The order-five Kummer model in characteristic three
DefinitionN25_ExplicitKummerThreemazur-n25
Over a characteristic-three field containing with , the file defines the two coefficients and , explicit forward and inverse eigenbasis maps, and the Segre cubic obtained from . On the dense chart it introduces the invariant and the cyclic Kummer equation
These definitions reduce the characteristic-three canonical-curve count from projective four-space to one-dimensional fifth-power fibres.
Definition code
import Mathlib.FieldTheory.Finite.Basic
import Mathlib.GroupTheory.SpecificGroups.Cyclic
import Definitions.Def_N25_ExplicitWeilThreeDefs
/-!
# The order-five Kummer model in characteristic three
The canonical level-25 curve carries a proved automorphism of order five.
Over a characteristic-three field containing a fifth root of unity `e`, its
four-dimensional representation diagonalizes. In the eigenbasis the
canonical quadric is the Segre quadric, so the substitution
```
(u₁,u₂,u₃,u₄) = (ac,ad,bc,bd)
```
solves the quadric identically. The cubic then becomes a four-term
bidegree-`(3,3)` equation. On the dense chart `a=c=1`, the invariant
parameter `t=r²s` converts that equation to a cyclic Kummer equation of
degree five.
This file proves those reductions symbolically over every field of
characteristic three. Finite computation is deliberately postponed until
the final one-dimensional quotient fibres.
-/
namespace MazurProof.RationalPointsN25QuotientKummerThree
open RationalPointsN25QuotientF2
section StructuralModel
variable {K : Type*} [Field K] [CharP K 3]
/-- The fifth-cyclotomic relation imposed on the eigenvalue `e` of the
order-five automorphism. -/
def IsCyclotomicFive25Three (e : K) : Prop :=
e ^ 4 + e ^ 3 + e ^ 2 + e + 1 = 0
/-- The first nontrivial coefficient in the four-term Segre cubic. -/
def segreCoefficientA25Three (e : K) : K :=
2 * e ^ 3 + e + 1
/-- The second nontrivial coefficient in the four-term Segre cubic. -/
def segreCoefficientB25Three (e : K) : K :=
e ^ 3 + 2 * e ^ 2 + 2 * e + 1
/-- The cubic left after diagonalizing the order-five action and solving the
canonical quadric by the Segre parametrization. Its four monomials all have
weight zero for `(r,s) ↦ (e²r,es)`. -/
def canonicalSegreCubic25Three (e a b c d : K) : K :=
a ^ 2 * b * c ^ 3 +
segreCoefficientA25Three e * b ^ 3 * c ^ 2 * d +
segreCoefficientB25Three e * a ^ 3 * c * d ^ 2 +
e ^ 2 * a * b ^ 2 * d ^ 3
/-- The inverse eigenbasis matrix, already reduced modulo the fifth
cyclotomic polynomial in characteristic three.
The inputs are the four eigen-coordinates and the outputs are the original
canonical coordinates `(x,y,z,w)`. Keeping this linear map separate from
the Segre parametrization makes its inverse available for the projective
chart comparison. -/
def canonicalCoordinatesFromEigen25Three
(e u₁ u₂ u₃ u₄ : K) : Coordinates4 K :=
⟨
(e ^ 2 + 2) * u₁ +
(2 * e ^ 3 + 2 * e ^ 2 + 2 * e + 1) * u₂ +
(e + 2) * u₃ + (e ^ 3 + 2) * u₄,
(2 * e ^ 3 + e ^ 2) * u₁ +
(2 * e ^ 3 + 2 * e ^ 2 + e + 2) * u₂ +
(e ^ 3 + e ^ 2 + 2 * e + 1) * u₃ +
(e ^ 3 + 2 * e ^ 2) * u₄,
(e ^ 3 + 2 * e) * u₁ + (2 * e ^ 2 + e) * u₂ +
(e ^ 3 + 2 * e ^ 2 + 2 * e + 2) * u₃ +
(e ^ 3 + 2 * e ^ 2 + e + 1) * u₄,
(e ^ 3 + 2 * e ^ 2 + e + 1) * u₁ +
(e ^ 3 + 2 * e ^ 2 + 2 * e + 2) * u₂ +
(2 * e ^ 2 + e) * u₃ + (e ^ 3 + 2 * e) * u₄
⟩
/-- The explicit forward eigenbasis matrix taking canonical coordinates
`(x,y,z,w)` back to the four order-five eigencoordinates.
These coefficients are the inverse of
`canonicalCoordinatesFromEigen25Three` over the fifth-cyclotomic locus. -/
def canonicalEigenCoordinates25Three (e : K)
(P : Coordinates4 K) : Coordinates4 K :=
⟨
P.x + (2 + 2 * e + 2 * e ^ 3) * P.y +
(2 + 2 * e + 2 * e ^ 2 + 2 * e ^ 3) * P.z + e * P.w,
P.x + (2 + 2 * e + 2 * e ^ 2) * P.y + e ^ 3 * P.z + e ^ 2 * P.w,
P.x + (e + e ^ 2) * P.y + e ^ 2 * P.z + e ^ 3 * P.w,
P.x + (e + e ^ 3) * P.y + e * P.z +
(2 + 2 * e + 2 * e ^ 2 + 2 * e ^ 3) * P.w
⟩
/-- Apply the inverse eigenbasis after the Segre map
`(u₁,u₂,u₃,u₄)=(ac,ad,bc,bd)`.
Avoiding projective normalization here keeps the algebraic change of
variables homogeneous and denominator-free. -/
def canonicalSegreCoordinates25Three (e a b c d : K) : Coordinates4 K :=
canonicalCoordinatesFromEigen25Three e (a * c) (a * d) (b * c) (b * d)
/-- The forward and inverse eigenbasis matrices compose to the identity on
eigen-coordinates.
The proof reduces powers of `e` with the fifth-cyclotomic relation. Every
remaining coefficient is a multiple of three, so characteristic three
finishes the four coordinate identities. -/
theorem canonicalEigenCoordinates_fromEigen {e u₁ u₂ u₃ u₄ : K}
(he : IsCyclotomicFive25Three e) :
canonicalEigenCoordinates25Three e
(canonicalCoordinatesFromEigen25Three e u₁ u₂ u₃ u₄) =
⟨u₁, u₂, u₃, u₄⟩ := by
unfold IsCyclotomicFive25Three at he
have he₄ : e ^ 4 = -e ^ 3 - e ^ 2 - e - 1 := by
linear_combination he
have he₅ : e ^ 5 = 1 := by
apply sub_eq_zero.mp
calc
e ^ 5 - 1 = (e - 1) * (e ^ 4 + e ^ 3 + e ^ 2 + e + 1) := by ring
_ = 0 := by rw [he]; ring
have he₆ : e ^ 6 = e := by
rw [show e ^ 6 = e ^ 5 * e by ring, he₅, one_mul]
simp only [canonicalEigenCoordinates25Three,
canonicalCoordinatesFromEigen25Three, Coordinates4.mk.injEq]
refine ⟨?_, ?_, ?_, ?_⟩ <;> ring_nf
all_goals simp only [he₆, he₅, he₄]
all_goals ring_nf
· linear_combination
(e * u₂ + e * u₃ - e * u₄ - e ^ 2 * u₁ + e ^ 2 * u₂ +
e ^ 2 * u₃ + e ^ 2 * u₄ + e ^ 3 * u₁ + 3 * e ^ 3 * u₂ +
2 * e ^ 3 * u₃ + 2 * e ^ 3 * u₄ - 2 * u₁ + u₄) *
CharP.cast_eq_zero K 3
· linear_combination
(-e * u₂ - 3 * e * u₁ - 2 * e * u₄ + e ^ 2 * u₂ -
2 * e ^ 2 * u₁ - e ^ 2 * u₄ + e ^ 3 * u₂ + e ^ 3 * u₃ -
e ^ 3 * u₁ + e ^ 3 * u₄ - u₁) * CharP.cast_eq_zero K 3
· linear_combination
(-e * u₂ - e * u₃ - e * u₁ - 2 * e * u₄ - e ^ 2 * u₂ -
e ^ 2 * u₁ - 2 * e ^ 2 * u₄ - e ^ 3 * u₄ - u₂ + u₁ - u₄) *
CharP.cast_eq_zero K 3
· linear_combination
(-e * u₁ + e * u₂ - e * u₃ + e ^ 2 * u₂ - e ^ 2 * u₄ +
3 * e ^ 3 * u₂ + e ^ 3 * u₄ - u₃ - u₄) * CharP.cast_eq_zero K 3
/-- The inverse and forward eigenbasis matrices compose to the identity on
canonical coordinates. -/
theorem canonicalCoordinatesFromEigen_eigen {e : K} {P : Coordinates4 K}
(he : IsCyclotomicFive25Three e) :
canonicalCoordinatesFromEigen25Three e
(canonicalEigenCoordinates25Three e P).x
(canonicalEigenCoordinates25Three e P).y
(canonicalEigenCoordinates25Three e P).z
(canonicalEigenCoordinates25Three e P).w = P := by
rcases P with ⟨x, y, z, w⟩
unfold IsCyclotomicFive25Three at he
have he₄ : e ^ 4 = -e ^ 3 - e ^ 2 - e - 1 := by
linear_combination he
have he₅ : e ^ 5 = 1 := by
apply sub_eq_zero.mp
calc
e ^ 5 - 1 = (e - 1) * (e ^ 4 + e ^ 3 + e ^ 2 + e + 1) := by ring
_ = 0 := by rw [he]; ring
have he₆ : e ^ 6 = e := by
rw [show e ^ 6 = e ^ 5 * e by ring, he₅, one_mul]
simp only [canonicalEigenCoordinates25Three,
canonicalCoordinatesFromEigen25Three, Coordinates4.mk.injEq]
refine ⟨?_, ?_, ?_, ?_⟩ <;> ring_nf
all_goals simp only [he₆, he₅, he₄]
all_goals ring_nf
· linear_combination
(e * x + 2 * e * y + e * z + e * w + e ^ 2 * x + 2 * e ^ 2 * y +
e ^ 2 * z + e ^ 3 * x + 4 * e ^ 3 * y + e ^ 3 * z + 2 * e ^ 3 * w +
2 * x + y + z + w) * CharP.cast_eq_zero K 3
· linear_combination
(e * x - e * y - e * z - 3 * e * w + 2 * e ^ 2 * x - 2 * e ^ 2 * z -
2 * e ^ 2 * w + 2 * e ^ 3 * x + 2 * e ^ 3 * y + e ^ 3 * z -
e ^ 3 * w + x - y - w) * CharP.cast_eq_zero K 3
· linear_combination
(2 * e * x - e * y - e * z - 2 * e * w + 2 * e ^ 2 * x - e ^ 2 * z -
e ^ 2 * w + e ^ 3 * x + x - 4 * y - 2 * z - w) *
CharP.cast_eq_zero K 3
· linear_combination
(2 * e * x - 2 * e * z - e * w + 2 * e ^ 2 * x + 2 * e ^ 2 * y -
e ^ 2 * z - e ^ 2 * w + e ^ 3 * x + 2 * e ^ 3 * y + x - y - z - 2 * w) *
CharP.cast_eq_zero K 3
/-- The two explicit matrices define a genuine linear coordinate
equivalence on the fifth-cyclotomic locus. -/
def canonicalEigenCoordinateEquiv25Three (e : K)
(he : IsCyclotomicFive25Three e) : Coordinates4 K ≃ Coordinates4 K where
toFun := canonicalEigenCoordinates25Three e
invFun U := canonicalCoordinatesFromEigen25Three e U.x U.y U.z U.w
left_inv P := canonicalCoordinatesFromEigen_eigen he
right_inv U := by
rcases U with ⟨u₁, u₂, u₃, u₄⟩
exact canonicalEigenCoordinates_fromEigen he
/-- The nonzero scalar relating the canonical quadric to the determinant
quadric in eigen-coordinates. -/
def canonicalSegreQuadricScalar25Three (e : K) : K :=
1 + 2 * e ^ 2 + 2 * e ^ 3
/-- The quadric scalar cannot vanish at a fifth-cyclotomic eigenvalue.
The displayed Bézout identity combines the fifth-cyclotomic polynomial with
the scalar to obtain one; the remaining coefficient is a multiple of three. -/
theorem canonicalSegreQuadricScalar25Three_ne_zero {e : K}
(he : IsCyclotomicFive25Three e) :
canonicalSegreQuadricScalar25Three e ≠ 0 := by
intro hs
unfold IsCyclotomicFive25Three at he
unfold canonicalSegreQuadricScalar25Three at hs
have hone : (1 : K) = 0 := by
linear_combination
(e ^ 2 + e - 1) * he + (e ^ 3 + e ^ 2 - 1) * hs -
(e ^ 6 + 2 * e ^ 5 + e ^ 4 - 1) * CharP.cast_eq_zero K 3
exact one_ne_zero hone
/-- In eigen-coordinates, the canonical quadric is a nonzero scalar times
the determinant `u₂u₃-u₁u₄`.
This single matrix identity is the structural content behind the Segre
parametrization. The proof is polynomial reduction modulo the
fifth-cyclotomic relation and characteristic three, not a point table. -/
theorem canonicalQuadric25_fromEigen {e u₁ u₂ u₃ u₄ : K}
(he : IsCyclotomicFive25Three e) :
canonicalQuadric25Three
(canonicalCoordinatesFromEigen25Three e u₁ u₂ u₃ u₄) =
canonicalSegreQuadricScalar25Three e * (u₂ * u₃ - u₁ * u₄) := by
unfold IsCyclotomicFive25Three at he
unfold canonicalQuadric25Three canonicalCoordinatesFromEigen25Three
canonicalSegreQuadricScalar25Three
linear_combination
(7 * e ^ 2 * u₁ ^ 2 + 7 * e ^ 2 * u₁ * u₂ +
8 * e ^ 2 * u₁ * u₃ + 7 * e ^ 2 * u₁ * u₄ +
2 * e ^ 2 * u₂ ^ 2 + 5 * e ^ 2 * u₂ * u₃ +
2 * e ^ 2 * u₂ * u₄ + 2 * e ^ 2 * u₃ ^ 2 +
4 * e ^ 2 * u₃ * u₄ + e ^ 2 * u₄ ^ 2 -
2 * e * u₁ ^ 2 + 6 * e * u₁ * u₂ + 9 * e * u₁ * u₃ +
10 * e * u₁ * u₄ + 2 * e * u₂ ^ 2 + 5 * e * u₂ * u₃ +
12 * e * u₂ * u₄ + 5 * e * u₃ ^ 2 + 9 * e * u₃ * u₄ +
7 * e * u₄ ^ 2 + u₁ ^ 2 - 5 * u₁ * u₂ + 5 * u₁ * u₃ -
2 * u₁ * u₄ - u₂ ^ 2 + 11 * u₂ * u₃ - u₂ * u₄ +
8 * u₃ ^ 2 + 10 * u₃ * u₄ + u₄ ^ 2) * he +
(-2 * e ^ 3 * u₁ ^ 2 - e ^ 3 * u₁ * u₃ -
4 * e ^ 3 * u₁ * u₄ + e ^ 3 * u₂ * u₃ -
2 * e ^ 3 * u₄ ^ 2 - 3 * e ^ 2 * u₁ ^ 2 -
5 * e ^ 2 * u₁ * u₂ - 7 * e ^ 2 * u₁ * u₃ -
4 * e ^ 2 * u₁ * u₄ + e ^ 2 * u₂ ^ 2 -
3 * e ^ 2 * u₂ * u₃ - 2 * e ^ 2 * u₂ * u₄ -
2 * e ^ 2 * u₃ ^ 2 - 6 * e ^ 2 * u₃ * u₄ -
3 * e ^ 2 * u₄ ^ 2 - e * u₁ ^ 2 - e * u₁ * u₂ -
7 * e * u₁ * u₃ - 6 * e * u₁ * u₄ - 2 * e * u₂ * u₃ -
5 * e * u₂ * u₄ - 3 * e * u₃ ^ 2 - 8 * e * u₃ * u₄ -
4 * e * u₄ ^ 2 - u₁ ^ 2 - 3 * u₁ * u₃ + u₂ ^ 2 -
2 * u₂ * u₃ - 3 * u₃ ^ 2 - 5 * u₃ * u₄ - u₄ ^ 2) *
CharP.cast_eq_zero K 3
/-- On the Segre quadric, the original canonical cubic is the four-term
Segre cubic multiplied by `e³+1`. Thus the diagonal change of coordinates
identifies the two curve equations once that harmless scalar is shown
nonzero. -/
theorem canonicalCubic25_segre {e a b c d : K}
(he : IsCyclotomicFive25Three e) :
let P := canonicalSegreCoordinates25Three e a b c d
canonicalCubic25Three P =
(e ^ 3 + 1) * canonicalSegreCubic25Three e a b c d := by
unfold IsCyclotomicFive25Three at he
have he₄ : e ^ 4 = -e ^ 3 - e ^ 2 - e - 1 := by
linear_combination he
have he₅ : e ^ 5 = 1 := by
apply sub_eq_zero.mp
calc
e ^ 5 - 1 = (e - 1) * (e ^ 4 + e ^ 3 + e ^ 2 + e + 1) := by ring
_ = 0 := by rw [he]; ring
have he₆ : e ^ 6 = e := by rw [show e ^ 6 = e ^ 5 * e by ring, he₅, one_mul]
have he₇ : e ^ 7 = e ^ 2 := by
rw [show e ^ 7 = e ^ 5 * e ^ 2 by ring, he₅, one_mul]
have he₈ : e ^ 8 = e ^ 3 := by
rw [show e ^ 8 = e ^ 5 * e ^ 3 by ring, he₅, one_mul]
have he₉ : e ^ 9 = e ^ 4 := by
rw [show e ^ 9 = e ^ 5 * e ^ 4 by ring, he₅, one_mul]
simp only [canonicalSegreCoordinates25Three,
canonicalCoordinatesFromEigen25Three, canonicalCubic25Three,
canonicalSegreCubic25Three, segreCoefficientA25Three,
segreCoefficientB25Three]
ring_nf
simp only [he₉, he₈, he₇, he₆, he₅, he₄]
ring_nf
linear_combination
(-5 * e ^ 3 * a ^ 3 * c ^ 3 - 2 * e ^ 3 * a ^ 2 * b * c ^ 3 +
6 * e ^ 3 * a * b ^ 2 * c ^ 3 + 2 * e ^ 3 * b ^ 3 * c ^ 3 -
2 * e ^ 3 * a ^ 3 * c ^ 2 * d + 11 * e ^ 3 * a ^ 2 * b * c ^ 2 * d -
3 * e ^ 3 * a * b ^ 2 * c ^ 2 * d - 5 * e ^ 3 * b ^ 3 * c ^ 2 * d +
e ^ 3 * a ^ 3 * c * d ^ 2 - 4 * e ^ 3 * a ^ 2 * b * c * d ^ 2 -
3 * e ^ 3 * a * b ^ 2 * c * d ^ 2 - e ^ 3 * b ^ 3 * c * d ^ 2 -
3 * e ^ 3 * a ^ 3 * d ^ 3 - 5 * e ^ 3 * a ^ 2 * b * d ^ 3 -
6 * e ^ 3 * a * b ^ 2 * d ^ 3 - 2 * e ^ 3 * b ^ 3 * d ^ 3 -
e ^ 2 * a ^ 2 * b * c ^ 3 + 5 * e ^ 2 * a * b ^ 2 * c ^ 3 +
3 * e ^ 2 * b ^ 3 * c ^ 3 + 3 * e ^ 2 * a ^ 3 * c ^ 2 * d +
14 * e ^ 2 * a ^ 2 * b * c ^ 2 * d +
8 * e ^ 2 * a * b ^ 2 * c ^ 2 * d + 2 * e ^ 2 * a ^ 3 * c * d ^ 2 -
11 * e ^ 2 * a ^ 2 * b * c * d ^ 2 -
13 * e ^ 2 * a * b ^ 2 * c * d ^ 2 - e ^ 2 * b ^ 3 * c * d ^ 2 +
3 * e ^ 2 * a ^ 2 * b * d ^ 3 + 3 * e ^ 2 * a * b ^ 2 * d ^ 3 -
e ^ 2 * b ^ 3 * d ^ 3 - e * a ^ 3 * c ^ 3 +
6 * e * a ^ 2 * b * c ^ 3 + 6 * e * a * b ^ 2 * c ^ 3 +
e * b ^ 3 * c ^ 3 + 2 * e * a ^ 3 * c ^ 2 * d +
11 * e * a ^ 2 * b * c ^ 2 * d + 2 * e * a * b ^ 2 * c ^ 2 * d +
3 * e * b ^ 3 * c ^ 2 * d + 5 * e * a ^ 3 * c * d ^ 2 +
13 * e * a ^ 2 * b * c * d ^ 2 + 16 * e * a * b ^ 2 * c * d ^ 2 +
9 * e * b ^ 3 * c * d ^ 2 + 2 * e * a ^ 3 * d ^ 3 +
3 * e * a ^ 2 * b * d ^ 3 + e * b ^ 3 * d ^ 3 -
5 * a ^ 3 * c ^ 3 - 8 * a ^ 2 * b * c ^ 3 + 2 * a * b ^ 2 * c ^ 3 +
4 * a ^ 3 * c ^ 2 * d + 28 * a ^ 2 * b * c ^ 2 * d +
12 * a * b ^ 2 * c ^ 2 * d + b ^ 3 * c ^ 2 * d +
2 * a ^ 3 * c * d ^ 2 + 2 * a ^ 2 * b * c * d ^ 2 +
11 * a * b ^ 2 * c * d ^ 2 + 7 * b ^ 3 * c * d ^ 2 +
2 * a ^ 3 * d ^ 3 + 3 * a ^ 2 * b * d ^ 3 +
9 * a * b ^ 2 * d ^ 3 + 2 * b ^ 3 * d ^ 3) *
CharP.cast_eq_zero K 3
omit [CharP K 3] in
/-- A fifth-cyclotomic eigenvalue is nonzero. -/
theorem cyclotomicFive25Three_ne_zero {e : K}
(he : IsCyclotomicFive25Three e) : e ≠ 0 := by
intro h
subst e
norm_num [IsCyclotomicFive25Three] at he
omit [CharP K 3] in
/-- The scalar `e³+1` relating the two cubics cannot vanish for a
fifth-cyclotomic eigenvalue in characteristic three. -/
theorem cyclotomicFive25Three_cubicScalar_ne_zero {e : K}
(he : IsCyclotomicFive25Three e) : e ^ 3 + 1 ≠ 0 := by
have he_ne := cyclotomicFive25Three_ne_zero he
intro hs
unfold IsCyclotomicFive25Three at he
have he₃ : e ^ 3 = -1 := by linear_combination hs
have he₄ : e ^ 4 = -e := by rw [show e ^ 4 = e ^ 3 * e by ring, he₃]; ring
rw [he₄, he₃] at he
have he₂ : e ^ 2 = 0 := by
linear_combination he
have he_zero : e = 0 := by
have hmul : e * e = 0 := by simpa [pow_two] using he₂
exact (mul_eq_zero.mp hmul).elim id id
exact he_ne he_zero
/-- The coefficient `A=2e³+e+1` is nonzero on the fifth-cyclotomic locus.
This prevents an entire boundary ruling of the Segre quadric from lying on
the cubic. -/
theorem segreCoefficientA25Three_ne_zero {e : K}
(he : IsCyclotomicFive25Three e) : segreCoefficientA25Three e ≠ 0 := by
intro hA
unfold IsCyclotomicFive25Three at he
unfold segreCoefficientA25Three at hA
have hone : (1 : K) = 0 := by
linear_combination
(e ^ 2 + e + 1) * he +
(e ^ 3 + 2 * e ^ 2 + e) * hA -
(e ^ 6 + 2 * e ^ 5 + 2 * e ^ 4 + 2 * e ^ 3 + 2 * e ^ 2 + e) *
CharP.cast_eq_zero K 3
exact one_ne_zero hone
/-- The coefficient `B=e³+2e²+2e+1` is nonzero on the
fifth-cyclotomic locus. It controls both the dense `r=0` fibre and the
single Kummer point introduced by clearing `r⁴`. -/
theorem segreCoefficientB25Three_ne_zero {e : K}
(he : IsCyclotomicFive25Three e) : segreCoefficientB25Three e ≠ 0 := by
intro hB
unfold IsCyclotomicFive25Three at he
unfold segreCoefficientB25Three at hB
have hone : (1 : K) = 0 := by
linear_combination
(2 * e) * he + (e ^ 2 + 2 * e + 1) * hB -
(e ^ 5 + 2 * e ^ 4 + 3 * e ^ 3 + 3 * e ^ 2 + 2 * e) *
CharP.cast_eq_zero K 3
exact one_ne_zero hone
/-! ## Boundary rulings and the dense Kummer chart -/
/-- On the boundary ruling `a=0`, the cubic has exactly the point `d=0`.
In projective coordinates this is one of the three points outside the dense
Segre chart. -/
theorem segre_boundary_a_zero {e d : K}
(he : IsCyclotomicFive25Three e) :
canonicalSegreCubic25Three e 0 1 1 d = 0 ↔ d = 0 := by
have hA := segreCoefficientA25Three_ne_zero he
simp [canonicalSegreCubic25Three, hA]
omit [CharP K 3] in
/-- On the boundary ruling `c=0`, the cubic has exactly the point `b=0`.
This is the second point outside the dense Segre chart. -/
theorem segre_boundary_c_zero {e b : K}
(he : IsCyclotomicFive25Three e) :
canonicalSegreCubic25Three e 1 b 0 1 = 0 ↔ b = 0 := by
have he0 := cyclotomicFive25Three_ne_zero he
simp [canonicalSegreCubic25Three, he0]
omit [CharP K 3] in
/-- The intersection of the two boundary rulings lies on the Segre cubic,
giving the third point outside the dense chart. -/
theorem segre_boundary_corner (e : K) :
canonicalSegreCubic25Three e 0 1 0 1 = 0 := by
simp [canonicalSegreCubic25Three]
/-- The dense Segre chart `a=c=1`, with affine coordinates `(r,s)`. -/
def canonicalDenseSegreCubic25Three (e r s : K) : K :=
canonicalSegreCubic25Three e 1 r 1 s
/-- In the dense chart, setting `r=0` forces `s=0`. Hence the genuine Segre
curve contributes only `(r,s)=(0,0)` above the zero value of `r`. -/
theorem denseSegre_at_zero {e s : K}
(he : IsCyclotomicFive25Three e) :
canonicalDenseSegreCubic25Three e 0 s = 0 ↔ s = 0 := by
have hB := segreCoefficientB25Three_ne_zero he
simp [canonicalDenseSegreCubic25Three, canonicalSegreCubic25Three, hB]
/-- The denominator-cleared Kummer equation on the quotient coordinate
`t=r²s`. For fixed `t` it is a degree-five equation in `r`. -/
def canonicalKummerPolynomial25Three (e r t : K) : K :=
r ^ 5 * (1 + segreCoefficientA25Three e * t) +
t ^ 2 * (segreCoefficientB25Three e + e ^ 2 * t)
/-- The coefficient of `r⁵` in the Kummer equation above the quotient
parameter `t`. -/
def canonicalKummerDenominator25Three (e t : K) : K :=
1 + segreCoefficientA25Three e * t
/-- The constant term in the Kummer equation above the quotient parameter
`t`. -/
def canonicalKummerNumerator25Three (e t : K) : K :=
t ^ 2 * (segreCoefficientB25Three e + e ^ 2 * t)
/-- When the coefficient of `r⁵` is nonzero, the Kummer equation is the
fifth-power equation `r⁵ = canonicalKummerTarget25Three e t`. -/
def canonicalKummerTarget25Three (e t : K) : K :=
-canonicalKummerNumerator25Three e t /
canonicalKummerDenominator25Three e t
omit [CharP K 3] in
/-- Multiplying the dense Segre cubic by `r⁴` produces the Kummer equation
with invariant parameter `t=r²s`. This exact identity is the second
dimension reduction in the characteristic-three point count. -/
theorem denseSegre_to_kummer_identity (e r s : K) :
r ^ 4 * canonicalDenseSegreCubic25Three e r s =
canonicalKummerPolynomial25Three e r (r ^ 2 * s) := by
simp only [canonicalDenseSegreCubic25Three, canonicalSegreCubic25Three,
canonicalKummerPolynomial25Three]
ring
omit [CharP K 3] in
/-- Away from `r=0`, the substitution `s=t/r²` is an equivalence between
the dense Segre equation and the Kummer equation. The excluded zero fibre is
handled separately because clearing `r⁴` creates one extraneous Kummer point. -/
theorem denseSegre_eq_zero_iff_kummer
(e : K) {r t : K} (hr : r ≠ 0) :
canonicalDenseSegreCubic25Three e r (t / r ^ 2) = 0 ↔
canonicalKummerPolynomial25Three e r t = 0 := by
have hrt : r ^ 2 * (t / r ^ 2) = t := by
field_simp
have hid := denseSegre_to_kummer_identity e r (t / r ^ 2)
rw [hrt] at hid
constructor
· intro h
rw [h, mul_zero] at hid
exact hid.symm
· intro h
rw [h] at hid
exact (mul_eq_zero.mp hid).resolve_left (pow_ne_zero 4 hr)
omit [CharP K 3] in
/-- At `r=0`, the Kummer equation factors into the two possible quotient
parameters. Only `t=0` comes from `t=r²s`; the second factor records the
single point introduced by clearing the power of `r`. -/
theorem kummer_at_zero (e t : K) :
canonicalKummerPolynomial25Three e 0 t =
t ^ 2 * (segreCoefficientB25Three e + e ^ 2 * t) := by
simp [canonicalKummerPolynomial25Three]
/-- The second zero-fibre parameter created by clearing `r⁴`. -/
def extraneousKummerParameter25Three (e : K) : K :=
-segreCoefficientB25Three e / e ^ 2
omit [CharP K 3] in
/-- The cleared Kummer equation at `r=0` has exactly two quotient parameters:
the genuine value `t=0` and one nonzero extraneous value. -/
theorem kummer_zero_fibre_iff {e t : K}
(he : IsCyclotomicFive25Three e) :
canonicalKummerPolynomial25Three e 0 t = 0 ↔
t = 0 ∨ t = extraneousKummerParameter25Three e := by
have he0 := cyclotomicFive25Three_ne_zero he
rw [kummer_at_zero, mul_eq_zero]
constructor
· rintro (ht | ht)
· left
have hmul : t * t = 0 := by simpa [pow_two] using ht
exact (mul_eq_zero.mp hmul).elim id id
· right
unfold extraneousKummerParameter25Three
apply (eq_div_iff (pow_ne_zero 2 he0)).2
linear_combination ht
· rintro (rfl | ht)
· left
simp
· right
unfold extraneousKummerParameter25Three at ht
have hmul := (eq_div_iff (pow_ne_zero 2 he0)).1 ht
linear_combination hmul
/-- The extraneous zero-fibre parameter is not zero because both `B` and the
cyclotomic eigenvalue are nonzero. -/
theorem extraneousKummerParameter25Three_ne_zero {e : K}
(he : IsCyclotomicFive25Three e) :
extraneousKummerParameter25Three e ≠ 0 := by
have he0 := cyclotomicFive25Three_ne_zero he
have hB := segreCoefficientB25Three_ne_zero he
simp [extraneousKummerParameter25Three, he0, hB]
end StructuralModel
/-! ## Fifth-power fibres over fields with 81 elements
The Kummer equation has the form `r⁵ = c` whenever its linear
coefficient is nonzero. The following lemmas justify the fibre classifier
structurally. They use that the unit group of an 81-element field is cyclic
of order 80; no elements of the field are enumerated.
-/
section FifthPowerFibres
open scoped Pointwise
/-- For a nonzero target in a field, taking fifth roots in the field is
equivalent to taking fifth roots in its unit group.
The only point requiring attention is that a fifth root of a nonzero target
cannot itself be zero. -/
def nonzeroFifthPowerFiberEquivUnits
{K : Type*} [Field K] {c : K} (hc : c ≠ 0) :
{r : K // r ^ 5 = c} ≃
{u : Kˣ // u ^ 5 = Units.mk0 c hc} where
toFun r := by
have hr : (r : K) ≠ 0 := by
intro hr
apply hc
rw [← r.property, hr]
norm_num
refine ⟨Units.mk0 r hr, ?_⟩
apply Units.ext
exact r.property
invFun u := by
refine ⟨(u.1 : K), ?_⟩
have hu := congrArg (↑· : Kˣ → K) u.property
exact hu
left_inv r := by
ext
rfl
right_inv u := by
ext
rfl
/-- The structural root-count classifier for one Kummer fibre over an
81-element field.
If the coefficient of `r⁵` vanishes, the equation is either inconsistent or
identically zero. Otherwise the normalized fifth-power target is classified
by `fifthPowerFiber_card_field81`. -/
def canonicalKummerFiberSizeField81
{K : Type*} [Field K] [DecidableEq K] (e t : K) : Nat :=
let denominator := canonicalKummerDenominator25Three e t
let numerator := canonicalKummerNumerator25Three e t
if denominator = 0 then
if numerator = 0 then 81 else 0
else
let target := -numerator / denominator
if target = 0 then 1 else if target ^ 16 = 1 then 5 else 0
end FifthPowerFibres
end MazurProof.RationalPointsN25QuotientKummerThree
Source
B. Mazur, Modular curves and the Eisenstein ideal, Publ. IHES 47 (1977); Lean formalization of the order-25 case for this mission. Modules in the MazurProof/Keystone/Tate families are adapted from anthropics/fermats-last-theorem (Apache-2.0) via the xiangyazi24/FLT fork.