跳转至

概率论第五周作业

习题 3 T3

\(C\) 应取何值才能使下列的数列称为概率分布?

  1. \(p_k = \dfrac{C}{N}, k=1,2,\cdots,N\) ;
  2. \(p_k = C \dfrac{\lambda^k}{k!},k=1,2,\cdots,\lambda>0\).

(1) 根据 \(\sum\limits_{k=1}^\infty p_k = 1\) ,有

\[ \sum\limits_{k=1}^\infty p_k = C \sum\limits_{k=1}^N \frac{1}{k} = 1 \]

\[ C = \frac{1}{\sum\limits_{k=1}^N \frac{1}{k}} \]

(2) 同理有

\[ \sum\limits_{k=1}^\infty p_k = C \sum\limits_{k=1}^\infty \dfrac{\lambda^k}{k!} = 1 \]

而由 \(\mathrm{e}^x\) 的 Maclaurin 展开有

\[ C = \mathrm{e}^{-\lambda} \]

\(\square\)

习题 3 T4

若分布函数定义为 \(F(x)= P \left\lbrace \xi \leqslant x \right\rbrace\) 试证明这时的 \(F(x)\) 具有下列性质:

  1. 非降
  2. \(F(- \infty) = 0, F(+\infty) =1\)
  3. 右连续.

(1) 设 \(x_1 < x_2\) ,那么

\[ F(x_2)-F(x_1) = P \left\lbrace x_1<\xi \leqslant x_2 \right\rbrace \geqslant 0 \]

从而 \(F(x)\) 非降.

(2) 考虑 \(P \left\lbrace -\infty < x < \infty \right\rbrace=1\) ,有

\[ \begin{aligned} P \left\lbrace -\infty < x < \infty \right\rbrace &= \sum\limits_{k=-\infty}^\infty [F(k)-F(k-1)] \\ &= F(+\infty)-F(-\infty) \\ &= 1 \end{aligned} \]

由于 \(F(x)\in [0,1]\) ,有 \(F(+\infty) = 1\)\(F(-\infty)=0\) .

(3) 考虑趋于 \(x\) 的单调降的序列:\(\left\lbrace x_n \right\rbrace\)\(x_0 > x_1> x_2 > \cdots, x_n \to x\) ,那么有

\[ \begin{aligned} F(x_0)-F(x) &= P \left\lbrace x < \xi \leqslant x_0 \right\rbrace \\ &= \sum\limits_{k=1}^\infty P \left\lbrace x_{k}< x \leqslant x_{k-1} \right\rbrace\\ &= \sum\limits_{k=1}^\infty [F(x_{k-1})-F(x_k)] \\ &= F(x_0) - \lim_{k\to \infty} F(x_k) \end{aligned} \]

那么 \(F(x) = \lim\limits_{k\to \infty}F(x_k)\) ,从而由 \(x_k\) 单调降可得右连续. \(\square\)

习题 3 T5

\(\zeta\sim N(0,1)\) ,试求常数 \(a,b,c\) 使得

  1. \(a=P \left\lbrace \zeta \geqslant 1.645 \right\rbrace\).
  2. \(P \left\lbrace |\zeta|<b \right\rbrace = 95\%\).
  3. \(P \left\lbrace |\zeta-c| >c \right\rbrace=0.51\).

(1) 由 \(\zeta \sim N(0,1)\) ,则有

\[ P \left\lbrace \zeta < x \right\rbrace = \dfrac{1}{\sqrt{2 \pi}}\int_{-\infty}^x\mathrm{e}^{- \frac{y^2}{2}}\mathrm{d}y \]

考虑

\[ P \left\lbrace \zeta < 1.645 \right\rbrace = \frac{1}{\sqrt{2 \pi}} \int_{-\infty}^{1.645}\mathrm{e}^{-\frac{x^2}{2}}\mathrm{d}x \approx 0.950015 \]

\(a\approx 1-0.950015 = 0.049985\) . 保留两位小数有 \(a\approx 0.05\) .

(2) 即

\[ P \left\lbrace -b<\zeta < b \right\rbrace = 0.95 \]

因此有 \(P \left\lbrace \zeta<b \right\rbrace = \dfrac{1-0.95}{2}+0.95= 0.975\) . 计算可得 \(b \approx 1.96\) .

(3) 考虑

\[ P \left\lbrace \zeta < 0 \right\rbrace+ P \left\lbrace \zeta> 2c \right\rbrace = 0.51 \]

其中由于 \(\mu =0\) 可知 \(P \left\lbrace \zeta<0 \right\rbrace = 0.5\) ,因此 \(P \left\lbrace \zeta>2c \right\rbrace = 0.01\) . \(c\approx 1.163\) . \(\square\)

习题 3 T10

设随机变量 \(\xi\) 取值于 \([0,1]\) ,若 \(P \left\lbrace x \leqslant \xi < y \right\rbrace\) 只与长度 \(y-x\) 有关,试证明 \(\xi\) 服从 \([0,1]\) 均匀分布.

设密度函数为 \(p(x)\) ,那么有

\[ P \left\lbrace x \leqslant \xi < y \right\rbrace = \int_x^y p(t)\mathrm{d}t \]

固定长度为 \(\eta\) ,那么有

\[ \int_x^{x+\eta} p(t) \mathrm{d} t \]

\(x\in [0,1-\eta]\) 无关,根据数学分析的知识, \(p(x)\) 为周期为 \(\eta\) 的周期函数,同时又由 \(\eta\) 的任意性,可知 \(p(x)\) 为常值函数.

根据

\[ \int_0^1 p(x) \mathrm{d} x = 1 \]

可知

\[ p(x) = \begin{cases} 1, & x\in [0,1] \\ 0, & x\in (-\infty,0)\cup (1,+\infty) \end{cases} \]

\(\xi\) 服从 \([0,1]\) 上的均匀分布. \(\square\)

习题 3 T12

定义二元函数:
$$ F(x,y) = \begin{cases}1, & x+y>0 \\ 0, & x+y \leqslant 0\end{cases} $$
验证此函数对每个变元都是非降、左连续的,且满足分布函数性质 (ii) ,但无法使 (3.2.5) 保持非负.

\(x\) 变元为例,\(y\) 类似,当 \(y\) 固定为 \(y_0\) 时,记 \(\varphi(x) = F(x,y_0)\) ,此时

\[ \varphi(x) = \begin{cases} 0, & x \leqslant -y_0, \\ 1, & x> -y_0. \\ \end{cases} \]

显然是非降的,左连续也仅需考虑 \(x=-y_0\) 点,易知

\[ \lim_{x\to -y_0^-} \varphi(x) = 0 =\varphi(-y_0) \]

从而左连续成立. \(X=(x,y)\to -\infty\) 时,\(F(X)=0\) 成立,\(+\infty\) 时情形相似,\(F(+\infty)=1\) . 因此分布函数性质 (ii) 也成立.

对 (3.2.5) 式,此时考虑

\[ F(1,1)-F(1,0)-F(0,1)+F(0,0) = 1-1-1+0 = -1 \]

因此无法保持非负. \(\square\)

评论