【学习笔记】数理统计习题二

Q1: Assume that $X_1\sim Gamma({\alpha }_1,\lambda )$, $X_2\sim Gamma({\alpha }_2,\lambda )$, and they are independent. Prove that:

1. $Y_1=X_1+X_2\sim Gamma({\alpha }_1+{\alpha }_2,\lambda )$,

2. $Y_2=X_1/(X_1+X_2)\sim Beta({\alpha }_1,{\alpha }_2)$,

3. $Y_1$ and $Y_2$ are independent.

Solution: 已知$X_1$ 与 $X_2$独立，故可得联合密度函数：
$$f(x_1,x_2)=f_{X_1}(x_1)f_{X_2}(x_2)=\frac{{\lambda }^{{\alpha }_1}}{\Gamma ({\alpha }_1)}{x}_1^{{\alpha }_1-1}{e}^{-\lambda x_1}\frac{{\lambda }^{{\alpha }_2}}{\Gamma ({\alpha }_2)}{x}_2^{{\alpha }_2-1}{e}^{-\lambda x_2}$$

令 $Y_1=X_1+X_2,Y_2=\frac{X_1}{X_1+X_2}$,则可推得$X_1=Y_1{Y}_2,X_2=Y_1(1-Y_2)$,由此得
$$J=\frac{\partial (X_1,X_2)}{\partial (Y_1,Y_2)}=\left|\begin{array}{cc}{Y}_2& Y_1\\ 1-Y_2& -Y_1\end{array}\right|=-Y_1,|J|=Y_1$$
于是得联合密度函数
$$\begin{array}{rl}q(y_1,y_2)& =f(x_1,x_2)|J|\\ & =\frac{{\lambda }^{{\alpha }_1+{\alpha }_2}}{\Gamma ({\alpha }_1)\Gamma ({\alpha }_2)}{y}_1^{{\alpha }_1-1}{y}_2^{{\alpha }_1-1}{y}_1^{{\alpha }_2-1}(1-y_2{)}^{{\alpha }_1-1}{e}^{-\lambda y_1}{y}_1\\ & =\frac{{\lambda }^{{\alpha }_1+{\alpha }_2}}{\Gamma ({\alpha }_1+{\alpha }_2)}{y}_1^{{\alpha }^{}}\\ & \end{array}$$