标准正态分布:
x
∽
N
(
0
,
1
)
:
f
(
x
)
=
1
2
π
e
−
x
2
2
x \backsim N(0,1):f(x)=\frac{1}{\sqrt{2\pi}}e^{\frac{-x^2}{2}}
x∽N(0,1):f(x)=2π1e2−x2
一般正态分布:
x
∽
N
(
μ
,
σ
2
)
:
f
(
x
)
=
1
2
π
σ
2
e
−
(
x
−
μ
)
2
2
σ
2
x \backsim N(\mu,\sigma^2):f(x)=\frac{1}{\sqrt{2\pi\sigma^2}}e^{\frac{-(x-\mu)^2}{2\sigma^2}}
x∽N(μ,σ2):f(x)=2πσ21e2σ2−(x−μ)2
多维正态分布:设
x
=
[
x
1
,
x
2
,
.
.
.
,
x
N
]
T
\boldsymbol{x}=[x_1,x_2,...,x_N]^T
x=[x1,x2,...,xN]T是一个
n
n
n维向量,第
n
n
n的维度上的随机变量
x
n
x_n
xn服从均值为
μ
n
\mu_n
μn,方差为
σ
n
2
\sigma^2_n
σn2的高斯分布,即
x
n
∽
N
(
μ
n
,
σ
n
2
)
x_n \backsim N(\mu_n,\sigma^2_n)
xn∽N(μn,σn2),且各个维度相互独立,则其联合概率密度等于各个维度上概率密度的乘积,所以
f
(
x
)
=
1
2
π
σ
1
2
e
−
(
x
1
−
μ
1
)
2
2
σ
1
2
1
2
π
σ
2
2
e
−
(
x
2
−
μ
2
)
2
2
σ
2
2
.
.
.
1
2
π
σ
N
2
e
−
(
x
N
−
μ
N
)
2
2
σ
N
2
(1)
f(\boldsymbol{x})=\frac{1}{\sqrt{2\pi\sigma^2_1}}e^{\frac{-(x_1-\mu_1)^2}{2\sigma^2_1}} \frac{1}{\sqrt{2\pi\sigma^2_2}}e^{\frac{-(x_2-\mu_2)^2}{2\sigma^2_2}}...\frac{1}{\sqrt{2\pi\sigma^2_N}}e^{\frac{-(x_N-\mu_N)^2}{2\sigma^2_N}} \tag{1}
f(x)=2πσ121e2σ12−(x1−μ1)22πσ221e2σ22−(x2−μ2)2...2πσN21e2σN2−(xN−μN)2(1)
令
Σ
=
d
i
a
g
(
σ
n
2
)
\boldsymbol{\Sigma}={\rm diag}(\sigma^2_n)
Σ=diag(σn2),
μ
=
[
μ
1
,
.
.
.
,
μ
N
]
T
\boldsymbol{\mu}=[\mu_1,...,\mu_N]^T
μ=[μ1,...,μN]T,则
x
−
μ
=
[
x
1
−
μ
1
,
.
.
.
,
x
N
−
μ
N
]
T
\boldsymbol{x}-\boldsymbol{\mu}=[x_1-\mu_1,...,x_N-\mu_N]^T
x−μ=[x1−μ1,...,xN−μN]T,则有
2
π
σ
1
2
2
π
σ
1
2
.
.
.
2
π
σ
N
2
=
(
2
π
)
N
/
2
[
d
e
t
(
Σ
)
]
1
/
2
(2)
\sqrt{2\pi\sigma^2_1}\sqrt{2\pi\sigma^2_1}...\sqrt{2\pi\sigma^2_N}=(2\pi)^{N/2}[{\rm det}(\boldsymbol{\Sigma})]^{1/2} \tag{2}
2πσ122πσ12...2πσN2=(2π)N/2[det(Σ)]1/2(2)
−
(
x
1
−
μ
1
)
2
2
σ
1
2
+
.
.
.
+
−
(
x
N
−
μ
N
)
2
2
σ
N
2
=
−
(
x
−
μ
)
T
Σ
−
1
(
x
−
μ
)
(3)
\frac{-(x_1-\mu_1)^2}{2\sigma^2_1}+...+\frac{-(x_N-\mu_N)^2}{2\sigma^2_N}=-(\boldsymbol{x}-\boldsymbol{\mu})^T \boldsymbol{\Sigma}^{-1}(\boldsymbol{x}-\boldsymbol{\mu}) \tag{3}
2σ12−(x1−μ1)2+...+2σN2−(xN−μN)2=−(x−μ)TΣ−1(x−μ)(3)
将
(
2
)
(
3
)
(2)(3)
(2)(3)代入
(
1
)
(1)
(1)可以得到
f
(
x
)
=
1
(
2
π
)
N
/
2
[
d
e
t
(
Σ
)
]
1
/
2
e
−
(
x
−
μ
)
T
Σ
−
1
(
x
−
μ
)
(4)
f(\boldsymbol{x})=\frac{1}{(2\pi)^{N/2}[{\rm det}(\boldsymbol{\Sigma})]^{1/2}}e^{-(\boldsymbol{x}-\boldsymbol{\mu})^T \boldsymbol{\Sigma}^{-1}(\boldsymbol{x}-\boldsymbol{\mu})} \tag{4}
f(x)=(2π)N/2[det(Σ)]1/21e−(x−μ)TΣ−1(x−μ)(4)
复高斯分布:设
X
∽
N
(
μ
x
,
σ
x
2
)
X \backsim N(\mu_x,\sigma^2_x)
X∽N(μx,σx2),
Y
∽
N
(
μ
y
,
σ
y
2
)
Y \backsim N(\mu_y,\sigma^2_y)
Y∽N(μy,σy2),若
X
X
X与
Y
Y
Y统计独立,则其联合概率密度函数可表示为:
f
X
Y
(
x
,
y
)
=
1
2
π
σ
x
2
e
−
(
x
−
μ
x
)
2
2
σ
x
2
1
2
π
σ
y
2
e
−
(
y
−
μ
y
)
2
2
σ
y
2
(5)
f_{XY}(x,y)=\frac{1}{\sqrt{2\pi\sigma^2_x}}e^{\frac{-(x-\mu_x)^2}{2\sigma^2_x}} \frac{1}{\sqrt{2\pi\sigma^2_y}}e^{\frac{-(y-\mu_y)^2}{2\sigma^2_y}} \tag{5}
fXY(x,y)=2πσx21e2σx2−(x−μx)22πσy21e2σy2−(y−μy)2(5)
令
μ
x
=
μ
y
\mu_x=\mu_y
μx=μy,
σ
=
σ
x
2
=
σ
y
2
\sigma=\sigma^2_x=\sigma^2_y
σ=σx2=σy2,则上式就可以化简为
f
X
Y
(
x
,
y
)
=
1
2
π
σ
2
e
−
(
x
−
μ
)
2
+
(
y
−
μ
)
2
2
σ
2
(6)
f_{XY}(x,y)=\frac{1}{2\pi\sigma^2}e^{-\frac{(x-\mu)^2+(y-\mu)^2}{2\sigma^2}} \tag{6}
fXY(x,y)=2πσ21e−2σ2(x−μ)2+(y−μ)2(6)
对应的复随机变量
Z
=
X
+
i
Y
Z=X+iY
Z=X+iY称为复高斯随机变量。
Z
Z
Z的均值
μ
z
\mu_z
μz和方差
σ
z
\sigma_z
σz分别为:
μ
z
=
E
[
Z
]
=
E
[
X
+
i
Y
]
=
E
[
X
]
+
i
E
[
Y
]
=
μ
x
+
i
μ
y
=
μ
+
i
μ
\mu_z={\rm E}[Z]={\rm E}[X+iY]={\rm E}[X]+i{\rm E}[Y]=\mu_x+i\mu_y=\mu+i\mu
μz=E[Z]=E[X+iY]=E[X]+iE[Y]=μx+iμy=μ+iμ
σ
z
=
E
[
(
Z
−
μ
z
)
(
Z
−
μ
z
)
′
]
=
E
[
∣
(
X
−
μ
x
)
+
i
(
Y
−
μ
y
)
∣
2
]
=
E
[
(
X
−
μ
x
)
2
]
+
E
[
(
Y
−
μ
y
)
2
]
=
σ
x
2
+
σ
y
2
=
2
σ
2
\sigma_z={\rm E}[(Z-\mu_z)(Z-\mu_z)']={\rm E}[|(X-\mu_x)+i(Y-\mu_y)|^2]={\rm E}[(X-\mu_x)^2]+{\rm E}[(Y-\mu_y)^2]=\sigma^2_x+\sigma^2_y=2\sigma^2
σz=E[(Z−μz)(Z−μz)′]=E[∣(X−μx)+i(Y−μy)∣2]=E[(X−μx)2]+E[(Y−μy)2]=σx2+σy2=2σ2
将
μ
z
\mu_z
μz和
σ
z
\sigma_z
σz代入
(
6
)
(6)
(6)可以得到
Z
Z
Z的概率密度函数为
f
Z
(
z
)
=
1
π
σ
z
2
e
−
(
z
−
μ
z
)
2
σ
z
2
(7)
f_Z(z)=\frac{1}{\pi\sigma_z^2}e^{-\frac{(z-\mu_z)^2}{\sigma_z^2}} \tag{7}
fZ(z)=πσz21e−σz2(z−μz)2(7)