Theorem Components under homogeneous maps and scalar multiplication
✓ theorem proved

Components under homogeneous maps and scalar multiplication

Let \(k\) be a field and \(A\) a commutative ring. Let \(M,N\) have direct-sum \(k\)-vector-space gradings and \(A\)-module structures. Suppose an \(A\)-linear map \(f:M\to N\) has degree \(r\): \(f(M_i)\subseteq N_{i+r}\). Then \[ \pi_n^N(f(m))=f(\pi_{n-r}^M(m)). \] Both the kernel and the image are closed under all homogeneous component maps.

For a degree-zero map, surjectivity gives homogeneous preimages in the same degree; injectivity implies that a vector whose image lies in \(N_n\) already lies in \(M_n\).

For a \(k\)-linear equivalence \(e:M\to N\) of degree \(r\), its inverse has degree \(-r\), and restriction gives \(k\)-linear isomorphisms \[ M_n\xrightarrow{\sim}N_{n+r}. \] In the finite-dimensional pieces used for characters, this gives \(\dim_k N_n=\dim_k M_{n-r}\).

Proof

Step 1. Apply the map to a homogeneous decomposition

Decompose \(m=\sum_i m_i\). Additivity gives \(f(m)=\sum_i f(m_i)\), and each summand belongs to \(N_{i+r}\).

Only the index \(i=n-r\) can contribute to the degree-\(n\) component. Uniqueness of decomposition therefore gives the displayed component identity.

Step 2. Close the kernel under components

If \(f(m)=0\), apply the identity at degree \(n+r\): \[ f(\pi_n^M(m))=\pi_{n+r}^N(f(m))=0. \] Thus every component of a kernel vector is again in the kernel.

Step 3. Close the image under components

For \(y=f(m)\), \[ \pi_n^N(y)=f(\pi_{n-r}^M(m)). \] The right side lies in the image, proving closure under components.

Step 4. Obtain homogeneous preimages and detect homogeneous vectors

Take \(r=0\). If \(f\) is surjective and \(y\in N_n\), choose \(m\) with \(f(m)=y\). Then \[ f(\pi_n^M(m))=\pi_n^N(y)=y, \] giving a preimage in \(M_n\).

If \(f\) is injective and \(f(m)\in N_n\), the same equality gives \(f(\pi_n^M(m))=f(m)\). Injectivity implies \(m=\pi_n^M(m)\in M_n\). In particular the inverse of a bijective degree-zero map preserves degrees.

Step 5. Apply the identity to a homogeneous scalar

In the compatibly graded \(A\)-module setting, if \(a\in A_r\), multiplication \(m\mapsto am\) is an \(A\)-linear map of degree \(r\). Thus \[ \pi_n(am)=a\,\pi_{n-r}(m). \] Its image \(aM\) is homogeneous by Step 3. This supplies the grading used on scalar quotients.

For \(A=k[X]\), multiplication by \(X\) has degree one, and the formula reads \(\pi_n(Xf)=X\pi_{n-1}(f)\).

Step 6. Restrict a homogeneous equivalence to its pieces

Let \(e:M\to N\) be a \(k\)-linear equivalence of degree \(r\). For \(y\in N_n\), put \(m=e^{-1}(y)\). The component identity from Step 1 gives \[ e(\pi_{n-r}^M(m))=\pi_n^N(e(m))=\pi_n^N(y)=y=e(m). \] Injectivity of \(e\) gives \(m=\pi_{n-r}^M(m)\), so \(e^{-1}(y)\in M_{n-r}\).

For each \(j\), the forward map therefore restricts to \(M_j\to N_{j+r}\), and its inverse restricts to \(N_{j+r}\to M_j\). Their two composite identities are inherited from \(e,e^{-1}\), and both restricted maps are \(k\)-linear. This constructs homogeneousPieceEquiv.

Taking \(j=n-r\) and dimensions gives the finite-dimensional formula \[ \dim_kN_n=\dim_kM_{n-r}. \] This is the finrank_homogeneousEquiv equality used in the word-character proof. The inverse-degree argument is the source's equiv_inverse_homogeneous, applied here with scalar ring \(k\).

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.Graded.map_component: GradedModules.lean:891.

OAI.KLInvariance.Graded.kernel_homogeneous: GradedModules.lean:916.

OAI.KLInvariance.Graded.range_homogeneous: GradedModules.lean:928.

OAI.KLInvariance.Graded.component_homogeneous_smul: GradedModules.lean:938.

OAI.KLInvariance.Graded.scalar_range_homogeneous: GradedModules.lean:948.

OAI.KLInvariance.Graded.ModuleData.Hom.component: GradedModules.lean:1157.

OAI.KLInvariance.Graded.ModuleData.Hom.mem_of_map_mem: GradedModules.lean:1327.

OAI.KLInvariance.Graded.ModuleData.Hom.homogeneous_preimage: GradedModules.lean:1336.

OAI.KLInvariance.Graded.equiv_inverse_homogeneous: BottSamelson.lean:567.

OAI.KLInvariance.Graded.homogeneousPieceEquiv: BottSamelson.lean:1654.

OAI.KLInvariance.Graded.finrank_homogeneousEquiv: BottSamelson.lean:1665.

Used by