The binary building block of the upper inner curve #
Predictor reflection of the source binary model; both directional coefficients are unchanged.
Equations
Instances For
theorem
Papers.AnsariRockel2026RhoFootrule.upperBinary_conditional_distribution
(a v : ↑unitInterval)
:
(fun (u : ↑unitInterval) => (upperBinary a).conditionalCDF u v) =ᵐ[MeasureTheory.volume] fun (u : ↑unitInterval) =>
if u ≤ ProbabilityTheory.Copula.unitHalf then ↑v + Verification.flatTent a v else ↑v - Verification.flatTent a v
The two conditional distribution functions are v plus or minus the flat-topped tent.
theorem
Papers.AnsariRockel2026RhoFootrule.upperBinary_conditionalMean
(a : ↑unitInterval)
(ha : ↑a ≤ 1 / 2)
:
Verification.conditionalMean (upperBinary a) =ᵐ[MeasureTheory.volume] fun (u : ↑unitInterval) =>
if u ≤ ProbabilityTheory.Copula.unitHalf then 1 / 2 - ↑a * (1 - ↑a) else 1 / 2 + ↑a * (1 - ↑a)
The exact conditional means of the two halves.
theorem
Papers.AnsariRockel2026RhoFootrule.upperBinary_coefficients
(a : ↑unitInterval)
(ha : ↑a ≤ 1 / 2)
:
(upperBinary a).chatterjeeXi = 2 * ↑a ^ 2 * (3 - 4 * ↑a) ∧ copulaCorrelationRatio (upperBinary a) = 12 * ↑a ^ 2 * (1 - ↑a) ^ 2
The n=1 branch in equations (45) and (120).
theorem
Papers.AnsariRockel2026RhoFootrule.upper_binary_attained
(a : ℝ)
(ha : a ∈ Set.Icc 0 (1 / 2))
:
Each parameter on the first upper branch is realized by an actual copula.