Negation of the Gelbrich-hull containment (false as formalized)
ProvedWassersteinDRO.Gelbrich.gelbrich_hull_containment_falseNegation of WassersteinDRO.Gelbrich.gelbrich_hull_containment (d5fc0224), which is false as formalized.
The formalized containment claims that every nominal distribution PN with matching mean/covariance has its Wasserstein ambiguity set inside the Gelbrich hull. But the nominal PN is not assumed to have finite second moments, and covarianceMatrix is a Bochner integral (junk value 0 on non-integrable entries), so it can be non-positive-semidefinite — which the mean-covariance uncertainty set explicitly forbids via a PosSemidef conjunct.
Counterexample: m = 2, Ξ = univ, ε = 1, p = 2, μhat = 0, PN = law of (sign X, X) for symmetric X with density proportional to 1/(1+|x|^3) (so E|X| < ∞ < E[X²]). Then meanVector PN = (0,0) and covarianceMatrix PN = !![1, c; c, 0] with c = E|X| > 0, det = -c^2 < 0, hence not PSD. PN ∈ ambiguitySet (diagonal coupling gives W₂(PN,PN) = 0 ≤ 1) but PN ∉ gelbrichHull. The inclusion fails. The paper's Theorem 13 assumes finite second moments; the formalization does not.
import Mathlib import Definitions.Def_WassersteinDRO_Gelbrich_ambiguitySet import Definitions.Def_WassersteinDRO_Gelbrich_gelbrichHull set_option autoImplicit false open MeasureTheory open scoped ENNReal
namespace WassersteinDRO.Gelbrich
/-- The Gelbrich-hull containment (Theorem 13, Kuhn et al. 2019) is FALSE as
formalized: the nominal distribution `PN` is not assumed to have finite
second moments, so `covarianceMatrix PN` (a Bochner integral, junk value `0`
on non-integrable entries) can fail to be positive semidefinite — which the
mean-covariance uncertainty set explicitly requires. Concretely, with
`m = 2`, `Ξ = univ`, `ε = 1`, `p = 2`, `μhat = 0`, the law `PN` of
`(sign X, X)` for symmetric `X` with `E|X| < ∞ < E[X²]` satisfies
`PN ∈ ambiguitySet 1 2 univ PN` (the diagonal coupling gives
`W₂(PN,PN) = 0 ≤ 1`) but `PN ∉ gelbrichHull 1 univ 0 (covarianceMatrix PN)`
because `covarianceMatrix PN = !![1, E|X|; E|X|, 0]` has negative
determinant, hence is not positive semidefinite. So the inclusion fails. -/
theorem gelbrich_hull_containment_false :
¬∀ {m : ℕ} (ε p : ℝ), 2 ≤ p → (Ξ : Set (EuclideanSpace ℝ (Fin m))) →
(PN : Measure (EuclideanSpace ℝ (Fin m))) →
(μhat : EuclideanSpace ℝ (Fin m)) → (SigmaHat : Matrix (Fin m) (Fin m) ℝ) →
meanVector PN = μhat → covarianceMatrix PN = SigmaHat →
ambiguitySet ε p Ξ PN ⊆ gelbrichHull ε Ξ μhat SigmaHat := by sorry
end WassersteinDRO.Gelbrich