Message Board

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

2020 Volume 42 Issue 9
Article Contents

Cheng-ying XIAO, Ming-ge YANG. Nonemptiness and Lower Semicontinuity of Solution Maps for Parametric Vector Optimization Problems in Banach Spaces[J]. Journal of Southwest University Natural Science Edition, 2020, 42(9): 95-100. doi: 10.13718/j.cnki.xdzk.2020.09.011
Citation: Cheng-ying XIAO, Ming-ge YANG. Nonemptiness and Lower Semicontinuity of Solution Maps for Parametric Vector Optimization Problems in Banach Spaces[J]. Journal of Southwest University Natural Science Edition, 2020, 42(9): 95-100. doi: 10.13718/j.cnki.xdzk.2020.09.011

Nonemptiness and Lower Semicontinuity of Solution Maps for Parametric Vector Optimization Problems in Banach Spaces

More Information
  • Corresponding author: Ming-ge YANG
  • Received Date: 18/10/2018
    Available Online: 20/09/2020
  • MSC: O224

  • In this paper, using the tools of variational analysis and generalized differentiation, we study the stability of solution maps for parametric vector optimization problems in Banach spaces, and give sufficient conditions in terms of Clarke coderivative to show that the restriction of the solution map on the open ball is nonempty and lower semicontinuous around the given point. Then, we provide sufficient conditions characterized by the objective function of the parametric vector optimization problem, the efficient point multifunction of the parametric vector optimization problem and Clarke coderivative. Finally, we discuss thesufficient conditions which ensure that the restriction of the inverse multifunction of the general multifunction on the open ball is nonempty and lower semicontinuous around the given point.
  • 加载中
  • [1] CHUONG T D, KRUGER A Y, YAO J C. Calmness of Efficient Solution Maps in Parametric Vector Optimization [J]. Journal of Global Optimization, 2011, 51(4):677-688. doi: 10.1007/s10898-011-9651-z

    CrossRef Google Scholar

    [2] YANG M G, XIAO Y B, HUANG N J. Coderivative Conditions for Calmness of Implicit Multifunctions and Applications [J]. Journal of Nonlinear Convex Analysis, 2018, 19(1):97-113.

    Google Scholar

    [3] MORDUKHOVICH B S. Variational Analysis and Generalized Differentiation, Vol. I:Basic Theory, Vol. II:Applications [M]. Berlin:Springer, 2006.

    Google Scholar

    [4] CLARKE F H. Optimization and Nonsmooth Analysis [M]. New York:Wiley, 1983.

    Google Scholar

    [5] CHUONG T D. Lipschitz-Like Property of an Implicit Multifunction and Its Applications [J]. Nonlinear Analysis:Theory, Methods & Applications, 2011, 74(17):6256-6264.

    Google Scholar

    [6] DONTCHEV A L, ROCKAFELLAR R T. Implicit Functions and Solution Mappings [M]. Berlin:Springer, 2009.

    Google Scholar

    [7] DUREA M, STRUGARIU R. On Parametric Vector Optimization via Metric Regularity of Constraint Systems [J]. Mathematical Methods of Operations Research, 2011, 74(3):409-425.

    Google Scholar

    [8] HUY N Q, KIM D S, NINH K V. Stability of Implicit Multifunctions in Banach Spaces [J]. Journal of Optimization Theory and Applications, 2012, 155(2):558-571. doi: 10.1007/s10957-012-0058-x

    CrossRef Google Scholar

    [9] NGHIA T T A. A Note on Implicit Multifunction Theorems [J]. Optimization Letters, 2014, 8(1):329-341.

    Google Scholar

    [10] YANG M G, HUANG N J. Random Implicit Function Theorems in Asplund Spaces with Applications [J]. Journal of Nonlinear and Convex Analysis, 2013, 14(3):497-517.

    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(1194) PDF downloads(5) Cited by(0)

Access History

Other Articles By Authors

Nonemptiness and Lower Semicontinuity of Solution Maps for Parametric Vector Optimization Problems in Banach Spaces

    Corresponding author: Ming-ge YANG

Abstract: In this paper, using the tools of variational analysis and generalized differentiation, we study the stability of solution maps for parametric vector optimization problems in Banach spaces, and give sufficient conditions in terms of Clarke coderivative to show that the restriction of the solution map on the open ball is nonempty and lower semicontinuous around the given point. Then, we provide sufficient conditions characterized by the objective function of the parametric vector optimization problem, the efficient point multifunction of the parametric vector optimization problem and Clarke coderivative. Finally, we discuss thesufficient conditions which ensure that the restriction of the inverse multifunction of the general multifunction on the open ball is nonempty and lower semicontinuous around the given point.

  • 参数向量优化问题的稳定性分析已经被许多学者广泛研究,其主要问题之一是研究有效解映射具有某种稳定性的充分条件[1-10].文献[1-2]利用隐函数的方法来研究参数向量优化问题的有效解映射的稳定性,即先讨论集值隐函数的稳定性,做为应用得到参数向量优化问题有效解映射的稳定性.本文在Banach空间中避开隐函数,直接讨论参数向量优化问题的有效解映射的非空性和下半连续性.

    XY是Banach空间,P是度量空间,$ F:P \times X_ \to ^ \to Y $是集值映射,定义集值隐函数$ G:P_ \to ^ \to X $如下:

    进一步,设fP×XY是单值映射,KY是顶点在原点的尖闭凸锥,AYyA,称yA关于K的有效点(efficient point)当且仅当(y-K)∩A={y}.A的有效点的集合记为EffKA,规定EffKØ=Ø.

    考虑参数向量优化问题:

    其中:x是未知的决策变量,p是参数.任意的pP,分别令

    $ \mathscr{F}:P_ \to ^ \to Y 和 \mathscr{S} :P_ \to ^ \to X $分别称为参数向量优化问题(2)的有效点映射(efficient point multifunction)和有效解映射(efficient solution map).注意到,有效解映射S可以写成集值隐函数的形式,即

    其中$ H:P \times X_ \to ^ \to Y $定义为

    事实上H是由目标函数f和有效点映射$ \mathscr{F} $构成的集值映射.以下记Hp(·):=H(p·).

    定理1  设XY是Banach空间,P是度量空间,fP×XY是单值映射,KY是顶点在原点的尖闭凸锥,$ \mathscr{F}:P_ \to ^ \to Y 和 \mathscr{S} :P_ \to ^ \to X $分别是参数向量优化问题(2)的有效点映射和有效解映射,$ H:P \times X_ \to ^ \to Y $是由(3)式定义的集值映射.进一步,(px)∈P×X且(px)∈gph$ \mathscr{S} $.若存在常数r>0,满足下列条件:

    (i) 任意的pB(pr),集值映射Hp是闭的;

    (ii) 常数

    其中: Dc*是Clarke上导数,任意的δ>0,Πδ(0;Hp(x)):={yHp(x)|║y║<d(0,Hp(x))+δ},Jδ(y):={y*SY*|d(y*J(y))<δ};

    (iii) 任意的(px)∈B(prB(xr),集值映射H(·x)在p是下半连续的.

    则存在常数s∈(0,r),使得集值映射$ \mathscr{\tilde S} :P_ \to ^ \to X $B(ps)上是非空下半连续的,其中$ \mathscr{\tilde S} $的定义为

    即有效解映射$ \mathscr{S} $在开球上的限制在p周围是非空下半连续的.

      因为(px)∈gph$ \mathscr{S} $,所以0∈H(px).由条件(iii),存在常数ρ>0满足

    从而

    任取常数s∈(0,min{rρ}),下面证明集值映射$ \mathscr{\tilde S} $B(ps)上是非空且下半连续的.

    1) 任意的p∈B(ps),我们证明$ \mathscr{\tilde S} $(p)非空.定义函数φX×Y如下:

    由条件(i),φX×Y上是下半连续的.

    φ(x,0)=0,则0∈Hp(x),故x$ \mathscr{ S} $(p),从而x$ \mathscr{ S} $(p)∩intB(xr),即$ \mathscr{\tilde S} $(p)≠Ø.

    φ(x,0)≠0,则0∉Hp(x),从而d(0,H(px))>0,令ε:=d(0,H(px)),则

    任取α∈(0,r-ε)且$ \frac{{\varepsilon + \alpha }}{r} < {k_r} $,由距离函数的定义,存在yHp(x)使得

    β:=φ(xy)=║y║,任意的$ t \in \left( {\frac{{\varepsilon + \alpha }}{r}, {k_r}} \right) $,易知

    显然

    任意$ \eta \in \left( {0, \frac{1}{{{k_r}}}} \right) $,在乘积空间X×Y中使用范数║(xy)║η:=║x║+ηy║.由文献[3]定理2.26中的Ekeland变分原理,存在$ \left( {\hat x, \hat y} \right) \in X \times Y $使得

    从而

    进一步,

    下面证明$ 0 \in {H_p}\left( {\hat x} \right) $.假设$ 0 \notin {H_p}\left( {\hat x} \right) $,则$ \hat y \ne 0 $.定义函数ψX×Y

    由(4)式易知,$ \left( {\hat x, \hat y} \right) $是函数ψ+δgphHpX×Y上的极小值点.注意到$ \hat y \ne 0 $,由文献[4]命题1.114得

    故存在$ {y_1}* \in J\left( {\hat y} \right) $和(x2*y2*)∈BX*×By*满足

    显然

    $ \left( {\tilde x^*, - \tilde y^*} \right) \in {N_c}\left( {\left( {\hat x, \hat y} \right);{\rm{gph}}{H_p}} \right) $,由Clarke上导数的定义可知$ \tilde x^* \in {D_c}^*{H_p}\left( {\hat x, \hat y} \right)\left( {\tilde y^*} \right) $.任意的$ y \in {H_p}\left( {\hat x} \right) $,由(4)式得

    从而

    进一步,

    由于

    从而

    任意的δ>0,只要η选取得充分小,由(6),(7)和(8)式得

    δ↓0,则$ \parallel \tilde x^*\parallel \le t < {k_r} $,这与条件(ii)矛盾,故$ 0 \in {H_p}\left( {\hat x} \right), 即 \hat x \in S\left( p \right) $.结合(5)式,可得$ \hat x \in \mathscr{\tilde S}\left( p \right) $,从而$ \mathscr{\tilde S} $(p)≠Ø.

    2) 任意的pB(ps),下面证明$ \mathscr{\tilde S} $p是下半连续的.事实上,只需证明:任意的$ x \in \mathscr{\tilde S}\left( p \right) $和任意的$ \tilde \varepsilon > 0 $,存在常数t>0使得

    因为x$ \mathscr{\tilde S} $(p),所以(px)∈gph$ \mathscr{ S} $x∈intB(xr).任取$ \gamma \in \left( {0, \min \left\{ {r, \tilde \varepsilon } \right\}} \right) $满足B(xγ)⊂B(xr)和B(pγ)⊂B(pr).用(px)代替(px),用常数γ代替常数r,用球B(xγ),B(0,γ)和B(pγ)分别代替球B(xr),B(0,r)和B(pr),由条件(i),(ii)和(iii),分别得到下列结论:

    ① 任意的p′B(pγ)⊂B(pr),集值映射Hp′是闭的;

    ② 任意的p′B(pγ)⊂B(pr),任意的$ x' \in B\left( {x, \gamma } \right)\backslash \mathscr{ S}\left( {p'} \right) \subset B\left( {\overline x , r} \right)\backslash \mathscr{ S}\left( {p'} \right) $,任意的y′Πδ(0;Hp′(x′))∩B(0,γ)⊂Πδ(0;Hp′(x′))∩B(0,r),任意的y*Jδ(y′),任意的x*Dc*Hp′(x′y′)(y*),均有

    从而

    ③ 任意的(p′x′)∈B(pγB(xγ)⊂B(prB(xr),集值映射H(·x′)在p′是下半连续的.

    由上述结论,类似1)中的证明,存在常数t∈(0,γ)使得

    因为$ {\mathop{\rm int}} B\left( {x, \gamma } \right) \subset {\mathop{\rm int}} B\left( {\overline x , r} \right) \cap {\mathop{\rm int}} B\left( {x, \tilde \varepsilon } \right) $,由(9)式可得

    注1  从定理1的证明过程可以看出,若将H换成一般的集值映射$ F:P \times X_ \to ^ \to Y $,将有效解映射$ \mathscr{ S} $换成(1)式中的集值隐函数G,则类似可以证明:集值隐函数G在开球上的限制在给定点p周围是非空下半连续的,即定理1的结论不仅对参数向量优化问题的有效解映射成立,对一般的集值隐函数也成立.

    由定理1可知,若H满足一定的条件,则有效解映射$ \mathscr{ S} $在开球上的限制在给定点周围是非空下半连续的.而H是由f$ \mathscr{ F} $构成的,故可以将定理1中的关于H的条件换为由f$ \mathscr{ F} $刻画的条件,从而得到定理2.

    定理2  设XY是Banach空间,P是度量空间,fP×XY是连续单值映射,KY是顶点在原点的尖闭凸锥,$ \mathscr{F}:P_ \to ^ \to Y 和 \mathscr{S} :P_ \to ^ \to X $分别是参数向量优化问题(2)的有效点映射和有效解映射$ H:P \times X_ \to ^ \to Y $是由(3)式定义的集值映射.进一步,(px)∈P×X且(px)∈gph$ \mathscr{ S} $.若存在常数r>0,满足定理1的条件(ii)和下列条件:

    (i) 集值映射$ \mathscr{ F} $是闭的;

    (ii) 任意的pB(pr),集值映射$ \mathscr{ F} $p是下半连续的.

    则存在常数s∈(0,r),使得集值映射$ \mathscr{\tilde S}:P_ \to ^ \to X $B(ps)上是非空下半连续的,其中$ \mathscr{\tilde S} $的定义为

    即有效解映射$ \mathscr{ S} $在开球上的限制在p周围是非空下半连续的.

      由集值映射$ \mathscr{ F} $是闭的,单值映射f连续,可以证明:任意的pB(pr),集值映射Hp是闭的,即定理1的条件(i)满足.进一步,任意的(px)∈B(prB(xr),由f连续和$ \mathscr{ F} $p下半连续,可以证明:H(·x)在p下半连续,即定理1的条件(iii)满足.由定理1可得定理2成立.

    最后,讨论一般的集值映射的逆映射在开球上的限制的非空下半连续性.

    定理3  设XP是Banach空间,$ Φ :X_ \to ^ \to P $是集值映射,(xp)∈X×PpΦ(x).若Φ是闭的且存在常数r>0,使得对任意的δ>0,有

    其中:Πδ(0;Φ(x)-p):={yΦ(x)-p|║y║<d(0,Φ(x)-p)+δ},Jδ(y):={y*SY*|d(y*J(y))<δ}.则存在常数s∈(0, r)使得由

    定义的集值映射$ {\tilde \Phi ^{ - 1}}:P_ \to ^ \to X $B(ps)上是非空下半连续的.

      令Y:=P,定义$ F:P \times X_ \to ^ \to Y和 G:P_ \to ^ \to X $分别为

    显然

    容易验证定理1的所有条件满足.事实上,pΦ(x)等价于(px)∈gphG.记Fp(·):=F(p·).易知gphFp=gphΦ-(0,p),∀pP.因为Φ是闭的,所以任意的pPFp是闭的,从而定理1的条件(i)满足.进一步,可以证明

    从而

    即定理1的条件(ii)满足.由F的定义易知,任意的(px)∈B(prB(xr),集值映射F(·x)在p是下半连续的,即定理1的条件(iii)满足,由定理1和注1可得定理3成立.

Reference (10)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return