Sampling a diagonal band from two independent uniforms #
noncomputable def
ProbabilityTheory.Copula.RankRegion.XiBlest.Support.bandSample
(b : ℝ)
(p : ↑unitInterval × ↑unitInterval)
:
Fin 2 → ↑unitInterval
The second rank is the CDF transform of b*U+Z.
Equations
Instances For
theorem
ProbabilityTheory.Copula.RankRegion.XiBlest.Support.bandSample_intercept_mem
{b : ℝ}
(hb : 0 ≤ b)
(u z : ↑unitInterval)
:
theorem
ProbabilityTheory.Copula.RankRegion.XiBlest.Support.bandSample_orthant
(b : ℝ)
(hb : 0 ≤ b)
(u v : ↑unitInterval)
:
(∫ (p : ↑unitInterval × ↑unitInterval), if bandSample b p ≤ ![u, v] then 1 else 0) = (diagonalBand b hb).cdf ![u, v]
The complete lower-orthant distribution of the sampled vector.
theorem
ProbabilityTheory.Copula.RankRegion.XiBlest.Support.diagonalBand_toMeasure_sample
(b : ℝ)
(hb : 0 ≤ b)
:
The constructed copula is exactly the law of the sampled vector.
theorem
ProbabilityTheory.Copula.RankRegion.XiBlest.Support.diagonalBand_integral_sample
(b : ℝ)
(hb : 0 ≤ b)
{f : (Fin 2 → ↑unitInterval) → ℝ}
(hf : Continuous f)
:
∫ (x : Fin 2 → ↑unitInterval), f x ∂(diagonalBand b hb).toMeasure = ∫ (u : ↑unitInterval) (z : ↑unitInterval), f (bandSample b (u, z))
Integrals under the band can be evaluated on two independent uniforms.
theorem
ProbabilityTheory.Copula.RankRegion.XiBlest.Support.diagonalBand_cdf_sample
(b : ℝ)
(hb : 0 ≤ b)
(u z : ↑unitInterval)
:
The CDF at the sampled point has a simple affine clamped integrand.