Documentation

Atlas.AlgebraicTopologyI.code.Section24

The chain complex obtained from $C_*$ by tensoring degreewise on the right with the $R$-module $M$: $(C_* \otimes M)_n = C_n \otimes M$ with the induced differential.

Instances For

    Functoriality of tensorComplex in the chain complex argument: a chain map $\varphi : C \to D$ tensored with the identity on $M$.

    Instances For

      Tensoring on the right with a fixed module $M$ preserves colimits; obtained by transporting along the symmetry isomorphism with tensorLeft M.

      In the downward $\mathbb{N}$-indexed complex shape, the successor n + 1 has next = n.

      In the downward $\mathbb{N}$-indexed complex shape, the successor n + 1 has prev = n + 2.

      The image of the cycles inclusion $Z_{n+1}(C) \hookrightarrow C_{n+1}$ under the functor $- \otimes M$ vanishes when followed by the differential of $C \otimes M$. Used to lift the cycles into the homology of the tensor complex.

      The map $Z_{n+1}(C) \otimes M \to H_{n+1}(C \otimes M)$ obtained by lifting through the cycles of the tensor complex and then quotienting by boundaries. This is the building block of the UCT map $\alpha$.

      Instances For

        The composition of the canonical map into cycles with cyclesToHomologyTC vanishes; i.e. boundaries are killed. This is the relation needed to descend cyclesToHomologyTC along the cokernel description of homology.

        The UCT map $\alpha : H_{n+1}(C) \otimes M \to H_{n+1}(C \otimes M)$, defined by descending the cycles-to-homology map through the cokernel presentation of $H_{n+1}(C)$ (valid because tensoring with $M$ preserves the relevant cokernel for free $C$).

        Instances For
          @[simp]

          The defining property of uctAlpha: precomposing with the canonical projection from cycles to homology recovers the basic cyclesToHomologyTC map. This is the universal property of the cokernel description of homology.

          @[simp]

          The defining property of uctAlpha: precomposing with the canonical projection from cycles to homology recovers the basic cyclesToHomologyTC map. This is the universal property of the cokernel description of homology.

          Naturality of the UCT map $\alpha$ in the chain complex variable: $\alpha$ commutes with the maps induced by a chain map $\varphi : C \to D$ between free chain complexes.

          Packaged data of the Universal Coefficient Theorem in degree $n+1$: for every free chain complex $C_*$ over a PID $R$ and every $R$-module $M$, a natural split short exact sequence $$0 \to H_{n+1}(C_*) \otimes M \xrightarrow{\alpha} H_{n+1}(C_* \otimes M) \xrightarrow{\beta} \operatorname{Tor}_1^R(H_n(C_*), M) \to 0,$$ together with naturality of $\alpha$ and $\beta$ in $C_*$.

          Instances For

            Tor as the homology of a projective resolution: for any projective resolution $P_\bullet$ of $M$, $$\operatorname{Tor}_1(X, M) \cong H_1(X \otimes P_\bullet).$$ Used to identify the connecting map of the UCT short exact sequence with the Tor term.

            Instances For

              In a projective resolution $P_\bullet \to M$, the composition $d_1 \circ \pi_0 = 0$: the augmentation kills the image of the first differential.

              The short complex of chain complexes obtained by tensoring $C_*$ degreewise with the short exact sequence $0 \to P_1 \to P_0 \to M \to 0$ extracted from a projective resolution $P_\bullet$ of $M$.

              Instances For

                Over a PID, every projective resolution of an $R$-module can be chosen to have length one, so $d_{2,1} = 0$. Used to upgrade the right end of resSES into a short exact sequence.

                Consequence of projRes_d_two_one_eq_zero: in a PID projective resolution the map $P_1 \to P_0$ is a monomorphism, so $0 \to P_1 \to P_0 \to M \to 0$ is short exact.

                For a free chain complex $C_*$ over a PID and a projective resolution $P_\bullet$ of $M$, the short complex resSES R C M P obtained by tensoring with the length-one resolution is short exact in every degree.

                The natural identification of the homology in degree $n$ of the chain complex $C_* \otimes P_1$ with $\operatorname{Tor}_1^R(H_n(C_*), M)$, where $P_\bullet$ is a projective (in fact length-one) resolution of $M$.

                Instances For

                  Naturality of torIdentification in the chain complex variable.

                  The UCT connecting map $\beta : H_{n+1}(C_* \otimes M) \to \operatorname{Tor}_1^R(H_n(C_*), M)$, built as the connecting homomorphism of resSES composed with torIdentification.

                  Instances For

                    Naturality of cyclesToHomologyTC in the module variable $M$: the construction commutes with maps $f : M_1 \to M_2$. Used to relate $\alpha$ with the connecting map of the tensored resolution.

                    theorem UniversalCoefficientTheorem.uctAlpha_comp_uctBeta (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C : ChainComplex (ModuleCat R) ) (hfree : ∀ (i : ), Module.Free R (C.X i)) :
                    CategoryTheory.CategoryStruct.comp (uctAlpha R M n C hfree) (uctBeta R M n C hfree) = 0

                    The composition $\beta \circ \alpha = 0$, i.e. the maps in the UCT short complex form a complex.

                    For a free chain complex over a PID, the cokernel of $Z_n \hookrightarrow C_n$ — i.e. the boundary group $B_{n-1}(C)$ identified inside $C_{n-1}$ — is projective. This rests on the fact that submodules of free modules over a PID are free.

                    A splitting of the inclusion $Z_n(C) \hookrightarrow C_n$ for a free chain complex over a PID, obtained from the short exact sequence $0 \to Z_n \to C_n \to \mathrm{cok}(Z_n \hookrightarrow C_n) \to 0$ together with projectivity of the cokernel.

                    Instances For

                      Noncomputable variant of iCycles_splitMono: the same splitting of $Z_n(C) \hookrightarrow C_n$, packaged for use in contexts where computability is not required.

                      Instances For
                        noncomputable def UniversalCoefficientTheorem.uctAlpha_retraction (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C : ChainComplex (ModuleCat R) ) (hfree : ∀ (i : ), Module.Free R (C.X i)) :

                        A retraction (split-mono witness) of the UCT map $\alpha_n$, constructed from a splitting of $Z_{n+1}(C) \hookrightarrow C_{n+1}$. This is the key ingredient making the UCT short exact sequence split.

                        Instances For

                          The identification of $\mathrm{Tor}_1(H_n(C), M)$ with a kernel of an induced homology map is an isomorphism on the relevant kernel, providing the bridge between the abstract Tor functor and the concrete connecting homomorphism appearing in the UCT.

                          theorem UniversalCoefficientTheorem.uctBeta_epi (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C : ChainComplex (ModuleCat R) ) (hfree : ∀ (i : ), Module.Free R (C.X i)) :
                          CategoryTheory.Epi (uctBeta R M n C hfree)

                          The UCT map $\beta_n : H_{n+1}(C \otimes M) \to \mathrm{Tor}_1(H_n(C), M)$ is an epimorphism. This is the surjectivity claim in the UCT short exact sequence.

                          theorem UniversalCoefficientTheorem.uctExact (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C : ChainComplex (ModuleCat R) ) (hfree : ∀ (i : ), Module.Free R (C.X i)) :
                          { X₁ := CategoryTheory.MonoidalCategoryStruct.tensorObj (HomologicalComplex.homology C (n + 1)) M, X₂ := HomologicalComplex.homology (tensorComplex R C M) (n + 1), X₃ := ((CategoryTheory.Tor (ModuleCat R) 1).obj (HomologicalComplex.homology C n)).obj M, f := uctAlpha R M n C hfree, g := uctBeta R M n C hfree, zero := }.Exact

                          Exactness of the UCT short complex at its middle term: the image of $\alpha_n$ equals the kernel of $\beta_n$, giving the exact middle of the universal coefficient sequence.

                          theorem UniversalCoefficientTheorem.uctSplitting (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C : ChainComplex (ModuleCat R) ) (hfree : ∀ (i : ), Module.Free R (C.X i)) :

                          The UCT short exact sequence splits: a (non-natural) splitting of the sequence $0 \to H_n(C) \otimes M \to H_n(C \otimes M) \to \mathrm{Tor}_1(H_{n-1}(C), M) \to 0$, obtained from the retraction uctAlpha_retraction together with exactness.

                          theorem UniversalCoefficientTheorem.uctShortExact (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C : ChainComplex (ModuleCat R) ) (hfree : ∀ (i : ), Module.Free R (C.X i)) :
                          { X₁ := CategoryTheory.MonoidalCategoryStruct.tensorObj (HomologicalComplex.homology C (n + 1)) M, X₂ := HomologicalComplex.homology (tensorComplex R C M) (n + 1), X₃ := ((CategoryTheory.Tor (ModuleCat R) 1).obj (HomologicalComplex.homology C n)).obj M, f := uctAlpha R M n C hfree, g := uctBeta R M n C hfree, zero := }.ShortExact

                          The UCT sequence is short exact: combining injectivity of $\alpha_n$, exactness in the middle, and surjectivity of $\beta_n$ into a single ShortExact package.

                          theorem UniversalCoefficientTheorem.uctBeta_natural (R : Type u_1) [CommRing R] [IsPrincipalIdealRing R] (M : ModuleCat R) (n : ) (C D : ChainComplex (ModuleCat R) ) (hfreeC : ∀ (i : ), Module.Free R (C.X i)) (hfreeD : ∀ (i : ), Module.Free R (D.X i)) (φ : C D) :

                          Naturality of the UCT connecting map $\beta_n$ in the chain complex variable: a chain map $\varphi : C \to D$ between free complexes induces a commuting square between the UCT $\beta$'s for $C$ and $D$.

                          Theorem 24.1 (Universal Coefficient Theorem). For a PID $R$, a chain complex of free $R$-modules $C$, and an $R$-module $M$, there is a natural split short exact sequence $$0 \to H_n(C) \otimes_R M \to H_n(C \otimes_R M) \to \mathrm{Tor}_1^R(H_{n-1}(C), M) \to 0.$$ This bundles the maps $\alpha_n, \beta_n$, exactness, splitting, and naturality.

                          Instances For