用子样均值取代期望时,证明所构成的卡方分布的自由度为(n-1)
若子样 x i x_i xi~ N ( μ , σ 2 ) N(\mu,\sigma^2) N(μ,σ2),且期望 μ \mu μ用子样均值 x ˉ \bar x xˉ取代时,会有子样无偏方差:
σ ^ 2 = 1 n − 1 ∑ i = 1 n ( x i − x ˉ ) 2 = 1 n − 1 V V T \hat{\sigma}^2=\frac{1}{n-1}\sum_{i=1}^{n}{(x_i-\bar x)^2}=\frac{1}{n-1}VV^T σ^2=n−11∑i=1n(xi−xˉ)2=n−11VVT则:
V V T σ 2 = ( n − 1 ) σ ^ 2 σ 2 = ∑ i = 1 n ( x i − x ˉ ) 2 σ 2 ∼ χ 2 ( n − 1 ) \frac{VV^T}{\sigma^2}=\frac{(n-1)\hat{\sigma}^2}{\sigma^2}=\sum_{i=1}^{n}{\frac{(x_i-\bar x)^2}{\sigma^2}}\thicksim\chi^2(n-1) σ2VVT=σ2(n−1)σ^2=∑i=1nσ2(xi−xˉ)2∼χ2(n−1)
∑ i = 1 n ( x i − x ˉ ) 2 σ 2 = 1 σ 2 ( ∑ i = 1 n x i 2 − n x ˉ 2 ) \sum_{i=1}^{n}{\frac{(x_i-\bar x)^2}{\sigma^2}}=\frac{1}{\sigma^2}\begin{pmatrix} \sum_{i=1}^{n}{x_i^2-n\bar x^2} \end{pmatrix} ∑i=1nσ2(xi−xˉ)2=σ21(∑i=1nxi2−nxˉ2)
证明上式:
令
Y
=
[
y
1
⋯
y
n
−
1
y
n
]
=
[
1
1
∗
2
−
1
1
∗
2
0
⋯
0
0
⋯
⋯
⋯
⋯
⋯
⋯
1
n
∗
(
n
−
1
)
1
n
∗
(
n
−
1
)
1
n
∗
(
n
−
1
)
1
n
∗
(
n
−
1
)
1
n
∗
(
n
−
1
)
−
(
n
−
1
)
n
∗
(
n
−
1
)
1
n
1
n
1
n
1
n
1
n
1
n
]
Y=\begin{bmatrix} y_1 \\\cdots \\ y_{n-1}\\ y_{n}\end{bmatrix}=\begin{bmatrix} \frac{1}{\sqrt{1*2}} & \frac{-1}{\sqrt{1*2}} & 0 & \cdots & 0 & 0 \\ \cdots & \cdots & \cdots & \cdots & \cdots & \cdots \\ \frac{1}{\sqrt{n*(n-1)}} & \frac{1}{\sqrt{n*(n-1)}} & \frac{1}{\sqrt{n*(n-1)}} & \frac{1}{\sqrt{n*(n-1)}} & \frac{1}{\sqrt{n*(n-1)}} & \frac{-(n-1)}{\sqrt{n*(n-1)}} \\ \frac{1}{\sqrt{n}} & \frac{1}{\sqrt{n}} & \frac{1}{\sqrt{n}} & \frac{1}{\sqrt{n}} & \frac{1}{\sqrt{n}} & \frac{1}{\sqrt{n}} \end{bmatrix}
Y=⎣⎢⎢⎡y1⋯yn−1yn⎦⎥⎥⎤=⎣⎢⎢⎢⎡1∗21⋯n∗(n−1)1n11∗2−1⋯n∗(n−1)1n10⋯n∗(n−1)1n1⋯⋯n∗(n−1)1n10⋯n∗(n−1)1n10⋯n∗(n−1)−(n−1)n1⎦⎥⎥⎥⎤
X
X
X
上述
X
X
X左乘的系数矩阵设为
A
A
A矩阵,
A
A
A为正交阵,满足
A
T
A
=
I
A^TA=I
ATA=I,运用误差传播定律可知,,
y
i
y_i
yi~
N
(
0
,
σ
2
)
N(0,\sigma^2)
N(0,σ2),且互独立,
i
=
1
,
2
,
⋯
,
n
−
1
i=1,2,\cdots,n-1
i=1,2,⋯,n−1,且
y
n
∼
N
(
n
μ
,
σ
2
)
y_n \thicksim N(\sqrt{n}\mu,\sigma^2)
yn∼N(nμ,σ2),
y
n
2
=
n
x
ˉ
2
y_n^2=n\bar x^2
yn2=nxˉ2
∑ i = 1 n x 2 = X T X \sum_{i=1}^{n} {x^2}=X^TX ∑i=1nx2=XTX
Y T Y = ( A X ) T ( A X ) = X T A A X = X T X = ∑ i = 1 n x 2 = ∑ i = 1 n y 2 Y^TY=(AX)^T(AX)=X^TAAX=X^TX=\sum_{i=1}^{n} {x^2}=\sum_{i=1}^{n} {y^2} YTY=(AX)T(AX)=XTAAX=XTX=∑i=1nx2=∑i=1ny2
V V T σ 2 = 1 σ 2 ∑ i = 1 n ( x i 2 − n x ˉ 2 ) = 1 σ 2 ∑ i = 1 n ( y i 2 − n x ˉ 2 ) = 1 σ 2 ∑ i = 1 n ( y i 2 − y n 2 ) = 1 σ 2 ∑ i = 1 n − 1 ( y i 2 ) ∼ χ 2 ( n − 1 ) \frac{VV^T}{\sigma^2}=\frac{1}{\sigma^2}\sum_{i=1}^{n}{(x_i^2-n\bar x^2)}=\frac{1}{\sigma^2}\sum_{i=1}^{n}{(y_i^2-n\bar x^2)}=\frac{1}{\sigma^2}\sum_{i=1}^{n}{(y_i^2-y_n^2)}=\frac{1}{\sigma^2}\sum_{i=1}^{n-1}{(y_i^2)}\thicksim \chi^2(n-1) σ2VVT=σ21∑i=1n(xi2−nxˉ2)=σ21∑i=1n(yi2−nxˉ2)=σ21∑i=1n(yi2−yn2)=σ21∑i=1n−1(yi2)∼χ2(n−1)
完美得证!
这是我大二时学习概率论时的疑惑,以上证明过程用到了测绘专业的知识证明,下午看到时醍醐灌顶,特此记录
参考资料:
(1)王穗辉老师的测量平差ppt
(2)第一次用Letax编辑公式,参考了这个博客 https://blog.csdn.net/weixin_44244154/article/details/104722239