Documentation

TauCeti.Analysis.Normed.Operator.RowCompact.PrModule

Row-compact maps between modules with property (Pr) #

A module M over a normed ring R has property (Pr) (TauCeti.HasPr R M) if it is a continuous linear retract M →ⁱ C₀(J, R) →ʳ M, r ∘ i = id, of a model space. A continuous linear map u : M →L[R] N between two such modules is row compact in the sense of (Pr) modules (ContinuousLinearMap.IsPrCompact) if for some retract presentations (iM, rM) of M and (iN, rN) of N the extension iN ∘ u ∘ rM : C₀(J, R) →L[R] C₀(I, R) is row compact (ContinuousLinearMap.IsRowCompact). This extends row compactness from model spaces to their continuous linear retracts, without choosing an orthonormal basis of M or N.

Over ultrametric Tate scalars the condition does not depend on the presentations. If iN ∘ u ∘ rM is row compact and j : N →L[R] C₀(I', R), s : C₀(J', R) →L[R] M are any continuous linear maps, then j ∘ u ∘ s is the composite (j ∘ rN) ∘ (iN ∘ u ∘ rM) ∘ (iM ∘ s), which is row compact because row-compact maps form a two-sided operator ideal (ContinuousLinearMap.isRowCompact_comp_of_retract). No finite-dimensional complement of the range of iM is presumed.

The ideal, closedness and approximation properties of row-compact maps transport to (Pr) modules. Over complete ultrametric Tate scalars, let M be a normed module with a bounded scalar action (IsBoundedSMul). If M is finitely generated with (Pr), every map into M from a module with (Pr) is row compact. The identity of M is row compact exactly when M has (Pr) and is finitely generated. Hence a row-compact endomorphism u of M with (1 - u) ^ n = 0 forces M to be finitely generated projective.

A row-compact idempotent e of a module with (Pr) has finitely generated projective range: through a retract presentation r ∘ i = id, the inclusion i identifies range e with the range of the row-compact idempotent i ∘ e ∘ r of the model space (ContinuousLinearMap.IsRowCompact.projective_range_of_isIdempotentElem).

Main definitions #

Main results #

References #

A continuous linear map u : M →L[R] N between modules with property (Pr) is row compact if there are continuous linear retract presentations rM ∘ iM = id of M through C₀(J, R) and rN ∘ iN = id of N through C₀(I, R) such that the extension iN ∘ u ∘ rM is row compact. Over ultrametric Tate scalars this does not depend on the presentations (ContinuousLinearMap.isPrCompact_iff_isRowCompact_comp). The index types live in the universes of M and N, as in TauCeti.HasPr.

Instances For

    A map is row compact in the sense of (Pr) modules exactly when it has row-compact extensions through some retract presentations of its source and target. This is the introduction and elimination interface for ContinuousLinearMap.IsPrCompact.

    Retract factorisation: if rM ∘ iM = id and rN ∘ iN = id present the source and target of u, then u factors through its extension as rN ∘ (iN ∘ u ∘ rM) ∘ iM.

    A map whose extension through retract presentations of its source and target is row compact is row compact in the sense of (Pr) modules.

    The source of a row-compact map between (Pr) modules has property (Pr).

    The target of a row-compact map between (Pr) modules has property (Pr).

    @[simp]

    The zero map between two modules is row compact exactly when both have property (Pr).

    Negatives of row-compact maps between (Pr) modules are row compact.

    @[simp]

    A negative is row compact between (Pr) modules exactly when the map is.

    Over Tate scalars, every map into a retract of a model space with a finite index type is row compact, provided its source has property (Pr).

    Scalar multiples of row-compact maps between (Pr) modules are row compact.

    Independence of the presentation. Over ultrametric Tate scalars, if the extension iN ∘ u ∘ rM of u through retract presentations of its source and target is row compact, then so is every composite j ∘ u ∘ s with continuous linear maps j into and s out of model spaces: it factors as (j ∘ rN) ∘ (iN ∘ u ∘ rM) ∘ (iM ∘ s).

    Over ultrametric Tate scalars, for a row-compact map u between (Pr) modules every composite j ∘ u ∘ s with continuous linear maps into and out of model spaces is row compact. In particular the extension of u through any retract presentations is row compact.

    Row compactness on (Pr) modules does not depend on the presentation: over ultrametric Tate scalars, u is row compact exactly when its extension through any one pair of retract presentations is.

    Transport along equivalences with model spaces: over ultrametric Tate scalars, a map between potentially orthonormalisable modules is row compact exactly when its conjugate eN ∘ u ∘ eM⁻¹ between the model spaces is.

    @[simp]

    On model spaces, row compactness in the sense of (Pr) modules is row compactness.

    Sums of row-compact maps between (Pr) modules are row compact.

    Differences of row-compact maps between (Pr) modules are row compact.

    Precomposing a row-compact map between (Pr) modules with a continuous linear map out of a module with property (Pr) gives a row-compact map.

    Postcomposing a row-compact map between (Pr) modules with a continuous linear map into a module with property (Pr) gives a row-compact map.

    A polynomial with zero constant coefficient preserves row compactness on modules with property (Pr), over ultrametric Tate scalars.

    theorem ContinuousLinearMap.isPrCompact_of_tendsto {R : Type w} [NormedCommRing R] [IsUltrametricDist R] [TauCeti.NormedRing.IsTate R] {M : Type u} {N : Type v} [SeminormedAddCommGroup M] [Module R M] [IsBoundedSMul R M] [SeminormedAddCommGroup N] [Module R N] [IsBoundedSMul R N] {ι : Type u_1} {l : Filter ι} [l.NeBot] {u : M →L[R] N} {v : ι → M →L[R] N} (h : Filter.Tendsto (fun (n : ι) => ‖u - v n‖) l (nhds 0)) (hv : ∀ᶠ (n : ι) in l, (v n).IsPrCompact) :

    Closedness: over ultrametric Tate scalars, an operator-norm limit of row-compact maps between (Pr) modules is row compact.

    Approximation by truncations: over ultrametric Tate scalars, if rN ∘ iN = id presents the target of a row-compact map u between (Pr) modules, then the truncations rN ∘ π_S ∘ iN ∘ u converge to u in operator norm as the finite set S of coordinates grows. Here π_S = ZeroAtInftyContinuousMap.indicatorCLM R S is the coordinate projection.

    Over ultrametric Tate scalars, row-compact maps between (Pr) modules are completely continuous.

    Over complete ultrametric Tate scalars, every continuous linear map from a module with property (Pr) into a finitely generated module with property (Pr) is row compact. A finitely generated module with (Pr) is a retract of C₀(S, R) for a finite set S of generators.

    The identity criterion on (Pr) modules: over complete ultrametric Tate scalars, the identity of a module is row compact exactly when the module has property (Pr) and is finitely generated.

    The finiteness germ: over complete ultrametric Tate scalars, a row-compact endomorphism u of a (Pr) module with (1 - u) ^ n = 0 forces the module to be finitely generated.

    Over complete ultrametric Tate scalars, a row-compact endomorphism u of a (Pr) module with (1 - u) ^ n = 0 forces the module to be finitely generated projective.

    A row-compact idempotent of a (Pr) module has finitely generated range.

    A row-compact idempotent of a (Pr) module has projective range.