Ordinary projective dimensions of actual incidence modules
Let k be any field and S a finite bounded lattice. Let P = PairPoset S
be the actual interval-pair poset, A its incidence algebra, and I its
diagonal ideal. Put R = Aᵐᵒᵖ and C = (A/I)ᵐᵒᵖ. The objects of ModuleCat R
and ModuleCat C are left modules; equivalently, they encode right modules
over A and A/I, respectively. Write
D(P) = maxComplementCard (S := P).
For every finite-dimensional object M of ModuleCat R,
TDA.Persistence.IncidenceVertexProjectiveDimension.hasProjectiveDimensionLE
proves HasProjectiveDimensionLE M (D(P)). The same bound holds for every
finite-dimensional object of ModuleCat C, measured using its canonical
restricted k-action, by quotient_hasProjectiveDimensionLE. No projective
resolution or projective-dimension bound is supplied as input.
The comparison theorem
hasProjectiveDimensionLE_of_vertex first transfers any bound on the ordinary
projective dimension of the actual all-vertex representation of a finite-
dimensional R-module back to the module. It uses finite-free covers, field
finite-dimensionality of their kernels, exactness of the actual vertex
functor, and projectivity of the regular module's vertex image. The latter is
identified with the graded-free module on interval-pair vertices by
IncidenceRegularVertex.regularVertexIso. The existing representation bound
then supplies D(PairPoset S) without an extra dimension hypothesis.
For an actual coordinate grid G : CoordinateGrid n with nonempty factors and
n ≥ 1, the actual incidence and quotient module theorems give bounds 2*n
for every finite-dimensional module over their respective rings. The
coefficient field remains arbitrary; finite-dimensionality is over the actual
restricted k-action, not a finite-field or finite-carrier assumption.
The representation result is for actual PModule objects, and this module
comparison is for finite-dimensional incidence modules. Together they do not
assert essential surjectivity of the vertex representation, an equivalence
with all persistence modules, or finite projective dimension for arbitrary
incidence modules. A separate result proves the ordinary projective-dimension
bounds for the actual Hom module and the actual cokernel of its two-injective
differential, including the finite-grid 2n−2 Hom-module bound; see
finite-grid Hom-module and Hom-cokernel dimensions.
The real-coordinate transfer and final stability result remain separate.
The source context is BOOS24-v3 §2.4, Lemma 2.1 and Remark 2.2, together with the ordinary-dimension ingredient around Theorem 5.16. The library gives an explicit finite-free dimension-reflection proof through its actual vertex representation and representation-category dimension result, without constructing the whole source equivalence. See BOOS24-v3 in the reference register.