Diagonal operators
DefinitionBounded diagonal operator on a Hilbert basis#✓fully proved
The bounded diagonal continuous linear map with coefficients a in the
Hilbert basis b. Its uniform coefficient bound is part of the constructor.
noncomputable def diagonalOperator (b : HilbertBasis ι 𝕜 H) (a : ι → 𝕜) {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) : H →L[𝕜] H :=
b.repr.symm.toLinearIsometry.toContinuousLinearMap.comp
((lp.mapCLM 2 (fun i => a i • ContinuousLinearMap.id 𝕜 𝕜) (abs_nonneg C) (fun i => by
simpa only [norm_smul, ContinuousLinearMap.norm_id, mul_one] using
(hC i).trans (le_abs_self C))).comp
b.repr.toLinearIsometry.toContinuousLinearMap)
Used by 8
TheoremCoordinates of a diagonal operator#✓fully proved
Let \((b_i)_{i \in \iota}\) be a Hilbert basis of an inner product space \(H\) over \(\mathbb{K}\) (\(\mathbb{R}\) or \(\mathbb{C}\)), with inner product \(\langle \cdot, \cdot \rangle\) conjugate-linear in the first argument, let \(a : \iota \to \mathbb{K}\) satisfy \(|a_i| \le C\) for all \(i\), and let \(D\) be the diagonal operator with coefficients \(a\). For every \(x \in H\) and \(i \in \iota\), the \(i\)-th coordinate of \(D x\) in \(\ell^2(\iota)\) with respect to the basis \((b_i)\) is \(a_i \langle b_i, x \rangle\).
theorem repr_diagonalOperator_apply {b : HilbertBasis ι 𝕜 H} {a : ι → 𝕜} {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) (x : H) (i : ι) :
b.repr (diagonalOperator b a hC x) i = a i * ⟪b i, x⟫_𝕜
noncomputable def diagonalOperator (b : HilbertBasis ι 𝕜 H) (a : ι → 𝕜) {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) : H →L[𝕜] H :=
b.repr.symm.toLinearIsometry.toContinuousLinearMap.comp
((lp.mapCLM 2 (fun i => a i • ContinuousLinearMap.id 𝕜 𝕜) (abs_nonneg C) (fun i => by
simpa only [norm_smul, ContinuousLinearMap.norm_id, mul_one] using
(hC i).trans (le_abs_self C))).comp
b.repr.toLinearIsometry.toContinuousLinearMap)
MathlibExt/Analysis/InnerProductSpace/HilbertBasisDiagonal.lean:50
Uses 1 · Used by 2
TheoremSeries formula for a diagonal operator#✓fully proved
Let \((b_i)_{i \in \iota}\) be a Hilbert basis of an inner product space \(H\) over \(\mathbb{K}\) (\(\mathbb{R}\) or \(\mathbb{C}\)), with inner product \(\langle \cdot, \cdot \rangle\) conjugate-linear in the first argument, and let \(a : \iota \to \mathbb{K}\) satisfy \(|a_i| \le C\) for all \(i\). The diagonal operator \(D\) with coefficients \(a\) satisfies, for every \(x \in H\),
where the sum is the unconditional sum over \(\iota\) (taken to be \(0\) if the family is not summable).
theorem diagonalOperator_eq_tsum {b : HilbertBasis ι 𝕜 H} {a : ι → 𝕜} {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) (x : H) :
diagonalOperator b a hC x = ∑' i, (a i * ⟪b i, x⟫_𝕜) • b i
noncomputable def diagonalOperator (b : HilbertBasis ι 𝕜 H) (a : ι → 𝕜) {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) : H →L[𝕜] H :=
b.repr.symm.toLinearIsometry.toContinuousLinearMap.comp
((lp.mapCLM 2 (fun i => a i • ContinuousLinearMap.id 𝕜 𝕜) (abs_nonneg C) (fun i => by
simpa only [norm_smul, ContinuousLinearMap.norm_id, mul_one] using
(hC i).trans (le_abs_self C))).comp
b.repr.toLinearIsometry.toContinuousLinearMap)
MathlibExt/Analysis/InnerProductSpace/HilbertBasisDiagonal.lean:50
TheoremAction of a diagonal operator on the basis#✓fully proved
Let \((b_i)_{i \in \iota}\) be a Hilbert basis of an inner product space \(H\) over \(\mathbb{K}\) (\(\mathbb{R}\) or \(\mathbb{C}\)), let \(a : \iota \to \mathbb{K}\) satisfy \(|a_i| \le C\) for all \(i\), and let \(D\) be the diagonal operator with coefficients \(a\). Then \(D b_j = a_j b_j\) for every \(j \in \iota\).
theorem diagonalOperator_apply_basis {b : HilbertBasis ι 𝕜 H} {a : ι → 𝕜} {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) (j : ι) :
diagonalOperator b a hC (b j) = a j • b j
noncomputable def diagonalOperator (b : HilbertBasis ι 𝕜 H) (a : ι → 𝕜) {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) : H →L[𝕜] H :=
b.repr.symm.toLinearIsometry.toContinuousLinearMap.comp
((lp.mapCLM 2 (fun i => a i • ContinuousLinearMap.id 𝕜 𝕜) (abs_nonneg C) (fun i => by
simpa only [norm_smul, ContinuousLinearMap.norm_id, mul_one] using
(hC i).trans (le_abs_self C))).comp
b.repr.toLinearIsometry.toContinuousLinearMap)
MathlibExt/Analysis/InnerProductSpace/HilbertBasisDiagonal.lean:50
Uses 2
TheoremSquare of a diagonal operator#✓fully proved
Let \((b_i)_{i \in \iota}\) be a Hilbert basis of an inner product space \(H\) over \(\mathbb{K}\) (\(\mathbb{R}\) or \(\mathbb{C}\)), and let \(a : \iota \to \mathbb{K}\) satisfy \(|a_i| \le C\) for all \(i\). Let \(D_a\) be the diagonal operator with coefficients \(a\), and let \(D_{a^2}\) be the diagonal operator with coefficients \(a_i^2\), which satisfy \(|a_i^2| \le C^2\). Then \(D_a(D_a x) = D_{a^2} x\) for every \(x \in H\).
theorem diagonalOperator_sq {b : HilbertBasis ι 𝕜 H} {a : ι → 𝕜} {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) (x : H) :
diagonalOperator b a hC (diagonalOperator b a hC x) =
diagonalOperator b (fun i => (a i) ^ 2) (C := C ^ 2) (fun i => by
rw [norm_pow]
exact pow_le_pow_left₀ (norm_nonneg _) (hC i) 2) x
noncomputable def diagonalOperator (b : HilbertBasis ι 𝕜 H) (a : ι → 𝕜) {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) : H →L[𝕜] H :=
b.repr.symm.toLinearIsometry.toContinuousLinearMap.comp
((lp.mapCLM 2 (fun i => a i • ContinuousLinearMap.id 𝕜 𝕜) (abs_nonneg C) (fun i => by
simpa only [norm_smul, ContinuousLinearMap.norm_id, mul_one] using
(hC i).trans (le_abs_self C))).comp
b.repr.toLinearIsometry.toContinuousLinearMap)
MathlibExt/Analysis/InnerProductSpace/HilbertBasisDiagonal.lean:50
Uses 2 · Used by 1
TheoremScalar multiplication on \(\ell^2\) is pointwise#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\). For every \(c \in \mathbb{K}\), every \(x \in \ell^2(\iota, \mathbb{K})\) and every \(i \in \iota\), the \(i\)-th coordinate of \(c x\) is \((c x)_i = c \, x_i\).
theorem lp_smul_apply (c : 𝕜) (x : ℓ²(ι, 𝕜)) (i : ι) : (c • x) i = c * x i
TheoremBounded multipliers preserve \(\ell^2\)#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\). If \(a : \iota \to \mathbb{K}\) is bounded, that is, \(|a_i| \le C\) for some real \(C\) and all \(i\), then for every \(x \in \ell^2(\iota, \mathbb{K})\) the sequence \((a_i x_i)_{i \in \iota}\) lies in \(\ell^2(\iota, \mathbb{K})\), that is, the family \((|a_i x_i|^2)_{i \in \iota}\) is summable.
theorem memℓp_diagonal (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C)
(x : ℓ²(ι, 𝕜)) : Memℓp (fun i => a i * x i) 2
AbbreviationDiagonal operator on ℓ²#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type and let \(a : \iota \to \mathbb{K}\) be a bounded sequence, that is, \(\|a_i\| \le C\) for some real \(C\) and all \(i \in \iota\). The diagonal operator \(D_a\) with coefficients \(a\) is the bounded linear operator on \(\ell^2(\iota, \mathbb{K})\) that sends \(x\) to the sequence \((a_i x_i)_{i \in \iota}\).
It is defined as the Hilbert-basis diagonal operator for the standard Hilbert basis of \(\ell^2(\iota, \mathbb{K})\), using the bound \(\max(C, 0)\) for a chosen bound \(C\), so that empty index types also admit a nonnegative bound.
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
noncomputable def diagonalOperator (b : HilbertBasis ι 𝕜 H) (a : ι → 𝕜) {C : ℝ}
(hC : ∀ i, ‖a i‖ ≤ C) : H →L[𝕜] H :=
b.repr.symm.toLinearIsometry.toContinuousLinearMap.comp
((lp.mapCLM 2 (fun i => a i • ContinuousLinearMap.id 𝕜 𝕜) (abs_nonneg C) (fun i => by
simpa only [norm_smul, ContinuousLinearMap.norm_id, mul_one] using
(hC i).trans (le_abs_self C))).comp
b.repr.toLinearIsometry.toContinuousLinearMap)
MathlibExt/Analysis/InnerProductSpace/HilbertBasisDiagonal.lean:50
Uses 1 · Used by 24
TheoremCoordinates of a diagonal operator on ℓ²#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type, let \(a : \iota \to \mathbb{K}\) be a bounded sequence and let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(a\). For every \(x \in \ell^2(\iota, \mathbb{K})\) and every \(i \in \iota\), the \(i\)-th coordinate of \(D_a x\) is \(a_i x_i\).
theorem diagonalCLM_apply (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C)
(x : ℓ²(ι, 𝕜)) (i : ι) :
diagonalCLM a hbdd x i = a i * x i
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 3 · Used by 11
AbbreviationDiagonal operator on ℓ² as a linear map#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type and let \(a : \iota \to \mathbb{K}\) be a bounded sequence. This is the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(a\), regarded as a \(\mathbb{K}\)-linear map \(\ell^2(\iota, \mathbb{K}) \to \ell^2(\iota, \mathbb{K})\) by forgetting its continuity.
noncomputable abbrev diagonalLinMap (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →ₗ[𝕜] ℓ²(ι, 𝕜) :=
(diagonalCLM a hbdd).toLinearMap
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 1 · Used by 2
TheoremCoordinates of the diagonal linear map on ℓ²#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type and let \(a : \iota \to \mathbb{K}\) be a bounded sequence. For every \(x \in \ell^2(\iota, \mathbb{K})\) and every \(i \in \iota\), the diagonal operator with coefficients \(a\), regarded as a linear map, sends \(x\) to a sequence whose \(i\)-th coordinate is \(a_i x_i\).
theorem diagonalLinMap_apply (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C)
(x : ℓ²(ι, 𝕜)) (i : ι) :
diagonalLinMap a hbdd x i = a i * x i
noncomputable abbrev diagonalLinMap (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →ₗ[𝕜] ℓ²(ι, 𝕜) :=
(diagonalCLM a hbdd).toLinearMap
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:105
Uses 2 · Used by 1
TheoremPointwise bound for diagonal operators#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be bounded, and let \(D_a\) be the diagonal operator \(x \mapsto (a_i x_i)_{i \in \iota}\) on \(\ell^2(\iota, \mathbb{K})\), viewed as a linear map. If \(C \ge 0\) and \(|a_i| \le C\) for every \(i \in \iota\), then \(\|D_a x\| \le C \|x\|\) for every \(x \in \ell^2(\iota, \mathbb{K})\).
theorem norm_diagonalLinMap_le (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C)
{C : ℝ} (hC0 : 0 ≤ C) (hC : ∀ i, ‖a i‖ ≤ C) (x : ℓ²(ι, 𝕜)) :
‖diagonalLinMap a hbdd x‖ ≤ C * ‖x‖
noncomputable abbrev diagonalLinMap (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →ₗ[𝕜] ℓ²(ι, 𝕜) :=
(diagonalCLM a hbdd).toLinearMap
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:105
Uses 3 · Used by 1
TheoremDiagonal operators on standard basis vectors#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type, let \(a : \iota \to \mathbb{K}\) be a bounded sequence and let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(a\). For \(i \in \iota\) let \(e_i \in \ell^2(\iota, \mathbb{K})\) be the sequence equal to \(1\) at \(i\) and \(0\) elsewhere. Then for every \(i \in \iota\) and every \(c \in \mathbb{K}\) we have \(D_a(c\,e_i) = a_i \, (c\,e_i)\).
theorem diagonalCLM_single [DecidableEq ι] (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C)
(i : ι) (c : 𝕜) :
diagonalCLM a hbdd (lp.single 2 i c) = a i • lp.single 2 i c
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 3
TheoremThe diagonal operator with all coefficients one is the identity#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\) and let \(\iota\) be a type. The diagonal operator on \(\ell^2(\iota, \mathbb{K})\) whose coefficients all equal \(1\) is the identity operator.
theorem diagonalCLM_one : diagonalCLM (fun _ : ι => (1 : 𝕜)) ⟨1, fun _ => by simp⟩ = 1
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 1
TheoremComposition of diagonal operators#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type, and for a bounded sequence \(c : \iota \to \mathbb{K}\) let \(D_c\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(c\). For bounded sequences \(a, b : \iota \to \mathbb{K}\), we have \(D_b \circ D_a = D_{ba}\), where \(ba\) is the pointwise product \(i \mapsto b_i a_i\).
theorem diagonalCLM_comp (a b : ι → 𝕜) (ha : ∃ C, ∀ i, ‖a i‖ ≤ C)
(hb : ∃ C, ∀ i, ‖b i‖ ≤ C) :
diagonalCLM b hb ∘L diagonalCLM a ha =
diagonalCLM (fun i => b i * a i) (by
obtain ⟨Ca, _hCa0, hCa⟩ := exists_nonneg_bound ha
obtain ⟨Cb, hCb0, hCb⟩ := exists_nonneg_bound hb
exact ⟨Cb * Ca, fun i => by
rw [norm_mul]
exact mul_le_mul (hCb i) (hCa i) (norm_nonneg _) hCb0⟩)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 1
TheoremDifference of diagonal operators#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type, and for a bounded sequence \(c : \iota \to \mathbb{K}\) let \(D_c\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(c\). For bounded sequences \(a, b : \iota \to \mathbb{K}\), we have \(D_a - D_b = D_{a - b}\), where \(a - b\) is the sequence \(i \mapsto a_i - b_i\).
theorem diagonalCLM_sub (a b : ι → 𝕜) (ha : ∃ C, ∀ i, ‖a i‖ ≤ C)
(hb : ∃ C, ∀ i, ‖b i‖ ≤ C) :
diagonalCLM a ha - diagonalCLM b hb =
diagonalCLM (fun i => a i - b i) (by
obtain ⟨Ca, _hCa0, hCa⟩ := exists_nonneg_bound ha
obtain ⟨Cb, _hCb0, hCb⟩ := exists_nonneg_bound hb
exact ⟨Ca + Cb, fun i => by
calc ‖a i - b i‖ ≤ ‖a i‖ + ‖b i‖ := norm_sub_le _ _
_ ≤ Ca + Cb := add_le_add (hCa i) (hCb i)⟩)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 2
TheoremOperator norm bound for diagonal operators#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be bounded, and let \(D_a\) be the diagonal operator \(x \mapsto (a_i x_i)_{i \in \iota}\) on \(\ell^2(\iota, \mathbb{K})\). If \(C \ge 0\) and \(|a_i| \le C\) for every \(i \in \iota\), then the operator norm satisfies \(\|D_a\| \le C\).
theorem norm_diagonalCLM_le (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C)
{C : ℝ} (hC0 : 0 ≤ C) (hC : ∀ i, ‖a i‖ ≤ C) :
‖diagonalCLM a hbdd‖ ≤ C
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 3
TheoremOperator norm of a diagonal operator#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be bounded, and let \(D_a\) be the diagonal operator \(x \mapsto (a_i x_i)_{i \in \iota}\) on \(\ell^2(\iota, \mathbb{K})\). Then the operator norm of \(D_a\) is \(\|D_a\| = \sup_{i \in \iota} |a_i|\), where the supremum is taken to be \(0\) when \(\iota\) is empty.
theorem norm_diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
‖diagonalCLM a hbdd‖ = ⨆ i, ‖a i‖
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 3 · Used by 2
TheoremAdjoint of a diagonal operator#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type and let \(a : \iota \to \mathbb{K}\) be a bounded sequence. The adjoint of the diagonal operator \(D_a\) on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(a\) is the diagonal operator with the conjugate coefficients \(i \mapsto \overline{a_i}\).
theorem adjoint_diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
(diagonalCLM a hbdd)† =
diagonalCLM (fun i => star (a i)) (by
obtain ⟨C, hC⟩ := hbdd
exact ⟨C, fun i => by rw [norm_star]; exact hC i⟩)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 2
TheoremSelf-adjoint diagonal operators on \(\ell^2\)#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be bounded, and let \(D_a\) be the diagonal operator \(x \mapsto (a_i x_i)_{i \in \iota}\) on \(\ell^2(\iota, \mathbb{K})\). Then \(D_a\) is self-adjoint if and only if \(\overline{a_i} = a_i\), that is, \(a_i\) is real, for every \(i \in \iota\).
theorem isSelfAdjoint_diagonalCLM_iff (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
IsSelfAdjoint (diagonalCLM a hbdd) ↔ ∀ i, star (a i) = a i
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 4
TheoremRank-one term of the finite-support decomposition on \(\ell^2\)#✓fully proved
A rank-one term of the finite-support decomposition, evaluated pointwise.
theorem singleCLM_comp_evalCLM_apply [DecidableEq ι] (b : ι → 𝕜) (i : ι) (x : ℓ²(ι, 𝕜)) :
(lp.singleContinuousLinearMap 𝕜 (fun _ : ι => 𝕜) 2 i ∘L
((b i) • lp.evalCLM 𝕜 (fun _ : ι => 𝕜) 2 i)) x
= lp.single 2 i (b i * x i)
TheoremCompact diagonal operators on ℓ²#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(\iota\) be a type and let \(a : \iota \to \mathbb{K}\) be a bounded sequence. The diagonal operator on \(\ell^2(\iota, \mathbb{K})\) with coefficients \(a\) is compact if and only if \(a\) tends to \(0\) along the cofinite filter on \(\iota\), that is, for every \(\varepsilon > 0\) only finitely many \(i \in \iota\) satisfy \(\|a_i\| \ge \varepsilon\).
theorem isCompactOperator_diagonalCLM_iff {a : ι → 𝕜}
(hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
IsCompactOperator (diagonalCLM a hbdd) ↔ Tendsto a cofinite (𝓝 0)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 5 · Used by 1
TheoremShifting a diagonal operator by a scalar#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be a bounded family of coefficients, let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) given by \((D_a x)_i = a_i x_i\), and let \(I\) be the identity operator. For every \(\mu \in \mathbb{K}\), \(\mu I - D_a = D_{\mu - a}\), the diagonal operator with coefficients \(\mu - a_i\).
theorem smul_one_sub_diagonalCLM_eq (μ : 𝕜) (a : ι → 𝕜)
(ha : ∃ C, ∀ i, ‖a i‖ ≤ C) :
μ • (1 : ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜)) - diagonalCLM a ha =
diagonalCLM (fun i => μ - a i) (by
obtain ⟨C, hC⟩ := ha
exact ⟨‖μ‖ + C, fun i => (norm_sub_le _ _).trans (add_le_add le_rfl (hC i))⟩)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 2
TheoremInvertibility of a shifted diagonal operator#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be a bounded family of coefficients, let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) given by \((D_a x)_i = a_i x_i\), and let \(I\) be the identity operator. If \(\mu \in \mathbb{K}\) does not lie in the closure of \(\{a_i : i \in \iota\}\), then \(\mu I - D_a\) is invertible in the algebra of bounded operators on \(\ell^2(\iota, \mathbb{K})\).
theorem isUnit_smul_one_sub_diagonalCLM_of_not_mem_closure_range {a : ι → 𝕜}
(ha : ∃ C, ∀ i, ‖a i‖ ≤ C) {μ : 𝕜} (hμ : μ ∉ closure (Set.range a)) :
IsUnit (μ • (1 : ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜)) - diagonalCLM a ha)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 4 · Used by 1
TheoremPoints off the closure of the coefficients are not in the spectrum#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be a bounded family of coefficients, and let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) given by \((D_a x)_i = a_i x_i\). If \(\mu \in \mathbb{K}\) does not lie in the closure of \(\{a_i : i \in \iota\}\), then \(\mu\) is not in the spectrum of \(D_a\) over \(\mathbb{K}\), that is, \(\mu I - D_a\) is invertible as a bounded operator, where \(I\) is the identity operator.
theorem not_mem_spectrum_of_not_mem_closure_range {a : ι → 𝕜}
(ha : ∃ C, ∀ i, ‖a i‖ ≤ C) {μ : 𝕜} (hμ : μ ∉ closure (Set.range a)) :
μ ∉ spectrum 𝕜 (diagonalCLM a ha)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 2 · Used by 1
TheoremEvery coefficient of a diagonal operator lies in its spectrum#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be a bounded family of coefficients, and let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) given by \((D_a x)_i = a_i x_i\). Then every \(a_j\) with \(j \in \iota\) lies in the spectrum of \(D_a\) over \(\mathbb{K}\), the set of \(\mu \in \mathbb{K}\) for which \(\mu I - D_a\) is not invertible as a bounded operator, where \(I\) is the identity operator.
For each \(j \in \iota\), the \(j\)-th coordinate vector lies in the kernel of \(a_j I - D_a\).
theorem range_subset_spectrum {a : ι → 𝕜} (ha : ∃ C, ∀ i, ‖a i‖ ≤ C) :
Set.range a ⊆ spectrum 𝕜 (diagonalCLM a ha)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Uses 3 · Used by 1
TheoremSpectrum of a diagonal operator#✓fully proved
Let \(\mathbb{K}\) be \(\mathbb{R}\) or \(\mathbb{C}\), let \(a : \iota \to \mathbb{K}\) be a bounded family of coefficients, and let \(D_a\) be the diagonal operator on \(\ell^2(\iota, \mathbb{K})\) given by \((D_a x)_i = a_i x_i\). The spectrum of \(D_a\) over \(\mathbb{K}\), the set of \(\mu \in \mathbb{K}\) for which \(\mu I - D_a\) is not invertible as a bounded operator (\(I\) the identity operator), is the closure of \(\{a_i : i \in \iota\}\).
theorem spectrum_diagonalCLM (a : ι → 𝕜) (ha : ∃ C, ∀ i, ‖a i‖ ≤ C) :
spectrum 𝕜 (diagonalCLM a ha) = closure (Set.range a)
noncomputable abbrev diagonalCLM (a : ι → 𝕜) (hbdd : ∃ C, ∀ i, ‖a i‖ ≤ C) :
ℓ²(ι, 𝕜) →L[𝕜] ℓ²(ι, 𝕜) :=
MathlibExt.Analysis.InnerProductSpace.HilbertBasisDiagonal.diagonalOperator
(HilbertBasis.ofRepr (LinearIsometryEquiv.refl 𝕜 (ℓ²(ι, 𝕜)))) a
(C := max hbdd.choose 0)
(fun i => (hbdd.choose_spec i).trans (le_max_left _ _))
MathlibExt/Analysis/InnerProductSpace/LpDiagonalOperator.lean:85
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.