Prescribing the images of the unit vectors of in by a finite modification of a fixed injective map
ProvedPlasticity.Ell1C0.exists_clamp_familyLet be a set, , the unit vectors of and . There is a family of maps , indexed by label families in , such that for all : (1) is nonexpansive; (2) ; (3) depends only on the labels with ; (4) if no exceeds then ; (5) if is not a unit vector , then and differ at only finitely many coordinates.
Preamble
import Definitions.Def_Plasticity_SeqSpace
Formal statement
namespace Plasticity
theorem Ell1C0.exists_clamp_family (J : Type*) [DecidableEq J] :
∃ G : (J → Metric.closedBall (0 : c0 (J × ℕ)) 1) →
Metric.closedBall (0 : lp (fun _ : J => ℝ) 1) 1 → Metric.closedBall (0 : c0 (J × ℕ)) 1,
(∀ y, LipschitzWith 1 (G y)) ∧
(∀ y j (x : Metric.closedBall (0 : lp (fun _ : J => ℝ) 1) 1),
(x : lp (fun _ : J => ℝ) 1) = lp.single 1 j 1 → G y x = y j) ∧
(∀ y y' (x : Metric.closedBall (0 : lp (fun _ : J => ℝ) 1) 1),
(∀ j, ‖(x : lp (fun _ : J => ℝ) 1)‖ / 4 < (x : lp (fun _ : J => ℝ) 1) j → y j = y' j) →
G y x = G y' x) ∧
(∀ y (x : Metric.closedBall (0 : lp (fun _ : J => ℝ) 1) 1),
(∀ j, (x : lp (fun _ : J => ℝ) 1) j ≤ ‖(x : lp (fun _ : J => ℝ) 1)‖ / 4) →
∀ i : J × ℕ, ((G y x : c0 (J × ℕ)) : Linf (J × ℕ)) i
= (2 : ℝ) ^ (-((i.2 : ℤ) + 2)) * (x : lp (fun _ : J => ℝ) 1) i.1) ∧
(∀ y (x : Metric.closedBall (0 : lp (fun _ : J => ℝ) 1) 1),
(∀ j, (x : lp (fun _ : J => ℝ) 1) ≠ lp.single 1 j 1) →
{i : J × ℕ | ((G y x : c0 (J × ℕ)) : Linf (J × ℕ)) i
≠ (2 : ℝ) ^ (-((i.2 : ℤ) + 2)) * (x : lp (fun _ : J => ℝ) 1) i.1}.Finite) := by sorry
end PlasticitySource
Proof draft 'Three constructions of nonplastic Banach unit balls' (ChatGPT Pro, OpenAI, 18 November 2026), Lemma 2.1 and Construction A (prescribing images, finite dependence)
Read-back
What the Lean code literally says, in plain math · claude-opus-5-5
Data: an arbitrary type (any universe) with decidable equality.
Spaces and balls:
- is the space of real functions on with ; denotes its closed unit ball , with the metric.
- is the subspace of of functions tending to along the cofinite filter (for every only finitely many coordinates have absolute value ), with the supremum norm; denotes its closed unit ball, with the supremum metric.
- is the unit vector with at and elsewhere.
- For let be the function on given by .
Conclusion: there is a map which assigns to every family of points a map , with the following five properties.
- For every , is -Lipschitz: for all .
- For every and every : .
- (Locality in .) For all families and every : if for every with , then . The comparison uses the signed coordinate , not .
- (Linear part.) For every and every with for all (again signed coordinates), for all , that is, .
- (Finite deviation.) For every and every that is not equal to any , the set is finite.
Edge cases:
- If is empty, , , there is exactly one family , and the statement reduces to ; it holds trivially.
- At property 4 applies and forces for every .
- For (all signed coordinates ) property 4 forces , independently of .
- Property 2 and property 4 never apply to the same point, since has a coordinate .
- The statement is purely existential with no hypotheses beyond decidable equality on , so there is no vacuity; its content is the simultaneous compatibility of the Lipschitz bound with properties 2 to 5 for arbitrary families .