Documentation

TauCeti.RepresentationTheory.Quiver.Representation.Injective.EulerForm

The Euler form against the dimension vector of a vertex injective #

The Euler form of a finite quiver is the numerical shadow of the homological pairing: for finite-dimensional representations of an acyclic quiver, ⟨dim M, dim N⟩ is dim Hom(M, N) - dim Ext¹(M, N). On the injective Iᵢ in the right-hand argument that identity needs no Ext, because Ext¹(-, Iᵢ) vanishes and Hom(M, Iᵢ) is the dual of Mᵢ. This file proves the resulting closed formula,

⟨d, dim Iᵢ⟩ = dᵢ for every d : Q → ℤ,

which holds over any finite quiver with finitely many paths into i — no acyclicity, no algebraically closed field, no finite-dimensionality of the representations involved.

It is the mirror image of TauCeti.RepresentationTheory.Quiver.Representation.Projective.EulerForm, which computes the Euler form against dim Pᵢ in the left-hand argument. The Euler form is not symmetric, so neither file's results follow from the other's; what makes them mirror images is that the path-count vector of Iᵢ is the transpose of that of Pᵢ (TauCeti.dimVector_indecInjRep_eq_dimVector_indecProjRep).

Main results #

Implementation notes #

The combinatorial core is TauCeti.eulerForm_card_path_right, in TauCeti.RepresentationTheory.Quiver.EulerForm, which pairs the Euler form against a path-count vector on the right; the results below just rewrite dim Iᵢ into that path count.

The hypothesis carried throughout is ∀ a, Finite (Quiver.Path a i), for the vertex i at hand: the finiteness that makes TauCeti.dimVector (indecInjRep k Q i) an honest path count rather than the 0 that Module.finrank returns on an infinite-dimensional space. For a finite quiver it follows from acyclicity, by TauCeti.finite_paths_of_isAcyclic; the loop quiver, where it fails, is exactly the boundary case the roadmap records. It is the transpose of the hypothesis ∀ a, Finite (Quiver.Path i a) carried by the projective file, and neither implies the other.

Dimension vectors take values in ℕ while the Euler form is defined on Q → ℤ, so every statement below feeds the Euler form the pointwise cast fun v ↦ (dimVector M v : ℤ).

References #

This implements the injective case of the homological interpretation of the Euler form, Layer 4 of TauCetiRoadmap/RepresentationTheory/QuiverRepresentations/README.md. See Derksen--Weyman, An Introduction to Quiver Representations, Ch. 1, and Assem--Simson--Skowroński, Elements of the Representation Theory of Associative Algebras I, Ch. III.

theorem TauCeti.eulerForm_dimVector_indecInjRep (Q : Type v) [Quiver Q] [Fintype Q] [(a b : Q) → Fintype (a ⟶ b)] (k : Type u) [Field k] (i : Q) [∀ (a : Q), Finite (Quiver.Path a i)] (d : Q → ℤ) :
(((eulerForm Q) d) fun (v : Q) => ↑(dimVector (indecInjRep k Q i) v)) = d i

The Euler form against the dimension vector of Iᵢ is evaluation at i. This is the homological identity ⟨dim M, dim N⟩ = dim Hom(M, N) - dim Ext¹(M, N) in the second case where it needs no Ext: the injective Iᵢ corepresents the dual of evaluation at i.

theorem TauCeti.eulerForm_dimVector_indecInjRep_eq_finrank_hom (Q : Type v) [Quiver Q] [Fintype Q] [(a b : Q) → Fintype (a ⟶ b)] (k : Type u) [Field k] (i : Q) [∀ (a : Q), Finite (Quiver.Path a i)] (M : QuiverRep k Q) :
(((eulerForm Q) fun (v : Q) => ↑(dimVector M v)) fun (v : Q) => ↑(dimVector (indecInjRep k Q i) v)) = ↑(Module.finrank k (M ⟶ indecInjRep k Q i))

The homological reading of the Euler form at an injective. For every representation M, ⟨dim M, dim Iᵢ⟩ is the dimension of Hom(M, Iᵢ); the Ext¹ term of the general identity is absent because Iᵢ is injective.

theorem TauCeti.eulerForm_dimVector_indecInjRep_indecInjRep (Q : Type v) [Quiver Q] [Fintype Q] [(a b : Q) → Fintype (a ⟶ b)] (k : Type u) [Field k] (i j : Q) [∀ (a : Q), Finite (Quiver.Path a j)] :
(((eulerForm Q) fun (v : Q) => ↑(dimVector (indecInjRep k Q i) v)) fun (v : Q) => ↑(dimVector (indecInjRep k Q j) v)) = ↑(Nat.card (Quiver.Path j i))

The Cartan pairing of two vertex injectives. ⟨dim Iᵢ, dim Iⱼ⟩ counts the paths j → i, which is dim Hom(Iᵢ, Iⱼ): the injectives reproduce the same path-counting Cartan matrix as the projectives.

theorem TauCeti.eulerForm_dimVector_indecProjRep_indecInjRep (Q : Type v) [Quiver Q] [Fintype Q] [(a b : Q) → Fintype (a ⟶ b)] (k : Type u) [Field k] (i j : Q) [∀ (a : Q), Finite (Quiver.Path a j)] :
(((eulerForm Q) fun (v : Q) => ↑(dimVector (indecProjRep k Q i) v)) fun (v : Q) => ↑(dimVector (indecInjRep k Q j) v)) = ↑(Nat.card (Quiver.Path i j))

The mixed pairing of a projective against an injective. ⟨dim Pᵢ, dim Iⱼ⟩ counts the paths i → j; reading it instead as dim Hom(Pᵢ, Iⱼ) is TauCeti.eulerForm_dimVector_indecInjRep_eq_finrank_hom at M := Pᵢ. Both TauCeti.eulerForm_dimVector_indecProjRep and TauCeti.eulerForm_dimVector_indecInjRep apply here, and they agree.

theorem TauCeti.eulerForm_single_dimVector_indecInjRep (Q : Type v) [Quiver Q] [Fintype Q] [(a b : Q) → Fintype (a ⟶ b)] (k : Type u) [Field k] [DecidableEq Q] (i : Q) [∀ (a : Q), Finite (Quiver.Path a i)] (j : Q) :
(((eulerForm Q) (Pi.single j 1)) fun (v : Q) => ↑(dimVector (indecInjRep k Q i) v)) = if i = j then 1 else 0

The Euler pairing of a vertex simple against an injective. ⟨αⱼ, dim Iᵢ⟩ = δᵢⱼ, the dimension of Hom(Sⱼ, Iᵢ): the injectives and the simples are dual bases for the Euler form, as the projectives and the simples are on the other side.

theorem TauCeti.titsForm_dimVector_indecInjRep (Q : Type v) [Quiver Q] [Fintype Q] [(a b : Q) → Fintype (a ⟶ b)] (k : Type u) [Field k] (i : Q) [∀ (a : Q), Finite (Quiver.Path a i)] :
((titsForm Q) fun (v : Q) => ↑(dimVector (indecInjRep k Q i) v)) = ↑(Nat.card (Quiver.Path i i))

The Tits norm of the dimension vector of Iᵢ counts the closed paths at i, the same count as for Pᵢ.