The p-adic split/inert dichotomy.
ProvedPadicBerggren.B2_null_eigenvector_iff_isSquare_twoaether-cataloggeometry
The p-adic split/inert dichotomy. B₂ has a nonzero eigenvector on the null cone
mod p if and only if 2 is a quadratic residue mod p. The eigenvectors are (1,1,±√2)
with eigenvalues 3 ± 2√2 — the p-adic shadow of the silver ratio.
theorem PadicBerggren.B₂_null_eigenvector_iff_isSquare_two(p : ℕ) [Fact p.Prime] (hp : p ≠ 2) :
(∃ (lam : ZMod p) (v : Fin 3 → ZMod p), v ≠ 0 ∧ lorentz (ZMod p) v = 0 ∧
B₂ (ZMod p) *ᵥ v = lam • v) ↔ IsSquare (2 : ZMod p) := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Geometry/PadicBerggrenDynamics.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Upload correction: only the declared theorem identifier was normalized from PadicBerggren.B₂_null_eigenvector_iff_isSquare_two to PadicBerggren.B2_null_eigenvector_iff_isSquare_two; mathematical statement unchanged.
Preamble
-- Thm stub generated from Geometry/PadicBerggrenDynamics.lean
import Mathlib
import Definitions.Def_Geometry_PadicBerggrenDynamics
import Mathlib.Algebra.CharP.Lemmas
import Mathlib.Data.ZMod.Basic
import Mathlib.GroupTheory.OrderOfElement
import Mathlib.LinearAlgebra.Matrix.Determinant.Basic
import Mathlib.LinearAlgebra.Matrix.Notation
import Mathlib.LinearAlgebra.Matrix.ToLinearEquiv
import Mathlib.NumberTheory.LegendreSymbol.QuadraticReciprocity
import Mathlib.Tactic.FieldSimp
import Mathlib.Tactic.FinCases
import Mathlib.Tactic.LinearCombination
import Mathlib.Tactic.NoncommRing
import Mathlib.Tactic.Ring
/-!
# p-adic Berggren dynamics
The Berggren (Barning–Hall) moves are the three integer matrices
```
B₁ = !![1,-2,2; 2,-1,2; 2,-2,3] B₂ = !![1,2,2; 2,1,2; 2,2,3] B₃ = !![-1,2,2; -2,1,2; -2,2,3]
```
acting on column vectors `(a,b,c)` and generating the ternary tree of primitive Pythagorean
triples. They preserve the Lorentz form `q(a,b,c) = a² + b² − c²`. The catalog already
contains the real/hyperbolic geometry of the tree and the exact spectral data of the
generators over `ℤ` (`Catalog/Cryptography/BerggrenSpectral/Generators.lean`:
`charpoly B₁ = charpoly B₃ = (X−1)³`, `charpoly B₂ = (X+1)(X²−6X+1)`).
This file develops the **p-adic / mod `p^k` dynamics** of the same three matrices. Everything
below is proved for the reductions of the *same* generators, over `ZMod (p^k)`.
## Main results
Structural (any commutative ring `R`):
* `lorentz_B₁`, `lorentz_B₂`, `lorentz_B₃` : the moves preserve `a² + b² − c²` over any `R`;
in particular the null cone mod `p^k` is invariant.
* `bijOn_nullCone_B₁/₂/₃` : each move is a *bijection* of the null cone; mod `p^k` the tree
therefore becomes a finite invertible dynamical system.
* `B₁_eq_one_add`, `N₁_cube`, `N₁_sq_ne_zero` : `B₁ = 1 + N₁` with `N₁³ = 0`, and the
nilpotency index is exactly `3` **iff `4 ≠ 0`** in `R`. Mod `2` and mod `4` the index drops
to `2`: the unipotent p-adic classification is uniform except at the prime `2`.
Unipotent generators (p odd):
* `pow_unipotent_formula` : `(1+X)^n = 1 + n X + C(n,2) X²` whenever `X³ = 0`.
* `B₁_pow_p_pow`, `B₃_pow_p_pow` : `B₁^(p^k) = 1` and `B₃^(p^k) = 1` in `ZMod (p^k)`.
* `B₁_order_exact` : conversely `B₁^m = 1` mod `p^k` forces `p^k ∣ m`, so the order of `B₁`
mod `p^k` is *exactly* `p^k` — a pure `p`-power ("pro-`p`", nilpotent behaviour).
* `B₁_pow_ne_one_padic` : sharpness of the depth — `B₁^(p^k) = 1` mod `p^k` but not mod
`p^(k+1)`.
* `B₁_fixes_null_line`, `B₁_fixed_iff` : `B₁` fixes exactly the null line spanned by `(0,1,1)`
(a boundary point of the light cone), and `B₃` fixes the null line `(1,0,1)`.
Hyperbolic generator (p odd):
* `Um_pow_card` : Frobenius applied to the hyperbolic `2×2` block `U = !![3,2;4,3] = 3 + 2J`,
`J² = 2` : `U^p = 3 + 2·(2^((p−1)/2))·J`.
* `B₂_pow_p_sub_one_of_isSquare_two` / `B₂_pow_p_add_one_of_not_isSquare_two` :
the order of `B₂` mod `p` divides `p − 1` if `2` is a square mod `p` and divides `p + 1`
otherwise. By quadratic reciprocity this is decided by `p mod 8`.
* `B₂_pow_card_sq_sub_one`, `B₂_orderOf_dvd` : in all cases the order divides `p² − 1`,
confirming the conjectured bound.
* `B₂_null_eigenvector_iff_isSquare_two`, `B₂_null_eigenvector_iff_mod_eight` :
`B₂` has a nonzero eigenvector **on the null cone** iff `2` is a square mod `p`, iff
`p ≡ ±1 (mod 8)`. This is the exact p-adic split/inert (hyperbolic/elliptic) dichotomy.
* `B₂_no_nonzero_fixed_point` : `B₂` has no nonzero fixed vector mod `p`, in sharp contrast
with the unipotent generators.
Depth (`p^k`) statements:
* `lift_pow` : an entrywise Hensel lift, `A ≡ 1 (mod p) → A^(p^k) ≡ 1 (mod p^(k+1))`.
* `B₂_pow_eq_one_padic` : `B₂^((p²−1)·p^(k−1)) = 1` in `ZMod (p^k)`.
* `B₂_padic_contraction` : the p-adic distance `|B₂^N v − v|_p ≤ p^(−k)` for `N = (p²−1)p^(k−1)`
and every integer vector `v`: the hyperbolic generator is *periodic to any p-adic precision*.
* `tree_collision_mod` : the `3^d` words of length `d` collide mod `m` as soon as `m³ < 3^d`,
so the reduction of the boundary of the tree is **not** injective: there is no p-adic Cantor
set inside a fixed finite level `ZMod (p^k)` (see `FUTURE_DIRECTIONS.md`).
-/
open PadicBerggren
open Matrix
/-! ## The generators and the Lorentz form -/
variable (R : Type*) [CommRing R]
variable {R : Type*} [CommRing R]
/-! ### Determinants and inverses -/
/-! ### The moves are bijections of the null cone
Over `ZMod (p^k)` the null cone is a finite set, so each Berggren move becomes a permutation
of a finite set: the tree reduces to an invertible finite dynamical system. -/
/-! ## Unipotent generators -/
/-! ### Order of the unipotent generators mod `p^k` -/
/-! ### The unipotent generators fix a single null line -/
/-! ## The hyperbolic generator
`B₂` is conjugate (over `ℤ[1/2]`) to `diag(−1) ⊕ U` with `U = !![3,2;4,3]`, the matrix of
multiplication by the square `3 + 2√2` of the silver ratio on `ℤ[√2]`. We prove everything
through the concrete conjugation `B₂ · W = W · (S · emb U)`. -/
/-! ### Frobenius on the hyperbolic block -/
/-! ### Fixed points and the split/inert dichotomy on the null cone -/Formal statement
theorem PadicBerggren.B2_null_eigenvector_iff_isSquare_two(p : ℕ) [Fact p.Prime] (hp : p ≠ 2) :
(∃ (lam : ZMod p) (v : Fin 3 → ZMod p), v ≠ 0 ∧ lorentz (ZMod p) v = 0 ∧
B₂ (ZMod p) *ᵥ v = lam • v) ↔ IsSquare (2 : ZMod p) := by sorrySource