A metric space is plastic if every nonexpansive bijection of it onto itself is an isometry; every totally bounded metric space is plastic (Freudenthal–Hurewicz 1936), and Cascales, Kadets, Orihuela and Wingler (2016) asked whether the closed unit ball of every real Banach space is plastic. They proved it for strictly convex spaces, whose unit sphere consists of one-point faces. Angosto, Kadets and Zavarzina (2019) proved it, in the stronger two-space form, when the unit sphere is a union of finite-dimensional polyhedral extreme subsets: then every nonexpansive bijection from the unit ball of any Banach space onto that ball is an isometry. This mission formalizes a theorem that removes the polyhedrality: it suffices that every point of the sphere lies in some finite-dimensional face, of any shape, for instance a round disc. The results come from an unpublished research project on the plasticity problem.
and are real normed spaces with closed unit balls and unit spheres ; the covering and plasticity forms assume them complete. A sphere face of is a nonempty convex set that is an extreme subset of : if a point of lies in the open segment between two points of , both points lie in . Its dimension is that of its affine hull; and are its relative boundary and its extreme points. is a nonexpansive bijection; nothing is assumed about the continuity of .
The result. For a polytope face, the relative boundary is a finite union of smaller polytopes, so an induction on dimension keeps the preimage of the boundary compact; this is the step of Angosto–Kadets–Zavarzina that needs polyhedrality. For a curved face the boundary has a continuum of extreme points, and the inverse map, not known to be continuous, could a priori scatter their preimages. The new step traps the whole boundary preimage in one compact face: the preimage of a relative-interior point is shown to lie strictly between any boundary preimage and another one, by following a path across the scaled boundary, which the induction hypothesis covers exactly. Spaces such as -sums of a Euclidean space with a three-dimensional space whose ball is a cylinder have only finite-dimensional faces, some of them discs, and are covered by the new theorem but not by the polyhedral one or by strict convexity.
Formalizing it. The proofs exist only in the project's notes and were checked by AI referees; a machine-checked proof is the independent confirmation sought.
The covering form ends with the project's theorem that a nonexpansive surjection between unit balls mapping the sphere into the sphere is linear, already formalized (card Plasticity.U10); this is the only step that uses completeness.
No regularity of the inverse map, no strict convexity, polyhedrality or separability is assumed, the face theorem does not assume completeness, and the faces may have any finite dimension. The topology needed is classical but absent from Mathlib: invariance of domain and of dimension, and the impossibility of retracting a ball onto its boundary sphere.
Real scalars. Faces are expressed with Mathlib's IsExtreme, dimensions as Module.finrank ℝ (vectorSpan ℝ C), relative boundaries as intrinsicFrontier ℝ C (definitions Plasticity_Faces).
A ball map is a function F : X → Y constrained on the closed unit ball by Set.BijOn and LipschitzOnWith 1; the preimage of is closedBall 0 1 ∩ F ⁻¹' C.
The plasticity form uses HasPlasticUnitBall (definitions Plasticity_Basic).
Statements are not to be weakened by assuming continuity of the inverse, polyhedrality or a bound on the face dimensions.
Published inputs (the basic facts of Zavarzina's Theorem 2.3, invariance of dimension, no retraction) are separate cards.
namespace Plasticity.Faces
theorem exists_linearIsometryEquiv_of_sphere_covered {X Y : Type*} [NormedAddCommGroup X] [NormedSpace ℝ X] [CompleteSpace X]
[NormedAddCommGroup Y] [NormedSpace ℝ Y] [CompleteSpace Y]
(F : X → Y)
(hbij : Set.BijOn F (Metric.closedBall (0 : X) 1) (Metric.closedBall (0 : Y) 1))
(hlip : LipschitzOnWith 1 F (Metric.closedBall (0 : X) 1))
(hcover : ∀ y ∈ Metric.sphere (0 : Y) 1, ∃ C : Set Y,
IsSphereFace C ∧ FiniteDimensional ℝ (vectorSpan ℝ C) ∧ y ∈ C) :
∃ T : X ≃ₗᵢ[ℝ] Y, Set.EqOn F T (Metric.closedBall (0 : X) 1) := by sorry
end Plasticity.FacesLet be real Banach spaces such that every point of the unit sphere of lies in a finite-dimensional face of contained in the sphere. Then every nonexpansive bijection is the restriction of a surjective linear isometry . This removes the polyhedrality hypothesis from Theorem 4.11 of Angosto, Kadets and Zavarzina, where the sphere is a union of finite-dimensional polyhedral extreme subsets.
No open leaves. Every sub-goal is proved or awaiting decomposition.