Wold decomposition
Wandering subspaces
DefinitionWandering subspace of an isometry#✓fully proved
The wandering subspace of an isometry V: the orthogonal complement of
its range. Vectors here are orthogonal to every vector in the range of V.
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
Used by 8
TheoremPowers of an isometry preserve inner products#✓fully proved
Pulling inner products back along powers of an isometry.
theorem inner_pow_pullback (V : H →ₗᵢ[𝕜] H) (k d : ℕ) (a b : H) :
⟪(V ^ k) a, (V ^ (k + d)) b⟫_𝕜 = ⟪a, (V ^ d) b⟫_𝕜
Used by 1
TheoremPositive powers of an isometry land in its range#✓fully proved
A positive power of an isometry lands in the range of the isometry.
theorem pow_mem_range (V : H →ₗᵢ[𝕜] H) {d : ℕ} (hd : 0 < d) (b : H) :
(V ^ d) b ∈ LinearMap.range (V : H →ₗ[𝕜] H)
Used by 1
TheoremOrthogonality of pushed-forward wandering vectors#✓fully proved
A wandering vector, pushed forward k times, is orthogonal to any vector
pushed forward strictly more times.
theorem inner_pow_wandering (V : H →ₗᵢ[𝕜] H) {k l : ℕ} (hkl : k < l) (a : H)
(ha : a ∈ wanderingSub V) (b : H) :
⟪(V ^ k) a, (V ^ l) b⟫_𝕜 = 0
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
TheoremOrthogonality of pushed-forward wandering vectors, symmetric form#✓fully proved
The symmetric form of inner_pow_wandering.
theorem inner_pow_wandering' (V : H →ₗᵢ[𝕜] H) {k l : ℕ} (hkl : k < l) (a : H)
(ha : a ∈ wanderingSub V) (b : H) :
⟪(V ^ l) b, (V ^ k) a⟫_𝕜 = 0
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 2 · Used by 1
Complete
TheoremWandering subspace and range are complementary#✓fully proved
The ambient space splits as the wandering subspace plus the range.
theorem isCompl_wandering (V : H →ₗᵢ[𝕜] H) :
IsCompl (wanderingSub V) (LinearMap.range (V : H →ₗ[𝕜] H))
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 1 · Used by 1
TheoremOne-step Wold splitting into wandering vector plus range#✓fully proved
One step of the Wold decomposition: every vector splits as a wandering vector plus a vector in the range.
theorem exists_wandering_add_range (V : H →ₗᵢ[𝕜] H) (x : H) :
∃ w ∈ wanderingSub V, ∃ y : H, x = w + V y
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 2 · Used by 1
TheoremFinite Wold decomposition#✓fully proved
Finite Wold decomposition: every vector is a finite sum of pushed-forward
wandering vectors plus a remainder in the range of V ^ n. The wandering
coefficients vanish at and beyond n.
theorem exists_wold_decomposition (V : H →ₗᵢ[𝕜] H) (x : H) (n : ℕ) :
∃ w : ℕ → H, ∃ y : H, (∀ k, w k ∈ wanderingSub V) ∧ (∀ k ≥ n, w k = 0) ∧
x = (∑ k ∈ Finset.range n, (V ^ k) (w k)) + (V ^ n) y
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 2
TheoremPythagorean identity for the Wold decomposition#✓fully proved
Pythagoras for a Wold decomposition: the squared norm of a vector is the sum of the squared norms of its wandering coefficients plus the squared norm of the remainder.
theorem wold_norm_sq (V : H →ₗᵢ[𝕜] H) (n : ℕ) (w : ℕ → H)
(hw : ∀ k, w k ∈ wanderingSub V) (y : H) (x : H)
(hx : x = (∑ k ∈ Finset.range n, (V ^ k) (w k)) + (V ^ n) y) :
‖x‖ ^ 2 = (∑ k ∈ Finset.range n, ‖w k‖ ^ 2) + ‖y‖ ^ 2
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 1
TheoremUniqueness of the finite Wold decomposition#✓fully proved
Uniqueness of the Wold decomposition at level n: the remainder and the
finite wandering sum are uniquely determined.
theorem wold_decomposition_unique (V : H →ₗᵢ[𝕜] H) (n : ℕ) (w₁ w₂ : ℕ → H)
(hw₁ : ∀ k, w₁ k ∈ wanderingSub V) (hw₂ : ∀ k, w₂ k ∈ wanderingSub V)
(y₁ y₂ : H) (x : H)
(hx₁ : x = (∑ k ∈ Finset.range n, (V ^ k) (w₁ k)) + (V ^ n) y₁)
(hx₂ : x = (∑ k ∈ Finset.range n, (V ^ k) (w₂ k)) + (V ^ n) y₂) :
y₁ = y₂ ∧
(∑ k ∈ Finset.range n, (V ^ k) (w₁ k)) =
∑ k ∈ Finset.range n, (V ^ k) (w₂ k)
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 2
DefinitionResidual subspace of an isometry#✓fully proved
The residual subspace of an isometry: vectors in the range of every
power of V.
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
Used by 6
TheoremAn isometry maps its residual subspace into itself#✓fully proved
The isometry maps its residual subspace into itself.
theorem mapsTo_residual (V : H →ₗᵢ[𝕜] H) :
∀ x ∈ residualSub V, V x ∈ residualSub V
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:292
TheoremResidual subspace is orthogonal to the wandering part#✓fully proved
The residual subspace is orthogonal to every pushed-forward wandering vector: the wandering part and the residual part of Wold's decomposition are orthogonal.
theorem inner_residual_wandering (V : H →ₗᵢ[𝕜] H) (x : H) (hx : x ∈ residualSub V)
(k : ℕ) (a : H) (ha : a ∈ wanderingSub V) :
⟪x, (V ^ k) a⟫_𝕜 = 0
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:292
noncomputable def wanderingSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
(LinearMap.range (V : H →ₗ[𝕜] H))ᗮ
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:73
Uses 3
TheoremIsometry restricted to the residual subspace is bijective#✓fully proved
The restriction of an isometry to its residual subspace is bijective.
theorem restrict_residual_bijective (V : H →ₗᵢ[𝕜] H) :
Function.Bijective (((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V) :
residualSub V →ₗ[𝕜] residualSub V))
theorem mapsTo_residual (V : H →ₗᵢ[𝕜] H) :
∀ x ∈ residualSub V, V x ∈ residualSub V
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:297
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:292
Uses 2 · Used by 1
TheoremIsometry restricted to the residual subspace preserves inner products#✓fully proved
The restriction of an isometry to its residual subspace preserves inner products.
theorem inner_restrict_residual (V : H →ₗᵢ[𝕜] H) (x y : residualSub V) :
⟪((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V) x : H),
((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V) y : H)⟫_𝕜 =
⟪(x : H), (y : H)⟫_𝕜
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:292
theorem mapsTo_residual (V : H →ₗᵢ[𝕜] H) :
∀ x ∈ residualSub V, V x ∈ residualSub V
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:297
Uses 2 · Used by 1
DefinitionUnitary part of an isometry on its residual subspace#✓fully proved
The isometry restricts to a unitary operator (a linear isometric equivalence) on its residual subspace.
noncomputable def woldResidualEquiv (V : H →ₗᵢ[𝕜] H) :
residualSub V ≃ₗᵢ[𝕜] residualSub V :=
LinearEquiv.isometryOfInner
(LinearEquiv.ofBijective ((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V))
(restrict_residual_bijective V))
(inner_restrict_residual V)
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:292
theorem mapsTo_residual (V : H →ₗᵢ[𝕜] H) :
∀ x ∈ residualSub V, V x ∈ residualSub V
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:297
theorem restrict_residual_bijective (V : H →ₗᵢ[𝕜] H) :
Function.Bijective (((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V) :
residualSub V →ₗ[𝕜] residualSub V))
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:330
theorem inner_restrict_residual (V : H →ₗᵢ[𝕜] H) (x y : residualSub V) :
⟪((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V) x : H),
((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V) y : H)⟫_𝕜 =
⟪(x : H), (y : H)⟫_𝕜
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:367
Uses 4 · Used by 1
TheoremAction of the residual unitary of an isometry#✓fully proved
Applying the residual unitary is the same as applying V and viewing the
result in the ambient space.
@[simp]
theorem woldResidualEquiv_apply_coe (V : H →ₗᵢ[𝕜] H) (x : residualSub V) :
((woldResidualEquiv V x : residualSub V) : H) = V (x : H)
noncomputable def residualSub (V : H →ₗᵢ[𝕜] H) : Submodule 𝕜 H :=
⨅ n, LinearMap.range ((V ^ n : H →ₗᵢ[𝕜] H) : H →ₗ[𝕜] H)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:292
noncomputable def woldResidualEquiv (V : H →ₗᵢ[𝕜] H) :
residualSub V ≃ₗᵢ[𝕜] residualSub V :=
LinearEquiv.isometryOfInner
(LinearEquiv.ofBijective ((V : H →ₗ[𝕜] H).restrict (mapsTo_residual V))
(restrict_residual_bijective V))
(inner_restrict_residual V)
MathlibExt/Analysis/InnerProductSpace/WoldDecomposition.lean:377
Uses 2
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.