跳转至

极限定理与估计

大数定律

中心极限定理(CLT)

Theorem

\(X_1, \cdots, X_n\) 独立同分布,\(E(X_i) = \mu\)\(Var(X_i) = \sigma^2 > 0\),则有:

\[\lim_{n\to \infty}P(\frac{X_1+\cdots+X_n - n\mu}{\sqrt n \sigma} \leq x) = \Phi(x)\]

也就是说: 1. \(X_1 + \cdots + X_n {\overset{近似 }{\sim}} N(n \mu, n \sigma^2)\) 2. \(\overline X { \overset{近似}{\sim}} N(n \mu, \frac{\sigma^2}{n})\)

也就是说,根据其应有的期望和方差称为正态分布。

应用:估计二项分布

\(X_i \sim B(p)\)\(\sum X_i \sim B(n,p)\),则:

\[P(t_1 \leq \sum_{i=1}^n X_i \leq t_2) \approx \Phi(y_2) - \Phi(y_1)\]

其中:

\[ \left\{ \begin{aligned} y_1 = \frac{t_1-np{\color{ff0000}-1/2}}{\sqrt{np(1-p)}}\\ y_2 = \frac{t_2-np{\color{ff0000}+1/2}}{\sqrt{np(1-p)}} \end{aligned} \right. \]