Lemma 4.9: weights on representation-finite posets #
lem:poset-weights (scalar-normalization.tex): "If the incidence
algebra of a finite poset R is representation-finite, every additive weight,
with values in any abelian group, has the form w(x,y) = ζ_y − ζ_x."
Proof of the source: "Remove a minimal element s and put U = {y : s < y}.
Distinct connected components of U lie in distinct components of R ∖ {s}.
Otherwise, put a one-dimensional space at every vertex, assign the map one to
every comparison outside s, and assign maps λ ∈ k* from s to one
component of U and one to the others. These are representations. An
isomorphism must have the same scalar at all vertices of each component of
R ∖ {s}. Two components of U in the same such component therefore force
equality of the parameters. This produces infinitely many isomorphism classes
of fixed dimension, contrary to finite representation type. […] By induction
normalize all weights on R ∖ {s} to zero. The remaining weights from s are
constant on each component of U. Adjust the object values separately on the
corresponding components of R ∖ {s} to remove these constants."
The representations are right modules over the incidence algebra on a convex
connected set D (thinRep); the source restricts them to the component of
s, where they are indecomposable. The removal of minimal elements runs
through the up-sets of R, whose components are convex in R, so the
representations of all steps are representations of the incidence algebra of
R itself (no corner argument is needed).
Comparability.
Instances For
Connectedness inside a set, through comparable elements of the set.
Instances For
A convex set.
Instances For
Thin representations #
The action of the incidence algebra on k^D through the scalars ρ.
Instances For
Multiplicativity of ρ along chains in D, with ρ x x = 1.
- refl (x : P) : x ∈ D → ρ x x = 1
- trans (x y z : P) : x ∈ D → y ∈ D → z ∈ D → x ≤ y → y ≤ z → ρ x z = ρ x y * ρ y z
Instances For
The carrier of the thin representation.
Instances For
The right module structure.
Instances For
The thin representation k^D with maps ρ.
Instances For
Homomorphisms between thin representations are diagonal #
The matrix unit at x ≤ y.
Instances For
The basis vector at x.
Instances For
Homomorphisms of thin representations are diagonal, with
c_x ρ'(x,y) = c_y ρ(x,y).
Finiteness and indecomposability #
The identification with k^D is k-linear.
Instances For
A connected set.
Instances For
Endomorphisms of a connected thin representation are scalars.
Isomorphic thin representations differ by object values.
Separation of the components of U #
A path inside Q from a point of a subset D closed under comparability
in Q stays in D.
Distinct components of U lie in distinct components of Q ∖ {s}
(scalar-normalization.tex), for an up-set Q and a minimal s ∈ Q.
Normalizing weights #
The induction of the source, over the up-sets of R.
Lemma 4.9 (lem:poset-weights), over an infinite field.
Lemma 4.9 for multiplicative k*-weights, with either orientation of the
incidence algebra: μ(a,b) = Z_a / Z_b.