数理统计与数据分析第三版习题 第3章 第54-69题

题目54

令$X$,$Y$和$Z$是独立的$N(0,{\sigma }^2)$.随机变量$\Theta ,\Phi 和R$是$(X,Y,Z)$的球形坐标：
$$\begin{array}{rl}x& =rsin\varphi cos\theta \\ y& =rsin\varphi sin\theta \\ z& =rcos\varphi \\ 0\le \varphi & \le \pi ,0\le \theta \le 2\pi \end{array}$$
计算$\Theta ,\Phi 和R$的联合密度和边际密度.(提示:$dxdydz=r^{2}sin\varphi drd\theta d\varphi$)

解题思路

根据球形坐标的转换提示，我们直接求联合密度：
$$\begin{array}{rl}{f}_{R,\Phi ,\Theta }(r,\varphi ,\theta )& =f_{X,Y,Z}(rsin\varphi cos\theta ,rsin\varphi sin\theta ,rcos\varphi )r^{2}sin\varphi \\ & =\frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(rsin\varphi cos\theta {)}^2}{2{\sigma }^2}}\cdot \frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(,rsin\varphi sin\theta {)}^2}{2{\sigma }^2}}\cdot \frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(rcos\varphi {)}^2}{2{\sigma }^2}}{r}^2\sin \varphi \\ & ={(2\pi )}^{-\frac{3}{2}}\cdot {\sigma }^{-3}\cdot e^{-\frac{r^{2}si{n}^2\varphi co{s}^2\theta +r^2{\sin }^2\varphi {\sin }^2\theta +r^2{\cos }^2\varphi }{2{\sigma }^2}}{r}^2\sin \varphi \\ & =\sin \varphi \cdot \frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}\cdot r^2\cdot e^{-\frac{r^2}{2{\sigma }^2}}\end{array}$$

求边际密度
$$\begin{array}{rl}{f}_r(r)& ={\int }_0^{\pi }{\int }_0^{2\pi }{f}_{R,\Phi ,\Theta }(r,\varphi ,\theta )d\theta d\varphi \\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}{\int }_0^{\pi }{\int }_0^{2\pi }\sin \varphi \cdot r^2\cdot e^{-\frac{r^2}{2{\sigma }^2}}d\theta d\varphi \\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}{\int }_0^{\pi }\sin \varphi d\varphi \cdot {\int }_0^{2\pi }d\theta \cdot r^2{e}^{-\frac{r^2}{2{\sigma }^2}}\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}\cdot 2\cdot 2\pi \cdot r^2{e}^{-\frac{r^2}{2{\sigma }^2}}\\ & =\frac{\sqrt{2}}{\sqrt{\pi }{\sigma }^3}{r}^2{e}^{-\frac{r^2}{2{\sigma }^2}}\end{array}$$
$$\begin{array}{rl}{f}_{\Phi }(\varphi )& ={\int }_0^{\infty }{\int }_0^{2\pi }{f}_{R,\Phi ,\Theta }(r,\varphi ,\theta )d\theta dr\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}{\int }_0^{\infty }{\int }_0^{2\pi }\sin \varphi \cdot r^2\cdot e^{-\frac{r^2}{2{\sigma }^2}}d\theta dr\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}\cdot \sin \varphi \cdot {\int }_0^{2\pi }d\theta {\int }_0^{\infty }{r}^2{e}^{-\frac{r^2}{2{\sigma }^2}}dr\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}\cdot \sin \varphi \cdot 2\pi \cdot \frac{\sqrt{\pi }}{\sqrt{2}}{\sigma }^3\\ & =\frac{\sin \varphi }{2}\end{array}$$
$$\begin{array}{rl}{f}_{\Theta }(\theta )& ={\int }_0^{\infty }{\int }_0^{\pi }{f}_{R,\Phi ,\Theta }(r,\varphi ,\theta )d\varphi dr\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}{\int }_0^{\pi }\sin \varphi d\varphi \cdot {\int }_0^{\infty }{r}^2\cdot e^{-\frac{r^2}{2{\sigma }^2}}dr\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{2\pi }}\cdot \frac{1}{{\sigma }^3}\cdot 2\cdot \frac{\sqrt{\pi }}{\sqrt{2}}{\sigma }^3\\ & =\frac{1}{2\pi }\end{array}$$
$$f_{R,\Phi ,\Theta }(r,\varphi ,\theta )=f_r(r)f_{\Phi }(\varphi )f_{\Theta }(\theta )$$三个变量转换为球形坐标后也是独立的。

计算上边三个边际密度时，有一个关键积分公式：
${\int }_0^{\infty }{x}^{2n}{e}^{-\frac{x^2}{{\alpha }^2}}dx=\sqrt{\pi }\frac{(2n)!}{n!}(\frac{\alpha }{2}{)}^{2n+1}$

题目55

从单位元内部按如下规则生成一点：半径R是[0,1]上的均匀随机变量，角度$\Theta 是[0,2\pi ]$上我均匀随机变量，并且与R独立

a.计算$X=R\cos \theta ,Y=R\sin \theta$的随便密度
b.计算$X和Y$的边际密度
c.密度是圆盘上的均匀分布吗？如果不是修正该方法是密度是均匀分布

解题思路

a.
$$\begin{array}{rl}& R\sim U[0,1]\\ & \Theta \sim U[0,2\pi ]\\ & (X=R\cos \Theta ,Y=R\sin \Theta )\Rightarrow (R=\sqrt{X^2+Y^2},\Theta ={\tan }^{-1}\frac{Y}{X})\\ & |J|=\Vert \begin{array}{cc}\frac{\partial r}{\partial x}& \frac{\partial r}{\partial y}\\ \frac{\partial \theta }{\partial x}& \frac{\partial \theta }{\partial y}\end{array}\Vert =\frac{1}{\sqrt{x^2+y^2}}\\ f_{XY}(x,y)& =f_{R\Theta }(\sqrt{x^2+y^2},{\tan }^{-1}\frac{y}{x})\cdot \frac{1}{\sqrt{x^2+y^2}}\\ & =\frac{1}{2\pi }\cdot \frac{1}{\sqrt{x^2+y^2}}{x}^2+y^2\le 1\end{array}$$
b.计算$X和Y$的边际密度
$$\begin{array}{rl}{f}_X(x)& ={\int }_{-\sqrt{1-x^2}}^{\sqrt{1-x^2}}{f}_{XY}(x,y)dy\\ & ={\int }_{-\sqrt{1-x^2}}^{\sqrt{1-x^2}}\frac{1}{2\pi }\cdot \frac{1}{\sqrt{x^2+y^2}}dy\\ & =\frac{1}{\pi }{\int }_0^{\sqrt{1-x^2}}\frac{1}{\sqrt{x^2+y^2}}dy\\ & =\frac{1}{\pi }[\ln (y+\sqrt{x^2+y^2})]{|}_0^{\sqrt{1-x^2}}\\ & =\frac{1}{\pi }\ln (\frac{\sqrt{1-x^2}+1}{|x|})-1\le x\le 1\\ 同样道理：& \\ f_Y(y)=& \frac{1}{\pi }\ln (\frac{\sqrt{1-y^2}+1}{|y|})-1\le y\le 1\end{array}$$
c.密度是圆盘上的均匀分布吗？如果不是修正该方法是密度是均匀分布
$$\begin{array}{r}{f}_R(r)=1\Rightarrow F_R(r)=r\end{array}$$
如果要在圆盘上均匀分布，则应该按面积进行均匀分布如果整个圆是1 ，则：
$F_R(r)=\frac{2\pi r^2}{2\pi }=r^2很明显r!=r^2$,不是均匀分布

$F_R(r)=r^2\Rightarrow f_R(r)=2r$

$f_{R\Theta }(r,\theta )=2r\cdot \frac{1}{2\pi }=\frac{r}{\pi }$

题目56

如果$X$和$Y$是独立的随机变量，计算点$（X，Y）$的极坐标R和$\Theta$的联合分布，R和$\Theta$独立吗？

解题思路

未找到答案

题目57

假设$Y_1$和$Y_2$服从二元正态分布，具有参数${\mu }_{Y_1}={\mu }_{Y_2}=0,{\sigma }_{Y_1}^2=1,{\sigma }_{Y_2}^2=2,且\rho =1/\sqrt{2}$.线性变换 $x_1=a_{11}{y}_1+a_{12}{y}_2,x_2=a_{21}{y}_1+a_{22}{y}_2$ 使得$x_1$和$x_2$是独立的标准的正态分布？

解题思路

几个关键公式
1.二元正态分布密度函数
$$f(y_1,y_2)=\frac{1}{2\pi {\sigma }_{Y_1}{\sigma }_{y_2}\sqrt{1-{\rho }^2}}exp\left[-\frac{1}{2(1-{\rho }^2)}\left(\frac{(y_1-{\mu }_{Y_1}{)}^2}{{\sigma }_{y_1}^2}-\frac{2\rho (y_1-{\mu }_{Y_1})(y_2-{\mu }_{Y_2})}{{\sigma }_{Y_1}{\sigma }_{Y_2}}+\frac{(y_2-{\mu }_{Y_2}{)}^2}{{\sigma }_{y_2}^2}\right)\right]$$
2.设$(Y_1,Y_2)\sim N({\mu }_{Y_1},{\mu }_{Y_2},{\sigma }_{Y_1}^2,{\sigma }_{Y_2}^2,\rho )$则$Y_1$与$Y_2$的线性组合仍然服从正态分布,且$a{Y}_1+b{Y}_2\sim N(a{\mu }_{Y_1}+b{\mu }_{Y_2},a^2{\sigma }_{Y_1}^2+b^2{\sigma }_{Y_2}^2+2ab\rho {\sigma }_{Y_1}{\sigma }_{Y_2})$

步骤1.求出$(X_1,X_2)$的联合密度函数
$$J(y_1,y_2)=\Vert \begin{array}{cc}\frac{\partial x_1}{\partial y_1}& \frac{\partial x_1}{\partial y_2}\\ \frac{\partial x_2}{\partial y_1}& \frac{\partial x_2}{\partial y_2}\end{array}\Vert =\Vert \begin{array}{cc}{a}_{11}& a_{12}\\ a_{21}& a_{22}\end{array}\Vert =a_{11}{a}_{22}-a_{12}{a}_{21}$$
$$\begin{array}{rl}{y}_1& =\frac{a_{12}{x}_2-a_{22}{x}_1}{a_{12}{a}_{21}-a_{22}{a}_{11}}\\ y_2& =\frac{a_{11}{x}_2-a_{21}{x}_1}{a_{22}{a}_{11}-a_{21}{a}_{12}}\\ f_{X_1{X}_2}(x_1,x_2)& =J^{-1}(y_1,y_2)\cdot f_{Y_1{Y}_2}(\frac{a_{12}{x}_2-a_{22}{x}_1}{a_{12}{a}_{21}-a_{22}{a}_{11}},\frac{a_{11}{x}_2-a_{21}{x}_1}{a_{22}{a}_{11}-a_{21}{a}_{12}})\\ 把题目的中参数代& 入:\\ & =\frac{1}{a_{11}{a}_{22}-a_{12}{a}_{21}}\cdot \frac{1}{2\pi }exp\left[-\left(\frac{(y_1{)}^2}{1}-\frac{2\frac{1}{\sqrt{2}}{y}_1{y}_2}{1\sqrt{2}}+\frac{(y_2{)}^2}{2}\right)\right]\\ & =\frac{1}{a_{11}{a}_{22}-a_{12}{a}_{21}}\cdot \frac{1}{2\pi }exp\left[-\left((\frac{a_{12}{x}_2-a_{22}{x}_1}{a_{12}{a}_{21}-a_{22}{a}_{11}}{)}^2-(\frac{a_{12}{x}_2-a_{22}{x}_1}{a_{12}{a}_{21}-a_{22}{a}_{11}})(\frac{a_{11}{x}_2-a_{21}{x}_1}{a_{22}{a}_{11}-a_{21}{a}_{12}})+\frac{(\frac{a_{11}{x}_2-a_{21}{x}_1}{a_{22}{a}_{11}-a_{21}{a}_{12}}{)}^2}{2}\right)\right]\end{array}$$
步骤2.求$f_{X_1}(x_1)和f_{X_2}(x_2)$
$$\begin{array}{rl}{f}_{X_1}& =\frac{1}{\sqrt{2\pi }\sqrt{a_{11}^2{\sigma }_{Y_1}^2+a_{12}^2{\sigma }_{Y_2}^2+2a_{11}{a}_{12}\rho {\sigma }_{Y_1}{\sigma }_{Y_2}}}exp\left[-\frac{x_1-(a_{11}{\mu }_{Y_1}+a_{12}{\mu }_{Y_2})}{2(a_{11}^2{\sigma }_{Y_1}^2+a_{12}^2{\sigma }_{Y_2}^2+2a_{11}{a}_{12}\rho {\sigma }_{Y_1}{\sigma }_{Y_2})}\right]\\ & =\frac{1}{\sqrt{2\pi }\sqrt{a_{11}^2+2a_{12}^2+2a_{11}{a}_{12}}}exp\left[-\frac{x_1}{2(a_{11}^2+a_{12}^2\cdot 2+2a_{11}{a}_{12})}\right]\\ f_{X_2}& =\frac{1}{\sqrt{2\pi }\sqrt{a_{21}^2{\sigma }_{Y_1}^2+a_{22}^2{\sigma }_{Y_2}^2+2a_{21}{a}_{22}\rho {\sigma }_{Y_1}{\sigma }_{Y_2}}}exp\left[-\frac{x_2-(a_{21}{\mu }_{Y_1}+a_{22}{\mu }_{Y_2})}{2(a_{21}^2{\sigma }_{Y_1}^2+a_{22}^2{\sigma }_{Y_2}^2+2a_{21}{a}_{22}\rho {\sigma }_{Y_1}{\sigma }_{Y_2})}\right]\\ & =\frac{1}{\sqrt{2\pi }\sqrt{a_{21}^2+2a_{22}^2+2a_{21}{a}_{22}}}exp\left[-\frac{x_2}{2(a_{21}^2+a_{22}^2\cdot 2+2a_{21}{a}_{22})}\right]\\ f_{X1}{f}_{X_2}& =\frac{1}{2\pi \sqrt{a_{11}^2+2a_{12}^2+2a_{11}{a}_{12}}\sqrt{a_{21}^2+2a_{22}^2+2a_{21}{a}_{22}}}exp\left[-\frac{x_1}{2(a_{11}^2+a_{12}^2\cdot 2+2a_{11}{a}_{12})}-\frac{x_2}{2(a_{21}^2+a_{22}^2\cdot 2+2a_{21}{a}_{22})}\right]\end{array}$$
根据题意要求，即标准的独立的二元正太分布，则标准差为1，$\rho =0$,整理后得到如下等式：
$$\begin{array}{r}{a}_{11}{a}_{22}-a_{12}{a}_{21}=1\\ a_{11}^2+2a_{12}^2+2a_{11}{a}_{12}=1\\ a_{21}^2+2a_{22}^2+2a_{21}{a}_{22}=1\\ 2a_{12}{a}_{22}+a_{12}{a}_{21}+a_{11}{a}_{22}+a_{11}{a}_{21}=0\end{array}$$
$$a_{11}=1,a_{12}=0,a_{21}=-1,a_{22}=1$$
当然这并不是的解。解可有很多，满足上边4个等式即可

题目58

如果$X_1$和$X_2$的联合分布是二元正态的，则$Y_1=a_1{X}_1+b_1,Y_2=a_2{X}_2+b_2$的联合分布也是二元正态的

解题思路

根据线性变换的雅可比行列式进行转换
$$J(x_1,x_2)=\Vert \begin{array}{cc}\frac{\partial y_1}{\partial x_1}& \frac{\partial y_1}{\partial x_2}\\ \frac{\partial y_2}{\partial x_1}& \frac{\partial y_2}{\partial x_2}\end{array}\Vert =\Vert \begin{array}{cc}{a}_1& 0\\ 0& a_2\end{array}\Vert =a_1{a}_2$$
$$x_1=\frac{y_1-b_1}{a_1},x_2=\frac{y_2-b_2}{a_2}$$
$$f_{X_1{X}_2}(x_1,x_2)=\frac{1}{2\pi {\sigma }_{X_1}{\sigma }_{X_2}\sqrt{1-{\rho }^2}}exp\left[-\frac{1}{2(1-{\rho }^2)}\left(\frac{(x_1-{\mu }_{X_1}{)}^2}{{\sigma }_{X_1}^2}-\frac{2\rho (x_1-{\mu }_{X_1})(x_2-{\mu }_{X_2})}{{\sigma }_{X_1}{\sigma }_{X_2}}+\frac{(x_2-{\mu }_{X_2}{)}^2}{{\sigma }_{X_2}^2}\right)\right]$$
$$\begin{array}{rl}{f}_{Y_1{Y}_2}(y_1,y_2)& =f_{X_1{X}_2}(\frac{y_1-b_1}{a_1},\frac{y_2-b_2}{a_2})\cdot |J{|}^{-1}\\ & =\frac{1}{a_1{a}_2}\frac{1}{2\pi {\sigma }_{X_1}{\sigma }_{X_2}\sqrt{1-{\rho }^2}}\exp \left[-\frac{1}{2(1-{\rho }^2)}\left(\frac{(\frac{y_1-b_1}{a_1}-{\mu }_{X_1}{)}^2}{{\sigma }_{X_1}^2}-\frac{2\rho (\frac{y_1-b_1}{a_1}-{\mu }_{X_1})(\frac{y_2-b_2}{a_2}-{\mu }_{X_2})}{{\sigma }_{X_1}{\sigma }_{X_2}}+\frac{(\frac{y_2-b_2}{a_2}-{\mu }_{X_2}{)}^2}{{\sigma }_{X_2}^2}\right)\right]\\ & =\frac{1}{2\pi a_1{\sigma }_{X_1}{a}_2{\sigma }_{X_2}\sqrt{1-{\rho }^2}}\exp \left[-\frac{1}{2(1-{\rho }^2)}\left(\frac{(y_1-(b_1+a_1{\mu }_{X_1}){)}^2}{(a_1{\sigma }_{X_1}{)}^2}-\frac{2\rho (y_1-(b_1+a_1{\mu }_{X_1}))(y_2-(b_2+a_2{\mu }_{X_2}))}{a_1{\sigma }_{X_1}{a}_2{\sigma }_{X_2}}+\frac{(y_2+(b_2+a_2{\mu }_{X_2}){)}^2}{(a_2{\sigma }_{X_2}{)}^2}\right)\right]\end{array}$$
令,${\mu }_{Y_1}=b_1+a_1{\mu }_{X_1},{\mu }_{Y_2}=b_2+a_2{\mu }_{X_2},{\sigma }_{Y_1}=a_1{\sigma }_{X_1},{\sigma }_{Y_2}=a_2{\sigma }_{X_2}$则:
$$(Y_1,Y_2)\sim N({\mu }_{Y_1},{\mu }_{Y_2},{\sigma }_{X_2}^2,{\sigma }_{X_2}^2,\rho )$$

题目59

如果$X_1$和$X_2$是独立的标准正态随机变量，证明：$Y_1$和$Y_2$的联合分布是二元正态的
$$\begin{array}{r}{Y}_1=a_{11}{X}_1+a_{12}{X}_2+b_1\\ Y_2=a_{21}{X}_1+a_{22}{X}_2+b_2\end{array}$$

解题思路

令:
$$\begin{array}{r}{Z}_1=a_{11}{X}_1+a_{12}{X}_2\\ Z_2=a_{21}{X}_1+a_{22}{X}_2\end{array}$$
则有:
$$\begin{array}{r}{Y}_1=Z_1+b_1\\ Y_2=Z_2+b_2\end{array}$$
如果X_1和X_2是标准正态随机变量则他们的联合分布是二元正态的即：
$(X_1,X_2)\sim N(0,0,1,1,0)$
二元正态随机变量的线性组合依然是二元正态的即：
$(Z_1,Z_2)$是二元正态的
再根据58题的结果即可以证明$(Y_1,Y_2)$也是二元正态的(线性变换)

题目60

利用上一题的结果，描述构造由独立伪随机均匀变量生成具有二元正态分布的伪随机变量的方法

解题思路

step1.采用Box Muller方法由两个均匀分布$(u,v)$，得到两个独立的正态分布：
$$\begin{gathered} z_1=\sqrt{-2logu}cos2\pi v \\ {z}_2=\sqrt{-2logu}sin2\pi \end{gathered}$$
step2.利用59题的结果将得到的两个正态分布样本进行变量生成的$(y_1,y_2)$即满足二元正态分布。

题目61

令$X$和$Y$是具有联合分布的连续随机变量，求$U=a+bX$和$V=c+dY$的联合分布密度的表达式

解:

令(X,Y)的联合密度为$f_{XY}(x,y)$
利用雅可比行列式$f_{UV}(u,v)=\frac{1}{J(x,y)}{f}_{XY}(U^{-1},V^{-1})$

$$J(x,y)=\Vert \begin{array}{cc}\frac{\partial u}{\partial x}& \frac{\partial u}{\partial y}\\ \frac{\partial v}{\partial x}& \frac{\partial v}{\partial y}\end{array}\Vert =\Vert \begin{array}{cc}b& 0\\ 0& d\end{array}\Vert =bd$$
所以答案：
$f_{UV}(u,v)=\frac{1}{bd}{f}_{XY}(\frac{u-a}{b},\frac{v-c}{d})$

题目62

令$X$和$Y$是独立标准正态随机变量，求$P(X^2+Y^2\le 1)$

解:

转化为极坐标：
$$\begin{array}{rl}{f}_{R\Theta }(r,\theta )& =r{f}_{XY}(r\cos \theta ,r\sin \theta )\\ & =\frac{1}{2\pi }r{e}^{-\frac{r^2}{2}}\end{array}$$

根据条件$X^2+Y^2\le 1$,即极坐标下$r\le 1$

$$\begin{array}{rl}P(X^2+Y^2\le 1)& =P(r\le 1)\\ & ={\int }_0^{2\pi }{\int }_0^1\frac{1}{2\pi }r{e}^{-\frac{r^2}{2}}drd\theta \\ & ={\int }_0^{2\pi }d\theta {\int }_0^1\frac{1}{2\pi }r{e}^{-\frac{r^2}{2}}dr\\ & =2\pi \cdot \frac{1}{2\pi }{\int }_0^{1}r{e}^{-\frac{r^2}{2}}dr\\ & ={\int }_0^1{e}^{-\frac{r^2}{2}}d\frac{r^2}{2}\\ 令u=\frac{r^2}{2}& \\ & ={\int }_0^{\frac{1}{2}}{e}^{-u}du\\ & =1-e^{-\frac{1}{2}}\end{array}$$

题目63

令$X$和$Y$是具有联合分布的连续随机变量
a.讨论$X+Y$和$X-Y$的联合密度表达式
b.讨论$XY$和$X/Y$的联合密度表达式
c.在$X$和$Y$独立的情况下,特殊表示a和b中的表达式

解:

a.令
$$\begin{array}{r}U=X+Y,V=X-Y\\ X=\frac{U+V}{2},Y=\frac{U-V}{2}\\ J(x,y)=\Vert \begin{array}{cc}\frac{\partial u}{\partial x}& \frac{\partial u}{\partial y}\\ \frac{\partial v}{\partial x}& \frac{\partial v}{\partial y}\end{array}\Vert =\Vert \begin{array}{cc}1& 1\\ 1& -1\end{array}\Vert =2\\ f_{UV}=\frac{1}{2}{f}_{XY}(\frac{u+v}{2},\frac{u-v}{2})\end{array}$$
b.令
$$\begin{array}{r}U=XY,V=X/Y\\ X=(UV{)}^{\frac{1}{2}}Y=(\frac{U}{V}{)}^{\frac{1}{2}}J(x,y)=\Vert \begin{array}{cc}\frac{\partial u}{\partial x}& \frac{\partial u}{\partial y}\\ \frac{\partial v}{\partial x}& \frac{\partial v}{\partial y}\end{array}\Vert =\Vert \begin{array}{cc}y& x\\ -\frac{1}{x^2}& \frac{1}{x}\end{array}\Vert =2\frac{x}{y}\\ f_{UV}=2|\frac{x}{y}|f_{XY}((xy{)}^{\frac{1}{2}},(\frac{x}{y}{)}^{\frac{1}{2}})\end{array}$$

题目64

计算$X+Y$和$X/Y$的联合密度，其中$X$和$Y$是参数为$\lambda$的独立指数随机变量，证明$X+Y$和$X/Y$独立的

解:

令$U=X+Y,V=X/Y\Rightarrow X=\frac{UV}{V+1},Y=\frac{U}{V+1}$

证明是独立需要证明:$f_U(u)\cdot f_V(v)=f_{UV}(u,v)$

有:
$$\begin{array}{rl}J(x,y)& =\Vert \begin{array}{cc}& \\ \frac{\partial u}{\partial x}& \frac{\partial u}{\partial y}\\ \frac{\partial v}{\partial x}& \frac{\partial v}{\partial y}\end{array}\Vert =\Vert \begin{array}{cc}1& 1\\ \frac{1}{y}& -\frac{x}{y^2}\end{array}\Vert =-\frac{x-y}{y^2}\\ |J(x,y){|}^{-1}& =\frac{y^2}{x+y}=\frac{u}{(v+1{)}^2}\end{array}$$
$$\begin{array}{rl}{f}_{UV}(u,v)& =|J(x,y){|}^{-1}{f}_{XY}(\frac{uv}{v+1},\frac{u}{v+1})\\ & =\frac{u}{(v+1{)}^2}\cdot \lambda e^{-\lambda \frac{uv}{v+1}}\cdot \lambda e^{-\lambda \frac{u}{v+1}}\\ & =\frac{u}{(v+1{)}^2}{\lambda }^2{e}^{-\lambda u}\end{array}$$
接下来，分别求 $f_U(u)和f_V(v)$
$$\begin{array}{rl}{f}_U(u)& ={\int }_{-\infty }^{\infty }{f}_X(x)f_Y(u-x)dx={\lambda }^{2}u{e}^{-\lambda u}\\ f_V(v)& ={\int }_{-\infty }^{\infty }|x|f_X(x)f_Y(xv)dx\\ & ={\lambda }^2{\int }_{-\infty }^{\infty }x{e}^{-\lambda x(1+v)}dx\\ 令a=-\lambda (1+v),则积分下限为0，上限为\infty & \\ & ={\lambda }^2{\int }_0^{\infty }x{e}^{ax}dx\\ & ={\lambda }^2\frac{1}{a^2}(ax-1)e^{ax}{|}_0^{\infty }\\ & =\frac{1}{1+v^2}\end{array}$$
所以$f_U(u)\cdot f_V(v)=f_{UV}(u,v)$是成立的，U和V是独立的。

题目65

假定系统元件串联在一起，且元件的寿命为参数为${\lambda }_i$独立指数随机变量，证明系统寿命服从参数是$\sum {\lambda }_i$的指数分布。

解:

参考书中例题
令$V$代表系统整体寿命,根据书中公式有
$$\begin{array}{rl}{f}_V(v)& =nf(v)[1-F(v){]}^{n-1}\\ 其中F(v)=1-e^{-\lambda v}& \\ & =n\lambda e^{-\lambda v}(e^{-\lambda v}{)}^{n-1}\\ & =n\lambda e^{-n\lambda v}\end{array}$$

题目66

下图系统中的每一个元件寿命都独立的服从参数为$\lambda$的指数分布，计算系统寿命的cdf和密度
图不画了，解释一下，6个元件分成3组，组内两个元件串连，3组并联

解:

令串联部分的寿命分布为$F_V(t)$
$F_V(t)=1-[1-e^{-\lambda t}{]}^2=1-e^{-2\lambda t}$
令整体系统寿命分布为$F_U(t)$
$F_U(t)=[F_V(v){]}^3=(1-e^{-2\lambda t}{)}^3$
求上式求导
$f(t)=6\lambda e^{-2\lambda t}(1-e^{-2\lambda t}{)}^2$

题目67

卡片含有n个芯片和一个纠错元件，这样如果只有一个芯片失效，卡片仍能正常工作；如果有两个或两个以上的芯片失效，卡片将不能正常工作。如果每个芯片的寿命服从参数为 $\lambda$的指数分布，计算卡片寿命的密度函数。

解:

令
$F_A(t)$系统的寿命分布
$F_0(t)$所有芯片都没有失效的概率分布
$F_1(t)$只有一个芯片失效的概率分布
$F(t)$单一的一个芯片失效的概率分布
则有:
$$\begin{array}{rl}{F}_A(t)& =1-F_0(t)-F(t)\\ & =1-(1-F(t){)}^n-nF(t)(1-F(t){)}^{n-1}\\ & =1-(1-(1-e^{-\lambda t}){)}^n-n(1-e^{-\lambda t})(1-(1-e^{-\lambda t}){)}^{n-1}\\ & =1-e^{-n\lambda t}-n(1-e^{-\lambda t})e^{-(n-1)\lambda t}\\ & =1-e^{-n\lambda t}-n{e}^{-(n-1)\lambda t}+n{e}^{-n\lambda t}\end{array}$$
对上式求导则为密度函数
$$\begin{array}{rl}{f}_A(t)& =F_A^{{}^{\prime }}(t)\\ & =-e^{-n\lambda t}\cdot -n\lambda -n{e}^{-(n-1)\lambda t}\cdot -(n-1)\lambda +n{e}^{-n\lambda t}\cdot -n\lambda \\ & =n\lambda e^{-n\lambda t}+n(n-1)\lambda e^{-(n-1)\lambda t}-n\cdot n\lambda e^{-n\lambda t}\\ & =n(n-1)\lambda e^{-(n-1)\lambda t}+n(1-n)\lambda e^{-n\lambda t}\\ & =n(n-1)\lambda (e^{-(n-1)\lambda t}-e^{-n\lambda t})\end{array}$$

题目68

令$U_1$、$U_2$和$U_3$是独立的均匀随机变量
a.计算$U_{(1)}$、$U_{(2)}$和$U_{(3)}$的联合密度
b.在一英里的高速公路上，独立且随机地建造三个加油站，任意两个加油站之间的距离不小于$\frac{1}{3}$英里的概率是多少

解:

a.模拟对最水值和最小值联合密度的计算方法
令$U=U_{(1)},V=U_{(2)},W=U_{(3)}$
$$\begin{array}{rl}f(u,v,w)& =\frac{n!}{1!1!1!}f(u)f(v)f(w)\\ 代入n=3,f(u)=f(v)=f(w)=1& \\ f(u,v,w)& =6,u\le v\le w\end{array}$$
b.在一英里的高速公路上，独立且随机地建造三个加油站，任意两个加油站之间的距离不小于$\frac{1}{3}$英里的概率是多少
我们可以认为3个加油站的位置服从$U(0,1)$,令X为第1个加油站的位置,Y为第二个加油站的位置Z为第三个加油站的位置,则X,Y,Z是3个均匀分布的顺序统计量，则$f_{XYZ}(x,y,z)=6(根据a问题的结果)$
任意两个加油站之间的距离大于$\frac{1}{3}$则必须有$Y-X>\frac{1}{3}并且Z-Y>\frac{1}{3}$
$$\begin{array}{rl}P(Y-X>\frac{1}{3}且Z-Y>\frac{1}{3})& ={\int }_0^{\frac{1}{3}}{\int }_{x+\frac{1}{3}}^{\frac{2}{3}}{\int }_{y+\frac{1}{3}}^{1}f(x,y,z)dzdydx\\ & ={\int }_0^{\frac{1}{3}}{\int }_{x+\frac{1}{3}}^{\frac{2}{3}}4-6ydy\\ & ={\int }_0^{\frac{1}{3}}\frac{1}{3}-2x-3x^{2}dx\\ & =\frac{1}{27}\end{array}$$
积分上下限是结合$0<x<y<z<1$且$y-x>\frac{1}{3}且z-t>\frac{1}{3}$得出

题目69

计算$n$个独立威布尔随机变量最小值的密度，每个变量具有密度：
$$f(t)=\beta {\alpha }^{-\beta }{t}^{-\beta -1}{e}^{-(t/\alpha {)}^{\beta }},t\ge 0$$

解:

令$V$代表$n$个独立威布尔变量的最小值随机变量
根据3.7节中的公式
$$f_V(v)=nf(v)[1-F(v){]}^{n-1}$$
其中：
$$f_{(}v)=\beta {\alpha }^{-\beta }{v}^{-\beta -1}{e}^{-(v/\alpha {)}^{\beta }}$$
$$F_{(}v)=1-e^{-(v/\alpha {)}^{\beta }}$$
代入上式经计算结果如下：
$$f_V(v)=n\beta {\alpha }^{-\beta }{v}^{\beta -1}{e}^{-n(v/\alpha {)}^{\beta }}$$