Skip to main content

Measure-Theoretic Probability · Weak convergence, characteristic functions, and CLT

Let μₙ=δ_(1/n) and μ=δ₀ be Dirac measures on ℝ, and write Fₙ and F for their…

Problem

Measure-Theoretic Probability practice problem

Let μn=δ1/n\mu_n=\delta_{1/n} and μ=δ0\mu=\delta_0 be Dirac measures on R\mathbb{R}, and write FnF_n and FF for their distribution functions (right-continuous, Fn(x)=1{1/n≤x}F_n(x)=1_{\{1/n\le x\}}, F(x)=1{0≤x}F(x)=1_{\{0\le x\}}). Do not replace the moving-atom Portmanteau ledger by a characteristic-function central-limit argument, by an abstract comparison of almost-sure / L1L^1 / L2L^2 modes repaired by uniform integrability, or by a tail-event 00–11 law.

1. Prove that μn⇒μ\mu_n\Rightarrow\mu by showing
∫f dμn=f(1/n)→f(0)=∫f dμ\int f\,d\mu_n=f(1/n)\to f(0)=\int f\,d\mu for every bounded continuous f:R→Rf:\mathbb{R}\to\mathbb{R}.

2. Prove that Fn(x)→F(x)F_n(x)\to F(x) for every x≠0x\neq 0, while
Fn(0)=0F_n(0)=0 for all nn and F(0)=1F(0)=1. Conclude that {0}\{0\} and (0,∞)(0,\infty) are not continuity sets of μ\mu, withμn({0})=0≠1=μ({0}),μn((0,∞))=1≠0=μ((0,∞)).\mu_n(\{0\})=0\neq 1=\mu(\{0\}),\qquad \mu_n\bigl((0,\infty)\bigr)=1\neq 0=\mu\bigl((0,\infty)\bigr).3. State the Portmanteau open-set and closed-set inequalities and
verify them on G=(0,∞)G=(0,\infty) and F={0}F=\{0\}:lim inf⁡nμn(G)=1≥0=μ(G),lim sup⁡nμn(F)=0≤1=μ(F).\liminf_n\mu_n(G)=1\ge 0=\mu(G),\qquad \limsup_n\mu_n(F)=0\le 1=\mu(F).Prove that the pp-Wasserstein distance satisfies Wp(μn,μ)=1/nW_p(\mu_n,\mu)=1/n for every p≥1p\ge 1.

4. Audit both sentences: (i) “distribution functions always
converge at every xx, so Fn(0)→F(0)F_n(0)\to F(0) and the atom {0}\{0\} is a continuity set”; (ii) “because μn⇒μ\mu_n\Rightarrow\mu, one has μn((0,∞))→μ((0,∞))\mu_n((0,\infty))\to\mu((0,\infty)), so the open ray is a continuity set and the probabilities 11 and 00 agree in the limit.”

Hint

Hint

A bounded continuous function is continuous at 00, so f(1/n)→f(0)f(1/n)\to f(0). This is the definition of weak convergence for these two Diracs.

Check your work

Work the problem yourself first. Then open it in Training to check your answer and read the full worked solution.

The answer check and full solution for this problem come with ProofAnvil Practice membership ($19 USD monthly). See membership. Or start with the free Measure-Theoretic Probability sample problem: Try the free sample problem.