mathlib3 documentation

probability.kernel.cond_cdf

Conditional cumulative distribution function #

THIS FILE IS SYNCHRONIZED WITH MATHLIB4. Any changes to this file require a corresponding PR to mathlib4.

Given ρ : measure (α × ℝ), we define the conditional cumulative distribution function (conditional cdf) of ρ. It is a function cond_cdf ρ : α → ℝ → ℝ such that if ρ is a finite measure, then for all a : α cond_cdf ρ a is monotone and right-continuous with limit 0 at -∞ and limit 1 at +∞, and such that for all x : ℝ, a ↦ cond_cdf ρ a x is measurable. For all x : ℝ and measurable set s, that function satisfies ∫⁻ a in s, ennreal.of_real (cond_cdf ρ a x) ∂ρ.fst = ρ (s ×ˢ Iic x).

Main definitions #

Main statements #

References #

The construction of the conditional cdf in this file follows the proof of Theorem 3.4 in O. Kallenberg, Foundations of modern probability.

TODO #

theorem directed.sequence_anti {α : Type u_1} {β : Type u_2} [encodable α] [inhabited α] [preorder β] {f : α β} (hf : directed ge f) :
theorem directed.sequence_le {α : Type u_1} {β : Type u_2} [encodable α] [inhabited α] [preorder β] {f : α β} (hf : directed ge f) (a : α) :
theorem prod_Inter {α : Type u_1} {β : Type u_2} {ι : Type u_3} {s : set α} {t : ι set β} [hι : nonempty ι] :
(s ×ˢ (i : ι), t i) = (i : ι), s ×ˢ t i
theorem real.Inter_Iic_rat  :
( (r : ), set.Iic r) =
theorem ennreal.of_real_cinfi {α : Type u_1} (f : α ) [nonempty α] :
ennreal.of_real ( (i : α), f i) = (i : α), ennreal.of_real (f i)
theorem lintegral_infi_directed_of_measurable {α : Type u_1} {β : Type u_2} {mα : measurable_space α} [countable β] {f : β α ennreal} {μ : measure_theory.measure α} (hμ : μ 0) (hf : (b : β), measurable (f b)) (hf_int : (b : β), ∫⁻ (a : α), f b a μ ) (h_directed : directed ge f) :
∫⁻ (a : α), ( (b : β), f b a) μ = (b : β), ∫⁻ (a : α), f b a μ

Monotone convergence for an infimum over a directed family and indexed by a countable type

noncomputable def measure_theory.measure.Iic_snd {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (r : ) :

Measure on α such that for a measurable set s, ρ.Iic_snd r s = ρ (s ×ˢ Iic r).

Equations
theorem measure_theory.measure.Iic_snd_apply {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (r : ) {s : set α} (hs : measurable_set s) :
(ρ.Iic_snd r) s = ρ (s ×ˢ set.Iic r)
theorem measure_theory.measure.Iic_snd_mono {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) {r r' : } (h_le : r r') :
ρ.Iic_snd r ρ.Iic_snd r'
theorem measure_theory.measure.Iic_snd_le_fst {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (r : ) :
ρ.Iic_snd r ρ.fst
theorem measure_theory.measure.infi_Iic_snd_gt {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (t : ) {s : set α} (hs : measurable_set s) [measure_theory.is_finite_measure ρ] :
( (r : {r' // t < r'}), (ρ.Iic_snd r) s) = (ρ.Iic_snd t) s

Auxiliary definitions #

We build towards the definition of probability_theory.cond_cdf. We first define probability_theory.pre_cdf, a function defined on α × ℚ with the properties of a cdf almost everywhere. We then introduce probability_theory.cond_cdf_rat, a function on α × ℚ which has the properties of a cdf for all a : α. We finally extend to .

noncomputable def probability_theory.pre_cdf {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (r : ) :

pre_cdf is the Radon-Nikodym derivative of ρ.Iic_snd with respect to ρ.fst at each r : ℚ. This function ℚ → α → ℝ≥0∞ is such that for almost all a : α, the function ℚ → ℝ≥0∞ satisfies the properties of a cdf (monotone with limit 0 at -∞ and 1 at +∞, right-continuous).

We define this function on and not because is countable, which allows us to prove properties of the form ∀ᵐ a ∂ρ.fst, ∀ q, P (pre_cdf q a), instead of the weaker ∀ q, ∀ᵐ a ∂ρ.fst, P (pre_cdf q a).

Equations
structure probability_theory.has_cond_cdf {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (a : α) :
Prop

A product measure on α × ℝ is said to have a conditional cdf at a : α if pre_cdf is monotone with limit 0 at -∞ and 1 at +∞, and is right continuous. This property holds almost everywhere (see has_cond_cdf_ae).

A measurable set of elements of α such that ρ has a conditional cdf at all a ∈ cond_cdf_set.

Equations
noncomputable def probability_theory.cond_cdf_rat {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) :

Conditional cdf of the measure given the value on α, restricted to the rationals. It is defined to be pre_cdf if a ∈ cond_cdf_set, and a default cdf-like function otherwise. This is an auxiliary definition used to define cond_cdf.

Equations
@[irreducible]
noncomputable def probability_theory.cond_cdf' {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) :

Conditional cdf of the measure given the value on α, as a plain function. This is an auxiliary definition used to define cond_cdf.

Equations

Conditional cdf #

noncomputable def probability_theory.cond_cdf {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (a : α) :

Conditional cdf of the measure given the value on α, as a Stieltjes function.

Equations
theorem probability_theory.cond_cdf_nonneg {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (a : α) (r : ) :

The conditional cdf is non-negative for all a : α.

theorem probability_theory.cond_cdf_le_one {α : Type u_1} {mα : measurable_space α} (ρ : measure_theory.measure × )) (a : α) (x : ) :

The conditional cdf is lower or equal to 1 for all a : α.

The conditional cdf tends to 0 at -∞ for all a : α.

The conditional cdf tends to 1 at +∞ for all a : α.

The conditional cdf is a measurable function of a : α for all x : ℝ.

Auxiliary lemma for set_lintegral_cond_cdf.

The conditional cdf is a strongly measurable function of a : α for all x : ℝ.

The function a ↦ (cond_cdf ρ a).measure is measurable.