高斯分布的全积分

高斯分布全积分：
$$\begin{array}{rl}I& ={\int }_x\frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(x-\mu {)}^2}{2{\sigma }^2}}dx={\int }_y\frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(y-\mu {)}^2}{2{\sigma }^2}}dy\\ I^2& ={\int }_x{\int }_y\frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(x-\mu {)}^2}{2{\sigma }^2}}\frac{1}{\sqrt{2\pi }\sigma }{e}^{-\frac{(y-\mu {)}^2}{2{\sigma }^2}}dxdy\\ & =\frac{1}{2\pi {\sigma }^2}{\int }_x{\int }_y{e}^{-\frac{(x-\mu {)}^2+(y-\mu {)}^2}{2{\sigma }^2}}dxdy\end{array}$$
做极坐标系变换 $x=r\cos \theta +\mu ,y=r\sin \theta +\mu$，有
$$\begin{array}{rl}{I}^2& =\frac{1}{2\pi {\sigma }^2}{\int }_x{\int }_y{e}^{-\frac{(x-\mu {)}^2+(y-\mu {)}^2}{2{\sigma }^2}}dxdy\\ & =\frac{1}{2\pi {\sigma }^2}{\int }_0^{2\pi }{\int }_0^{+\infty }{e}^{-\frac{r^2}{2{\sigma }^2}}rdrd\theta \\ & =\frac{1}{2\pi {\sigma }^2}2\pi {\sigma }^2=1\end{array}$$
又由于$I>0$，所以$I=1$。

极坐标变换涉及雅克比行列式的计算：
$$\begin{array}{r}\left|\begin{array}{cc}\cos \theta & \sin \theta \\ -r\sin \theta & r\cos \theta \end{array}\right|=r\end{array}$$

二. 多元高斯分布

多元高斯分布，详见：https://zhuanlan.zhihu.com/p/36522776

$$\begin{array}{r}{\int }_{x_1}\cdots {\int }_{x_K}\frac{1}{(2\pi {)}^{\frac{K}{2}}|\Sigma {|}^{\frac{1}{2}}}{e}^{-\frac{1}{2}(x-\mu {)}^T{\Sigma }^{-1}(x-\mu )}d{x}_1\cdots d{x}_K\end{array}$$
把矩阵写成标量积分：
$$\begin{array}{rl}& {\int }_{x_1}\cdots {\int }_{x_K}\frac{1}{(2\pi {)}^{\frac{K}{2}}|\Sigma {|}^{\frac{1}{2}}}{e}^{-\frac{1}{2}{\sum }_{i,j}{\hat{\sigma }}_{i,j}(x_i-{\mu }_i)(x_j-{\mu }_j)}d{x}_1\cdots d{x}_K\end{array}$$
这么做是一种思路，但是做不出来。

$\Sigma$为半正定对称矩阵，设${\Sigma }^{-1}=U^{T}U$, $y=U(x-{\mu }_1)$，由于线性变换矩阵就是雅克比矩阵，因此
$$d{y}_1\cdots d{y}_K=|U|d{x}_1\cdots d{x}_K$$
由$|{\Sigma }^{-1}|=|U{|}^2$，可知$|{\Sigma }^{-\frac{1}{2}}|=|\Sigma {|}^{-\frac{1}{2}}=|U|$，因此
$$\begin{array}{rl}& \frac{1}{|\Sigma {|}^{\frac{1}{2}}}{\int }_{y_1}\cdots {\int }_{y_K}\frac{1}{(2\pi {)}^{\frac{K}{2}}}{e}^{-\frac{1}{2}{y}^{T}y}|U{|}^{-1}d{y}_1\cdots d{y}_K\\ & =\frac{1}{|\Sigma {|}^{\frac{1}{2}}}{\int }_{y_1}\cdots {\int }_{y_K}\frac{1}{(2\pi {)}^{\frac{K}{2}}}{e}^{-\frac{1}{2}(y_1^2+y_2^2+\cdots +y_K^2)}|U{|}^{-1}d{y}_1\cdots d{y}_K\\ & =\frac{1}{|\Sigma {|}^{\frac{1}{2}}}|U{|}^{-1}{\int }_{y_1}\frac{1}{\sqrt{2\pi }}{e}^{-\frac{y_1^2}{2}}d{y}_1\cdots {\int }_{y_K}\frac{1}{\sqrt{2\pi }}{e}^{-\frac{y_K^2}{2}}d{y}_K\\ & =\frac{1}{|\Sigma {|}^{\frac{1}{2}}}|\Sigma {|}^{\frac{1}{2}}\cdot 1=1\end{array}$$