【Fuzzy】不确定规划：模糊相关机会规划与模糊随机变量

不确定原理

定义1

将模糊约束
$$g_j(x,\xi )\le 0,j=1,2,\dots ,p$$
称为不确定环境(模糊环境)，其中$x$为决策向量，$\xi$为模糊向量.

定义2

称模糊不等式组
$$h_k(x,\xi )\le 0,k=1,2,\dots ,q$$
称为一个事件，其中$x$为决策向量，$\xi$为模糊向量

定义3

设$\epsilon$表示事件，它的机会函数由事件的可能性测度定义给出
f ( x ) = C r { h k ( x , ξ ) ≤ 0 , k = 1 , 2 , … , q } s . t . g j ( x , ξ ) ≤ 0 , j = 1 , 2 , … , p f(x)=Cr\{h_k(x, \xi)\leq 0, k=1,2, \dots, q\}\\ s.t.\quad g_j(x, \xi)\leq 0, j=1,2, \dots, p f(x)=Cr{hk​(x,ξ)≤0,k=1,2,…,q}s.t.gj​(x,ξ)≤0,j=1,2,…,p
不确定原理
在不确定环境下，一个模糊事件的机会等于此事件相容的可信性. 设在不确定环境$g_j(x,\xi )\le 0,j=1,2,\dots ,p$下有$m$个事件$h_{ik}(x,\xi )\le 0,k=1,2,\dots ,q;k=1,2,\dots ,m$，分别记做为${\epsilon }_i$，根据不确定原理，在不确定环境在，第$i$个事件的${\epsilon }_i$的机会函数为
$$f_i(x)=Cr\left\{\begin{array}{cc}& h_{ik}(x,\xi )\le 0,k=1,2,\dots ,q_i\\ & g_j(x,\xi )\le 0,j\in J_i\end{array}\right\}$$

相关机会规划

在模糊环境下，标准的相关机会规划形式为
max ⁡ C r { h k ( x , ξ ) ≤ 0 , k = 1 , 2 , … , q } s . t . g j ( x , ξ ) ≤ 0 , j = 1 , 2 , … , p \max Cr\{h_k(x, \xi)\leq 0, k=1,2, \dots, q\}\\ s.t.\quad g_j(x, \xi)\leq 0, j=1, 2, \dots, p maxCr{hk​(x,ξ)≤0,k=1,2,…,q}s.t.gj​(x,ξ)≤0,j=1,2,…,p
可以理解为，在不确定环境$g_j(x,\xi )\le 0,j=1,2,\dots ,p$下极大化模糊事件$h_k(x,\xi )\le 0,k=1,2,\dots ,q$的可信性.

模糊随机变量

模糊随机变量是从概率空间到模糊变量构成的集类的可测函数，本事取值为模糊变量的随机变量.

定义1

设$\xi$是一个从概率空间$(\Omega ,A,Pr)$的模糊变量集合的函数，如果对于$\mathbb{R}$上的任何Borel集$B$， P o s { ξ ( w ) ∈ B } Pos\{\xi(w)\in B\} Pos{ξ(w)∈B}是$w$的可测函数，则称$\xi$为一个模糊随机变量.

定理1

设$\xi$是概率空间$(\Omega ,A,Pr)$上的模糊随机变量，则对于$\mathbb{R}$上的任何Borel集$B$，可能性 P o s { ξ ( w ) ∈ B } Pos\{\xi(w)\in B\} Pos{ξ(w)∈B}是一个随机变量，由$Nec(B)$和$Pos$的计算公式可以知道，其均为随机变量.

Related Paper Read

In the sense of mathematical programming, since the proposed model is formulated as ill-defined problem due to fuziness, we need to set some certain optimization criterion so as to transform into well-defined problems.

Definition

For any two interval numbers $a=[\underline{a},\overline{a}]$ and $b=[\underline{b},\overline{b}]$, if $m(a)<m(b)$ and $\overline{a}<\underline{b}$, the interval inequality relation $a<b$ os called optimistic satisfactory. And the optimistic satisfaction index of the interval inequality relation between $a\le b$ can be defined as
$$OSD(a\le b)=\frac{\underline{b}-\overline{a}}{w(a)+w(b)}$$

Type-2 Fuzzy Set

In the real world, there are many problems that the decision maker cannot determine the precise membership function because of noise in the data and the insufficient time to receive the effective information, and so need to assume the membership function based on the long-term experience and its versatile function to respond to the move of financial market flexibly.

example: Gaussian membership function
$$\begin{gathered} \mu (x)=\exp \{-\frac{1}{2}(\frac{x-m}{\sigma }{)}^2\} \\ \sigma \in [{\sigma }_1,{\sigma }_2] \end{gathered}$$

A possibility Maximization Model

The investor often has optimistic and pessimistic subjectivities with respect to the target return $f$ represented the optimistic tolerated profit level and the pessimistic tolerated profit level, respectively. We propose the more flexible model considering the profit tolerance interval to the goal for the total future returns as $[\underline{f},\overline{f}]$ follows.
max ⁡ P o s { ∑ j = 1 n r ~ j x j ≥ f } s . t . { ∑ i = 1 n = 1 0 ≤ x j ≤ p j α j ∈ [ α ‾ j , α ‾ j ] f ∈ [ f ‾ , f ‾ ] \max Pos\{\sum_{j=1}^n\tilde{r}_j x_j\geq f\}\\ s.t. \begin{cases} \sum_{i=1}^n=1\\ 0\leq x_j\leq p_j\\ \alpha_j\in [\underline{\alpha}_j, \overline{\alpha}_j]\\ f\in [\underline{f}, \overline{f}] \end{cases} maxPos{j=1∑n​r~j​xj​≥f}s.t.⎩⎪⎪⎪⎨⎪⎪⎪⎧​∑i=1n​=10≤xj​≤pj​αj​∈[α​j​,αj​]f∈[f​,f​]​

参考资料

不确定规划及应用 清华大学出版社
A portfolio selection problem with Type-2 Fuzzy return based on Possibility measure and Interval Programming