Finite spectral decomposition with canonical summation equivalence
ProvedWeierstrassEllipticZeta.finite_generalized_eigenspace_decompositioncommutative-algebraelliptic-functionstranscendencezero-estimates
Let be an algebraically closed field, let be a finite-dimensional -vector space, and let be a linear endomorphism of . For , write for the maximal generalized eigenspace of at .
Suppose that is finite and that
Then the subspaces indexed by form an internal direct sum of . More precisely, there is a linear equivalence
Thus every vector has a unique decomposition as the sum of one vector from each indexed generalized eigenspace. Diagonalizability is not assumed. The set may include values whose generalized eigenspaces are zero; the empty set is allowed when has dimension zero.
Preamble
import Mathlib.LinearAlgebra.Eigenspace.Triangularizable import Mathlib.Algebra.DirectSum.Module open scoped Classical
Formal statement
theorem WeierstrassEllipticZeta.finite_generalized_eigenspace_decomposition
(K M : Type*) [Field K] [IsAlgClosed K]
[AddCommGroup M] [Module K M] [FiniteDimensional K M]
(f : Module.End K M) (s : Finset K)
(hzero : ∀ z : K, z ∉ s →
Module.finrank K (f.maxGenEigenspace z) = 0) :
DirectSum.IsInternal (fun z : s => f.maxGenEigenspace z.val) ∧
∃ e : (∀ z : s, f.maxGenEigenspace z.val) ≃ₗ[K] M,
∀ x, e x = ∑ z : s, (x z : M) := by sorry
Source
Derived linear-algebra lemma for the approach associated with Senthil Kumar K (2026), Appendix A and Theorem A.2, https://doi.org/10.1017/S001309152610145X. This lemma is proved here and is not quoted from the article. In a finite-dimensional vector space over an algebraically closed field, if maximal generalized eigenspaces have dimension zero outside a finite set, those indexed by that set form an internal direct sum. Their finite product is linearly equivalent to the ambient space by summation. Uses Mathlib generalized-eigenspace spanning and independence. No Prove2Me theorem dependencies or new definitions.