Forgot password?
 Register account
View 1258|Reply 2

[概率/统计] 以概率1收敛(a.s.收敛)问题

[Copy link]

83

Threads

434

Posts

5419

Credits

Credits
5419

Show all posts

tommywong Posted 2016-5-16 07:25 |Read mode
Last edited by hbghlyj 2025-5-6 20:14设 $f(x)$ 是 $[0,1]$ 上有定义的连续函数,且 $0 \leqslant f(x) \leqslant 1$ .若 $\xi_1, \eta_1, \xi_2, \eta_2, \cdots$ 是一列服从 $[0,1]$ 上均勺分布的相互独立的随机变量序列,令
\[
\rho_i= \begin{cases}1, & f\left(\xi_i\right) \geqslant \eta_i, \quad(i \geqslant 1) \\ 0, & f\left(\xi_i\right)<\eta_i\end{cases}
\]
试证明:当 $n \rightarrow \infty$ 时,
\[
\frac{1}{n} \sum_{i=1}^n \rho_i \xrightarrow{\text{a.s.}} \int_0^1 f(x) \mathrm{d} x .
\]

24

Threads

1010

Posts

110K

Credits

Credits
12655

Show all posts

战巡 Posted 2016-5-16 13:19
回复 1# tommywong

\[P(\rho=1)=P(f(\xi)\ge\eta)=\int_0^1P(f(\xi)\ge\eta|f(\xi))P(f(\xi))=\int_0^1f(\xi)·\frac{df(\xi)}{f'(\xi)}=\int_0^1f(\xi)d\xi=p\]
而后显然
\[\sum_{k=1}^n\rho_k\sim BIN(n,p)\]


这个延伸下去可以得出Metropolis–Hastings算法,可以对一些不那么常见的分布进行抽样模拟,往往用在MCMC里面

83

Threads

434

Posts

5419

Credits

Credits
5419

Show all posts

 Author| tommywong Posted 2016-5-17 06:11
收皮性呢?

Mobile version|Discuz Math Forum

2025-6-5 07:39 GMT+8

Powered by Discuz!

× Quick Reply To Top Edit