Developable gauge iff
ProvedImpossibleFigures.NonAbelian.developable_gauge_iffaether-cataloggeometry
Formal statement of ImpossibleFigures.NonAbelian.developable_gauge_iff from the Aether Catalog (Geometry). The mathematical content is given by the Lean statement below; a human-readable write-up is pending.
theorem ImpossibleFigures.NonAbelian.developable_gauge_iff{s t : E → V} (H : V → G) (ω : E → G) (base : V)
(hconn : ∀ v : V, ∃ l, IsWalk s t base v l) :
Developable s t (gauge s t H ω) ↔ Developable s t ω := by sorry
Formalization Note Transplanted verbatim from the Aether Catalog source Geometry/NonAbelianHolonomy.lean; the statement is byte-identical to the source declaration, elaborated with autoImplicit disabled in the platform environment.
Preamble
-- Thm stub generated from Geometry/NonAbelianHolonomy.lean
import Mathlib
import Definitions.Def_Geometry_NonAbelianHolonomy
/-!
# Impossible Figures VI: non-abelian holonomy and rotational developability
The previous files in this thread
(`Geometry.CellularDevelopability`, `Geometry.CycleCertificates`,
`Geometry.GridCertificates`, `Geometry.TwistedDevelopability`) classify *additive*
increment fields: an increment field `ω : E → A` with values in an abelian group is
developable iff its cellular curvature vanishes and its periods vanish on a
generating family of one-cycles.
Direction 5 of the previous cycle observed that additive height integrability is only
the "shadow" of a genuinely geometric realisation problem: for a piecewise linear
developable realisation in three-space the increments are *rotations*, and they
compose non-commutatively. This file carries out the first testable step listed
there: the **non-abelian Poincaré lemma**.
## Setup
* A combinatorial one-skeleton is a pair of maps `s t : E → V` (source, target).
* A *step* is a pair `(b, e) : Bool × E`: the edge `e` traversed forwards (`b = true`)
or backwards (`b = false`).
* A *walk* from `a` to `b` is a list of steps, chained by `IsWalk`.
* An increment field with values in a (possibly non-abelian) group `G` is a map
`ω : E → G`; its **holonomy** `hol ω l` along a walk is the ordered product of the
steps, later steps multiplying on the left.
* `Developable s t ω` means `ω` is a coboundary: `ω e = H (t e) * (H (s e))⁻¹`.
## Main results
* `hol_of_coboundary` — discrete Stokes: the holonomy of a coboundary along a walk
from `a` to `b` is `H b * (H a)⁻¹`; in particular closed walks have trivial
holonomy (`hol_eq_one_of_developable`).
* `developable_iff_holonomy_trivial` — the non-abelian Poincaré lemma: on a connected
one-skeleton, `ω` is developable **iff** every closed walk based at a fixed vertex
has trivial holonomy.
* `hol_gauge` / `holonomy_conj_of_gauge` — the holonomy of a gauge-transformed field
is conjugate to the original; so the *conjugacy class* of the holonomy of a closed
walk is a gauge invariant, the non-abelian replacement for `period_gauge_invariant`.
* `hol_map` — a group homomorphism `φ : G →* G'` transports holonomy; hence an
additive obstruction (a nonzero period after abelianisation) is a special case of a
non-abelian one (`not_developable_of_map_holonomy_ne_one`).
* `triangle_developable_iff` and `penrose_rotational_not_developable`: on the
three-cycle (the underlying loop of the Penrose triangle) developability is
equivalent to the triviality of the single product `ω 2 * ω 1 * ω 0`, and the field
which turns each beam by one and the same transposition of `Equiv.Perm (Fin 3)`
is not developable — an impossible figure whose obstruction is purely rotational
and invisible to any abelian period.
-/
open ImpossibleFigures.NonAbelian
variable {V E G G' : Type*} [Group G] [Group G']
/-! ## Steps and walks -/
/-! ## Developability and discrete Stokes -/
/-! ## Gauge invariance -/Formal statement
theorem ImpossibleFigures.NonAbelian.developable_gauge_iff{s t : E → V} (H : V → G) (ω : E → G) (base : V)
(hconn : ∀ v : V, ∃ l, IsWalk s t base v l) :
Developable s t (gauge s t H ω) ↔ Developable s t ω := by sorrySource