The function field is generated by the y-coordinate #
F(W) = F(x)(y): the function field is generated over the rational functions of the affine
coordinate x by the single element y. The coordinate ring is free of rank two over F[X]
with basis {1, y}, so every one of its elements is p + q y with p and q polynomial in
x; the function field is its fraction field, so a subfield containing the image of the
coordinate ring is everything.
Together with finrank_ratFuncRange, which puts the degree at two, this is what makes y a
primitive element of the extension — the form in which questions about F(W)/F(x) reduce to
questions about the minimal polynomial of y.
The base is any field L between F[X] and F(W), following finrank_functionField in
Finrank.lean, so that the statement serves RatFunc F and FractionRing F[X] alike. What it
needs of L is the algebra structures displayed below together with their scalar-tower
compatibility; being a fraction field of F[X] is not among them.
Main results #
WeierstrassCurve.Affine.adjoin_genericY_eq_top:F(x)⟮y⟯ = F(W).WeierstrassCurve.Affine.minpoly_genericY: the minimal polynomial ofyis the Weierstrass polynomial itself. This is generation and degree read together, so it lives here; it needsLto be a fraction field ofF[X], which the statement above does not.WeierstrassCurve.Affine.adjoin_genericX_genericY_eq_top: over the constants,K⟮x, y⟯ = K(W), andWeierstrassCurve.Affine.fieldRange_eq_adjoin_genericX_genericY: the image ofK(W)under an embedding overKis generated by the images ofxandy.WeierstrassCurve.Affine.fieldRange_eq_adjoin_range_pow: an embeddingK(W') → K(W)carrying the generic point ofW'to theq-th powers of that ofW, and whose image contains everyq-th power, has imageK(K(W)^q). The one-step and the iterated relative Frobenius each supply their own values at the generic point and read off their pulled-back function field.
Provenance #
minpoly_genericY corresponds to the inline step h_eq inside functionField_isSeparable of
projects/HasseWeil/HasseWeil/Ramification.lean in
AINTLIB at revision 513e83879e2f, Apache-2.0, exported
here as a statement in its own right and proved from generation rather than from Gauss's lemma.
The coordinate-ring containment follows the plan of algebraMap_mem_adjoin_genericX_genericY in
Affine/FunctionField/GenericPoint/Basic.lean, which proves the analogous statement for the
F-subalgebra generated by both coordinates: the same basis decomposition and scalar-action step.
It is not derived from that theorem, because passing from an F-subalgebra to an L-subfield
would force an Algebra F L instance that nothing else here needs. The fraction-field step is
Mathlib's
IsFractionRing.closure_range_algebraMap (Mathlib/RingTheory/Localization/FractionRing.lean),
that the image of the ring generates its fraction field as a subfield. What is new here is only
their combination, and the observation that no IsFractionRing F[X] L hypothesis is needed for
it.
Every element of the coordinate ring lies in a subfield of the function field that contains
the image of F[X] and the generic y-coordinate: it is p + q y with p and q polynomial in
x.
The function field is generated by the generic y-coordinate over any field L carrying
the image of F[X]: L⟮y⟯ = F(W). Taking L = F(x) gives F(x)⟮y⟯ = F(W), which is the form
that makes y a primitive element of the quadratic extension.
The Weierstrass polynomial is the minimal polynomial of y. The equation defining the
curve is exactly the relation y satisfies over L, and it has no smaller one.
The function field is generated over the constants by the generic point: K⟮x, y⟯ = K(W).
This is the form of generation that computes the image of an embedding of K(W) over K from
its two values at x and y (fieldRange_eq_adjoin_genericX_genericY).
The image of the function field under an embedding over the constants is generated by the
images of the generic coordinates: f(K(W)) = K⟮f x, f y⟯. This computes the image of a
function-field embedding — the pullback of an isogeny, say — from its values at the two
coordinates.
An embedding K(W') → K(W) carrying the generic point to q-th powers has image
K(K(W)^q), the subfield generated over the constants by the q-th powers, provided every
q-th power lies in its image. The image is generated by the two values x ^ q, y ^ q at the
generic point, and conversely every q-th power is in it by hypothesis.