The function field of a geometrically integral scheme #
Let X be an integral scheme over a field k that is geometrically integral over k. Then for
every field extension L / k the ring k(X) ⊗[k] L is a domain, and consequently k is
algebraically closed in the function field k(X), that is, k is the full field of constants of
k(X).
For the first statement, choose a nonempty affine open U of X. The base change
Spec (Γ(X, U) ⊗[k] L) of U to L is a nonempty open subscheme of the integral scheme
X ×_k Spec L, so Γ(X, U) ⊗[k] L is a domain, and k(X) ⊗[k] L is a localization of it because
k(X) is the fraction field of Γ(X, U). The second statement follows by taking for L an
algebraic closure of k (TauCeti.algebraicClosure_eq_bot_of_isDomain_tensorProduct).
The constant-field hypothesis IsIntegrallyClosedIn k X.functionField is how the function-field
theory of curves (Weil differentials, the genus of the function field, Serre duality for divisor
sheaves) is applied to a curve. This file derives it from geometric integrality, which in turn
holds for smooth geometrically connected schemes over k
(TauCeti.AlgebraicGeometry.Smooth.geometricallyIntegral).
The proof is built on Andrew Yang's formalization of geometrically integral morphisms in Mathlib
(AlgebraicGeometry.GeometricallyIntegral, in Mathlib/AlgebraicGeometry/Geometrically/Integral),
which supplies the integrality of the base change X ×_k Spec L.
Main results #
TauCeti.AlgebraicGeometry.isDomain_functionField_tensorProduct_of_geometricallyIntegral:k(X) ⊗[k] Lis a domain for every field extensionL / k;TauCeti.AlgebraicGeometry.isIntegrallyClosedIn_functionField_of_geometricallyIntegral:kis algebraically closed ink(X).
The function field of a geometrically integral scheme over k stays a domain after extending
scalars to any field extension L of k.
The constant field of a geometrically integral scheme. If X is integral and
geometrically integral over k, then k is algebraically closed in the function field k(X).