Normed spaces isometric to their duals
DefinitionMazur's Problem 91 (Scottish Book), Banach-space form#✓fully proved
This is the proposition that every finite-dimensional real normed space \(E\) that is linearly isometric to its dual space \(E^*\) (the continuous linear functionals on \(E\) with the operator norm) is an inner product space, that is, its norm is induced by some inner product on \(E\).
Mazur's original formulation asks whether, for every \(n\), a convex body in \(\mathbb{R}^n\) that is affinely equivalent to its conjugate (polar) body is necessarily an ellipsoid.
def ScottishBookProblem91Claim : Prop :=
∀ (n : ℕ) (E : Type*) [NormedAddCommGroup E] [NormedSpace ℝ E]
[FiniteDimensional ℝ E],
Module.finrank ℝ E = n →
(∃ _ : E ≃ₗᵢ[ℝ] StrongDual ℝ E, True) →
∃ _ : InnerProductSpace ℝ E, True
Used by 1
TheoremMazur's Problem 91 has a negative answer#●ready to prove
There is a finite-dimensional real normed space \(E\) that is linearly isometric to its dual space \(E^*\) (the continuous linear functionals on \(E\) with the operator norm) but whose norm is not induced by any inner product on \(E\). Thus the Banach-space form of Mazur's Problem 91 is false.
theorem_wanted mazur91_refuted : ¬ ScottishBookProblem91Claim
def ScottishBookProblem91Claim : Prop :=
∀ (n : ℕ) (E : Type*) [NormedAddCommGroup E] [NormedSpace ℝ E]
[FiniteDimensional ℝ E],
Module.finrank ℝ E = n →
(∃ _ : E ≃ₗᵢ[ℝ] StrongDual ℝ E, True) →
∃ _ : InnerProductSpace ℝ E, True
WantedExt/Analysis/FunctionalAnalysis/ScottishBook91Wanted.lean:27
Sources
- S. Mazur, Problem 91 of the Scottish Book. The typescript itself records the negative answer when
nis even (for example,ℓ¹ 2, whose unit ball is a diamond, is linearly isometric to its dualℓ^∞ 2but is not a Hilbert space, since its norm does not satisfy the parallelogram law); the odd-dimensional case is not needed to refute the universal claim.
Uses 1
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.