概率论考研复习

火舞炫风

已于 2023-12-14 15:41:03 修改

阅读量90

点赞数

文章标签：概率论

于 2023-09-02 16:21:19 首次发布

本文链接：https://blog.csdn.net/qq_51884709/article/details/132063194

版权

随机事件和概率

事件的关系与运算

互斥（互不相容）

如果A和B的关系为AB=ф，即AB不能同时发生

则称A与B互斥或互不相容

互逆（对立）

如果A与B有且仅有一个发生，A∪B=Ω，A∩B=ф

则称A与B为对立事件或互逆事件

差

A发生而B不发生，称A与B的差，记A-B

$eq?A-B%3DA%5Cbar%7BB%7D$

运算规律

$eq?%5Coverline%7BA%5Ccup%20B%7D%3D%5Cbar%20A%5Cbar%20B%2C%5Coverline%7BAB%7D%3D%5Cbar%20A%20%5Ccup%20%5Cbar%20B$

概率及概率公式

独立

当0<P(A)<1,A与B独立相当于P(B|A)=P(B)或 $eq?P%28B%7CA%29%3DP%28B%7C%5Cbar%7BA%7D%29$

多个事件相互独立，则其中任意几个事件相互独立

多个事件两两独立，则这几个事件不一定相互独立

五大概率公式

$eq?P%28A%5Ccup%20B%29%3DP%28A%29+P%28B%29-P%28AB%29$

$eq?P%28A%5Ccup%20B%20%5Ccup%20C%29%3DP%28A%29+P%28B%29+P%28C%29-P%28AB%29-P%28AC%29-P%28BC%29+P%28ABC%29$

$eq?P%28A-B%29%3DP%28A%29-P%28AB%29%3DP%28A%5Cbar%7BB%7D%29$

$eq?P%28A%5Ccup%20B%29%3D1-P%28%5Cbar%7BA%7D%20%5Cbar%7BB%7D%29$

全概率公式

$eq?P%28A%29%3D%5Csum_%7Bi%3D1%7D%5E%7Bn%7DP%28B_i%29P%28A%7CB_i%29$

贝叶斯公式

设 $eq?B_1%2CB_2%2C....B_n$ 满足 $eq?%5Cbigcup_%7Bi%3D1%7D%5E%7Bn%7DB_i%3D%5COmega%20%2CB_iB_j%3D%5Cvarnothing%20%28i%5Cneq%20j%29$

则 $eq?P%28B_j%7CA%29%3D%5Cfrac%7BP%28B_j%29P%28A%7CB_j%29%7D%7B%5Csum_%7Bi%3D1%7D%5E%7Bn%7DP%28B_i%29P%28A%7CB_i%29%7D$

补充

从n个不同不同元素中每次取出一个放回后在取下一个

连续m次所得的组合称为重复组合，则重复组合总数为 $eq?C%5E%7Bm%7D_%7Bn+m-1%7D$

如果把n个不同元素分成k组，使得第i组有 $eq?n_i$ 个元素

$eq?%5Csum_%7Bi%3D1%7D%5E%7Bk%7Dn_i%3Dn$ ，不考虑元素的顺序，那么不同的分法 $eq?%5Cfrac%7Bn%21%7D%7Bn_1%21n_2%21...n_k%21%7D$

常用组合公式

$eq?C%5Ek_n%3DC%5E%7Bn-k%7D_n%2CC%5Ek_%7Bn+1%7D%3DC%5Ek_n+C%5E%7Bk-1%7D_n$

$eq?%5Csum_%7Bi%3D0%7D%5E%7Bn%7DC%5Ei_n%3D2%5En%2CC%5Ek_%7Bn+m%7D%3D%5Csum_%7Bi%3D0%7D%5E%7Bk%7DC%5Ei_nC%5E%7Bk-i%7D_m$

随机变量及其概率分布

分布函数的性质

$eq?f%28-%5Cinfty%29%3D0%2Cf%28+%5Cinfty%29%3D1$ ，单调非减，右连续

常用分布

泊松分布

$eq?P%5C%7BX%3Dk%5C%7D%3D%5Cfrac%7B%5Clambda%5Ek%7D%7Bk%21%7De%5E%7B-%5Clambda%7D%2C%28%5Clambda%3E0%29$

记作 $eq?X%5Csim%20P%28%5Clambda%29$

泊松定理

在伯努利试验中， $eq?p_n$ 代表事件A在一次实验中出现的概率

它与实验总数n有关，且随n的增大， $eq?p_n$ 在减小

$eq?%5Clim_%7Bn%5Cto%20%5Cinfty%7Dnp_n%3D%5Clambda$ ,则，出现k次A发生的概率

$eq?%5Clim_%7Bn%5Cto%20%5Cinfty%7DC%5Ek_np%5Ek_n%281-p_n%29%5E%7Bn-k%7D%3D%5Cfrac%7B%5Clambda%5Ek%7D%7Bk%21%7De%5E%7B-%5Clambda%7D$

指数分布

连续型随机变量X的概率密度为

$eq?f%28x%29%3D%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%20%5Clambda%20e%5E%7B-%5Clambda%7D%2Cx%3E0%5C%5C%200%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2Cx%5Cleq%200%20%5Cend%7Bmatrix%7D%5Cright.$

则称X服从参数 $eq?%5Clambda$ 的指数分布，记作 $eq?X%5Csim%20E%28%5Clambda%29$

X的分布函数

$eq?F%28x%29%3D%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%201-e%5E%7B%5Clambda%20x%7D%2Cx%3E0%5C%5C%200%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2Cx%5Cleq%200%20%5Cend%7Bmatrix%7D%5Cright.$

正态分布

如果随机变量X的概率密度为

$eq?f%28x%29%3D%5Cfrac%7B1%7D%7B%5Csqrt%7B2%5Cpi%7D%5Csigma%20%7De%5E%7B-%5Cfrac%7B%28x-%5Cmu%20%29%5E2%7D%7B2%5Csigma%20%5E2%7D%7D%2C-%5Cinfty%3Cx%3C+%5Cinfty$

则称X服从参数为 $eq?%5Cmu%20%2C%5Csigma$ 的正态分布，记作 $eq?X%5Csim%20N%28%5Cmu%2C%5Csigma%5E2%29$

概率密度

$eq?%5Cvarphi%20%28x%29%3D%5Cfrac%7B1%7D%7B%5Csqrt%7B2%5Cpi%7D%7De%5E%7B-%5Cfrac%7B%28x-%5Cmu%29%5E2%7D%7B2%5Csigma%5E2%7D%7D%2C-%5Cinfty%3Cx%3C+%5Cinfty$

随机变量函数的分布

公式法

设y=g(x)是单调函数，导数不为0的可导函数，h(y)

为它的反函数，则

$eq?f_Y%28y%29%3D%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%20%7Ch%27%28y%29%7Cf_X%28h%28y%29%29%2C%5Calpha%3Cy%3C%5Cbeta%5C%5C%200%2C%20%5Cend%7Bmatrix%7D%5Cright.$

定义法

先求Y的分布函数 $eq?F_Y%28y%29$

$eq?F_Y%28y%29%3DP%5C%7BY%3Cy%5C%7D%3DP%5C%7Bg%28X%29%3Cy%5C%7D%3D%5Cint_%7Bg%28x%29%5Cleq%20y%7Df_X%28x%29dx$

甲乙两人进行射击比赛，约定甲先射 $eq?P%5C%7BX%3Dn%5C%7D%3D%5Cfrac%7B30%5En%7D%7Bn%21%7De%5E%7B-30%7D$

若射不中，乙射，若射不中再由甲射

以此类推，谁先射中谁获胜，比赛终止

甲乙射中的概率分别为0.4，0.6，求

甲乙获胜的概率

$eq?P%5C%7BY%3D0%5C%7D%3D0.36%281+0.24+0.24%5E2+...+0.24%5En%29%3D%5Cfrac%7B0.36%7D%7B1-0.24%7D%3D%5Cfrac%7B9%7D%7B19%7D$

$eq?P%5C%7BY%3D1%5C%7D%3D0.4%281+0.24+0.24%5E2+...+0.24%5En%29%3D%5Cfrac%7B0.4%7D%7B1-0.24%7D%3D%5Cfrac%7B10%7D%7B19%7D$

设某时段进入某景区游客人数服从

参数为 $eq?%5Clambda%3D30$ 的泊松分布，每位游客

乘坐观光缆车的概率均为0.6，且每位

游客乘坐观光缆车时=是相互独立的

求进入该景区的游客乘坐观光缆车

人数的概率分布及数学期望与方差

设X=n的条件下，游客人数 $eq?Y%5Csim%20B%28n%2C0.6%29$

$eq?P%5C%7BX%3Dn%5C%7D%3D%5Cfrac%7B30%5En%7D%7Bn%21%7De%5E%7B-30%7D$

$eq?P%5C%7BY%3Dk%7CX%3Dn%5C%7D%3DC%5Ek_n0.6%5Ek%5Ccdot%200.4%5E%7Bn-k%7D$

当k>n时 $eq?P%5C%7BY%3Dk%7CX%3Dn%5C%7D%3D0$

由全概率公式得

$eq?P%5C%7BY%3Dk%5C%7D%3D%5Csum_%7Bn%3D0%7D%5E%7B%5Cinfty%7DP%5C%7BX%3Dn%5C%7DP%5C%7BY%3Dk%7CX%3Dn%5C%7D$

$eq?%3D%5Csum_%7Bn%3Dk%7D%5E%7B%5Cinfty%7DP%5C%7BX%3Dn%5C%7DP%5C%7BY%3Dk%7CX%3Dn%5C%7D$

$eq?%3D%5Cfrac%7B18%5Ek%7D%7Bk%21%7De%5E%7B-30%7D$

于是Y服从参数为18的泊松分布，EY=DY=18

若随机变量X的概率分布为

X 1 2 3 4
P 0.1 0.2 0.3 0.4

令随机变量 $eq?Y%3Dg%28X%29%3D%5Cfrac%7B1%7D%7B2%7D%5B1+%28-1%29%5EX%5D$

求Y的概率分布

若 $eq?X%5Csim%20B%28n%2Cp%29$ 求X的取值为偶数时的概率P

$eq?P%5C%7BY%3D1%5C%7D%3DEY%3DE%5Bg%28X%29%5D%3D%5Cfrac%7B1%7D%7B2%7D%5C%7B1+E%5B%28-1%29%5EX%5D%5C%7D$

$eq?%3D%5Cfrac%7B1%7D%7B2%7D%5B1+%5Csum_%7Bk%3D0%7D%5En%28-1%29%5EkC%5Ek_np%5Ek%281-p%29%5E%7Bn-k%7D%5D$

$eq?%3D%5Cfrac%7B1%7D%7B2%7D%5B1+%281-2p%29%5En%5D$

多维随机变量及其分布

分布函数的性质

$eq?F%28x%2Cy%29%3DP%5C%7BX%5Cleq%20x%2CY%5Cleq%20y%20%5C%7D%2C-%5Cinfty%3Cx%2Cy%3C+%5Cinfty$

$eq?P%5C%7Ba%3C%20X%5Cleq%20b%2Cc%3CY%5Cleq%20d%5C%7D%3DF%28b%2Cd%29+F%28a%2Cc%29-F%28b%2Cc%29-F%28a%2Cd%29$

F(x,y)均关于x，y右连续，单调不减

条件分布

如果给定的ε>0, $eq?P%5C%7By-%5Cvarepsilon%20%3CY%5Cleq%20y+%5Cvarepsilon%20%5C%7D%3E0$

$eq?%5Clim_%7B%5Cvarepsilon%20%5Cto%200%5E+%7DP%5C%7BX%5Cleq%20x%7Cy-%5Cvarepsilon%20%3CY%5Cleq%20y+%5Cvarepsilon%5C%7D$

$eq?%3D%5Clim_%7B%5Cvarepsilon%20%5Cto%200%5E+%7D%5Cfrac%7BP%5C%7BX%5Cleq%20x%2Cy-%5Cvarepsilon%20%3CY%5Cleq%20y+%5Cvarepsilon%5C%7D%7D%7BP%5C%7By-%5Cvarepsilon%20%3CY%5Cleq%20y+%5Cvarepsilon%5C%7D%7D$

此极限为Y=y条件下X的分布

记 $eq?F_%7BX%7CY%7D%28x%7Cy%29%2CP%5C%7BX%5Cleq%20x%7CY%3Dy%5C%7D$

边缘概率密度

$eq?f_X%28x%29%3D%5Cint%5E%7B+%5Cinfty%7D_%7B-%5Cinfty%7Df%28x%2Cy%29dy$

$eq?f_Y%28y%29%3D%5Cint%5E%7B+%5Cinfty%7D_%7B-%5Cinfty%7Df%28x%2Cy%29dx$

两个随机变量函数的分布

X,Y均连续

$eq?F_Z%28z%29%3DP%5C%7BZ%5Cleq%20z%5C%7D%3DP%5C%7Bg%28X%2CY%29%5Cleq%20z%5C%7D%3D%5Ciint_%7Bg%28x%2Cy%29%5Cleq%20z%7Df%28x%2Cy%29dxdy$

X,Y相互独立

Z=X+Y

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7Df_X%28x%29f_Y%28z-x%29dx$

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7Df_X%28z-y%29f_Y%28y%29dy$

令 $eq?%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%20u%3Dx%5C%5C%20v%3Dx+y%20%5Cend%7Bmatrix%7D%5Cright.$

$eq?F_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7Bz%7D%5Cint_%7B-%5Cinfty%7D%5E%7Bz-x%7Df%28u%2Cv%29dudv$

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7Df_X%28x%29f_Y%28z-x%29dx$

Z=XY

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7D%5Cfrac%7B1%7D%7B%7Cx%7C%7Df_X%28x%29f_Y%28z-x%29dx$

令 $eq?%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%20u%3Dx%5C%5C%20v%3Dxy%20%5Cend%7Bmatrix%7D%5Cright.$

$eq?F_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7Bz%7D%5Cint_%7B-%5Cinfty%7D%5E%7B%5Cfrac%7Bz%7D%7Bx%7D%7Df%28u%2Cv%29dudv$

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7D%5Cfrac%7B1%7D%7B%7Cx%7C%7Df_X%28x%29f_Y%28%5Cfrac%7Bz%7D%7Bx%7D%29dx$

Z=Y/X

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7D%7Cx%7Cf_X%28x%29f_Y%28z-x%29dx$

令 $eq?%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%20u%3Dx%5C%5C%20v%3D%5Cfrac%7By%7D%7Bx%7D%20%5Cend%7Bmatrix%7D%5Cright.$

$eq?F_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7Bz%7D%5Cint_%7B-%5Cinfty%7D%5E%7Bxz%7Df%28u%2Cv%29dudv$

$eq?f_Z%28z%29%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7D%7Cx%7Cf_X%28x%29f_Y%28%5Cfrac%7Bz%7D%7Bx%7D%29dx$

Z=max{X,Y}

$eq?F_Z%28z%29%3DF_X%28x%29F_Y%28y%29$

Z=min{X,Y}

$eq?F_Z%28z%29%3D1-%5B1-F_X%28x%29%5D%5B1-F_Y%28y%29%5D$

设二维连续型随机变量(X,Y)的分布函数为

$eq?%5Cleft%5C%7B%5Cbegin%7Bmatrix%7D%200%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2Cx%3C0%20%5C%2Cor%20y%3C0%20%5C%5C%20x%5E2y%5E2%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C0%5Cleq%20x%3C1%2C0%5Cleq%20y%3C1%5C%5C%20x%5E2%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C0%5Cleq%20x%3C1%2Cy%5Cgeq1%5C%5C%20y%5E2%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2Cx%5Cgeq%201%2C0%5Cleq%20y%20%3C1%5C%5C%201%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2C%5C%2Cx%5Cgeq1%2Cy%5Cgeq1%20%5Cend%7Bmatrix%7D%5Cright.$

(1)求 $eq?P%5C%7B0%3CX%3C%5Cfrac%7B1%7D%7B2%7D%2C%5Cfrac%7B1%7D%7B4%7D%3CY%3C1%5C%7D%2CP%5C%7B0%5Cleq%20X%3C%5Cfrac%7B1%7D%7B2%7D%7C%5Cfrac%7B1%7D%7B4%7D%3CY%3C1%5C%7D$

f(x,y)=4xy 0≤x≤1,0≤y≤1

$eq?P%5C%7B0%3CX%3C%5Cfrac%7B1%7D%7B2%7D%2C%5Cfrac%7B1%7D%7B4%7D%3CY%3C1%5C%7D%3D%5Cint_0%5E%7B%5Cfrac%7B1%7D%7B2%7D%7Ddx%5Cint_%7B%5Cfrac%7B1%7D%7B4%7D%7D%5E14xydy%3D%5Cfrac%7B15%7D%7B64%7D$

$eq?P%5C%7B0%5Cleq%20X%3C%5Cfrac%7B1%7D%7B2%7D%7C%5Cfrac%7B1%7D%7B4%7D%3CY%3C1%5C%7D%3D%5Cint_0%5E%7B%5Cfrac%7B1%7D%7B2%7D%7D2xdx%3D%5Cfrac%7B1%7D%7B4%7D$

(2)求P{X≤Y},P{X=Y}

$eq?P%5C%7B%20X%5Cleq%20Y%5C%7D%3D%5Cint_0%5E1dy%5Cint_0%5Ey4xydx%3D%5Cfrac%7B1%7D%7B2%7D$

P{X=Y}为一条线积分为0

设 $eq?D%5C%7B%28x%2Cy%29%7C0%5Cleq%20y%20%5Cleq%201%2C-y%5Cleq%20x%5Cleq%20y%5C%7D$

二维随机变量(X,Y)服从D上的均匀分布

（1）(X,Y)关于X,Y的边缘概率密度

$eq?f_Y%3D%5Cint_%7B-y%7D%5E%7By%7Ddx%3D2y$

$eq?1.-1%5Cleq%20x%5Cleq0%2Cf_X%3D%5Cint_%7B-x%7D%5E%7B1%7Ddx%3D1+x$

$eq?2.%5C%2C%5C%2C0%5Cleq%20x%5Cleq1%2Cf_X%3D%5Cint_%7Bx%7D%5E%7B1%7Ddx%3D1-x$

（2）求Z=X-Y的概率密度

当-2≤Z≤0时

$eq?F_Z%28z%29%3DP%5C%7BX-Y%5Cleq%20z%5C%7D%3D%5Cint_%7B-%5Cfrac%7B1%7D%7B2%7Dz%7D%5E%7B1%7Ddy%5Cint_%7B-y%7D%5E%7Bz+y%7Ddx%3D%5Cfrac%7B1%7D%7B4%7Dz%5E2+z+1$

$eq?f_z%3D%5Cfrac%7B1%7D%7B2%7Dz+1$

设二维随机变量(X,Y)的概率密度为f(x,y)=4x+2y, 0≤x≤1, 0≤y≤1-x

记Z=min{X,Y}求Z的概率密度

$eq?F_Z%28z%29%3DP%5C%7Bmin%5C%7BX%2CY%5C%7D%3Cz%5C%7D%3DP%5C%7BX%5Cleq%20z%5C%7D+P%5C%7BY%5Cleq%20z%5C%7D-P%5C%7BX%5Cleq%20z%2CY%5Cleq%20z%5C%7D$

$eq?%3D%5Cint_%7B0%7D%5E%7Bz%7Ddx%5Cint_%7B0%7D%5E%7B1-x%7D%284x+2y%29dy+%5Cint_%7B0%7D%5E%7Bz%7Ddy%5Cint_%7B0%7D%5E%7B1-y%7D%284x+2y%29dx+%5Cint_%7B0%7D%5E%7Bz%7Ddx%5Cint_%7B0%7D%5E%7Bz%7D%284x+2y%29dy$

$eq?%3D3z-4z%5E2$

设某商品一周的需求量X是随机变量，已知其概率密度为

$eq?f%28x%29%3Dxe%5E%7B-x%7D%2Cx%3E0$

设各周的需求量相互独立，以 $eq?U_k$ 表示k周的总需求量

（1） $eq?U_2$ 和 $eq?U_3$ 的概率密度 $eq?f_k%28x%29$

$eq?U_2%3DX_1+X_2%5CRightarrow%20f_2%28x%29%3D%5Cint_%7B0%7D%5E%7Bx%7Df%28t%29f%28x-t%29dt%3De%5E%7B-x%7D%5Cint_%7B0%7D%5E%7Bx%7Dt%28x-t%29dt%3D%5Cfrac%7B1%7D%7B6%7Dx%5E3e%5E%7B-x%7D$

$eq?U_3%3DU_2+X_3%5CRightarrow%20f_3%28x%29%3D%5Cint_%7B0%7D%5E%7Bx%7Df_2%28t%29f%28x-t%29dt%3D%5Cfrac%7B1%7D%7B6%7De%5E%7B-x%7D%5Cint_%7B0%7D%5E%7Bx%7Dt%5E3%28x-t%29dt%3D%5Cfrac%7B1%7D%7B120%7Dx%5E5e%5E%7B-x%7D$

随机变量的数字特征

数学期望

（1）离散型： $eq?EX%3D%5Csum_%7Bk%3D1%7D%5E%7B%5Cinfty%7Dx_kp_k$

$eq?Y%3Dg%28X%29%2CEY%3D%5Csum_%7Bk%3D1%7D%5E%7B%5Cinfty%7Dg%28x_k%29p_k$

（2）连续型： $eq?EX%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7Dxf%28x%29dx$

$eq?Y%3Dg%28X%29%2CEY%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7Dg%28x%29f%28x%29dx$

$eq?Z%3Dg%28X%2CY%29%2CEZ%3D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7D%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7Dg%28x%2Cy%29f%28x%2Cy%29dxdy$

方差

$eq?D%28X%29%3DEX%5E2-E%5E2X$

$eq?D%28X%5Cpm%20Y%29%3DD%28X%29+D%28Y%29%5Cpm2Cov%28X%2CY%29$

常用随机变量数学期望方差

分布	EX	DX
0-1分布	p	p(1-p)
二项分布，X~B(n,p)	np	np(1-p)
超几何分布	$eq?n%5Cfrac%7BN_1%7D%7BN%7D$	$eq?n%5Cfrac%7BN_1%28N-N_1%29%28N-n%29%7D%7BN%5E2%28N-1%29%7D$
泊松分布，X~P(λ)	λ	λ
几何分布，P{X=k}=p(1-p)^(k-1)	$eq?%5Cfrac%7B1%7D%7Bp%7D$	$eq?%5Cfrac%7B1-p%7D%7Bp%5E2%7D$
均匀分布，X~U(a,b)	$eq?%5Cfrac%7Ba+b%7D%7B2%7D$	$eq?%5Cfrac%7B%28b-a%29%5E2%7D%7B12%7D$
指数分布，X~E(λ)	$eq?%5Cfrac%7B1%7D%7B%5Clambda%7D$	$eq?%5Cfrac%7B1%7D%7B%5Clambda%5E2%7D$
正态分布，X~N(μ,σ^2)	μ	σ^2

协方差的计算公式和性质

$eq?Cov%28X%2CY%29%3DE%28XY%29-E%28X%29E%28Y%29$

$eq?Cov%28aX%2CbY%29%3DabCov%28X%2CY%29$

$eq?Cov%28X_1+X_2%2CY%29%3DCov%28X_1%2CY%29+Cov%28X_2%2CY%29$

独立与不相关

独立必不相关，不相关也不一定独立

对于二维正态随机变量(X,Y)，相互独立的充要条件是ρ=0

对于二维正态随机变量(X,Y)，独立与不相关是等价的

某人进行独立重复射击，直到命中与不命中两个结果都出现为止

设每次命中概率为p，X表示射击次数，EX=3，则p

$eq?P%5C%7BX%3Dn%5C%7D%3Dpq%5E%7Bn-1%7D+qp%5E%7Bn-1%7D$

$eq?EX%3D%5Csum_%7Bn%3D2%7D%5E%7B%5Cinfty%7Dn%28pq%5E%7Bn-1%7D+qp%5E%7Bn-1%7D%29%3Dp%5Csum_%7Bn%3D1%7D%5E%7B%5Cinfty%7Dnq%5E%7Bn-1%7D+q%5Csum_%7Bn%3D1%7D%5E%7B%5Cinfty%7Dnp%5E%7Bn-1%7D-%28p+q%29$

构造幂级数 $eq?%5Csum_%7Bn%3D1%7D%5E%7B%5Cinfty%7Dnx%5E%7Bn-1%7D%2CS%28x%29%3D%28%5Csum_%7Bn%3D0%7D%5E%7B%5Cinfty%7Dx%5En%29%27%3D%28%5Cfrac%7Bx%7D%7B1-x%7D%29%27%3D%5Cfrac%7B1%7D%7B%281-x%29%5E2%7D$

$eq?EX%3DpS%28q%29+qS%28p%29-%28p+q%29%3D%5Cfrac%7B1%7D%7Bp%7D+%5Cfrac%7B1%7D%7B1-p%7D-1%3D3%5CRightarrow%20p%3D%5Cfrac%7B1%7D%7B2%7D$

设随机变量X,Y相互独立，且都服从标准正态分布N(0,1)

记U=X-Y，V=XY，则U,V的相关系数

$eq?Cov%28U%2CV%29%3DE%28UV%29-E%28U%29E%28V%29$

$eq?%3DE%5B%28X-Y%29XY%5D%3DEX%5E2Y-EXY%5E2%3D0$

若连续型随机变量X服从参数为1的指数分布

令随机变量Y=[X+1](取整)，EY=

$eq?F%28X%29%3D1-e%5E%7B-x%7D%2Cx%5Cgeq0$

$eq?P%5C%7B%5BX+1%5D%3Dk%5C%7D%3DP%5C%7Bk%5Cleq%20X+1%3Ck+1%5C%7D$

$eq?%3DF%28k%29-F%28k-1%29%3D%281-e%5E%7B-1%7D%29%28e%5E%7B-1%7D%29%5E%7Bk-1%7D$

所以随机变量Y服从几何分布

$eq?EY%3D%5Cfrac%7B1%7D%7B1-e%5E%7B-1%7D%7D%3D%5Cfrac%7Be%7D%7Be-1%7D$

若一设备在任何长为t的时间段内发生故障的次数N(t)

服从参数为λt的泊松分布，设两次故障之间时间间隔T

ET

t<0时，F(t)=P{T≤t}=0，t≥0，T≥t与N(t)=0等价

此时 $eq?F%28t%29%3DP%5C%7BT%5Cleq%20t%5C%7D%3D1-P%5C%7BN%28t%29%3D0%5C%7D%3D1-e%5E%7B%5Clambda%20t%7D$

时间间隔T服从参数为λ的指数分布

故 $eq?ET%3D%5Cfrac%7B1%7D%7B%5Clambda%7D$

设随机变量(X,Y)的概率密度f(x,y)满足f(x,y)=f(-x,y)且ρ存在

ρ的值

$eq?EXY%3D%5Cint%20ydy%5Cint%20xf%28x%2Cy%29dx$

令t=-x

$eq?EXY%3D%5Cint%20ydy%5Cint_%7B+%5Cinfty%7D%5E%7B-%5Cinfty%7D%20tf%28t%2Cy%29dt%3D-%5Cint%20yfy%5Cint_%7B-%5Cinfty%7D%5E%7B+%5Cinfty%7D%20xf%28-x%2Cy%29dx%3D-EXY$

ρ=0

结论

非矩形区域不独立

积分区域无对称性则必定相关

产品寿命X是一个随机变量，其分布函数与概率密度分别为

F(x)，f(x)产品已工作到时刻x，在时刻x后的单位时间去△x

内发生失效的概率称为产品在时刻x的顺势失效率，记λ(x)

（1）证明 $eq?%5Clambda%28x%29%3D%5Cfrac%7Bf%28x%29%7D%7B1-F%28x%29%7D$

$eq?%5Clambda%28x%29%3D%5Clim_%7B%5CDelta%20x%5Cto%200%5E+%7D%5Cfrac%7BP%5C%7Bx%3CX%5Cleq%20x+%5CDelta%20x%7CX%3Ex%5C%7D%7D%7B%5CDelta%20x%7D%3D%20%5Cfrac%7BP%5C%7Bx%3CX%5Cleq%20x+%5CDelta%20x%2CX%3Ex%5C%7D%7D%7B%5CDelta%20x.P%5C%7BX%3Ex%5C%7D%7D$

$eq?%3D%5Clim_%7B%5CDelta%20x%5Cto%200%5E+%7D%5Cfrac%7BF%28x+%5CDelta%20x%29-F%28x%29%7D%7B%5CDelta%20x%7D.%5Cfrac%7B1%7D%7B1-F%28x%29%7D%3D%20%5Cfrac%7Bf%28x%29%7D%7B1-F%28x%29%7D$

（2）λ(x)=α，求产品寿命X的数学期望

$eq?%5Calpha%3D%5Cfrac%7BF%27%28x%29%7D%7B1-F%28x%29%7D%5CRightarrow%20%5Calpha%20x+C%3D-%5Cln%20%5B1-F%28x%29%5D$

$eq?F%280%29%3D0%5CRightarrow%20C%3D0%5CRightarrow%20F%28x%29%3D1-e%5E%7B%5Calpha%20x%7D%5CRightarrow%20EX%3D%5Cfrac%7B1%7D%7B%5Calpha%7D$

大数定律与中心极限定理

切比雪夫不等式

随机变量X的数学期望EX和方差DX存在，对任意 $eq?%5Cvarepsilon$ >0

$eq?P%5C%7B%7CX-EX%7C%5Cgeq%5Cvarepsilon%20%5C%7D%5Cleq%5Cfrac%7BDX%7D%7B%5Cvarepsilon%20%5E2%7D$

拉普拉斯中心极限定理

设随机变量 $eq?X_n%5Csim%20B%28n%2Cp%29$ 对于任意实数x有

$eq?%5Clim_%7Bn%5Cto%20%5Cinfty%7DP%5C%7B%5Cfrac%7BX_n-np%7D%7B%5Csqrt%7Bnp%281-p%29%7D%7D%5Cleq%20x%5C%7D%3D%5CPhi%20%28x%29$

当n充分大时近似服从N(0,1)，或 $eq?X_n%5Csim%20N%28np%2Cnp%281-p%29%29$

林德伯格中心极限定理

设随机变量 $eq?X_1%2CX_2...X_n$ 独立同分布，具有数学期望与方差

$eq?E%28X_n%29%3D%5Cmu%20%2CD%28X_n%29%3D%5Csigma%20%5E2$ ,则对于任意x

$eq?%5Clim_%7Bn%5Cto%20%5Cinfty%7DP%5C%7B%5Cfrac%7B%5Csum%20_%7Bi%3D1%7D%5E%7Bn%7DX_i-n%5Cmu%7D%7B%5Csqrt%7Bn%7D%5Csigma%7D%5Cleq%20x%5C%7D%3D%5CPhi%20%28x%29$

当n充分大时近似服从N(0,1)，或 $eq?%5Csum_%7Bi%3D1%7D%5E%7Bn%7D%20X_i%5Csim%20N%28n%5Cmu%2Cn%5Csigma%5E2%29$

数理统计

样本分布

如果随机变量 $eq?X_1%2CX_2...X_n$ 独立与总体X同分布

则称 $eq?X_1%2CX_2...X_n$ 为来自总体的简单随机样本

n为样本容量，样本的具体观测值 $eq?x_1%2Cx_2...x_n$

为样本值，或称总体X的n个独立观测值

1.如果总体X的分布为F(x), $eq?X_1%2CX_2...X_n$ 的分布

$eq?F_n%28x_1%2Cx_2...x_n%29%3D%5Cprod_%7Bi%3D1%7D%5E%7Bn%7DF%28x_i%29$

2.如果总体X的概率密度为f(x), $eq?X_1%2CX_2...X_n$ 的概率密度

$eq?f_n%28x_1%2Cx_2...x_n%29%3D%5Cprod_%7Bi%3D1%7D%5E%7Bn%7Df%28x_i%29$

数字特征

样本均值： $eq?%5Cbar%7BX%7D%3D%5Cfrac%7B1%7D%7Bn%7D%5Csum_%7Bi%3D1%7D%5E%7Bn%7DX_i$

样本方差： $eq?S%5E2%3D%5Cfrac%7B1%7D%7Bn-1%7D%5Csum_%7Bi%3D1%7D%5E%7Bn%7D%28X_i-%5Cbar%7BX%7D%29%5E2$

样本k阶原点矩： $eq?A_k%3D%5Cfrac%7B1%7D%7Bn%7D%5Csum_%7Bi%3D1%7D%5E%7Bn%7DX_i%5Ek%2CA_1%3D%5Cbar%7BX%7D$

样本k阶中心矩： $eq?B_k%3D%5Cfrac%7B1%7D%7Bn%7D%5Csum_%7Bi%3D1%7D%5E%7Bn%7D%28X-%5Cbar%7BX%7D%29%5Ek%2CB_2%3D%5Cfrac%7Bn-1%7D%7Bn%7DS%5E2$

常用抽样分布

$eq?%5Cchi%20%5E2$ 分布

t分布

F分布

两个正态总体

$eq?X%5Csim%20N%28%5Cmu_1%2C%5Csigma_1%5E2%29%2CY%5Csim%20N%28%5Cmu_2%2C%5Csigma_2%5E2%29$ ， $eq?X_1%2CX_2...X_%7Bn_1%7D$

$eq?Y_1%2CY_2...Y_%7Bn_2%7D$ 是分别来自总体X,Y的样本且相互独立

样本均值分别为 $eq?%5Cbar%7BX%7D%2C%5Cbar%7BY%7D$ ,样本方差分别为 $eq?S_1%5E2%2CS_2%5E2$ ,有

$eq?%5Cbar%7BX%7D-%5Cbar%7BY%7D%5Csim%20N%28%5Cmu_1-%5Cmu_2%2C%5Cfrac%7B%5Csigma_1%5E2%7D%7Bn_1%7D+%5Cfrac%7B%5Csigma_2%5E2%7D%7Bn_2%7D%29$

$eq?U%3D%5Cfrac%7B%28%5Cbar%7BX%7D-%5Cbar%7BY%7D%29-%28%5Cmu_1-%5Cmu_2%29%7D%7B%5Csqrt%7B%5Cfrac%7B%5Csigma_1%5E2%7D%7Bn_1%7D+%5Cfrac%7B%5Csigma_2%5E2%7D%7Bn_2%7D%7D%7D%5Csim%20N%280%2C1%29$

如果 $eq?%5Csigma_1%5E2%3D%5Csigma_2%5E2$

$eq?U%3D%5Cfrac%7B%28%5Cbar%7BX%7D-%5Cbar%7BY%7D%29-%28%5Cmu_1-%5Cmu_2%29%7D%7B%5Csqrt%7B%5Cfrac%7B1%7D%7Bn_1%7D+%5Cfrac%7B1%7D%7Bn_2%7D%7D%7D%5Csim%20N%280%2C1%29$

$eq?S_w%3D%5Cfrac%7B%28n_1-1%29S_1%5E2+%28n_2-1%29S_2%5E2%7D%7Bn_1+n_2-2%7D$

设总体X服从分布 $eq?P%5C%7BX%3Dk%5C%7D%3Dp%5Ek%281-p%29%5E%7B1-k%7D$

$eq?X_1%2CX_2...X_n$ 是来自总体X的简单随机样本，记

$eq?Y_1%3Dmax%7BX_i%7D%2CY_2%3Dmin%7BX_j%7D%2CY_3%3DY_1-Y_2$

求 $eq?EY_3$

$eq?P%5C%7BY_3%3D0%5C%7D%3DP%5C%7BY_1%3D0%2CY_2%3D0%5C%7D+P%5C%7BY_1%3D1%2CY_2%3D1%5C%7D$

$eq?%3Dp%5En+%281-p%29%5En$

$eq?EY_3%3D1-p%5En-%281-p%29%5En$

参数估计与假设检验

点估计

设 $eq?%5Chat%7B%5Ctheta%7D$ 是 $eq?%5Ctheta$ 的估计量，如果 $eq?E%28%5Chat%5Ctheta%29%3D%5Ctheta$ ,则称

$eq?%5Chat%5Ctheta%3D%5Chat%5Ctheta%28X_1%2CX_2...X_n%29$ 是未知参数θ的无偏估计量

$eq?%5Csum_%7Bi%3D1%7D%5E%7Bn%7D%28X_i-%5Cbar%7BX%7D%29%5E2%3D%5Csum_%7Bi%3D1%7D%5E%7Bn%7DX_i%5E2-n%5Cbar%7BX%7D%5E2$

$eq?%5Csum_%7Bi%3D1%7D%5E%7Bn%7D%28X_i-%5Cbar%7BX%7D%29%5E2%3D%5Csum_%7Bi%3D1%7D%5E%7Bn%7D%28X_i-%5Cmu%29%5E2-n%28%5Cbar%20X-%5Cmu%29%5E2$

设 $eq?%5Chat%5Ctheta_1%2C%5Chat%5Ctheta_2$ 都是θ的无偏估计量，且 $eq?D%28%5Chat%5Ctheta_1%29%5Cleq%20D%28%5Chat%5Ctheta_2%29$

则称 $eq?%5Chat%5Ctheta_1$ 是比 $eq?%5Chat%5Ctheta_2$ 更有效的θ无偏估计量

矩估计

样本的k阶原点矩 $eq?%5Cfrac%7B1%7D%7Bn%7D%5Csum_%7Bi%3D1%7D%5E%7Bn%7DX_i%5Ek$ 存在

用样本k阶原点矩表示待估参数

最大似然估计

构造似然函数

（1）离散型

概率分布： $eq?P%5C%7BX%3Da_i%5C%7D%3Dp%28a_i%3B%5Ctheta%29$

$eq?L%28%5Ctheta%29%3DL%28x_1%2Cx_2...x_n%3B%5Ctheta%29%3D%5Cprod_%7Bi%3D1%7D%5E%7Bn%7Dp%28x_i%3B%5Ctheta%29$

（2）连续型

概率密度： $eq?f%28x%3B%5Ctheta%29$

$eq?L%28%5Ctheta%29%3DL%28x_1%2Cx_2...x_n%3B%5Ctheta%29%3D%5Cprod_%7Bi%3D1%7D%5E%7Bn%7Df%28x_i%3B%5Ctheta%29$

求待求参数分别求偏导令偏导方程为0

即可得到使似然函数取最大值的参数值

区间估计

待定参数	1-α置信区间
μ	$eq?%5Csigma%5E2$ 已知	$eq?%28%5Cbar%7BX%7D-u_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%5Cfrac%7B%5Csigma%7D%7B%5Csqrt%20n%7D%2C%5Cbar%7BX%7D+u_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%5Cfrac%7B%5Csigma%7D%7B%5Csqrt%20n%7D%29$
μ	$eq?%5Csigma%5E2$ 未知	$eq?%28%5Cbar%7BX%7D-t%28n-1%29_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%5Cfrac%7BS%7D%7B%5Csqrt%20n%7D%2C%5Cbar%7BX%7D+t%28n-1%29_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%5Cfrac%7BS%7D%7B%5Csqrt%20n%7D$
$eq?%5Csigma%5E2$	μ未知	$eq?%28%5Cfrac%7B%28n-1%29S%5E2%7D%7B%5Cchi%20%5E2_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%28n-1%29%7D%2C%5Cfrac%7B%28n-1%29S%5E2%7D%7B%5Cchi%20%5E2_%7B1-%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%28n-1%29%7D%29$

待定参数	1-α置信区间
$eq?%5Cmu_1-%5Cmu_2$	$eq?%5Csigma%5E2_1%2C%5Csigma%5E2_2$ 已知	$eq?%28%5Cbar%7BX%7D-%5Cbar%7BY%7D-u_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%5Csqrt%7B%5Cfrac%7B%5Csigma%5E2_1%7D%7Bn_1%7D+%5Cfrac%7B%5Csigma_2%5E2%7D%7Bn_2%7D%7D%2C%5Cbar%7BX%7D-%5Cbar%7BY%7D+u_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%5Csqrt%7B%5Cfrac%7B%5Csigma%5E2_1%7D%7Bn_1%7D+%5Cfrac%7B%5Csigma_2%5E2%7D%7Bn_2%7D%7D%29$
$eq?%5Cmu_1-%5Cmu_2$	$eq?%5Csigma%5E2_1%2C%5Csigma%5E2_2$ 未知, $eq?%5Csigma%5E2_1%3D%5Csigma%5E2_2$	$eq?%28%5Cbar%7BX%7D-%5Cbar%7BY%7D-t_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%28n_1+n_2-2%29S_w%5Csqrt%7B%5Cfrac%7B1%7D%7Bn_1%7D+%5Cfrac%7B1%7D%7Bn_2%7D%7D%2C$ $eq?%5Cbar%7BX%7D-%5Cbar%7BY%7D+t_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%28n_1+n_2-2%29S_w%5Csqrt%7B%5Cfrac%7B1%7D%7Bn_1%7D+%5Cfrac%7B1%7D%7Bn_2%7D%7D%29$
$eq?%5Cfrac%7B%5Csigma%5E2_1%7D%7B%5Csigma%5E2_2%7D$	$eq?%5Cmu_1%2C%5Cmu_2$ 未知	$eq?%28%5Cfrac%7BS_1%5E2%7D%7BS_2%5E2%7D%5Ccdot%20%5Cfrac%7B1%7D%7BF_%7B%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%28n_1-1%2Cn_2-1%29%7D%2C%5Cfrac%7BS_1%5E2%7D%7BS_2%5E2%7D%5Ccdot%20%5Cfrac%7B1%7D%7BF_%7B1-%5Cfrac%7B%5Calpha%7D%7B2%7D%7D%28n_1-1%2Cn_2-1%29%7D%29$

火舞炫风

关注

0
点赞
踩
1

收藏

觉得还不错? 一键收藏
0
评论
概率论考研复习

先求Y的分布函数甲乙两人进行射击比赛，约定甲先射若射不中，乙射，若射不中再由甲射以此类推，谁先射中谁获胜，比赛终止甲乙射中的概率分别为0.4，0.6，求甲乙获胜的概率设某时段进入某景区游客人数服从参数为的泊松分布，每位游客乘坐观光缆车的概率均为0.6，且每位游客乘坐观光缆车时=是相互独立的求进入该景区的游客乘坐观光缆车人数的概率分布及数学期望与方差设X=n的条件下，游客人数当k>n时由全概率公式得于是Y服从参数为18的泊松分布，EY=DY=18若随机变量X的概率分布为。
复制链接

扫一扫