s-finite measures can be written as withDensity of a finite measure #
If μ is an s-finite measure, then there exists a finite measure μ.toFinite and a measurable
function densityToFinite μ such that μ = μ.toFinite.withDensity μ.densityToFinite. If μ is
zero this is the zero measure, and otherwise we can choose a probability measure for μ.toFinite.
That measure is not unique, and in particular our implementation leads to μ.toFinite ≠ μ even if
μ is a probability measure.
We use this construction to define a set μ.sigmaFiniteSet, such that μ.restrict μ.sigmaFiniteSet
is sigma-finite, and for all measurable sets s ⊆ μ.sigmaFiniteSetᶜ, either μ s = 0
or μ s = ∞.
Main definitions #
In these definitions and the results below, μ is an s-finite measure (SFinite μ).
MeasureTheory.Measure.toFinite: a finite measure withμ ≪ μ.toFiniteandμ.toFinite ≪ μ. Ifμ ≠ 0, this is a probability measure.MeasureTheory.Measure.densityToFinite: a measurable function such thatμ = μ.toFinite.withDensity μ.densityToFinite.
Main statements #
absolutelyContinuous_toFinite:μ ≪ μ.toFinite.toFinite_absolutelyContinuous:μ.toFinite ≪ μ.withDensity_densitytoFinite:μ.toFinite.withDensity μ.densityToFinite = μ.
Auxiliary definition for MeasureTheory.Measure.toFinite.
Equations
- μ.toFiniteAux = MeasureTheory.Measure.sum fun (n : ℕ) => (2 ^ (n + 1) * (MeasureTheory.sFiniteSeq μ n) Set.univ)⁻¹ • MeasureTheory.sFiniteSeq μ n
Instances For
A finite measure obtained from an s-finite measure μ, such that
μ = μ.toFinite.withDensity μ.densityToFinite (see withDensity_densitytoFinite).
If μ is non-zero, this is a probability measure.
Equations
- μ.toFinite = ProbabilityTheory.cond μ.toFiniteAux Set.univ
Instances For
Equations
- ⋯ = ⋯
Equations
- ⋯ = ⋯
Equations
- ⋯ = ⋯
A measurable function such that μ.toFinite.withDensity μ.densityToFinite = μ.
See withDensity_densitytoFinite.
Equations
- μ.densityToFinite a = ∑' (n : ℕ), (MeasureTheory.sFiniteSeq μ n).rnDeriv μ.toFinite a