Documentation

TauCeti.Geometry.Manifold.MFDeriv.RestrictScalars

Restricting the scalars of manifold derivatives #

Let 𝕜' be a normed algebra over 𝕜, and let E and E' be normed spaces over both fields, compatibly. A manifold charted by E carries the trivial models 𝓘(𝕜', E) and 𝓘(𝕜, E) with the same charts. This file shows that a map between such manifolds which is differentiable for the models over 𝕜' is differentiable for the models over 𝕜, and that its derivative over 𝕜 is the derivative over 𝕜' with scalars restricted to 𝕜.

The tangent spaces TangentSpace 𝓘(𝕜', E) x and TangentSpace 𝓘(𝕜, E) x are both the model space E, but they are distinct types carrying module structures over different fields. The comparison of derivatives is therefore stated after reading both derivatives in the model spaces through the identifications tangentSpaceCastModel, which is also how HasMFDerivWithinAt reads a derivative in charts.

The main example is a holomorphic map between complex manifolds modelled on 𝓘(ℂ, E) and 𝓘(ℂ, E'): its real derivative is its complex derivative regarded as an ℝ-linear map. Likewise, the tangent coordinate changes of a complex manifold, regarded as a real manifold, are its complex tangent coordinate changes regarded as ℝ-linear maps.

The fibrewise linear equivalence identifies the native self-model tangent spaces over the two fields and intertwines their manifold and vector-valued derivatives.

Main results #

theorem MDifferentiableWithinAt.restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {s : Set M} {x : M} (h : MDiffAt[s] f x) :
MDiffAt[s] f x

A map which is differentiable within a set at a point for the models over 𝕜' is differentiable within that set at that point for the models over the smaller field 𝕜.

theorem MDifferentiableAt.restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {x : M} (h : MDiffAt f x) :
MDiffAt f x

A map which is differentiable at a point for the models over 𝕜' is differentiable at that point for the models over the smaller field 𝕜.

theorem MDifferentiableOn.restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {s : Set M} (h : MDiff[s] f) :
MDiff[s] f

A map which is differentiable on a set for the models over 𝕜' is differentiable on that set for the models over the smaller field 𝕜.

theorem MDifferentiable.restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} (h : MDiff f) :
MDiff f

A map which is differentiable for the models over 𝕜' is differentiable for the models over the smaller field 𝕜.

theorem HasMFDerivWithinAt.restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {s : Set M} {x : M} {f' : TangentSpace (modelWithCornersSelf 𝕜' E) x →L[𝕜'] TangentSpace (modelWithCornersSelf 𝕜' E') (f x)} {g' : TangentSpace (modelWithCornersSelf 𝕜 E) x →L[𝕜] TangentSpace (modelWithCornersSelf 𝕜 E') (f x)} (h : HasMFDerivAt[s] f x f') (hg : ContinuousLinearMap.restrictScalars 𝕜 (↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E') (f x)) ∘SL f' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E) x).symm) = ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E') (f x)) ∘SL g' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E) x).symm) :
HasMFDerivAt[s] f x g'

If f' is a derivative of f within a set at a point for the models over 𝕜', and g' is a 𝕜-linear map between the tangent spaces over 𝕜 which agrees with f' once both are read in the model spaces, then g' is a derivative of f within that set at that point for the models over 𝕜.

theorem HasMFDerivAt.restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {x : M} {f' : TangentSpace (modelWithCornersSelf 𝕜' E) x →L[𝕜'] TangentSpace (modelWithCornersSelf 𝕜' E') (f x)} {g' : TangentSpace (modelWithCornersSelf 𝕜 E) x →L[𝕜] TangentSpace (modelWithCornersSelf 𝕜 E') (f x)} (h : HasMFDerivAt% f x f') (hg : ContinuousLinearMap.restrictScalars 𝕜 (↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E') (f x)) ∘SL f' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E) x).symm) = ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E') (f x)) ∘SL g' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E) x).symm) :
HasMFDerivAt% f x g'

If f' is a derivative of f at a point for the models over 𝕜', and g' is a 𝕜-linear map between the tangent spaces over 𝕜 which agrees with f' once both are read in the model spaces, then g' is a derivative of f at that point for the models over 𝕜.

theorem HasMFDerivWithinAt.of_restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {s : Set M} {x : M} {f' : TangentSpace (modelWithCornersSelf 𝕜' E) x →L[𝕜'] TangentSpace (modelWithCornersSelf 𝕜' E') (f x)} {g' : TangentSpace (modelWithCornersSelf 𝕜 E) x →L[𝕜] TangentSpace (modelWithCornersSelf 𝕜 E') (f x)} (h : HasMFDerivAt[s] f x g') (hg : ContinuousLinearMap.restrictScalars 𝕜 (↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E') (f x)) ∘SL f' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E) x).symm) = ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E') (f x)) ∘SL g' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E) x).symm) :
HasMFDerivAt[s] f x f'

If g' is a derivative of f within a set at a point for the models over 𝕜, and f' is a 𝕜'-linear map between the tangent spaces over 𝕜' which agrees with g' once both are read in the model spaces, then f' is a derivative of f within that set at that point for the models over 𝕜'. This is the converse of HasMFDerivWithinAt.restrictScalars.

theorem HasMFDerivAt.of_restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {x : M} {f' : TangentSpace (modelWithCornersSelf 𝕜' E) x →L[𝕜'] TangentSpace (modelWithCornersSelf 𝕜' E') (f x)} {g' : TangentSpace (modelWithCornersSelf 𝕜 E) x →L[𝕜] TangentSpace (modelWithCornersSelf 𝕜 E') (f x)} (h : HasMFDerivAt% f x g') (hg : ContinuousLinearMap.restrictScalars 𝕜 (↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E') (f x)) ∘SL f' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜' E) x).symm) = ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E') (f x)) ∘SL g' ∘SL ↑(tangentSpaceCastModel (modelWithCornersSelf 𝕜 E) x).symm) :
HasMFDerivAt% f x f'

If g' is a derivative of f at a point for the models over 𝕜, and f' is a 𝕜'-linear map between the tangent spaces over 𝕜' which agrees with g' once both are read in the model spaces, then f' is a derivative of f at that point for the models over 𝕜'. This is the converse of HasMFDerivAt.restrictScalars.

theorem MDifferentiableWithinAt.restrictScalars_mfderivWithin (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {s : Set M} {x : M} (h : MDiffAt[s] f x) (hs : UniqueMDiffAt[s] x) :

At a point where f is differentiable within a set for the models over 𝕜', and where the set has unique derivatives over the smaller field 𝕜, the derivative of f within the set over 𝕜 is its derivative over 𝕜' with scalars restricted to 𝕜, both read in the model spaces.

theorem MDifferentiableAt.mfderiv_restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {x : M} (h : MDiffAt f x) :

At a point where f is differentiable for the models over 𝕜', the derivative of f over the smaller field 𝕜 is its derivative over 𝕜' with scalars restricted to 𝕜, both read in the model spaces. For instance, the real derivative of a holomorphic map is its complex derivative.

theorem TauCeti.injective_mfderiv_restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {E' : Type u_4} [NormedAddCommGroup E'] [NormedSpace 𝕜 E'] [NormedSpace 𝕜' E'] [IsScalarTower 𝕜 𝕜' E'] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {N : Type u_6} [TopologicalSpace N] [ChartedSpace E' N] {f : M → N} {x : M} (hinj : Function.Injective ⇑(mfderiv% f x)) :
Function.Injective ⇑(mfderiv% f x)

If the derivative of f at x over 𝕜' is injective, then so is its derivative over the smaller field 𝕜. For instance, a holomorphic immersion between complex manifolds is a real immersion. Where f is not differentiable over 𝕜' its derivative over 𝕜' is 0, and injectivity of that derivative forces the tangent space to be trivial.

theorem TauCeti.tangentCoordChange_restrictScalars (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {x : M} [IsManifold (modelWithCornersSelf 𝕜' E) 1 M] [IsManifold (modelWithCornersSelf 𝕜 E) 1 M] {y z : M} (h : z ∈ (chartAt E x).source ∩ (chartAt E y).source) :

On the overlap of the charts at x and y, the tangent coordinate change for the model over 𝕜 is the tangent coordinate change for the model over 𝕜', with scalars restricted to 𝕜: both are the derivative of the same chart transition. For instance, the tangent coordinate changes of a complex manifold, regarded as a real manifold, are complex linear.

theorem TauCeti.tangentCoordChange_smul (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {x : M} [IsManifold (modelWithCornersSelf 𝕜' E) 1 M] [IsManifold (modelWithCornersSelf 𝕜 E) 1 M] {y z : M} (h : z ∈ (chartAt E x).source ∩ (chartAt E y).source) (c : 𝕜') (w : E) :

On the overlap of the charts at x and y, the tangent coordinate change for the model over 𝕜 is 𝕜'-linear. For instance, the real tangent coordinate changes of a complex manifold are complex linear.

theorem TauCeti.mfderivWithin_extChartAt_symm_smul (𝕜 : Type u_1) {𝕜' : Type u_2} [NontriviallyNormedField 𝕜] [NontriviallyNormedField 𝕜'] [NormedAlgebra 𝕜 𝕜'] {E : Type u_3} [NormedAddCommGroup E] [NormedSpace 𝕜 E] [NormedSpace 𝕜' E] [IsScalarTower 𝕜 𝕜' E] {M : Type u_5} [TopologicalSpace M] [ChartedSpace E M] {x : M} [IsManifold (modelWithCornersSelf 𝕜' E) 1 M] {z : E} (hz : z ∈ (extChartAt (modelWithCornersSelf 𝕜 E) x).target) (c : 𝕜') (w : E) :
(mfderiv[Set.range ↑(modelWithCornersSelf 𝕜 E)] ↑(extChartAt (modelWithCornersSelf 𝕜 E) x).symm z) ((c • ContinuousLinearMap.id 𝕜 E) w) = (c • ContinuousLinearMap.id 𝕜 E) ((mfderiv[Set.range ↑(modelWithCornersSelf 𝕜 E)] ↑(extChartAt (modelWithCornersSelf 𝕜 E) x).symm z) w)

On the target of the chart at x, the derivative over 𝕜 of the inverse of the chart is 𝕜'-linear: it is a tangent coordinate change. For instance, on a complex manifold, regarded as a real manifold, the real derivative of the inverse of a chart commutes with multiplication by i.

Scalar restriction identifies a self-model tangent fibre with the native self-model fibre over the smaller field.

Instances For

    The native self-model fibre comparison intertwines derivatives over the two fields.

    Vector-valued derivatives agree through the native self-model fibre comparison.