory cross the non-self-adjoint boundary in `GeneralIdeal.lean`/`GeneralProduct.lean`.

The partial-isometry adjoint identity

24 declarations

definition

Absolute value ∣T∣|T| of a bounded operator TT

For a bounded linear operator T∈L(H)T \in \mathcal{L}(H) acting on a complex Hilbert space HH, the function returns the absolute value of the operator ∣T∣|T|, which is defined by applying the absolute value function to TT via the continuous functional calculus.

definition

Closed range of ∣T∣|T|

For a bounded linear operator T∈L(H)T \in \mathcal{L}(H) on a complex Hilbert space HH, let ∣T∣|T| denote its absolute value. This definition represents the topological closure of the range of ∣T∣|T|, which is the closed subspace of HH denoted by ran(∣T∣)‾\overline{\text{ran}(|T|)}.

definition

Absolute value map ∣T∣|T| into its closed range ran(∣T∣)‾\overline{\text{ran}(|T|)}

For a bounded linear operator T∈L(H)T \in \mathcal{L}(H) acting on a complex Hilbert space HH, let ∣T∣|T| denote its absolute value. The map `absIntoRange` is the linear map ∣T∣:H→ran(∣T∣)‾|T| : H \to \overline{\text{ran}(|T|)} defined by restricting the codomain of the absolute value operator to the closure of its range.

definition

Inclusion of ran(∣T∣)\text{ran}(|T|) into ran(∣T∣)‾\overline{\text{ran}(|T|)}

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, ran(∣T∣)\text{ran}(|T|) denote its range, and ran(∣T∣)‾\overline{\text{ran}(|T|)} denote the topological closure of that range. This definition is the canonical continuous linear inclusion map from ran(∣T∣)\text{ran}(|T|) into ran(∣T∣)‾\overline{\text{ran}(|T|)}.

theorem

rangeInClosure(T)(x)=x\text{rangeInClosure}(T)(x) = x

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, ran(∣T∣)\text{ran}(|T|) denote its range, and ran(∣T∣)‾\overline{\text{ran}(|T|)} denote the topological closure of that range. Let ι:ran(∣T∣)→ran(∣T∣)‾\iota: \text{ran}(|T|) \to \overline{\text{ran}(|T|)} be the canonical continuous linear inclusion map (denoted as `rangeInClosure T`). For any vector x∈ran(∣T∣)x \in \text{ran}(|T|), it holds that ι(x)=x\iota(x) = x when the result is viewed as an element of the Hilbert space HH.

theorem

The range ran(∣T∣)\text{ran}(|T|) is dense in its closure ran(∣T∣)‾\overline{\text{ran}(|T|)}

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, ran(∣T∣)\text{ran}(|T|) denote its range, and ran(∣T∣)‾\overline{\text{ran}(|T|)} denote the topological closure of that range. The canonical inclusion map ι:ran(∣T∣)→ran(∣T∣)‾\iota : \text{ran}(|T|) \to \overline{\text{ran}(|T|)} has a dense range.

theorem

∣T∣clx=ι(∣T∣x)|T|_{\text{cl}} x = \iota(|T|x)

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, ran(∣T∣)\text{ran}(|T|) its range, and ran(∣T∣)‾\overline{\text{ran}(|T|)} the closure of its range. Let ∣T∣cl:H→ran(∣T∣)‾|T|_{\text{cl}} : H \to \overline{\text{ran}(|T|)} be the operator ∣T∣|T| with its codomain restricted to its closed range, and let ι:ran(∣T∣)→ran(∣T∣)‾\iota : \text{ran}(|T|) \to \overline{\text{ran}(|T|)} be the canonical inclusion map. For any vector x∈Hx \in H, it holds that ∣T∣clx=ι(∣T∣x)|T|_{\text{cl}} x = \iota(|T|x) where ∣T∣x|T|x in the right-hand side is treated as an element of the subspace ran(∣T∣)\text{ran}(|T|).

theorem

The range of ∣T∣|T| is dense in its closure ran(∣T∣)‾\overline{\text{ran}(|T|)}

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, and let ran(∣T∣)‾\overline{\text{ran}(|T|)} denote the closure of its range. The range of the linear map ∣T∣:H→ran(∣T∣)‾|T| : H \to \overline{\text{ran}(|T|)} is dense in ran(∣T∣)‾\overline{\text{ran}(|T|)}.

theorem

(∣T∣→ran(∣T∣)‾)x=∣T∣x(|T|_{\to \overline{\text{ran}(|T|)}}) x = |T| x

For a bounded linear operator T∈L(H)T \in \mathcal{L}(H) acting on a complex Hilbert space HH, let ∣T∣|T| denote its absolute value. Let ∣T∣→ran(∣T∣)‾|T|_{\to \overline{\text{ran}(|T|)}} be the linear map from HH to the closure of the range of ∣T∣|T| (denoted as `absIntoRange T`). For any vector x∈Hx \in H, applying this map to xx and viewing the result as an element of HH is equal to ∣T∣x|T|x.

theorem

∥Tx∥=∥∣T∣x∥\|T x\| = \||T| x\|

Let HH be a complex Hilbert space. For any bounded linear operator T∈L(H)T \in \mathcal{L}(H) and any vector x∈Hx \in H, the norm of TxT x is equal to the norm of ∣T∣x|T| x, where ∣T∣|T| denotes the absolute value of the operator TT. That is, ∥Tx∥=∥∣T∣x∥\|T x\| = \||T| x\|.

theorem

∥∣T∣x∥ran(∣T∣)‾=∥∣T∣x∥\||T|x\|_{\overline{\text{ran}(|T|)}} = \||T|x\|

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT. For any vector x∈Hx \in H, the norm of ∣T∣x|T|x viewed as an element of the closed range ran(∣T∣)‾\overline{\text{ran}(|T|)} is equal to its norm viewed as an element of HH. That is, ∥∣T∣x∥=∥∣T∣x∥\||T|x\| = \||T|x\|.

instance

ran(∣T∣)‾\overline{\text{ran}(|T|)} is a complete space

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT. The topological closure of the range of ∣T∣|T|, denoted by ran(∣T∣)‾\overline{\text{ran}(|T|)}, is a complete space.

definition

Partial isometry U:ran(∣T∣)‾→HU: \overline{\text{ran}(|T|)} \to H such that U∣T∣=TU|T| = T

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT and ran(∣T∣)‾\overline{\text{ran}(|T|)} denote the topological closure of the range of ∣T∣|T|. The map `polarOnRange T` is the continuous linear map U:ran(∣T∣)‾→HU : \overline{\text{ran}(|T|)} \to H obtained by extending the assignment U(∣T∣x)=TxU(|T|x) = Tx from the range of ∣T∣|T| to its closure. This extension is well-defined because ∥Tx∥=∥∣T∣x∥\|Tx\| = \||T|x\| for all x∈Hx \in H.

theorem

The partial isometry UU satisfies U(∣T∣x)=TxU(|T|x) = Tx

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, and let U:ran(∣T∣)‾→HU: \overline{\text{ran}(|T|)} \to H be the partial isometry (denoted as `polarOnRange T`) obtained by extending the assignment U(∣T∣x)=TxU(|T|x) = Tx to the closure of the range of ∣T∣|T|. Then for any x∈Hx \in H, it holds that U(∣T∣x)=TxU(|T|x) = Tx where ∣T∣x|T|x is viewed as an element of the closed subspace ran(∣T∣)‾\overline{\text{ran}(|T|)}.

theorem

The partial isometry UU on ran(∣T∣)‾\overline{\text{ran}(|T|)} is contractive: ∥Ux∥≤∥x∥\|Ux\| \le \|x\|

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT and ran(∣T∣)‾\overline{\text{ran}(|T|)} be the closure of its range. Let U:ran(∣T∣)‾→HU : \overline{\text{ran}(|T|)} \to H be the continuous linear map (denoted by `polarOnRange T`) obtained by extending the assignment U(∣T∣v)=TvU(|T|v) = Tv for v∈Hv \in H. Then for any x∈ran(∣T∣)‾x \in \overline{\text{ran}(|T|)}, the norm of UxU x is bounded by the norm of xx: ∥Ux∥≤∥x∥\|U x\| \le \|x\|

definition

Partial isometry UU in the polar decomposition T=U∣T∣T = U|T|

For a bounded linear operator T∈L(H)T \in \mathcal{L}(H) on a complex Hilbert space HH, the `polarFactor` is the bounded linear operator U:H→HU : H \to H defined as the composition U=V∘PU = V \circ P, where P:H→ran(∣T∣)‾P: H \to \overline{\text{ran}(|T|)} is the orthogonal projection onto the closure of the range of the absolute value ∣T∣|T|, and V:ran(∣T∣)‾→HV: \overline{\text{ran}(|T|)} \to H is the partial isometry (`polarOnRange T`) that maps ∣T∣x|T|x to TxTx for all x∈Hx \in H. This construction results in a partial isometry UU that satisfies T=U∣T∣T = U|T| and acts as zero on the orthogonal complement (ran(∣T∣))⊥(\text{ran}(|T|))^\perp.

theorem

U∣T∣=TU|T| = T for the polar factor UU and absolute value ∣T∣|T|

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT and UU (the `polarFactor` of TT) be the partial isometry associated with TT in its polar decomposition. Then for any vector x∈Hx \in H, the operator UU maps ∣T∣x|T|x to TxTx: U(∣T∣x)=TxU(|T|x) = Tx

theorem

∥Ux∥≤∥x∥\|U x\| \le \|x\| for the polar factor UU of a bounded operator TT

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let UU (denoted as `polarFactor T`) be the partial isometry obtained from the polar decomposition of TT. For any vector x∈Hx \in H, the norm of the image of xx under UU is bounded by the norm of xx: ∥Ux∥≤∥x∥\|U x\| \leq \|x\|

theorem

The operator norm of the polar factor UU satisfies ∥U∥≤1\|U\| \leq 1

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let UU (denoted as `polarFactor T`) be the partial isometry in the polar decomposition of TT, such that T=U∣T∣T = U|T|. Then the operator norm of UU satisfies ∥U∥≤1\|U\| \le 1.

theorem

Polar Decomposition U∣T∣=TU|T| = T

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT and UU denote the partial isometry given by the polar factor of TT. Then the composition of UU and ∣T∣|T| satisfies U∣T∣=TU |T| = T.

theorem

∥Uy∥=∥y∥\|U y\| = \|y\| for y∈ran(∣T∣)‾y \in \overline{\text{ran}(|T|)}

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT and ran(∣T∣)‾\overline{\text{ran}(|T|)} denote the topological closure of the range of ∣T∣|T|. Let U:ran(∣T∣)‾→HU: \overline{\text{ran}(|T|)} \to H be the continuous linear map (denoted as `polarOnRange T`) obtained by extending the assignment U(∣T∣x)=TxU(|T|x) = Tx to the closure of the range. For any vector y∈ran(∣T∣)‾y \in \overline{\text{ran}(|T|)}, it holds that ∥Uy∥=∥y∥\|Uy\| = \|y\|.

theorem

The adjoint of U=polarOnRange TU = \text{polarOnRange } T is its left inverse: U∗U=IU^* U = I

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT, and let ran(∣T∣)‾\overline{\text{ran}(|T|)} be the topological closure of its range. Let U:ran(∣T∣)‾→HU: \overline{\text{ran}(|T|)} \to H be the continuous linear map `polarOnRange T` (the isometry that extends U(∣T∣x)=TxU(|T|x) = Tx). Then U∗∘U=IU^* \circ U = I, where U∗U^* is the adjoint of UU and II is the identity operator on ran(∣T∣)‾\overline{\text{ran}(|T|)}.

theorem

U∗U=Pran(∣T∣)‾U^* U = P_{\overline{\text{ran}(|T|)}}

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let UU be the partial isometry (the `polarFactor`) such that T=U∣T∣T = U|T|, where ∣T∣|T| is the absolute value of TT. Then the operator U∗UU^* U is equal to the orthogonal projection onto the topological closure of the range of ∣T∣|T|, denoted by ran(∣T∣)‾\overline{\text{ran}(|T|)}.

theorem

U∗T=∣T∣U^* T = |T| for the Polar Factor UU of a Bounded Operator TT

Let HH be a complex Hilbert space and T∈L(H)T \in \mathcal{L}(H) be a bounded linear operator. Let ∣T∣|T| denote the absolute value of TT and UU denote the polar factor of TT (the partial isometry provided by the polar decomposition). Then the product of the adjoint of the polar factor and the operator TT recovers the absolute value, i.e., U∗T=∣T∣U^* T = |T|.