UA MATH564 概率论 多元随机变量的变换 理论与应用2
UA MATH564 概率論 多元隨機變量的變換 幾個例題
例5 X1,X2,X3~iidEXP(λ)X_1,X_2,X_3 \sim_{iid} EXP(\lambda)X1?,X2?,X3?~iid?EXP(λ),Y1=X1+X2+X3,Y2=X1X1+X2,Y3=X1+X2X1+X2+X3Y_1 = X_1+X_2+X_3,Y_2 = \frac{X_1}{X_1+X_2},Y_3 = \frac{X_1+X_2}{X_1+X_2+X_3}Y1?=X1?+X2?+X3?,Y2?=X1?+X2?X1??,Y3?=X1?+X2?+X3?X1?+X2??,求Y1,Y2,Y3Y_1,Y_2,Y_3Y1?,Y2?,Y3?的聯合概率密度,并判斷他們是否互相獨立。
先把X1,X2,X3X_1,X_2,X_3X1?,X2?,X3?表示成Y1,Y2,Y3Y_1,Y_2,Y_3Y1?,Y2?,Y3?的函數,
Y3=X1+X2X1+X2+X3=X1+X2Y1?X1+X2=Y1Y3Y2=X1X1+X2=X1Y1Y3?X1=Y1Y2Y3Y_3 = \frac{X_1+X_2}{X_1+X_2+X_3} = \frac{X_1+X_2}{Y_1} \Rightarrow X_1+X_2 = Y_1Y_3 \\ Y_2 = \frac{X_1}{X_1+X_2} = \frac{X_1}{Y_1Y_3} \Rightarrow X_1 = Y_1Y_2Y_3Y3?=X1?+X2?+X3?X1?+X2??=Y1?X1?+X2???X1?+X2?=Y1?Y3?Y2?=X1?+X2?X1??=Y1?Y3?X1???X1?=Y1?Y2?Y3?
因此X2=Y1Y3(1?Y2),X3=Y1(1?Y3)X_2 = Y_1Y_3(1-Y_2),X_3 = Y_1(1-Y_3)X2?=Y1?Y3?(1?Y2?),X3?=Y1?(1?Y3?),
?(X1,X2,X3)?(Y1,Y2,Y3)=∣Y2Y3Y3(1?Y2)1?Y3Y1Y3?Y1Y30Y1Y2Y1(1?Y2)?Y1∣=(1?Y3)[Y12(1?Y2)Y3+Y12Y2Y3]?Y1[?Y1Y2Y32?Y1(1?Y2)Y32]=(1?Y3)Y12Y3+Y12Y32=Y12\frac{\partial (X_1,X_2,X_3)}{\partial (Y_1,Y_2,Y_3)} = \left| \begin{matrix}Y_2Y_3 & Y_3(1-Y_2) & 1-Y_3 \\ Y_1Y_3 & -Y_1Y_3 & 0 \\ Y_1Y_2 & Y_1(1-Y_2) &-Y_1 \end{matrix}\right| \\ = (1-Y_3)[Y_1^2(1-Y_2)Y_3+Y_1^2Y_2Y_3]-Y_1[-Y_1Y_2Y_3^2-Y_1(1-Y_2)Y_3^2] \\ = (1-Y_3)Y_1^2Y_3+Y_1^2Y_3^2 = Y_1^2?(Y1?,Y2?,Y3?)?(X1?,X2?,X3?)?=∣∣∣∣∣∣?Y2?Y3?Y1?Y3?Y1?Y2??Y3?(1?Y2?)?Y1?Y3?Y1?(1?Y2?)?1?Y3?0?Y1??∣∣∣∣∣∣?=(1?Y3?)[Y12?(1?Y2?)Y3?+Y12?Y2?Y3?]?Y1?[?Y1?Y2?Y32??Y1?(1?Y2?)Y32?]=(1?Y3?)Y12?Y3?+Y12?Y32?=Y12?
X1,X2,X3X_1,X_2,X_3X1?,X2?,X3?的聯合概率密度是
fX1,X2,X3(x1,x2,x3)=λ3e?λ(x1+x2+x3),x1,x2,x3>0f_{X_1,X_2,X_3}(x_1,x_2,x_3) = \lambda^3e^{-\lambda(x_1+x_2+x_3)},x_1,x_2,x_3>0fX1?,X2?,X3??(x1?,x2?,x3?)=λ3e?λ(x1?+x2?+x3?),x1?,x2?,x3?>0
因此Y1,Y2,Y3Y_1,Y_2,Y_3Y1?,Y2?,Y3?的聯合概率密度是
fY1,Y2,Y3(y1,y2,y3)=λ3y12e?λy1,y1>0f_{Y_1,Y_2,Y_3}(y_1,y_2,y_3) = \lambda^3 y_1^2 e^{-\lambda y_1},y_1>0fY1?,Y2?,Y3??(y1?,y2?,y3?)=λ3y12?e?λy1?,y1?>0
顯然Y1,Y2,Y3Y_1,Y_2,Y_3Y1?,Y2?,Y3?獨立。
例5 續 Z1=X1/Y1Z_1 = X_1/Y_1Z1?=X1?/Y1?,Z2=X2/Y1Z_2 = X_2/Y_1Z2?=X2?/Y1?,求Z1,Z2Z_1,Z_2Z1?,Z2?的聯合概率密度。
先對Z1,Z2Z_1,Z_2Z1?,Z2?變形一下,
Z1=X1Y1=Y1Y2Y3Y1=Y2Y3Z2=X2Y1=Y1Y3(1?Y2)Y1=Y3(1?Y2)Z_1 = \frac{X_1}{Y_1} = \frac{Y_1Y_2Y_3}{Y_1} = Y_2Y_3 \\ Z_2 = \frac{X_2}{Y_1} = \frac{Y_1Y_3(1-Y_2)}{Y_1} = Y_3(1 - Y_2)Z1?=Y1?X1??=Y1?Y1?Y2?Y3??=Y2?Y3?Z2?=Y1?X2??=Y1?Y1?Y3?(1?Y2?)?=Y3?(1?Y2?)
再把Y2,Y3Y_2,Y_3Y2?,Y3?表示為Z1,Z2Z_1,Z_2Z1?,Z2?的函數
1?Y2Y2=1Y2?1=Z2Z1?Y2=Z1Z1+Z2\frac{1-Y_2}{Y_2} = \frac{1}{Y_2}-1= \frac{Z_2}{Z_1} \Rightarrow Y_2 = \frac{Z_1}{Z_1+Z_2}Y2?1?Y2??=Y2?1??1=Z1?Z2???Y2?=Z1?+Z2?Z1??
所以Y3=Z1+Z2Y_3 = Z_1+Z_2Y3?=Z1?+Z2?,計算Jacobi行列式
?(Y2,Y3)?(Z1,Z2)=∣Z2(Z1+Z2)21?Z1(Z1+Z2)21∣=1Z1+Z2\frac{\partial (Y_2,Y_3)}{\partial (Z_1,Z_2)} = \left| \begin{matrix} \frac{Z_2}{(Z_1+Z_2)^2}&1\\-\frac{Z_1}{(Z_1+Z_2)^2} & 1 \end{matrix} \right| = \frac{1}{Z_1+Z_2}?(Z1?,Z2?)?(Y2?,Y3?)?=∣∣∣∣∣?(Z1?+Z2?)2Z2???(Z1?+Z2?)2Z1???11?∣∣∣∣∣?=Z1?+Z2?1?
Y2,Y3Y_2,Y_3Y2?,Y3?的聯合概率密度為
fY2,Y3(y2,y3)=∫0∞λ3y12e?λy1dy1=?∫0∞λ2y12e?λy1d(?λy1)=?λ2y12e?λy1∣0∞+∫0∞2λ2y1e?λy1dy1=?∫0∞2λy1e?λy1d(?λy1)=?2λy1e?λy1∣0∞+∫0∞2λe?λy1dy1=?2e?λy1∣0∞=2f_{Y_2,Y_3}(y_2,y_3) = \int_{0}^{\infty} \lambda^3 y_1^2 e^{-\lambda y_1}dy_1 =-\int_{0}^{\infty} \lambda^2 y_1^2 e^{-\lambda y_1}d(-\lambda y_1) \\ = -\lambda^2 y_1^2 e^{-\lambda y_1}|_0^{\infty} + \int_0^{\infty} 2\lambda^2 y_1 e^{-\lambda y_1} dy_1 = -\int_0^{\infty} 2\lambda y_1 e^{-\lambda y_1} d(-\lambda y_1) \\ = -2\lambda y_1e^{-\lambda y_1}|_0^{\infty} + \int_{0}^{\infty} 2\lambda e^{-\lambda y_1}dy_1 = -2e^{-\lambda y_1}|_0^{\infty} = 2fY2?,Y3??(y2?,y3?)=∫0∞?λ3y12?e?λy1?dy1?=?∫0∞?λ2y12?e?λy1?d(?λy1?)=?λ2y12?e?λy1?∣0∞?+∫0∞?2λ2y1?e?λy1?dy1?=?∫0∞?2λy1?e?λy1?d(?λy1?)=?2λy1?e?λy1?∣0∞?+∫0∞?2λe?λy1?dy1?=?2e?λy1?∣0∞?=2
因此Z1,Z2Z_1,Z_2Z1?,Z2?的聯合概率密度是
fZ1,Z2(z1,z2)=2z1+z2,z1,z2∈(0,1)f_{Z_1,Z_2}(z_1,z_2) = \frac{2}{z_1+z_2},z_1,z_2 \in (0,1)fZ1?,Z2??(z1?,z2?)=z1?+z2?2?,z1?,z2?∈(0,1)
總結
以上是生活随笔為你收集整理的UA MATH564 概率论 多元随机变量的变换 理论与应用2的全部內容,希望文章能夠幫你解決所遇到的問題。
- 上一篇: UA PHYS515 电磁理论II 静电
- 下一篇: UA SIE545 优化理论基础1 例题