Levy continuity theorem을 위해 필요

Lemma

If \(\mu\) is a probability measure with characteristic function \(\phi\), then for any \(u>0\), \[ \frac{1}{u}\int^u_{-u}[1-\phi(t)]dt\ge \mu(\{x:|x|\ge 2/u\}). \]



중요하다

Theorem (Levy Continuity Theorem)

Let \(\mu_n,n\ge 1\) be probability measures with characterisric functions \(\phi_n\). Then,

  1. \(\mu_n\leadsto \mu\iff \phi_n(t)\rightarrow \phi(t)\) for all \(t\in \mathbb{R}\) where \(\phi(t)\) is char.func of \(\mu\).

  2. \(\phi_n(t)\rightarrow g(t)\) for all \(t\in \mathbb{R}\) and \(g\) is continuous at \(t=0\implies g\) is the characterisric function of \(\mu\), and \(\mu_n\leadsto \mu\).


\[ \phi_n(t)=\int e^{itx}\mbox{ }d\mu_n = \int \cos(tx)\mbox{ }d\mu_n + i \int \sin(tx)\mbox{ }d\mu_n \longrightarrow \int \cos(tx)\mbox{ }d\mu + i \int \sin(tx)\mbox{ }d\mu=\phi(t). \]




매우 중요하고 자주 출제되는 내용이다.

Corollary

Suppose that \(g(t)=\lim_{n\rightarrow \infty}\phi_n(t)\) exists for each \(t\in \mathbb{R}\), and that \(\{\mu_n,n\ge 1\}\) is tight. Then there exists a prob measure \(\mu\) s.t. \(\mu_n\leadsto \mu\) and \(\mu\) has characteristic function \(g\).





back