Documentation

TauCeti.Analysis.Normed.Operator.RingOpNorm.Banach

The Banach module and Banach ring of operators over Tate scalars #

Let R be a seminormed ring admitting a multiplicative pseudo-uniformiser (TauCeti.NormedRing.IsTate R), and let E, F be seminormed R-modules with contractive scalar actions. Over such scalars every continuous linear map has a uniform bound, so the operator norm ContinuousLinearMap.ringOpNorm of TauCeti/Analysis/Normed/Operator/RingOpNorm.lean satisfies the usual laws. This file installs the resulting structures on E →L[R] F as scoped instances in the scope TauCeti.RingOpNorm:

Mathlib's topology of bounded convergence on continuous linear maps is only defined over normed fields, so over other scalar rings these instances introduce no diamond. Over a nontrivially normed field the scoped normed group structure equals Mathlib's (TauCeti.RingOpNorm.seminormedAddCommGroup_eq), but not definitionally, so the scope should not be opened there. On the zero module every operator has norm zero (norm_of_subsingleton).

For a complete normed R-module E, the endomorphism ring is a Banach ring, so Neumann inversion applies: for ‖t‖ < 1 the operator 1 - t is invertible with inverse the Neumann series ∑ tⁿ. This is packaged as the continuous linear automorphism ContinuousLinearMap.oneSub t ht. When E is ultrametric, 1 - t and its inverse are isometries, so the inverse has operator norm one on a module with a vector of nonzero norm.

Main definitions #

Main results #

References #

The completeness proof follows Mathlib's ContinuousLinearMap.ofTendstoOfBoundedRange and ContinuousLinearMap.tendsto_of_tendsto_pointwise_of_cauchySeq in Mathlib/Analysis/Normed/Operator/Completeness.lean.

@[instance_reducible]

Over Tate scalars, continuous linear maps between seminormed modules form a seminormed group for the operator norm.

Instances For

    The operator norm is ultrametric when the target is.

    A scalar action on the target commuting with R acts contractively on the operators. For commutative R this makes E →L[R] F an R-module with contractive action.

    Evaluation (f, x) ↦ f x is jointly continuous for the operator norm.

    @[instance_reducible]

    Over Tate scalars the continuous endomorphisms of a seminormed module form a seminormed ring for the operator norm.

    Instances For

      Composition of operators is jointly continuous for the operator norm. This is the continuity of multiplication in TauCeti.RingOpNorm.seminormedRing, stated directly for the multiplication of continuous linear maps so that instance search need not unfold the ring structure.

      The operators form a topological ring for the operator norm. As for TauCeti.RingOpNorm.continuousMul, this is stated directly for continuous linear maps so that instance search need not unfold the ring structure.

      The identity has operator norm one on a module with a vector of nonzero norm.

      Scalars move out of the left factor of a composition of operators. This is stated directly: for some modules, such as C₀(I, R), instance search does not derive it from ContinuousLinearMap.algebra through IsScalarTower.right once TauCeti.RingOpNorm.seminormedRing is in scope.

      Scalars move out of the right factor of a composition of operators. As for TauCeti.RingOpNorm.isScalarTower, this is stated directly.

      @[instance_reducible]

      Over Tate scalars, continuous linear maps into a normed module form a normed group for the operator norm.

      Instances For

        Over Tate scalars, continuous linear maps into a normed module form a Hausdorff space for the operator norm. This is stated directly, since instance search does not find it through TauCeti.RingOpNorm.normedAddCommGroup, whose metric is not reducibly the one of TauCeti.RingOpNorm.seminormedAddCommGroup.

        @[instance_reducible]

        Over Tate scalars the continuous endomorphisms of a normed module form a normed ring for the operator norm.

        Instances For

          Completeness of the operator space. Over Tate scalars, continuous linear maps into a complete normed module form a complete space for the operator norm.

          Over a nontrivially normed field, the scoped operator-norm group structure is Mathlib's normed group structure on continuous linear maps.

          Composition with fixed operators on both sides, X ↦ g ∘ X ∘ f, is Lipschitz for the operator norm, with constant ‖g‖ * ‖f‖.

          theorem ContinuousLinearMap.apply_comp_apply_of_mem_closure_adjoin {R : Type u_1} [NormedCommRing R] [TauCeti.NormedRing.IsTate R] {M : Type u_2} {N : Type u_3} [SeminormedAddCommGroup M] [Module R M] [NormedAddCommGroup N] [Module R N] [IsBoundedSMul R N] {i : M →L[R] N} {r : N →L[R] M} {u : M →L[R] M} (hri : r ∘SL i = ContinuousLinearMap.id R M) {X : N →L[R] N} (hX : X ∈ closure ↑R[i ∘SL u ∘SL r]) (y : N) :
          i (r (X y)) = X (i (r y))

          For a retract r ∘ i = id of M through N, every operator-norm limit of polynomials in the extension i ∘ u ∘ r of an endomorphism u of M commutes with the idempotent i ∘ r.

          theorem ContinuousLinearMap.comp_comp_mem_closure_adjoin {R : Type u_1} [NormedCommRing R] [TauCeti.NormedRing.IsTate R] {M : Type u_2} {N : Type u_3} [SeminormedAddCommGroup M] [Module R M] [IsBoundedSMul R M] [NormedAddCommGroup N] [Module R N] [IsBoundedSMul R N] {i : M →L[R] N} {r : N →L[R] M} {u : M →L[R] M} (hri : r ∘SL i = ContinuousLinearMap.id R M) {X : N →L[R] N} (hX : X ∈ closure ↑R[i ∘SL u ∘SL r]) :
          r ∘SL X ∘SL i ∈ closure ↑R[u]

          For a retract r ∘ i = id of M through N, compression X ↦ r ∘ X ∘ i sends the operator-norm closure of the polynomials in the extension i ∘ u ∘ r of an endomorphism u of M into the closure of the polynomials in u. On the polynomials it is multiplicative, since they commute with the idempotent i ∘ r.

          noncomputable def ContinuousLinearMap.oneSub {R : Type u_1} {E : Type u_2} [SeminormedRing R] [TauCeti.NormedRing.IsTate R] [NormedAddCommGroup E] [Module R E] [IsBoundedSMul R E] [CompleteSpace E] (t : E →L[R] E) (ht : ‖t‖ < 1) :
          E ≃L[R] E

          Neumann inversion. On a complete normed module over Tate scalars, an operator t of operator norm less than one makes 1 - t a continuous linear automorphism, whose inverse is the Neumann series ∑ tⁿ.

          Instances For
            @[simp]
            theorem ContinuousLinearMap.coe_oneSub {R : Type u_1} {E : Type u_2} [SeminormedRing R] [TauCeti.NormedRing.IsTate R] [NormedAddCommGroup E] [Module R E] [IsBoundedSMul R E] [CompleteSpace E] (t : E →L[R] E) (ht : ‖t‖ < 1) :
            ↑(t.oneSub ht) = 1 - t

            The automorphism oneSub t ht is the operator 1 - t.

            @[simp]
            theorem ContinuousLinearMap.oneSub_apply {R : Type u_1} {E : Type u_2} [SeminormedRing R] [TauCeti.NormedRing.IsTate R] [NormedAddCommGroup E] [Module R E] [IsBoundedSMul R E] [CompleteSpace E] (t : E →L[R] E) (ht : ‖t‖ < 1) (x : E) :
            (t.oneSub ht) x = x - t x

            The automorphism oneSub t ht sends x to x - t x.

            The inverse of 1 - t is the ring inverse of 1 - t in the Banach ring of operators.

            theorem ContinuousLinearMap.hasSum_oneSub_symm {R : Type u_1} {E : Type u_2} [SeminormedRing R] [TauCeti.NormedRing.IsTate R] [NormedAddCommGroup E] [Module R E] [IsBoundedSMul R E] [CompleteSpace E] (t : E →L[R] E) (ht : ‖t‖ < 1) :
            HasSum (fun (n : ℕ) => t ^ n) ↑(t.oneSub ht).symm

            The inverse of 1 - t is the sum of the Neumann series ∑ tⁿ.

            In an ultrametric module, 1 - t preserves norms when ‖t‖ < 1.

            @[simp]

            In an ultrametric module, the inverse of 1 - t preserves norms when ‖t‖ < 1.

            @[simp]

            In an ultrametric module with a nonzero vector, the inverse of 1 - t has operator norm one when ‖t‖ < 1.