Complete future null infinity (Christodoulou) and genericity
Definitionwcc_complete_null_infinityThis file defines complete future null infinity in Christodoulou's sense and the notion of generic initial data used by the goal.
sphereNormal3 h x is the outward -unit normal to the Euclidean sphere in (the Euclidean radial covector raised by and normalized); devNormal is the future unit normal to the initial surface of a development. Pushed forward by , these give the null tangents (outgoing) and (incoming) of the source.
HasCompleteFutureNullInfinity h k D asserts: there is a radius such that every outgoing null geodesic from with tangent is defined on all of (the boundary has future-complete generators), and for every there is such that every incoming null geodesic from with tangent meets at some parameter and extends as a null geodesic to parameter (affine length at least beyond the meeting). This is Christodoulou's definition (CQG 16 (1999) A26–A27), with Euclidean balls as the cofinal family of domains.
CompletesNullInfinity h k says every maximal development of has complete future null infinity; GoodData is the set of asymptotically flat data with this property; dataDist is the weighted distance; GenericOpenDense S says is open and dense in the asymptotically flat data for dataDist.
Formalization Note. The affine origin is taken at (parameter ) rather than at ; affine length is translation-invariant, so the condition "extends to " is exactly "affine length from the meeting". dataDist uses ENNReal.toReal, which is the genuine value on asymptotically flat pairs (finite weighted size).
import Mathlib
import Definitions.Def_scc_coordinate_framework
import Definitions.Def_wcc_initial_data
import Definitions.Def_wcc_spacetime
open scoped Matrix
open Set StrongCosmicCensorship
namespace WeakCosmicCensorship
noncomputable def sphereNormal3 (h : Field3) (x : Coords3) : Coords3 :=
let v : Coords3 := fun a => x a / ‖x‖
let nv : Coords3 := fun a => ∑ b : Fin 3, ginv3 h x a b * v b
let norm := Real.sqrt (∑ a : Fin 3, ∑ b : Fin 3, ginv3 h x a b * v a * v b)
fun a => nv a / norm
noncomputable def devNormal (h k : Field3) (D : Development h k) : Coords3 → Coords :=
Classical.choose D.induced
noncomputable def HasCompleteFutureNullInfinity (h k : Field3) (D : Development h k) : Prop :=
let n := devNormal h k D
∃ R₀ : ℝ, 0 < R₀ ∧
(∀ x : Coords3, ‖x‖ = R₀ → ∃ γ : ℝ → Coords,
IsNullGeodesicOn D.U D.g (Ici 0) γ ∧ γ 0 = D.ι x ∧
vel γ 0 = (fun A => n x A + (jac34 D.ι x).mulVec (sphereNormal3 h x) A)) ∧
(∀ A : ℝ, 0 < A → ∃ R : ℝ, R₀ ≤ R ∧
∀ x : Coords3, ‖x‖ = R →
∃ γ : ℝ → Coords, ∃ τc : ℝ, 0 ≤ τc ∧
IsNullGeodesicOn D.U D.g (Icc 0 (τc + A)) γ ∧
γ 0 = D.ι x ∧
vel γ 0 = (fun A => n x A - (jac34 D.ι x).mulVec (sphereNormal3 h x) A) ∧
γ τc ∈ frontier (causalFuture D.U D.g D.T (D.ι '' {y : Coords3 | ‖y‖ ≤ R₀})))
noncomputable def dataDist (p q : Field3 × Field3) : ℝ :=
(weightedSize3 3 1 (fun x => p.1 x - q.1 x)).toReal +
(weightedSize3 3 2 (fun x => p.2 x - q.2 x)).toReal
def CompletesNullInfinity (h k : Field3) : Prop :=
∀ D : Development h k, IsMaximalDevelopment h k D → HasCompleteFutureNullInfinity h k D
def GoodData : Set (Field3 × Field3) :=
{p | IsAFInitialData p.1 p.2 ∧ CompletesNullInfinity p.1 p.2}
def GenericOpenDense (S : Set (Field3 × Field3)) : Prop :=
(∀ p ∈ S, ∃ ε : ℝ, 0 < ε ∧
∀ q : Field3 × Field3, IsAFInitialData q.1 q.2 → dataDist p q < ε → q ∈ S) ∧
(∀ p : Field3 × Field3, IsAFInitialData p.1 p.2 →
∀ ε : ℝ, 0 < ε → ∃ q ∈ S, dataDist p q < ε)
end WeakCosmicCensorship
Read-back
What the Lean code literally says, in plain math · claude (drafting agent — NON-INDEPENDENT blind pass; not a separate auditor)
NON-INDEPENDENT READ-BACK. Written by the same agent that drafted the Lean, as a literal rendering of the code only. It is NOT the independent sub-agent testimony required by the skill; an independent auditor (references/mission_auditor.md) should re-read the code before submission.
Literal content of wcc_complete_null_infinity.
sphereNormal3 h x : with v a = x a / ‖x‖, nv a = Σ_b ginv3 h x a b · v b, norm = sqrt(Σ_{a,b} ginv3 h x a b · v a · v b), returns nv a / norm (junk at x=0). devNormal h k D = Classical.choose D.induced (a chosen unit normal).
HasCompleteFutureNullInfinity h k D (with n = devNormal h k D) asserts ∃ R₀>0 such that:
(1) for every x with ‖x‖=R₀ there is γ a null geodesic on [0,∞) in D.U, γ 0 = D.ι x, vel γ 0 = n x + jac34 D.ι x · sphereNormal3 h x;
(2) for every A>0 there is R≥R₀ such that for every x with ‖x‖=R there exist γ and τc≥0 with γ a null geodesic on [0, τc+A], γ 0 = D.ι x, vel γ 0 = n x − jac34 D.ι x · sphereNormal3 h x, and γ τc ∈ frontier(causalFuture D.U D.g D.T (D.ι '' {y | ‖y‖≤R₀})).
dataDist p q = (weightedSize3 3 1 (p.1−q.1)).toReal + (weightedSize3 3 2 (p.2−q.2)).toReal. CompletesNullInfinity h k = for every D : Development h k, IsMaximalDevelopment h k D → HasCompleteFutureNullInfinity h k D. GoodData = {p | IsAFInitialData p.1 p.2 ∧ CompletesNullInfinity p.1 p.2}. GenericOpenDense S = (open) for every p∈S there is ε>0 such that every AF q with dataDist p q < ε lies in S; and (dense) for every AF p and ε>0 there is q∈S with dataDist p q < ε.