Documentation

DerivedAlgGeo.CategoryTheory.Triangulated.FourierMukai.CounitKernel

Kernel realizations of adjunction counits #

Let Φ_P : X ⥤ Y have a kernel-presented right adjoint Φ_Q. With a convolution kernel and a unit kernel on Y, the counit has the form

Φ_{conv Q P} ⟶ Φ_{O_Δ}.

Here conv Q P is the kernel of Φ_Q ⋙ Φ_P (ConvolutionData.compIso reads composition diagrammatically, so this is Huybrechts' P ∘ Q).

The missing geometric statement is that this natural transformation is induced by an actual kernel morphism conv Q P ⟶ O_Δ. CounitKernelData records that ordinary kernel arrow and the equality identifying its transform with the counit. Any enhancement of the kernel category then supplies a noncanonical closed representative and a noncanonical dg cone, packaged by CounitKernelData.toConeData. Exactness in the kernel variable sends that dg cone to a source-natural family of pointwise distinguished triangles. CounitKernelConeData.counitTriangleInSource transports their first two vertices to the composite adjoint transform and the identity functor, with first map literally the adjunction counit and third vertex the Fourier--Mukai twist candidate. The selected cone kernel is named twistKernel and definitionally presents that twist, so the candidate is explicitly a kernel functor without any exactness or invertibility conclusion.

The constructor CounitKernelData.ofFull produces the ordinary arrow when the kernel transform is explicitly assumed full. No fullness theorem is asserted here: without that strong hypothesis, producing the kernel arrow remains the genuine Fourier--Mukai realization obligation involving convolution, the diagonal kernel, and the adjunction trace.

structure CategoryTheory.Triangulated.FourierMukai.CounitKernelData {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] (C : Correspondence X Y W₁) (C' : Correspondence Y X W₂) (E : Correspondence Y Y W) (P : W₁) (R : RightAdjointKernelData C C' P) (D : ConvolutionData C' C E) (U : UnitKernelData E) :
Type vW

An ordinary kernel arrow realizing the counit of a kernel-presented adjunction, together with the exact comparison between its transform and the counit. It deliberately contains no enhancement or cone choice.

Instances For
    structure CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] (C : Correspondence X Y W₁) (C' : Correspondence Y X W₂) (E : Correspondence Y Y W) (e : Enhancement W) (P : W₁) (R : RightAdjointKernelData C C' P) (D : ConvolutionData C' C E) (U : UnitKernelData E) :
    Type (max uE vE)

    A dg kernel arrow realizing the counit of a kernel-presented adjunction, together with a chosen cone of that arrow, for an enhancement e of the kernel category.

    The arrow lives between the enhancement's lifts e.equiv.inverse.obj _ of the convolution kernel and the unit kernel. The equality transform_arrow includes all the changes of presentation: the comparison equivalence identifies the transform of the arrow with a map between the two kernels, convolution identifies the source transform with Φ_Q ⋙ Φ_P, and the unit-kernel isomorphism identifies the target transform with 𝟭 Y.

    Instances For

      Read counit-kernel data through the generic kernel-transformation realization interface.

      Equations
      Instances For
        theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelData.ext {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} {S T : CounitKernelData C C' E P R D U} (h : S.arrow = T.arrow) :
        S = T

        Ordinary counit-kernel data are determined by their kernel arrow.

        theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelData.ext_iff {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} {S T : CounitKernelData C C' E P R D U} :
        S = T ↔ S.arrow = T.arrow
        noncomputable def CategoryTheory.Triangulated.FourierMukai.CounitKernelData.ofFull {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} [E.kernelTransform.Full] :
        CounitKernelData C C' E P R D U

        Fullness of the kernel transform is a sufficient abstract hypothesis for lifting the adjunction counit comparison to an ordinary kernel morphism.

        Equations
        Instances For
          noncomputable def CategoryTheory.Triangulated.FourierMukai.CounitKernelData.liftedArrow {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelData C C' E P R D U) (e : Enhancement W) :
          ↥(cocycles (have this := e.equiv.inverse.obj (D.conv R.adjKernel P); this) (have this := e.equiv.inverse.obj U.unitKernel; this))

          A noncanonical closed degree-zero representative of the ordinary kernel arrow in an enhancement.

          Equations
          Instances For
            @[simp]
            theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelData.homMk_liftedArrow {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelData C C' E P R D U) (e : Enhancement W) :

            The selected cocycle represents the inverse image of the ordinary kernel arrow in H⁰ of the enhancement.

            Conjugating the selected representative by the counit of the enhancement equivalence recovers the original kernel arrow.

            noncomputable def CategoryTheory.Triangulated.FourierMukai.CounitKernelData.toConeData {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelData C C' E P R D U) (e : Enhancement W) :
            CounitKernelConeData C C' E e P R D U

            Choose a closed representative and a dg cone for an ordinary counit kernel arrow. Both choices are noncanonical; the resulting comparison equation is the exact equation stored in S.

            Equations
            • One or more equations did not get rendered due to their size.
            Instances For

              Read enhanced counit-kernel data through the generic kernel-transformation realization interface.

              Equations
              Instances For
                def CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.toCounitKernelData {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :
                CounitKernelData C C' E P R D U

                Forget the chosen cocycle and cone while retaining the ordinary kernel arrow represented through the enhancement equivalence.

                Equations
                Instances For

                  The kernel arrow and its cone as a reusable dg cone presentation.

                  Equations
                  Instances For
                    @[simp]
                    theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.presentation_source {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :
                    @[simp]
                    @[simp]
                    theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.presentation_cone {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :
                    @[reducible, inline]
                    abbrev CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.twistKernel {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :
                    W

                    The ordinary kernel represented by the selected enhanced counit cone.

                    Equations
                    Instances For
                      @[reducible, inline]
                      abbrev CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.twist {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :

                      The Fourier--Mukai transform of the counit-cone kernel, read in the kernel category through the enhancement. This is the kernel-presented twist candidate.

                      Equations
                      Instances For
                        def CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.twistKernelIso {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :

                        The named twist kernel presents the twist definitionally.

                        Equations
                        Instances For
                          theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelConeData.isKernelFunctor_twist {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {e : Enhancement W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelConeData C C' E e P R D U) :

                          The twist candidate is a kernel functor, witnessed by its selected cone kernel. This asserts neither exactness nor invertibility.

                          Identify the enhanced lift of the convolution kernel first with the ordinary convolution kernel and then with the composite transform presenting the source of the adjunction counit.

                          Equations
                          Instances For

                            Identify the enhanced lift of the unit kernel with the identity transform, using the enhancement comparison and the supplied unit-kernel presentation.

                            Equations
                            Instances For

                              The stored kernel-transform equation, normalized to the naturality square whose bottom edge is literally the adjunction counit.

                              The generic endpoint-normalization package underlying the literal counit triangle.

                              Equations
                              Instances For

                                The counit-kernel cone as a source-natural family of transform triangles. The first two vertices are the transforms of the enhancement's lifts of the convolution and unit kernels; the third is S.twist.

                                Equations
                                Instances For

                                  Every objectwise triangle obtained from the counit-kernel cone is distinguished.

                                  The first map of the raw transform triangle before endpoint normalization.

                                  The source endpoint comparison, typed against the first projection of the raw triangle functor.

                                  Equations
                                  Instances For

                                    The target endpoint comparison, typed against the second projection of the raw triangle functor.

                                    Equations
                                    Instances For

                                      The transported second map from the identity functor to the twist candidate. It retains the supplied enhancement and cone choices.

                                      Equations
                                      Instances For

                                        The transported connecting map from the twist candidate to the shifted composite of the right adjoint and the original transform.

                                        Equations
                                        Instances For

                                          The source-natural counit triangle R P ⋙ P ⟶ 𝟙 Y ⟶ twist ⟶ (R P ⋙ P)⟦1⟧.

                                          Its first map and all three vertices are literal. Its remaining maps retain the choices in S; this is only an objectwise distinguished family, not a distinguished triangle in a functor category.

                                          Equations
                                          Instances For

                                            The raw exact-family triangle and the literal counit triangle are naturally isomorphic through the two endpoint comparisons and the identity on the twist candidate.

                                            Equations
                                            Instances For

                                              Specialize a generic kernel realization of the adjunction counit to the stable counit-kernel interface.

                                              Equations
                                              Instances For

                                                Specialize a generic enhanced realization of the adjunction counit to the stable counit-kernel cone interface.

                                                Equations
                                                Instances For
                                                  @[simp]
                                                  theorem CategoryTheory.Triangulated.FourierMukai.CounitKernelData.toConeData_toCounitKernelData {X : Type uX} {Y : Type uY} {W₁ : Type u₁} {W₂ : Type u₂} {W : Type uW} [Category.{vX, uX} X] [Category.{vY, uY} Y] [Category.{v₁, u₁} W₁] [Category.{v₂, u₂} W₂] [Category.{vW, uW} W] {C : Correspondence X Y W₁} {C' : Correspondence Y X W₂} {E : Correspondence Y Y W} {P : W₁} {R : RightAdjointKernelData C C' P} {D : ConvolutionData C' C E} {U : UnitKernelData E} (S : CounitKernelData C C' E P R D U) (e : Enhancement W) :

                                                  Forgetting the choices made by toConeData recovers the original ordinary kernel datum. This does not identify the selected cocycle or cone.