Conjecture 2 in rank one.
ProvedDiophantineLattice.rank_one_multiplicity_even_iffaether-catalognovelty
Conjecture 2 in rank one. For the standard form on ℤ, all coefficients of the
shifted theta series of x ↦ (x - t)² are even iff 2t is an integer and t is not.
theorem DiophantineLattice.rank_one_multiplicity_even_iff(t : Fin 1 → ℚ) :
(∀ (c : ℚ) (S : Finset (Fin 1 → ℤ)),
(∀ m : Fin 1 → ℤ, m ∈ S ↔ form (1 : Matrix (Fin 1) (Fin 1) ℚ)
(fun i => t i - emb m i) = c) → Even S.card)
↔ ((∃ v : Fin 1 → ℤ, ∀ i, (2 : ℚ) * t i = (v i : ℚ)) ∧ ∀ k : Fin 1 → ℤ, t ≠ emb k) := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Novelty/DiophantineLatticeMultiplicity.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from Novelty/DiophantineLatticeMultiplicity.lean
import Mathlib
import Definitions.Def_Novelty_DiophantineLatticeExactOrder
import Definitions.Def_Novelty_DiophantineLatticeSpectralGap
/-!
# Cycle 7: the parity criterion for shifted theta coefficients
Cycle 2 proved that all coefficients of the shifted theta series of `x ↦ Q(x - v/2)` are even
when `v` is a *shortest* vector (`halfPt_multiplicity_even`). Conjecture 2 of
`FUTURE_DIRECTIONS.md` asserted that evenness is in fact an exact criterion:
`(∀ c, r_t(c) is even) ⟺ 2t ∈ L and t ∉ L`.
This file proves the `⟸` direction **in full generality** — for an arbitrary rational
positive-semidefinite-free setting, in fact for an arbitrary rational matrix `B`, with no
minimality, positivity or symmetry hypothesis at all — and settles the `⟹` direction in
rank one, where the criterion becomes a theorem.
* `halfPt_multiplicity_even_of_primitive` : the hypothesis "`v` is shortest" in
`halfPt_multiplicity_even` is superfluous; all that the antipodal involution `m ↦ v - m`
needs is `v ∉ 2L`.
* `two_torsion_multiplicity_even` : consequently every coefficient is even whenever `2t ∈ L`
and `t ∉ L`. This closes the `⟸` half of Conjecture 2.
* `rank_one_multiplicity_odd_of_not_half` : in rank one, if `2t ∉ ℤ` then the coefficient at
`c = t²` equals `1`, so it is odd.
* `rank_one_multiplicity_even_iff` : **Conjecture 2 in rank one**, an exact `iff`.
-- !-- Lab Notes -- !--
Hypothesis (Hypothesizer): evenness of the shifted theta coefficients is equivalent to the
existence of the antipodal symmetry, i.e. to `2t ∈ L, t ∉ L`; no other mechanism can force it.
Experiment (Experimenter): stripping `hv : Q(v) = λ₁` from `halfPt_multiplicity_even` leaves a
proof that uses only `v - 2m ≠ 0`, so the `⟸` direction holds for every `2`-torsion shift.
For `⟹`, the rank-one computation is decisive: `(t - m)² = (t - m')²` with `m ≠ m'` forces
`m + m' = 2t`, so `2t ∉ ℤ` makes *every* coefficient at most `1`, and the coefficient at
`c = t²` is exactly `1`.
Analysis (Analyst): the obstruction to a rank-`n` proof of `⟹` is that a coefficient can be
even "by accident" (two unrelated pairs of lattice points at the same distance); the rank-one
argument works because the fibre of `x ↦ Q(x)` over `c` has at most two points. So the
general `⟹` is "true but hard", needing a global count rather than a fibrewise one.
Critique (Critic): `rank_one_multiplicity_even_iff` is an honest `iff` with no side condition,
and its two halves are proved by genuinely different arguments (involution vs. fibre count);
`two_torsion_multiplicity_even` is strictly stronger than the cycle-2 theorem, which it
re-derives (`halfPt_multiplicity_even_reproved`).
Synthesis (PI): the parity of the shifted theta series is an exact `2`-torsion criterion in
rank one, and the sufficiency half holds in every rank.
-/
open DiophantineLattice
open Finset
variable {n : ℕ}
/-! ## Evenness without minimality -/
/-! ## Rank one: the criterion is exact -/Formal statement
theorem DiophantineLattice.rank_one_multiplicity_even_iff(t : Fin 1 → ℚ) :
(∀ (c : ℚ) (S : Finset (Fin 1 → ℤ)),
(∀ m : Fin 1 → ℤ, m ∈ S ↔ form (1 : Matrix (Fin 1) (Fin 1) ℚ)
(fun i => t i - emb m i) = c) → Even S.card)
↔ ((∃ v : Fin 1 → ℤ, ∀ i, (2 : ℚ) * t i = (v i : ℚ)) ∧ ∀ k : Fin 1 → ℤ, t ≠ emb k) := by sorrySource