The converse ordinal-sum theorem #
At each interior diagonal fixed point a copula is the ordinal sum of its two rescaled restrictions. The pair of components is unique for that fixed split. This is the binary form of Nelsen's Theorem 3.2.1; the split itself need not be unique.
theorem
ProbabilityTheory.Copula.ordinalSum_components
(C : Copula 2)
(a : ↑unitInterval)
(ha0 : 0 < a)
(ha1 : a < 1)
(ha : C.diagonal a = ↑a)
:
@[simp]
theorem
ProbabilityTheory.Copula.lowerOrdinalComponent_ordinalSum
(C D : Copula 2)
(a : ↑unitInterval)
(ha0 : 0 < a)
:
@[simp]
theorem
ProbabilityTheory.Copula.upperOrdinalComponent_ordinalSum
(C D : Copula 2)
(a : ↑unitInterval)
(ha1 : a < 1)
:
@[simp]
@[simp]
theorem
ProbabilityTheory.Copula.diagonal_eq_iff_exists_ordinalSum
(C : Copula 2)
(a : ↑unitInterval)
(ha0 : 0 < a)
(ha1 : a < 1)
:
A prescribed interior point is an ordinal-sum split exactly when the diagonal is fixed there.
The binary ordinal-sum characterization, Nelsen, second edition, Theorem 3.2.1.
theorem
ProbabilityTheory.Copula.IsNQD.not_exists_ordinalSum
{C : Copula 2}
(h : C.IsNQD)
:
¬∃ (a : ↑unitInterval) (D : Copula 2) (E : Copula 2), 0 < a ∧ a < 1 ∧ D.ordinalSum E a = C