Problem1
设投资组合 P P P是由上述 n n n个风险证券构成的一个已知投资组合，其组合比例为 X P X_P XP，收益率为 r P r_P rP，期望收益率为 E ( r P ) E\left(r_P\right) E(rP)，收益率标准差为 σ P \sigma_P σP。
a ) \quad a)\, a)试求投资组合 Q Q Q的组合比例 X Q \textbf X_Q XQ，其中 X Q \textbf X_Q XQ为满足下述模型的解：
m i n ( X Q − X P ) T V ( X Q − X P ) {\rm min}\,\left(\textbf X_Q-\textbf X_P\right)^T\textbf V\left(\textbf X_Q-\textbf X_P\right) min(XQ−XP)TV(XQ−XP)
s
.
t
.
{
(
X
Q
−
X
P
)
T
R
=
μ
X
Q
T
e
=
1
s.t.\,\begin{cases}\left(\textbf X_Q-\textbf X_P\right)^T\textbf R=\mu\\\textbf X_Q^T\textbf e=1\end{cases}
s.t.{(XQ−XP)TR=μXQTe=1
b
)
\quad b)\,
b)试求在
(
σ
2
,
E
(
r
)
)
\left(\sigma^2,\,E\left(r\right)\right)
(σ2,E(r))的坐标系中，通过坐标原点及全局最小方差组合
(
M
V
P
)
\,\left(MVP\right)\,
(MVP)的射线与上述
n
n
n个风险证券构成的投资组合边界曲线的交点
M
M
M（非全局最小方差组合）所对应的边界组合的方差
σ
M
2
\sigma_M^2
σM2、期望收益
E
(
X
M
)
E\left(X_M\right)
E(XM)和组合比例
X
M
X_M
XM.
c ) \quad c)\, c)证明在 a ) a) a)得出的解中， ( X Q − X P ) \,\left(\textbf X_Q-\textbf X_P\right)\, (XQ−XP)为 ( X M − X M V P ) \,\left(\textbf X_M-\textbf X_{MVP}\right)\, (XM−XMVP)的倍数， ( X Q − X P ) T V ( X Q − X P ) \left(\textbf X_Q-\textbf X_P\right)^T\textbf V\left(\textbf X_Q-\textbf X_P\right) (XQ−XP)TV(XQ−XP)为 ( σ M 2 − σ M V P 2 ) \left(\sigma_M^2-\sigma_{MVP}^2\right) (σM2−σMVP2)的倍数.
Solve1
a
)
\quad \textbf a)\,
a)记
L
=
(
X
Q
−
X
P
)
T
V
(
X
Q
−
X
P
)
−
λ
1
[
(
X
Q
−
X
P
)
T
R
−
μ
]
−
λ
2
(
X
Q
T
e
−
1
)
L=\left(\textbf X_Q-\textbf X_P\right)^T\textbf V\left(\textbf X_Q-\textbf X_P\right)-\lambda_1\left[\left(\textbf X_Q-\textbf X_P\right)^T\textbf R-\mu\right]-\lambda_2\left(\textbf X_Q^T\textbf e-1\right)
L=(XQ−XP)TV(XQ−XP)−λ1[(XQ−XP)TR−μ]−λ2(XQTe−1)
\quad\quad
则
⟹
{
∂
L
∂
X
=
(
∂
L
∂
X
1
,
⋯
,
∂
L
∂
X
n
)
T
=
2
V
(
X
Q
−
X
P
)
−
λ
1
R
−
λ
2
e
=
0
∂
L
∂
λ
1
=
−
[
(
X
Q
−
X
P
)
T
R
−
μ
]
=
0
∂
L
∂
λ
2
=
−
(
X
Q
T
e
−
1
)
=
0
\Longrightarrow\left\{\begin{aligned}&\frac{\partial L}{\partial\textbf X}=\left(\frac{\partial L}{\partial\textbf X_1},\cdots,\frac{\partial L}{\partial\textbf X_n}\right)^T=2\textbf V\left(\textbf X_Q-\textbf X_P\right)-\lambda_1\textbf R-\lambda_2\textbf e=\textbf 0\\&\frac{\partial L}{\partial \lambda_1}=-\left[\left(\textbf X_Q-\textbf X_P\right)^T\textbf R-\mu\right]=0\\&\frac{\partial L}{\partial \lambda_2}=-\left(\textbf X_Q^T\textbf e-1\right)=0\end{aligned}\right.
⟹⎩⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎧∂X∂L=(∂X1∂L,⋯,∂Xn∂L)T=2V(XQ−XP)−λ1R−λ2e=0∂λ1∂L=−[(XQ−XP)TR−μ]=0∂λ2∂L=−(XQTe−1)=0
\quad\quad
由第一个等式可得
X
Q
=
λ
1
2
V
−
1
R
+
λ
2
2
V
−
1
e
+
X
P
\textbf X_Q=\frac{\lambda_1}{2}\textbf V^{-1}\textbf R+\frac{\lambda_2}{2}\textbf V^{-1}\textbf e+\textbf X_P
XQ=2λ1V−1R+2λ2V−1e+XP
\quad\quad
分别带入第二个和第三个等式，得
μ
=
(
X
Q
−
X
P
)
T
R
=
(
λ
1
2
V
−
1
R
+
λ
2
2
V
−
1
e
)
T
R
=
λ
1
2
R
T
V
−
1
R
+
λ
2
2
e
T
V
−
1
R
=
λ
1
2
A
+
λ
2
2
B
\begin{aligned}\mu&=\left(\textbf X_Q-\textbf X_P\right)^T\textbf R=\left(\frac{\lambda_1}{2}\textbf V^{-1}\textbf R+\frac{\lambda_2}{2}\textbf V^{-1}\textbf e\right)^T\textbf R\\&=\frac{\lambda_1}{2}\textbf R^T\textbf V^{-1}\textbf R+\frac{\lambda_2}{2}\textbf e^T\textbf V^{-1}\textbf R\\&=\frac{\lambda_1}{2}A+\frac{\lambda_2}{2}B\end{aligned}
μ=(XQ−XP)TR=(2λ1V−1R+2λ2V−1e)TR=2λ1RTV−1R+2λ2eTV−1R=2λ1A+2λ2B
1 = X Q T e = ( λ 1 2 V − 1 R + λ 2 2 V − 1 e + X P ) T e = λ 1 2 R T V − 1 e + λ 2 2 e T V − 1 e + X P T e = λ 1 2 B + λ 2 2 C + 1 \begin{aligned}1&=\textbf X_Q^T\textbf e=\left(\frac{\lambda_1}{2}\textbf V^{-1}\textbf R+\frac{\lambda_2}{2}\textbf V^{-1}\textbf e+\textbf X_P\right)^T\textbf e\\&=\frac{\lambda_1}{2}\textbf R^T\textbf V^{-1}\textbf e+\frac{\lambda_2}{2}\textbf e^T\textbf V^{-1}\textbf e+\textbf X_P^T\textbf e\\&=\frac{\lambda_1}{2}B+\frac{\lambda_2}{2}C+1\end{aligned} 1=XQTe=(2λ1V−1R+2λ2V−1e+XP)Te=2λ1RTV−1e+2λ2eTV−1e+XPTe=2λ1B+2λ2C+1
\quad\quad
即
λ
1
2
A
+
λ
2
2
B
=
μ
\frac{\lambda_1}{2}A+\frac{\lambda_2}{2}B=\mu
2λ1A+2λ2B=μ
λ 1 2 B + λ 2 2 C = 0 \frac{\lambda_1}{2}B+\frac{\lambda_2}{2}C=0 2λ1B+2λ2C=0
\quad\quad
解得
λ
1
=
2
C
D
μ
\lambda_1=\frac{2C}D\mu
λ1=D2Cμ
λ 2 = − 2 B D μ \lambda_2=-\frac{2B}D\mu λ2=−D2Bμ
\quad\quad
故
X
Q
=
λ
1
2
V
−
1
R
+
λ
2
2
V
−
1
e
+
X
P
=
C
D
μ
V
−
1
R
−
B
D
μ
V
−
1
e
+
X
P
\begin{aligned}\textbf X_Q&=\frac{\lambda_1}{2}\textbf V^{-1}\textbf R+\frac{\lambda_2}{2}\textbf V^{-1}\textbf e+\textbf X_P\\&=\frac CD\mu\textbf V^{-1}\textbf R-\frac BD\mu\textbf V^{-1}\textbf e+\textbf X_P\end{aligned}
XQ=2λ1V−1R+2λ2V−1e+XP=DCμV−1R−DBμV−1e+XP
b
)
\quad \textbf b)\,
b)由
(
30
)
\,\left(30\right)\,
(30)，在
(
σ
2
,
E
(
r
)
)
\left(\sigma^2,\,E\left(r\right)\right)
(σ2,E(r))的坐标系中，
M
V
P
MVP\,
MVP坐标为
(
1
C
,
B
C
)
\left(\frac 1C,\,\frac{B}{C}\right)
(C1,CB)
\quad\quad
则通过坐标原点及
M
V
P
\,MVP\,
MVP的射线方程为
E
(
r
)
=
B
⋅
σ
2
E\left(r\right)=B\cdot\sigma^2
E(r)=B⋅σ2
\quad\quad
投资组合边界曲线方程为
σ
2
1
C
−
(
E
(
r
)
−
B
C
)
2
D
C
2
=
1
\frac{\sigma^2}{\frac{1}{C}}-\frac{\left(E\left(r\right)-\frac{B}{C}\right)^2}{\frac{D}{C^2}}=1
C1σ2−C2D(E(r)−CB)2=1
\quad\quad
联立上两式，得
σ
2
1
C
−
(
B
⋅
σ
2
−
B
C
)
2
D
C
2
=
1
\frac{\sigma^2}{\frac{1}{C}}-\frac{\left(B\cdot\sigma^2-\frac{B}{C}\right)^2}{\frac{D}{C^2}}=1
C1σ2−C2D(B⋅σ2−CB)2=1
\quad\quad
即
B
2
C
2
D
(
σ
2
−
1
C
)
2
=
C
(
σ
2
−
1
C
)
\frac{B^2C^2}D\left(\sigma^2-\frac{1}{C}\right)^2=C\left(\sigma^2-\frac{1}{C}\right)
DB2C2(σ2−C1)2=C(σ2−C1)
⟹ σ 2 − 1 C = D B 2 C \Longrightarrow\sigma^2-\frac{1}{C}=\frac{D}{B^2C} ⟹σ2−C1=B2CD
⟹ σ M 2 = D + B 2 B 2 C = A C B 2 C = A B 2 \Longrightarrow\sigma_M^2=\frac{D+B^2}{B^2C}=\frac{AC}{B^2C}=\frac A{B^2} ⟹σM2=B2CD+B2=B2CAC=B2A
⟹ E ( r M ) = B ⋅ σ M 2 = A B \Longrightarrow E\left(r_M\right)=B\cdot\sigma_M^2=\frac AB ⟹E(rM)=B⋅σM2=BA
\quad\quad
又由
(
25
)
\,\left(25\right)\,
(25)
X
M
=
E
(
r
M
)
C
−
B
D
V
−
1
R
−
E
(
r
M
)
B
−
A
D
V
−
1
e
=
A
B
C
−
B
D
V
−
1
R
−
A
B
B
−
A
D
V
−
1
e
=
1
B
V
−
1
R
\begin{aligned}\textbf X_M&=\frac{E\left(r_M\right)C-B}D\textbf V^{-1}\textbf R-\frac{E\left(r_M\right)B-A}D\textbf V^{-1}\textbf e\\&=\frac{\frac ABC-B}D\textbf V^{-1}\textbf R-\frac{\frac ABB-A}D\textbf V^{-1}\textbf e\\&=\frac 1B\textbf V^{-1}\textbf R\end{aligned}
XM=DE(rM)C−BV−1R−DE(rM)B−AV−1e=DBAC−BV−1R−DBAB−AV−1e=B1V−1R
\quad\quad
综上，有
σ
M
2
=
A
B
2
\sigma_M^2=\frac{A}{B^2}
σM2=B2A
E ( r M ) = A B E\left(r_M\right)=\frac{A}{B} E(rM)=BA
X
M
=
1
B
V
−
1
R
\textbf X_M=\frac 1B\textbf V^{-1}\textbf R
XM=B1V−1R
c
)
\quad\textbf c)\,
c)由
b
)
\,b)\,
b)及
(
35
)
\,\left(35\right)\,
(35)知
X
M
−
X
M
V
P
=
1
B
V
−
1
R
−
1
C
V
−
1
e
\textbf X_M-\textbf X_{MVP}=\frac 1B\textbf V^{-1}\textbf R-\frac 1C\textbf V^{-1}\textbf e
XM−XMVP=B1V−1R−C1V−1e
\quad\quad
又由
a
)
\,a)
a)得
X
Q
−
X
P
=
C
D
μ
V
−
1
R
−
B
D
μ
V
−
1
e
=
C
B
D
μ
(
1
B
V
−
1
R
−
1
C
V
−
1
e
)
=
C
B
D
μ
(
X
M
−
X
M
V
P
)
\begin{aligned}\textbf X_Q-\textbf X_P&=\frac CD\mu\textbf V^{-1}\textbf R-\frac BD\mu\textbf V^{-1}\textbf e\\&=\frac {CB}D\mu\left(\frac 1B\textbf V^{-1}\textbf R-\frac 1C\textbf V^{-1}\textbf e\right)\\&=\frac {CB}D\mu\left(\textbf X_M-\textbf X_{MVP}\right)\end{aligned}
XQ−XP=DCμV−1R−DBμV−1e=DCBμ(B1V−1R−C1V−1e)=DCBμ(XM−XMVP)
\quad\quad 即 ( X Q − X P ) \,\left(\textbf X_Q-\textbf X_P\right)\, (XQ−XP)为 ( X M − X M V P ) \,\left(\textbf X_M-\textbf X_{MVP}\right)\, (XM−XMVP)的倍数
\quad\quad\, 由 b ) \,b)\, b)及 ( 36 ) \,\left(36\right)\, (36)知
σ M 2 − σ M V P 2 = A B 2 − 1 C = A C − B 2 B 2 C = D B 2 C \sigma_M^2-\sigma_{MVP}^2=\frac A{B^2}-\frac 1C=\frac{AC-B^2}{B^2C}=\frac D{B^2C} σM2−σMVP2=B2A−C1=B2CAC−B2=B2CD
\quad\quad 又由 a ) \,a) a)得
( X Q − X P ) T V ( X Q − X P ) = ( X Q − X P ) T V C B D μ ( 1 B V − 1 R − 1 C V − 1 e ) = C B D μ ( X Q − X P ) T ( 1 B R − 1 C e ) = ( C B D μ ) 2 ( 1 B V − 1 R − 1 C V − 1 e ) T ( 1 B R − 1 C e ) = ( C B D μ ) 2 [ 1 B 2 R T V − 1 R − 1 B C ( R T V − 1 e + e T V − 1 R ) + 1 C 2 e T V − 1 e ] = ( C B D μ ) 2 ( A B 2 − 2 B B C + C C 2 ) = ( C B D μ ) 2 ⋅ A C − B 2 B 2 C = ( C B D μ ) 2 ⋅ D B 2 C = ( C B D μ ) 2 ( σ M 2 − σ M V P 2 ) = C D μ 2 \begin{aligned}\left(\textbf X_Q-\textbf X_P\right)^T\textbf V\left(\textbf X_Q-\textbf X_P\right)&=\left(\textbf X_Q-\textbf X_P\right)^T\textbf V\,\frac {CB}D\mu\left(\frac 1B\textbf V^{-1}\textbf R-\frac 1C\textbf V^{-1}\textbf e\right)\\&=\frac{CB}D\mu\left(\textbf X_Q-\textbf X_P\right)^T\left(\frac 1B\textbf R-\frac 1C\textbf e\right)\\&=\left(\frac{CB}D\mu\right)^2\left(\frac 1B\textbf V^{-1}\textbf R-\frac 1C\textbf V^{-1}\textbf e\right)^T\left(\frac 1B\textbf R-\frac 1C\textbf e\right)\\&=\left(\frac{CB}D\mu\right)^2\left[\frac 1{B^2}\textbf R^T\textbf V^{-1}\textbf R-\frac 1{BC}\left(\textbf R^T\textbf V^{-1}\textbf e+\textbf e^T\textbf V^{-1}\textbf R\right)+\frac 1{C^2}\textbf e^T\textbf V^{-1}\textbf e\right]\\&=\left(\frac{CB}D\mu\right)^2\left(\frac A{B^2}-\frac{2B}{BC}+\frac C{C^2}\right)=\left(\frac{CB}D\mu\right)^2\cdot\frac{AC-B^2}{B^2C}=\left(\frac{CB}D\mu\right)^2\cdot\frac D{B^2C}\\\\&=\left(\frac{CB}D\mu\right)^2\left(\sigma_M^2-\sigma_{MVP}^2\right)=\frac CD\mu^2\end{aligned} (XQ−XP)TV(XQ−XP)=(XQ−XP)TVDCBμ(B1V−1R−C1V−1e)=DCBμ(XQ−XP)T(B1R−C1e)=(DCBμ)2(B1V−1R−C1V−1e)T(B1R−C1e)=(DCBμ)2[B21RTV−1R−BC1(RTV−1e+eTV−1R)+C21eTV−1e]=(DCBμ)2(B2A−BC2B+C2C)=(DCBμ)2⋅B2CAC−B2=(DCBμ)2⋅B2CD=(DCBμ)2(σM2−σMVP2)=DCμ2
\quad\quad 即 ( X Q − X P ) T V ( X Q − X P ) \,\left(\textbf X_Q-\textbf X_P\right)^T\textbf V\left(\textbf X_Q-\textbf X_P\right) (XQ−XP)TV(XQ−XP)为 ( σ M 2 − σ M V P 2 ) \left(\sigma_M^2-\sigma_{MVP}^2\right) (σM2−σMVP2)的倍数
Problem2
已知风险证券 A A A、 B B B、 C C C的相关数据如下：
| 证券 | 收益率 | 期望收益率 | 标准差 |
|---|---|---|---|
| A A A | r A r_A rA | μ A \mu_A μA | σ A \sigma_A σA |
| B B B | r B r_B rB | μ B \mu_B μB | σ B \sigma_B σB |
| C C C | r C r_C rC | μ C \mu_C μC | σ C \sigma_C σC |
且三证券两两收益率的相关系数为0，现将证券 A A A、 B B B、 C C C构造投资组合，试求投资组合中 M V P \,MVP\, MVP的组合比例及期望收益与方差。
Solve2
易得
V
=
(
σ
A
2
0
0
0
σ
B
2
0
0
0
σ
C
2
)
\textbf V=\left(\begin{matrix}\sigma_A^2&0&0\\0&\sigma_B^2&0\\0&0&\sigma_C^2\end{matrix}\right)
V=⎝⎛σA2000σB2000σC2⎠⎞
R = ( μ A μ B μ C ) \textbf R=\left(\begin{matrix}\mu_A\\\mu_B\\\mu_C\end{matrix}\right) R=⎝⎛μAμBμC⎠⎞
e = ( 1 1 1 ) \textbf e=\left(\begin{matrix}1\\1\\1\\\end{matrix}\right) e=⎝⎛111⎠⎞
⟹ V − 1 = ( 1 σ A 2 0 0 0 1 σ B 2 0 0 0 1 σ C 2 ) \Longrightarrow\textbf V^{-1}=\left(\begin{matrix}\displaystyle\frac 1{\sigma_A^2}&0&0\\0&\displaystyle\frac 1{\sigma_B^2}&0\\0&0&\displaystyle\frac 1{\sigma_C^2}\end{matrix}\right) ⟹V−1=⎝⎜⎜⎜⎜⎜⎛σA21000σB21000σC21⎠⎟⎟⎟⎟⎟⎞
⟹ B = R T V − 1 e = ( μ A μ B μ C ) ( 1 σ A 2 0 0 0 1 σ B 2 0 0 0 1 σ C 2 ) ( 1 1 1 ) = ( μ A μ B μ C ) ( 1 σ A 2 1 σ B 2 1 σ C 2 ) = μ A σ A 2 + μ B σ B 2 + μ C σ C 2 C = e T V − 1 e = ( 1 1 1 ) ( 1 σ A 2 0 0 0 1 σ B 2 0 0 0 1 σ C 2 ) ( 1 1 1 ) = ( 1 1 1 ) ( 1 σ A 2 1 σ B 2 1 σ C 2 ) = 1 σ A 2 + 1 σ B 2 + 1 σ C 2 \begin{aligned}\Longrightarrow B=\textbf R^T\textbf V^{-1}\textbf e&=\left(\begin{matrix}\mu_A&\mu_B&\mu_C\end{matrix}\right)\left(\begin{matrix}\displaystyle\frac 1{\sigma_A^2}&0&0\\0&\displaystyle\frac 1{\sigma_B^2}&0\\0&0&\displaystyle\frac 1{\sigma_C^2}\end{matrix}\right)\left(\begin{matrix}1\\1\\1\\\end{matrix}\right)\\&=\left(\begin{matrix}\mu_A&\mu_B&\mu_C\end{matrix}\right)\left(\begin{matrix}\displaystyle\frac 1{\sigma_A^2}\\\\\displaystyle\frac 1{\sigma_B^2}\\\\\displaystyle\frac 1{\sigma_C^2}\end{matrix}\right)\\&=\frac{\mu_A}{\sigma_A^2}+\frac{\mu_B}{\sigma_B^2}+\frac{\mu_C}{\sigma_C^2}\\\\C=\textbf e^T\textbf V^{-1}\textbf e&=\left(\begin{matrix}1&1&1\end{matrix}\right)\left(\begin{matrix}\displaystyle\frac 1{\sigma_A^2}&0&0\\0&\displaystyle\frac 1{\sigma_B^2}&0\\0&0&\displaystyle\frac 1{\sigma_C^2}\end{matrix}\right)\left(\begin{matrix}1\\1\\1\\\end{matrix}\right)\\&=\left(\begin{matrix}1&1&1\end{matrix}\right)\left(\begin{matrix}\displaystyle\frac 1{\sigma_A^2}\\\\\displaystyle\frac 1{\sigma_B^2}\\\\\displaystyle\frac 1{\sigma_C^2}\end{matrix}\right)\\&=\frac 1{\sigma_A^2}+\frac 1{\sigma_B^2}+\frac 1{\sigma_C^2}\end{aligned} ⟹B=RTV−1eC=eTV−1e=(μAμBμC)⎝⎜⎜⎜⎜⎜⎛σA21000σB21000σC21⎠⎟⎟⎟⎟⎟⎞⎝⎛111⎠⎞=(μAμBμC)⎝⎜⎜⎜⎜⎜⎜⎜⎜⎜⎛σA21σB21σC21⎠⎟⎟⎟⎟⎟⎟⎟⎟⎟⎞=σA2μA+σB2μB+σC2μC=(111)⎝⎜⎜⎜⎜⎜⎛σA21000σB21000σC21⎠⎟⎟⎟⎟⎟⎞⎝⎛111⎠⎞=(111)⎝⎜⎜⎜⎜⎜⎜⎜⎜⎜⎛σA21σB21σC21⎠⎟⎟⎟⎟⎟⎟⎟⎟⎟⎞=σA21+σB21+σC21
则由
(
35
)
∼
(
37
)
\,(35)\sim(37)
(35)∼(37)，
M
V
P
MVP\,
MVP期望收益为
E
(
r
)
=
B
C
=
μ
A
σ
A
2
+
μ
B
σ
B
2
+
μ
C
σ
C
2
1
σ
A
2
+
1
σ
B
2
+
1
σ
C
2
E\left(r\right)=\frac BC=\frac{\displaystyle\frac{\mu_A}{\sigma_A^2}+\frac{\mu_B}{\sigma_B^2}+\frac{\mu_C}{\sigma_C^2}}{\displaystyle\frac 1{\sigma_A^2}+\frac 1{\sigma_B^2}+\frac 1{\sigma_C^2}}
E(r)=CB=σA21+σB21+σC21σA2μA+σB2μB+σC2μC
方差为
σ
2
=
1
C
=
1
1
σ
A
2
+
1
σ
B
2
+
1
σ
C
2
\sigma^2=\frac 1C=\frac 1{\displaystyle\frac 1{\sigma_A^2}+\frac 1{\sigma_B^2}+\frac 1{\sigma_C^2}}
σ2=C1=σA21+σB21+σC211
组合比例为
X
=
V
−
1
e
C
=
(
1
σ
A
2
0
0
0
1
σ
B
2
0
0
0
1
σ
C
2
)
(
1
1
1
)
C
=
(
1
σ
A
2
C
1
σ
B
2
C
1
σ
C
2
C
)
=
(
1
σ
A
2
1
σ
A
2
+
1
σ
B
2
+
1
σ
C
2
1
σ
B
2
1
σ
A
2
+
1
σ
B
2
+
1
σ
C
2
1
σ
C
2
1
σ
A
2
+
1
σ
B
2
+
1
σ
C
2
)
T
\begin{aligned}\textbf X=\frac{\textbf V^{-1}\textbf e}C&=\frac{\left(\begin{matrix}\displaystyle\frac 1{\sigma_A^2}&0&0\\0&\displaystyle\frac 1{\sigma_B^2}&0\\0&0&\displaystyle\frac 1{\sigma_C^2}\end{matrix}\right)\left(\begin{matrix}1\\1\\1\\\end{matrix}\right)}C=\left(\begin{matrix}\displaystyle\frac{1}{\sigma_A^2 C}\\\\\displaystyle\frac{1}{\sigma_B^2 C}\\\\\displaystyle\frac{1}{\sigma_C^2 C}\end{matrix}\right)\\\\&=\left(\begin{matrix}\displaystyle\frac {\displaystyle\frac{1}{\sigma_A^2}}{\displaystyle\frac 1{\sigma_A^2}+\frac 1{\sigma_B^2}+\frac 1{\sigma_C^2}}&\displaystyle\frac {\displaystyle\frac{1}{\sigma_B^2}}{\displaystyle\frac 1{\sigma_A^2}+\frac 1{\sigma_B^2}+\frac 1{\sigma_C^2}}&\displaystyle\frac {\displaystyle\frac{1}{\sigma_C^2}}{\displaystyle\frac 1{\sigma_A^2}+\frac 1{\sigma_B^2}+\frac 1{\sigma_C^2}}\end{matrix}\right)^T\end{aligned}
X=CV−1e=C⎝⎜⎜⎜⎜⎜⎛σA21000σB21000σC21⎠⎟⎟⎟⎟⎟⎞⎝⎛111⎠⎞=⎝⎜⎜⎜⎜⎜⎜⎜⎜⎜⎛σA2C1σB2C1σC2C1⎠⎟⎟⎟⎟⎟⎟⎟⎟⎟⎞=⎝⎜⎜⎛σA21+σB21+σC21σA21σA21+σB21+σC21σB21σA21+σB21+σC21σC21⎠⎟⎟⎞T
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
