前言
博客《一种线性方程约束下生成随机数修正的一般性方法(上)》、《一种线性方程约束下生成随机数修正的一般性方法(中)》、《一种线性方程约束下生成随机数修正的一般性方法(下)》给出了生成随机数不满足线性方程约束时的修正公式,该修正公式既适用于单线性方程约束,也适用于多线性方程约束,可以作为线性方程约束下生成随机数修正的一般性方法。但上述博客未能判断修正后的结果是否满足变量上下限约束和线性方程约束,即未能完成对修正方法所得结果的可行性分析。为进一步完善生成随机数修正方法的理论框架,本系列博客将首先给出单线性方程约束问题可行性定理和多线性方程约束问题可行性定理,作为分别判断线性方程约束问题的充分必要条件,然后利用上述定理研究修正结果与线性方程约束问题可行性之间的关系,通过修正结果的可行性判断线性方程约束问题的可行性,并基此设计线性约束下生成随机数修正方法的可执行程序。
可行性分析系列博客:
1. 《线性方程约束下修正方法的可行性分析(演绎推理,一)》
2. 《线性方程约束下修正方法的可行性分析(演绎推理,二)》
3. 《线性方程约束下修正方法的可行性分析(演绎推理,三)》
4. 《线性方程约束下修正方法的可行性分析(演绎推理,四)》
5. 《线性方程约束下修正方法的可行性分析(演绎推理,五)》
6. 《线性方程约束下修正方法的可行性分析(演绎推理,六)》
7. 《线性方程约束下修正方法的可行性分析(程序设计)》
8. 《线性方程约束下修正方法的可行性分析(实例应用)》
单线性方程约束下修正方法的可行性分析
单线性方程约束对应于博客《一种线性方程约束下生成随机数修正的一般性方法(上)》中的情况2.2.1,首先需要求证单线性方程约束问题可行性定理,并结合利用博客《一种线性方程约束下生成随机数修正的一般性方法(上)》所提修正方法得到的修正结果取值范围,探究其与单线性方程约束问题可行性定理之间的联系,进而达到通过修正解的结果即可判定方程可行性的目的。
单线性方程约束问题可行性定理
定理1:对于博客《一种线性方程约束下生成随机数修正的一般性方法(上)》式(1)所示问题,若经系数矩阵行最简形变换后的常数向量 b R R E F \boldsymbol{b_{RREF}} bRREF自第 ( R + 1 ) \left( R + 1 \right) (R+1)列到第 M M M列不存在非零元素,且线性方程组对应系数矩阵的秩 R = 1 R = 1 R=1,则问题有解的充分必要条件为 ∑ j = 1 , a 1 , j > 0 N a 1 , j x ‾ j + ∑ j = 1 , a 1 , j < 0 N a 1 , j x ‾ j ≤ b 1 ≤ ∑ j = 1 , a 1 , j > 0 N a 1 , j x ‾ j + ∑ j = 1 , a 1 , j < 0 N a 1 , j x ‾ j \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j ≤ b_1 ≤ \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \underline{x}_j j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj≤b1≤j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj。
证明1:首先求证必要性,由于博客《一种线性方程约束下生成随机数修正的一般性方法(上)》式(1)所示问题中,变量
x
j
x_j
xj存在上下界约束
x
j
∈
[
x
‾
j
,
x
‾
j
]
,
j
∈
{
1
,
⋯
,
N
}
x_j ∈ \left[ \underline{x}_j, \overline{x}_j \right], j ∈ \{ 1, \cdots, N \}
xj∈[xj,xj],j∈{1,⋯,N},因此欲使该问题有解,则所求的
x
j
x_j
xj必须在上下界范围内,将
x
j
x_j
xj乘以方程中对应的系数
a
1
,
j
a_{1, j}
a1,j,不难得到
a
1
,
j
x
j
a_{1, j} x_j
a1,jxj的取值范围为
a
1
,
j
x
‾
j
(
a
1
,
j
>
0
)
0
(
a
1
,
j
=
0
)
a
1
,
j
x
‾
j
(
a
1
,
j
<
0
)
}
≤
a
1
,
j
x
j
≤
{
a
1
,
j
x
‾
j
(
a
1
,
j
>
0
)
0
(
a
1
,
j
=
0
)
a
1
,
j
x
‾
j
(
a
1
,
j
0
)
\begin{equation} \begin{rcases} a_{1, j} \underline{x}_j & \left( a_{1, j} > 0 \right) \\ 0 & \left( a_{1, j} = 0 \right) \\ a_{1, j} \overline{x}_j & \left( a_{1, j} < 0 \right) \end{rcases} ≤ a_{1, j} x_j ≤ \begin{cases} a_{1, j} \overline{x}_j & \left( a_{1, j} > 0 \right) \\ 0 & \left( a_{1, j} = 0 \right) \\ a_{1, j}\underline{x}_j & \left( a_{1, j} 0 \right) \\ \end{cases} \end{equation}
a1,jxj0a1,jxj(a1,j>0)(a1,j=0)(a1,j<0)⎭
⎬
⎫≤a1,jxj≤⎩
⎨
⎧a1,jxj0a1,jxj(a1,j>0)(a1,j=0)(a1,j0)
将
a
1
,
j
x
j
a_{1, j} x_j
a1,jxj从
j
=
1
j = 1
j=1到
j
=
N
j = N
j=N进行求和可得
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
≤
∑
j
=
1
N
a
1
,
j
x
j
=
b
1
≤
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
\begin{equation} \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j ≤ \displaystyle \sum_{j = 1}^{N} a_{1, j} x_j = b_1 ≤ \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \underline{x}_j \end{equation}
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj≤j=1∑Na1,jxj=b1≤j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
此时必要性可证。
然后求证充分性,由于该问题中的系数矩阵行最简形变换后的常数向量
b
R
R
E
F
\boldsymbol{b_{RREF}}
bRREF自第
(
R
+
1
)
\left( R + 1 \right)
(R+1)列到第
M
M
M列不存在非零元素,且线性方程组对应系数矩阵的秩
R
=
1
R = 1
R=1,因此根据线性方程组理论可知,该问题在
N
N
N维实数空间内必然有解,此时需求证
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
≤
b
1
≤
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
\displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j ≤ b_1 ≤ \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \underline{x}_j
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj≤b1≤j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj时,必能得到解向量
x
=
[
x
1
⋯
x
j
⋯
x
N
]
T
\boldsymbol{x} = \left[ \begin{matrix} x_1 & \cdots & x_j & \cdots & x_N \end{matrix} \right]^T
x=[x1⋯xj⋯xN]T,使得对于
∀
j
∈
{
1
,
⋯
,
N
}
\forall j ∈ \{ 1, \cdots, N\}
∀j∈{1,⋯,N},都有
x
j
∈
[
x
‾
j
,
x
‾
j
]
x_j ∈ \left[ \underline{x}_j, \overline{x}_j \right]
xj∈[xj,xj]。
以下采用数学归纳法进行进一步证明:
当
N
=
1
N = 1
N=1时,由于
R
=
1
R = 1
R=1,因此
a
1
,
1
≠
0
a_{1, 1} \neq 0
a1,1=0。若
a
1
,
1
>
0
a_{1, 1} > 0
a1,1>0,则此时
a
1
,
1
x
‾
1
≤
a
1
,
1
x
1
=
b
1
≤
a
1
,
1
x
‾
1
a_{1, 1} \underline{x}_1 ≤ a_{1, 1} x_1 = b_1 ≤ a_{1, 1} \overline{x}_1
a1,1x1≤a1,1x1=b1≤a1,1x1;若
a
1
,
1
<
0
a_{1, 1} < 0
a1,1<0,则此时
a
1
,
1
x
‾
1
≤
a
1
,
1
x
1
=
b
1
≤
a
1
,
1
x
‾
1
a_{1, 1} \overline{x}_1 ≤ a_{1, 1} x_1 = b_1 ≤ a_{1, 1} \underline{x}_1
a1,1x1≤a1,1x1=b1≤a1,1x1。两种情况均可求得
x
‾
1
≤
x
1
=
b
1
a
1
,
1
≤
x
‾
1
\underline{x}_1 ≤ x_1 = \dfrac{b_1} {a_{1, 1}} ≤ \overline{x}_1
x1≤x1=a1,1b1≤x1,故必能在
[
x
‾
1
,
x
‾
1
]
\left[ \underline{x}_1, \overline{x}_1 \right]
[x1,x1]上求得
x
1
=
b
1
a
1
,
1
x_1 = \dfrac{b_1} {a_{1, 1}}
x1=a1,1b1,使得待求量既满足该问题中的所有约束条件,问题的充分性成立。
假设当
N
=
K
(
K
>
1
)
N = K \left( K > 1 \right)
N=K(K>1)时,命题的充分性成立。则当
N
=
K
+
1
N = K + 1
N=K+1时,有
∑
j
=
1
,
a
1
,
j
>
0
K
+
1
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
+
1
a
1
,
j
x
‾
j
≤
∑
j
=
1
K
+
1
a
1
,
j
x
j
=
b
1
≤
∑
j
=
1
,
a
1
,
j
>
0
K
+
1
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
+
1
a
1
,
j
x
‾
j
\begin{equation} \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K + 1} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{K + 1} a_{1, j} \overline{x}_j ≤ \displaystyle \sum_{j = 1}^{K + 1} a_{1, j} x_j = b_1 ≤ \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K + 1} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K + 1} a_{1, j} \underline{x}_j \end{equation}
j=1,a1,j>0∑K+1a1,jxj+j=1,a1,j<0∑K+1a1,jxj≤j=1∑K+1a1,jxj=b1≤j=1,a1,j>0∑K+1a1,jxj+j=1,a1,j<0∑K+1a1,jxj
将式(3)所示和式方程部分分解为前
K
K
K项和式与第
(
K
+
1
)
\left( K + 1 \right)
(K+1)项相加的形式,则式(3)变为
∑
j
=
1
K
a
1
,
j
x
j
+
a
1
,
K
+
1
x
K
+
1
=
b
1
\begin{equation} \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j + a_{1, K + 1} x_{K + 1} = b_1 \end{equation}
j=1∑Ka1,jxj+a1,K+1xK+1=b1
结合式(4)和题设中的充分性条件可知,若
∑
j
=
1
K
a
1
,
j
x
j
\displaystyle \sum_{j = 1}^{K} a_{1, j} x_j
j=1∑Ka1,jxj满足
{
b
1
−
a
1
,
K
+
1
x
‾
K
+
1
≤
∑
j
=
1
K
a
1
,
j
x
j
=
b
1
−
a
1
,
K
+
1
x
K
+
1
≤
b
1
−
a
1
,
K
+
1
x
‾
K
+
1
(
a
1
,
K
+
1
>
0
)
b
1
−
a
1
,
K
+
1
x
‾
K
+
1
≤
∑
j
=
1
K
a
1
,
j
x
j
=
b
1
−
a
1
,
K
+
1
x
K
+
1
≤
b
1
−
a
1
,
K
+
1
x
‾
K
+
1
(
a
1
,
K
+
1
<
0
)
∑
j
=
1
K
a
1
,
j
x
j
=
b
1
(
a
1
,
K
+
1
=
0
)
\begin{equation} \begin{cases} b_1 - a_{1, K + 1} \overline{x}_{K + 1} ≤ \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j = b_1 - a_{1, K + 1} x_{K + 1} ≤ b_1 - a_{1, K + 1} \underline{x}_{K + 1} & \left( a_{1, K + 1} > 0 \right) \\ b_1 - a_{1, K + 1} \underline{x}_{K + 1} ≤ \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j = b_1 - a_{1, K + 1} x_{K + 1} ≤ b_1 - a_{1, K + 1} \overline{x}_{K + 1} & \left( a_{1, K + 1} < 0 \right) \\ \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j = b_1 & \left( a_{1, K + 1} = 0 \right) \end{cases} \end{equation}
⎩
⎨
⎧b1−a1,K+1xK+1≤j=1∑Ka1,jxj=b1−a1,K+1xK+1≤b1−a1,K+1xK+1b1−a1,K+1xK+1≤j=1∑Ka1,jxj=b1−a1,K+1xK+1≤b1−a1,K+1xK+1j=1∑Ka1,jxj=b1(a1,K+1>0)(a1,K+1<0)(a1,K+1=0)
则必然有
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
≤
∑
j
=
1
K
a
1
,
j
x
j
=
b
1
−
a
1
,
K
+
1
x
K
+
1
≤
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
\begin{equation} \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \overline{x}_j ≤ \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j = b_1 - a_{1, K + 1} x_{K + 1} ≤ \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \end{equation}
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj≤j=1∑Ka1,jxj=b1−a1,K+1xK+1≤j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
根据前述在
N
=
K
N = K
N=K时的假设,必能得到解向量
x
1
−
K
=
[
x
1
⋯
x
j
⋯
x
K
]
T
\boldsymbol{x}_{1 - K} = \left[ \begin{matrix} x_1 & \cdots & x_j & \cdots & x_K \end{matrix} \right]^T
x1−K=[x1⋯xj⋯xK]T,使得对于
∀
j
∈
{
1
,
⋯
,
K
}
\forall j ∈ \{ 1, \cdots, K\}
∀j∈{1,⋯,K},都有
x
j
∈
[
x
‾
j
,
x
‾
j
]
x_j ∈ \left[ \underline{x}_j, \overline{x}_j \right]
xj∈[xj,xj]。将其代入式(4),可得
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
)
≤
a
1
,
K
+
1
x
K
+
1
=
b
1
−
∑
j
=
1
K
a
1
,
j
x
j
≤
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
)
\begin{equation} b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \right) ≤ a_{1, K + 1} x_{K + 1} = b_1 - \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j ≤ b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \right) \end{equation}
b1−
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
≤a1,K+1xK+1=b1−j=1∑Ka1,jxj≤b1−
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
1) 当
a
1
,
K
+
1
>
0
a_{1, K + 1} > 0
a1,K+1>0时,式(7)变为
a
1
,
K
+
1
x
‾
K
+
1
≤
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
)
≤
a
1
,
K
+
1
x
K
+
1
≤
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
)
≤
a
1
,
K
+
1
x
‾
K
+
1
\begin{equation} a_{1, K + 1} \underline{x}_{K + 1} ≤ b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \right) ≤ a_{1, K + 1} x_{K + 1} ≤ b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \right) ≤ a_{1, K + 1} \overline{x}_{K + 1} \end{equation}
a1,K+1xK+1≤b1−
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
≤a1,K+1xK+1≤b1−
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
≤a1,K+1xK+1
结合式(6)可知,此时必能得到
x
K
+
1
=
b
1
−
∑
j
=
1
K
a
1
,
j
x
j
a
1
,
K
+
1
x_{K + 1} = \dfrac{b_1 - \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j} {a_{1, K + 1}}
xK+1=a1,K+1b1−j=1∑Ka1,jxj,使得
x
‾
K
+
1
≤
x
K
+
1
=
b
1
−
∑
j
=
1
K
a
1
,
j
x
j
a
1
,
K
+
1
≤
x
‾
K
+
1
\underline{x}_{K + 1} ≤ x_{K + 1} = \dfrac{b_1 - \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j} {a_{1, K + 1}} ≤ \overline{x}_{K + 1}
xK+1≤xK+1=a1,K+1b1−j=1∑Ka1,jxj≤xK+1。
2) 当
a
1
,
K
+
1
<
0
a_{1, K + 1} < 0
a1,K+1<0时,式(7)变为
a
1
,
K
+
1
x
‾
K
+
1
≤
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
)
≤
a
1
,
K
+
1
x
K
+
1
≤
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
K
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
K
a
1
,
j
x
‾
j
)
≤
a
1
,
K
+
1
x
‾
K
+
1
\begin{equation} a_{1, K + 1} \overline{x}_{K + 1} ≤ b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \right) ≤ a_{1, K + 1} x_{K + 1} ≤ b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{K} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{K} a_{1, j} \underline{x}_j \right) ≤ a_{1, K + 1} \underline{x}_{K + 1} \end{equation}
a1,K+1xK+1≤b1−
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
≤a1,K+1xK+1≤b1−
j=1,a1,j>0∑Ka1,jxj+j=1,a1,j<0∑Ka1,jxj
≤a1,K+1xK+1
结合式(6)可知,此时必能得到
x
K
+
1
=
b
1
−
∑
j
=
1
K
a
1
,
j
x
j
a
1
,
K
+
1
x_{K + 1} = \dfrac{b_1 - \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j} {a_{1, K + 1}}
xK+1=a1,K+1b1−j=1∑Ka1,jxj,使得
x
‾
K
+
1
≤
x
K
+
1
=
b
1
−
∑
j
=
1
K
a
1
,
j
x
j
a
1
,
K
+
1
≤
x
‾
K
+
1
\underline{x}_{K + 1} ≤ x_{K + 1} = \dfrac{b_1 - \displaystyle \sum_{j = 1}^{K} a_{1, j} x_j} {a_{1, K + 1}} ≤ \overline{x}_{K + 1}
xK+1≤xK+1=a1,K+1b1−j=1∑Ka1,jxj≤xK+1。
3) 当 a 1 , K + 1 = 0 a_{1, K + 1} = 0 a1,K+1=0时,结合式(6)可得,式(7)变为 0 = 0 0 = 0 0=0的恒等式,此时可在区间 [ x ‾ K + 1 , x ‾ K + 1 ] \left[ \underline{x}_{K + 1}, \overline{x}_{K + 1}\right] [xK+1,xK+1]上任意取值均可作为 x K + 1 x_{K + 1} xK+1的解。
因此可证,当
N
=
K
+
1
N = K + 1
N=K+1时,必能得到解向量
x
=
[
x
1
⋯
x
j
⋯
x
K
+
1
]
T
\boldsymbol{x} = \left[ \begin{matrix} x_1 & \cdots & x_j & \cdots & x_{K + 1} \end{matrix} \right]^T
x=[x1⋯xj⋯xK+1]T,使得对于
∀
j
∈
{
1
,
⋯
,
K
+
1
}
\forall j ∈ \{ 1, \cdots, K + 1\}
∀j∈{1,⋯,K+1},都有
x
j
∈
[
x
‾
j
,
x
‾
j
]
x_j ∈ \left[ \underline{x}_j, \overline{x}_j \right]
xj∈[xj,xj],命题的充分性成立。综合对
R
=
1
R = 1
R=1时的讨论和对
R
=
K
R = K
R=K时的假设可知,命题的充分性对于
∀
R
\forall R
∀R恒成立,命题的充分性得证。
综上所述,命题的必要性和充分性均成立,命题得证,该命题可作为对修正方法进行可行性分析时的预备定理——单线性方程约束问题可行性定理。
修正结果与单线性方程约束问题可行性定理之间的联系
当生成的随机向量 x g e n \boldsymbol{x_{gen}} xgen不满足单线性方程约束时,可通过对 ∑ j = 1 N a 1 , j x g e n , j > b 1 \displaystyle \sum_{j = 1}^{N} a_{1, j} x_{gen, j} > b_1 j=1∑Na1,jxgen,j>b1和 ∑ j = 1 N a 1 , j x g e n , j < b 1 \displaystyle \sum_{j = 1}^{N} a_{1, j} x_{gen, j} < b_1 j=1∑Na1,jxgen,j<b1的情况分类讨论,并根据博客《一种线性方程约束下生成随机数修正的一般性方法(上)》所提方法修正为 x c o r \boldsymbol{x_{cor}} xcor。通过下述讨论可知,单线性约束问题的可行性可通过修正后 x c o r \boldsymbol{x_{cor}} xcor是否在变量上下界约束范围内的特征直接体现:当单线性约束问题可行时,修正结果 x c o r \boldsymbol{x_{cor}} xcor必然在变量上下界约束范围内;当单线性约束问题不可行时,修正结果 x c o r \boldsymbol{x_{cor}} xcor必然超出变量上下界约束范围。
情况1: 当
∑
j
=
1
N
a
1
,
j
x
g
e
n
,
j
>
b
1
\displaystyle \sum_{j = 1}^{N} a_{1, j} x_{gen, j} > b_1
j=1∑Na1,jxgen,j>b1时,
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
>
0
\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right) > 0
j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)>0,其修正解在实数空间上的分布可分如下情况进行讨论:
情况1.1:当单线性方程约束问题有解时,根据单线性方程约束问题可行性定理可知
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
≤
b
1
≤
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
\displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j ≤ b_1 ≤ \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \underline{x}_j
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj≤b1≤j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj,将其代入修正结果,可得
x
c
o
r
,
j
=
{
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
)
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
(
x
g
e
n
,
j
−
x
‾
j
)
+
x
‾
j
∈
[
x
‾
j
,
x
‾
j
]
(
a
1
,
j
>
0
)
x
g
e
n
,
j
∈
[
x
‾
j
,
x
‾
j
]
(
a
1
,
j
=
0
)
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
)
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
(
x
g
e
n
,
j
−
x
‾
j
)
+
x
‾
j
∈
[
x
‾
j
,
x
‾
j
]
(
a
1
,
j
<
0
)
\begin{equation} x_{cor, j} = \begin{cases} \dfrac{b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j \right)} {\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right)} \left( x_{gen, j} - \underline{x}_j \right) + \underline{x}_j ∈ \left[ \underline{x}_j, \overline{x}_j \right] & \left( a_{1, j} > 0 \right) \\ x_{gen, j} ∈ \left[ \underline{x}_j, \overline{x}_j \right] & \left( a_{1, j} = 0 \right) \\ \dfrac{b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j\right)} {\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right)} \left( x_{gen, j} - \overline{x}_j \right) + \overline{x}_j ∈ \left[ \underline{x}_j, \overline{x}_j \right] & \left( a_{1, j} < 0 \right) \\ \end{cases} \end{equation}
xcor,j=⎩
⎨
⎧j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)b1−
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
(xgen,j−xj)+xj∈[xj,xj]xgen,j∈[xj,xj]j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)b1−
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
(xgen,j−xj)+xj∈[xj,xj](a1,j>0)(a1,j=0)(a1,j<0)
情况1.2:当单线性方程约束问题无解时,根据单线性方程约束问题可行性定理可知 b 1 < ∑ j = 1 , a 1 , j > 0 N a 1 , j x ‾ j + ∑ j = 1 , a 1 , j < 0 N a 1 , j x ‾ j b_1 < \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j b1<j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj或 b 1 > ∑ j = 1 , a 1 , j > 0 N a 1 , j x ‾ j + ∑ j = 1 , a 1 , j < 0 N a 1 , j x ‾ j b_1 > \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \overline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \underline{x}_j b1>j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj,将其代入修正结果,可得
情况1.2.1:当
b
1
<
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
b_1 < \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j
b1<j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj时
x
c
o
r
,
j
=
{
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
)
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
(
x
g
e
n
,
j
−
x
‾
j
)
+
x
‾
j
<
x
‾
j
(
a
1
,
j
>
0
)
x
g
e
n
,
j
∈
[
x
‾
j
,
x
‾
j
]
(
a
1
,
j
=
0
)
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
)
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
(
x
g
e
n
,
j
−
x
‾
j
)
+
x
‾
j
>
x
‾
j
(
a
1
,
j
<
0
)
\begin{equation} x_{cor, j} = \begin{cases} \dfrac{b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j \right)} {\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right)} \left( x_{gen, j} - \underline{x}_j \right) + \underline{x}_j < \underline{x}_j & \left( a_{1, j} > 0 \right) \\ x_{gen, j} ∈ \left[ \underline{x}_j, \overline{x}_j \right] & \left( a_{1, j} = 0 \right) \\ \dfrac{b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j\right)} {\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right)} \left( x_{gen, j} - \overline{x}_j \right) + \overline{x}_j > \overline{x}_j & \left( a_{1, j} < 0 \right) \\ \end{cases} \end{equation}
xcor,j=⎩
⎨
⎧j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)b1−
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
(xgen,j−xj)+xj<xjxgen,j∈[xj,xj]j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)b1−
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
(xgen,j−xj)+xj>xj(a1,j>0)(a1,j=0)(a1,j<0)
情况1.2.2:当
b
1
>
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
b_1 > \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \overline{x}_j + \displaystyle\sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \underline{x}_j
b1>j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj时
x
c
o
r
,
j
=
{
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
)
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
(
x
g
e
n
,
j
−
x
‾
j
)
+
x
‾
j
>
x
‾
j
(
a
1
,
j
>
0
)
x
g
e
n
,
j
∈
[
x
‾
j
,
x
‾
j
]
(
a
1
,
j
=
0
)
b
1
−
(
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
x
‾
j
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
x
‾
j
)
∑
j
=
1
,
a
1
,
j
>
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
+
∑
j
=
1
,
a
1
,
j
<
0
N
a
1
,
j
(
x
g
e
n
,
j
−
x
‾
j
)
(
x
g
e
n
,
j
−
x
‾
j
)
+
x
‾
j
<
x
‾
j
(
a
1
,
j
<
0
)
\begin{equation} x_{cor, j} = \begin{cases} \dfrac{b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j \right)} {\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right)} \left( x_{gen, j} - \underline{x}_j \right) + \underline{x}_j > \overline{x}_j & \left( a_{1, j} > 0 \right) \\ x_{gen, j} ∈ \left[ \underline{x}_j, \overline{x}_j \right] & \left( a_{1, j} = 0 \right) \\ \dfrac{b_1 - \left( \displaystyle \sum_{j = 1, a_{1, j} > 0}^{N} a_{1, j} \underline{x}_j + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \overline{x}_j\right)} {\displaystyle \sum_{j = 1,a_{1, j} > 0}^{N} a_{1, j} \left( x_{gen, j} - \underline{x}_j \right) + \displaystyle \sum_{j = 1, a_{1, j} < 0}^{N} a_{1, j} \left( x_{gen, j} - \overline{x}_j \right)} \left( x_{gen, j} - \overline{x}_j \right) + \overline{x}_j < \underline{x}_j & \left( a_{1, j} < 0 \right) \\ \end{cases} \end{equation}
xcor,j=⎩
⎨
⎧j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)b1−
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
(xgen,j−xj)+xj>xjxgen,j∈[xj,xj]j=1,a1,j>0∑Na1,j(xgen,j−xj)+j=1,a1,j<0∑Na1,j(xgen,j−xj)b1−
j=1,a1,j>0∑Na1,jxj+j=1,a1,j<0∑Na1,jxj
(xgen,j−xj)+xj<xj(a1,j>0)(a1,j=0)(a1,j<0)
剩余内容参见博客 《线性方程约束下修正方法的可行性分析(演绎推理,二)》