设已知XXX的分布函数FX(x)F_X(x)FX(x)或概率密度函数fX(x)f_X(x)fX(x),则随机变量函数Y=g(X)Y=g(X)Y=g(X)的分布函数可按如下方法求得:FY(y)=P{Y⩽y}=P{g(X)⩽y}=P{X⩽h(y)}=FX(h(y))\begin{aligned} F_Y(y) & =P\{Y \leqslant y\} \\ & = P\{g(X)\leqslant y\} \\ & = P\{X \leqslant h(y)\} \\ & = F_X(h(y)) \\ \end{aligned}FY(y)=P{Y⩽y}=P{g(X)⩽y}=P{X⩽h(y)}=FX(h(y))
公式法求连续型随机变量函数的分布
设连续型随机变量XXX的密度函数为fX(x)f_X(x)fX(x),若y=g(x)y=g(x)y=g(x)是处处可导且恒有g′(x)>0g'(x) > 0g′(x)>0或g′(x)<0g'(x) < 0g′(x)<0的函数(即g(x)g(x)g(x)单调),其反函数为x=h(y)x=h(y)x=h(y),则Y=g(X)Y=g(X)Y=g(X)的概率密度函数为fY(y)=FX′(h(y))={fX(h(y))h′(y),α<y<β0,otherwisef_Y(y) =F_X'(h(y))=
\begin{cases}
f_X(h(y))h'(y), & \alpha < y < \beta \\
0, & otherwise
\end{cases}fY(y)=FX′(h(y))={fX(h(y))h′(y),0,α<y<βotherwise
其中α=min{g(−∞),g(∞)}\alpha = \min\{g(-\infty), g(\infty)\}α=min{g(−∞),g(∞)}β=max{g(−∞),g(∞)}\beta = \max\{g(-\infty), g(\infty)\}β=max{g(−∞),g(∞)}