Message Board

Dear readers, authors and reviewers,you can add a message on this page. We will reply to you as soon as possible!

2017 Volume 39 Issue 1
Article Contents

Xiao-qing OU, Jin-fu LI, Jia LIU, et al. An Alternative Theorem for a Class of Cone Constrained Multiobjective Optimization Problems[J]. Journal of Southwest University Natural Science Edition, 2017, 39(1): 109-113. doi: 10.13718/j.cnki.xdzk.2017.01.017
Citation: Xiao-qing OU, Jin-fu LI, Jia LIU, et al. An Alternative Theorem for a Class of Cone Constrained Multiobjective Optimization Problems[J]. Journal of Southwest University Natural Science Edition, 2017, 39(1): 109-113. doi: 10.13718/j.cnki.xdzk.2017.01.017

An Alternative Theorem for a Class of Cone Constrained Multiobjective Optimization Problems

More Information
  • Corresponding author: Jia-wei CHEN
  • Received Date: 05/12/2014
    Available Online: 20/01/2017
  • MSC: O177.91

  • Multiobjective optimization is a common problem in economic management and game theory. However, a great number of practical economic management problems are subject to the constraints of external and internal conditions. That is why constrained multiobjective optimization problems have aroused wide concern. Optimality conditions of multiobjective optimization problems are one of the important contents in the optimization theory. This paper is devoted to the study of the optimality conditions of a class of cone constrained multiobjective optimization problems by image space analysis. A class of regular weak separation functions is introduced by the oriented distance function, and an alternative theorem is established. Finally, some sufficient and necessary optimality conditions for cone constrained multiobjective optimization problems are obtained by the alternative theorem, without the convexity of the involved mappings.
  • 加载中
  • [1] CHEN J W, CHO Y J, KIM J, et al. Multiobjective Optimization Problems with Modified Objective Functions and Cone Constraints and Applications [J]. J Glob Optim, 2011, 49(1): 137-147. doi: 10.1007/s10898-010-9539-3

    CrossRef Google Scholar

    [2] 陈加伟, 李军, 王金南.锥约束非光滑多目标优化问题的对偶及最优性条件[J].数学物理学报, 2012, 32(1): 1-12.

    Google Scholar

    [3] 周志昂.强G-预不变凸向量优化问题的最优性条件[J].西南大学学报(自然科学版), 2013, 35(1): 1-7.

    Google Scholar

    [4] GIANNESSI F. Constrained Optimization and Image Space Analysis [M]. London: Springer, 2005.

    Google Scholar

    [5] GIANNESSI F, MASTROENI G, PELLEGRINI L. On the Theory of Vector Optimization and Variational Inequalities. Image Space Analysis and Seperation [M] //GIANNESSI F. Vector Variational Inequalities and Vector Equilibria. London: Kluwer Academic, 2000: 153-215.

    Google Scholar

    [6] LI J, HUANG N J. Image Space Analysis for Vector Variational Inequalites with Matrix Inequality Constraints and Applications [J]. J Optim Theory Appl, 2010, 145(3): 459-477. doi: 10.1007/s10957-010-9691-4

    CrossRef Google Scholar

    [7] LI J, HUANG N J. Image Space Analysis for Variational Inequalites with Cone Constraints and Applications to Traffic Equilibria [J]. Sci China Math, 2012, 55(4): 851-868. doi: 10.1007/s11425-011-4287-5

    CrossRef Google Scholar

    [8] LI S J, XU Y D, ZHU S K. Nonlinear Seperation Approach to Constrained Extremum Problems [J]. J Optim Theory Appl, 2012, 154(3): 842-856. doi: 10.1007/s10957-012-0027-4

    CrossRef Google Scholar

    [9] XU Y D, LI S J. Nonlinear Seperation Functions and Constrained Extremum Problems [J]. Optim Lett, 2014, 8(3): 1149-1160. doi: 10.1007/s11590-013-0644-3

    CrossRef Google Scholar

    [10] XU Y D, LI S J. Gap Functions and Error Bounds for Weak Vector Variational Inequalities [J]. Optim, 2014, 63(9): 1339-1352. doi: 10.1080/02331934.2012.721115

    CrossRef Google Scholar

    [11] 罗彬, 王莲明, 张谋.约束向量优化问题的像空间分析[J].吉林大学学报(理学版), 2013, 51(6): 1068-1072.

    Google Scholar

    [12] HIRIART-URRUTY J B. Tangent Cones, Generalized Gradients and Mathematical Programming in Banach Spaces [J]. Math Oper Res, 1979, 4(1): 79-97. doi: 10.1287/moor.4.1.79

    CrossRef Google Scholar

    [13] ZAFFARONI A. Degrees of Efficiency and Degrees of Minimality [J]. SIAM J Control Optim, 2003, 42(3): 1071-1086. doi: 10.1137/S0363012902411532

    CrossRef Google Scholar

    [14] JEYAKUMAR V, OETTLI W, NATIVIDAD M. A Solvability Theorem for a Class of Quasiconvex Mappings with Applications to Optimization [J]. J Math Anal Appl, 1993, 179(2): 537-546. doi: 10.1006/jmaa.1993.1368

    CrossRef Google Scholar

  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Article Metrics

Article views(781) PDF downloads(51) Cited by(0)

Access History

Other Articles By Authors

An Alternative Theorem for a Class of Cone Constrained Multiobjective Optimization Problems

    Corresponding author: Jia-wei CHEN

Abstract: Multiobjective optimization is a common problem in economic management and game theory. However, a great number of practical economic management problems are subject to the constraints of external and internal conditions. That is why constrained multiobjective optimization problems have aroused wide concern. Optimality conditions of multiobjective optimization problems are one of the important contents in the optimization theory. This paper is devoted to the study of the optimality conditions of a class of cone constrained multiobjective optimization problems by image space analysis. A class of regular weak separation functions is introduced by the oriented distance function, and an alternative theorem is established. Finally, some sufficient and necessary optimality conditions for cone constrained multiobjective optimization problems are obtained by the alternative theorem, without the convexity of the involved mappings.

  • 弱有效解是经济、决策理论、多目标优化理论以及最优控制与博弈论中的重要概念之一.对于多目标优化问题弱有效解的研究常常涉及目标函数与约束函数的凸性[1-3].众所周知,择一定理在研究多目标优化问题的最优性条件与对偶性中起着至关重要的作用.文献[4]提出像空间分析理论对约束优化问题进行统一研究.这类优化问题常常可等价地表示成一个参数系统的不可行性以及约束优化问题像空间中两个集合的分离性.随后,许多学者用像空间分析理论研究了向量优化问题、向量变分不等式等约束优化问题的最优性条件、对偶性、误差界、间隙函数以及非线性分离性等[5-11].在Banach空间中,Hiriart-Urruty[12]用定向距离函数(oriented distance function)Δ刻画非光滑优化问题的几何性质,从而得到了非凸优化问题的最优性必要条件.定向距离函数Δ也被称为一类非线性标量化函数[8-11],正是其显著的几何特征而被广泛应用于研究各类优化问题的最优性条件.后来,Zaffaroni[13]进一步研究了定向距离函数Δ的性质.

    本文利用像空间分析理论研究一类锥约束多目标优化问题的最优性条件.首先,通过定向距离函数Δ引入一类不同于文献[11]的正则弱分离函数;然后借助该类正则弱分离函数建立了一个择一定理;最后,通过择一定理在不涉及函数凸性的条件下得到了锥约束多目标优化问题弱有效解的充分和必要最优性条件.

1.   预备知识
  • 无特别说明,本文总假设 $\mathbb{R}$ 为实数集, ${{\mathbb{R}}^{n}}$ n维欧氏空间,X ${{\mathbb{R}}^{n}}$ 为非空闭凸子集,CD分别为 ${{\mathbb{R}}^{m}}$ ${{\mathbb{R}}^{l}}$ 的闭凸点锥且具有非空内部intC,intD,映像f ${{\mathbb{R}}^{n}}$ ${{\mathbb{R}}^{m}}$ ,g: ${{\mathbb{R}}^{n}}$ ${{\mathbb{R}}^{l}}$ 为向量值映射.设M ${{\mathbb{R}}^{m}}$ 的任意非空凸锥,则M的对偶锥定义为

    对于函数ψX $\mathbb{R}$ ∪{±∞},α $\mathbb{R}$ ,集合

    分别称为ψ的非负水平集与正水平集.

    考虑如下约束多目标优化问题(简记为M OP)

    记M OP的可行域为 $\mathscr{F}$ ={xXg(x)∈D}.

    下面介绍M OP的像及相关符号:设xX.定义映像 $\mathscr{A}$ x ${{\mathbb{R}}^{n}}$ ${{\mathbb{R}}^{m}}$ × ${{\mathbb{R}}^{l}}$

    定义集合:

    称集合 $\mathscr{K}$ x为M OP在点x的像,如果空间 ${{\mathbb{R}}^{m}}$ × ${{\mathbb{R}}^{l}}$ 为M OP的像空间.

    定义1x $\mathscr{F}$ 称为M OP的弱有效解,如果

    记M OP的弱有效解集为 $\mathscr{F}$ w.

    定义2[5] 若函数ω ${{\mathbb{R}}^{m}}$ × ${{\mathbb{R}}^{l}}$ ×∏→ $\mathbb{R}$ 满足:

    则称ω为正则弱分离函数,其中∏为参数集合.记所有正则弱分离函数的集合为 $\mathscr{W}$ $\mathscr{R}$ (∏).

    定义3[12] 设 $\mathscr{M}$ ∉Y.函数Δ $\mathscr{M}$ Y $\mathbb{R}$ ∪{±∞},

    称为定向距离函数,其中 ${d_{\mathscr{M}}}\left( y \right) = \mathop {\inf }\limits_{m \in {\cal \mathscr{M}}} {\mkern 1mu} \left\| {y - m} \right\|$ yY $\mathscr{M}$ 的距离.

    引理1[13] 设 $\mathscr{M}$ Y的非空子集且 $\mathscr{M}$ Y,则有:

    (ⅰ)Δ $\mathscr{M}$ 为1-李普希兹实值函数,即对任意y1y2Y,|Δ $\mathscr{M}$ (y1)-Δ $\mathscr{M}$ (y2)|≤‖y1-y2 ‖;

    (ⅱ)对任意y ∈int $\mathscr{M}$ Δ $\mathscr{M}$ (y)<0;

    (ⅲ)对任意y ∈∂ $\mathscr{M}$ Δ $\mathscr{M}$ (y)=0,其中∂ $\mathscr{M}$ $\mathscr{M}$ 的边界;

    (ⅳ)对任意y ∈int(Y\ $\mathscr{M}$ ),Δ $\mathscr{M}$ (y)>0;

    (ⅴ) $\mathscr{M}$ ={yΔ $\mathscr{M}$ (y)≤0}当 $\mathscr{M}$ 为闭集;

    (ⅵ)Δ $\mathscr{M}$ 为凸的,若 $\mathscr{M}$ 为凸集;

    (ⅶ)Δ $\mathscr{M}$ 为正齐次的,若 $\mathscr{M}$ 为锥;

    (ⅷ)如果 $\mathscr{M}$ 为闭凸锥,则对任意y1y2Yy1-y2 $\mathscr{M}$ Δ $\mathscr{M}$ (y1)≤Δ $\mathscr{M}$ (y2);

    特别地,若int $\mathscr{M}$ y1-y2∈int $\mathscr{M}$ Δ $\mathscr{M}$ (y1)<Δ $\mathscr{M}$ (y2).

    引理2[14] 设CY为闭凸点锥,且intC.则有:

    (ⅰ)yCy*(y)≥0,∀y*C*

    (ⅱ)y∈intCy*(y)>0,∀y*C*\{0}.

    由MOP弱有效解及其像的定义,我们有如下结论:

    引理3x $\mathscr{F}$ w当且仅当 $\mathscr{K}$ x $\mathscr{H}$ =,即不存在xX满足 $\mathscr{A}$ x(x)∈ $\mathscr{H}$ .

2.   主要结果
  • 在本节中,我们用定向距离函数Δ引入一类不同于文献[11]的正则弱分离函数,借助该类正则弱分离函数建立了一个择一定理.最后,通过择一定理在不涉及函数凸性的条件下得到了锥约束多目标优化问题弱有效解的充分和必要最优性条件.首先,定义如下非线性函数:

    下面引理说明函数ω(uvλ)为一类正则弱分离函数.

    引理4ω $\mathscr{W}$ $\mathscr{R}$ (∏),其中∏=D*.

     对任意(uv)∈ $\mathscr{H}$ λD*u∈intCvD,进而,由引理1与引理2,可得

    于是有

    反之,假设存在 $\left( \hat{u},\hat{v} \right)$ ${{\mathbb{R}}^{m}}$ × ${{\mathbb{R}}^{l}}$ \ $\mathscr{H}$ , 使得

    由于 $\left( \hat{u},\hat{v} \right)$ ${{\mathbb{R}}^{m}}$ × ${{\mathbb{R}}^{l}}$ \ $\mathscr{H}$ ,则 ${\hat{u}}$ ∉intC ${\hat{v}}$ D.

    ${\hat{u}}$ ∉intC,由引理1(ⅲ),(ⅳ)可得-ΔC( ${\hat{u}}$ )≤0.取λ=0Z*,从而

    与(1) 式矛盾!

    ${\hat{v}}$ D,则存在 ${\hat{\lambda }}$ D* ${\hat{\lambda }}$ ≠0使得 ${{{\hat{\lambda }}}^{\top}}\hat{v}$ <0.因为D*为锥,则对任意4>0,c ${\hat{\lambda }}$ D*,进而

    与(1) 式矛盾!

    综上所述,

    从而

    由定义2知,ω $\mathscr{W}$ $\mathscr{R}$ (∏),其中∏=D*.

    引理5 系统 $\mathscr{A}$ x(x)∈ $\mathscr{H}$ xX与系统:

    不可能同时成立.

     若系统 $\mathscr{A}$ x(x)∈ $\mathscr{H}$ , xX成立,则存在 ${\hat{x}}$ X使得g( ${\hat{x}}$ )∈D

    由引理1与引理2,对任意μD*,有

    于是有

    故系统(2) 不成立.

    反之,若系统(2) 成立.由引理4,有

    故系统 $\mathscr{A}$ x(x)∈ $\mathscr{H}$ xX不成立.

    下面我们通过引理5研究约束多目标优化问题MOP弱有效解的充分和必要条件.

    定理1 设xX.若存在 ${\hat{\mu }}$ D*,使得

    则有x $\mathscr{F}$ w.

     由命题3可得结论.

    定理2 设xX.则x $\mathscr{F}$ w的充要条件为

     充分性若(4) 式成立.于是有

    假设x $\mathscr{F}$ w,则存在 ${\tilde{x}}$ $\mathscr{F}$ ,使得

    引理1(ⅱ),有

    与(5) 式矛盾!故x $\mathscr{F}$ w.

    必要性 若x $\mathscr{F}$ w,则

    由于

    所以

    易知(6) 式等价于

    故系统 $\mathscr{A}$ x(x)=(f(x)-f(x),g(x))∈ $\mathscr{H}$ xX不成立.由命题3, $\mathscr{K}$ x $\mathscr{H}$ =.联合引理4,对每一xX,存在μD*,使得

    于是有

    故由(7),(8) 式,有

    成立.

Reference (14)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return