The linear-algebra inequality in the Cech proof of Riemann's inequality #
Let A ≤ A' and B be subspaces of a vector space V. The canonical map
A' / A → V / (A ⊔ B)
has kernel canonically represented by
(A' ⊓ B) / (A ⊓ B).
Rank--nullity therefore gives
finrank (A' / A) ≤ finrank ((A' ⊓ B) / (A ⊓ B)) + finrank (V / (A ⊔ B)).
Only the three quotient spaces occurring in this argument need to be
finite-dimensional; the ambient space V and the subspaces themselves may
be infinite-dimensional. This is the abstract finite-dimensional step used
by the Cech proof of Riemann's inequality.
The copy of A inside A'. When A ≤ A', this is the denominator in
the relative quotient A' / A.
Equations
Instances For
The canonical map A' / A → V / (A ⊔ B) used in the Cech argument.
Equations
- BGS.HasseWeil.cechRelativeQuotientMap A A' B = (BGS.HasseWeil.relativeSubmodule A A').mapQ (A ⊔ B) A'.subtype ⋯
Instances For
The quotient (A' ⊓ B) / (A ⊓ B) maps naturally into A' / A.
Equations
- BGS.HasseWeil.cechIntersectionQuotientMap A A' B = (Submodule.comap (A' ⊓ B).subtype (A ⊓ B)).liftQ ((BGS.HasseWeil.relativeSubmodule A A').mkQ ∘ₗ Submodule.inclusion ⋯) ⋯
Instances For
The intersection quotient identifies exactly with the kernel of the Cech relative quotient map.
Rank--nullity with the Cech kernel written as the intersection quotient. This equality is the strongest finite-dimensional form of the abstract linear-algebra argument.
The abstract finite-dimensional inequality behind the Cech proof of Riemann's inequality. No finite-dimensionality hypothesis is imposed on the ambient vector space.
Subtractive form of
finrank_relativeQuotient_le_infQuotient_add_supQuotient, convenient when
the intersection quotient is the desired lower bound.