Weak convergence in the space of continuous functions
TheoremContinuous-path laws are determined by finite-dimensional marginals#✓fully proved
MathlibExt.Probability.ContinuousPathConvergence.continuousPath_measure_ext_of_finset_eval
Probability measures on continuous real paths over [0, T] are
determined by all of their finite evaluation pushforwards. The empty Finset
is included, and repeated coordinates are identified by Finset.
theorem continuousPath_measure_ext_of_finset_eval (T : ℝ)
(μ ν : ProbabilityMeasure C(Set.Icc (0 : ℝ) T, ℝ))
(h : ∀ I : Finset (Set.Icc (0 : ℝ) T),
μ.map (fun f => I.restrict (fun t => f t)) =
ν.map (fun f => I.restrict (fun t => f t))) :
μ = ν
TheoremTightness plus finite-dimensional convergence gives weak convergence#✓fully proved
MathlibExt.Probability.ContinuousPathConvergence.tendsto_continuousPath_of_tight_of_finset_eval
Let T > 0. A tight sequence of probability laws on
C(Set.Icc 0 T, ℝ) converges weakly to μ in the uniform topology if every
finite evaluation pushforward converges weakly to the corresponding
pushforward of μ.
This is a condition-to-convergence criterion, not a process convergence theorem: applications must separately prove tightness and all finite-dimensional convergences for their constructed laws.
MathlibExt.Probability.ContinuousPathConvergence.tendsto_continuousPath_of_tight_of_finset_eval · 16 lines
theorem tendsto_continuousPath_of_tight_of_finset_eval (T : ℝ) (_hT : 0 < T)
(μs : ℕ → ProbabilityMeasure C(Set.Icc (0 : ℝ) T, ℝ))
(μ : ProbabilityMeasure C(Set.Icc (0 : ℝ) T, ℝ))
(htight : IsTightMeasureSet
{((ν : ProbabilityMeasure C(Set.Icc (0 : ℝ) T, ℝ)) :
Measure C(Set.Icc (0 : ℝ) T, ℝ)) | ν ∈ Set.range μs})
(hfdd : ∀ I : Finset (Set.Icc (0 : ℝ) T),
Tendsto
(fun n => (μs n).map
(fun f : C(Set.Icc (0 : ℝ) T, ℝ) =>
I.restrict (fun t => f t)))
atTop
(𝓝 (μ.map
(fun f : C(Set.Icc (0 : ℝ) T, ℝ) =>
I.restrict (fun t => f t))))) :
Tendsto μs atTop (𝓝 μ)
Uses 1 · Used by 1
2 fully proved
Loading dependency explorer…
The dependency explorer needs JavaScript and its graph data.