Probability identities for seed-indexed output families #
The retained-seed weighting is a deterministic image of a source and seed pair. Its finite tests agree with weighted seeded tests, including repeated output values. Normalized conditional distributions give a normalized joint distribution whenever the seed type is nonempty.
theorem
Algebraic.Cutwidth.Extractor.Internal.weightedSeededOutput_eq_mapWeight
{α : Type u_1}
{Seed : Type u_2}
{Ω : Type u_3}
[Fintype α]
[Fintype Seed]
(p : α → ℝ)
(C : α → Seed → Ω)
:
weightedSeededOutput p C = mapWeight (fun (xy : α × Seed) => (xy.2, C xy.1 xy.2)) fun (xy : α × Seed) => p xy.1 / ↑(Fintype.card Seed)
theorem
Algebraic.Cutwidth.Extractor.Internal.probabilityWeight_seedFamilyWeight
{Seed : Type u_1}
{Ω : Type u_2}
[Fintype Seed]
[Fintype Ω]
[Nonempty Seed]
(q : Seed → Ω → ℝ)
(probability : ∀ (y : Seed), IsProbabilityWeight (q y))
:
theorem
Algebraic.Cutwidth.Extractor.Internal.probabilityWeight_weightedSeededOutput
{α : Type u_1}
{Seed : Type u_2}
{Ω : Type u_3}
[Fintype α]
[Fintype Seed]
[Fintype Ω]
[Nonempty Seed]
{p : α → ℝ}
(probability : IsProbabilityWeight p)
(C : α → Seed → Ω)
:
theorem
Algebraic.Cutwidth.Extractor.Internal.weightTestProb_seedFamilyWeight_mixture
{ι : Type u_1}
{Seed : Type u_2}
{Ω : Type u_3}
[Fintype ι]
[Fintype Seed]
(w : ι → ℝ)
(q : ι → Seed → Ω → ℝ)
(T : Finset (Seed × Ω))
:
weightTestProb (seedFamilyWeight fun (y : Seed) (z : Ω) => ∑ i : ι, w i * q i y z) T = ∑ i : ι, w i * weightTestProb (seedFamilyWeight (q i)) T
theorem
Algebraic.Cutwidth.Extractor.Internal.cappedWeight_map_injective
{α : Type u_1}
{β : Type u_2}
[Fintype α]
{p : α → ℝ}
{K : ℕ}
(cap : CappedWeight p K)
(f : α → β)
(injective : Function.Injective f)
:
CappedWeight (mapWeight f p) K