Você está na página 1de 9

J Optim Theory Appl (2010) 147: 507515 DOI 10.

1007/s10957-010-9736-8

Lower Semicontinuity of the Solution Mappings to a Parametric Generalized Ky Fan Inequality


S.J. Li Z.M. Fang

Published online: 21 July 2010 Springer Science+Business Media, LLC 2010

Abstract In this paper, we investigate weak vector solutions and global vector solutions to a generalized Ky Fan inequality. Under new assumptions, which are weaker than the assumption of strict C -mappings, we establish the lower semicontinuity of the solution mappings to a parametric generalized Ky Fan inequality by using a scalarization method. These results extend the corresponding ones in the literature. Some examples are given to illustrate our results. Keywords Lower semicontinuity Parametric generalized Ky Fan inequality Weak vector solution Global vector solution Scalarization 1 Introduction Since Ky Fan inequality was introduced in [1], it has been extended and generalized to vector-valued mappings. The Ky Fan inequality for a vector valued mapping is known as the generalized Ky Fan inequality (GKFI, in short). The existence and stability results for parametric GKFIs have been investigated intensively; see [210], where GKFIs are called generalized systems or vector equilibrium problems.
Communicated by F. Giannessi. This research was partially supported by the National Natural Science Foundation of China (Grant number: 10871216), Chongqing University Postgraduates Science and Innovation Fund (Project Number: 201005B1A0010338) and Innovative Talent Training Project, the Third Stage of 211 Project, Chongqing University (Project number: S-09110). The authors would like to thank Professor Franco Giannessi for valuable comments and suggestions, which helped to improve the paper. S.J. Li ( ) Z.M. Fang College of Mathematics and Science, Chongqing University, Chongqing, 400044, China e-mail: lisj@cqu.edu.cn Z.M. Fang e-mail: fangzhimiao1983@163.com

508

J Optim Theory Appl (2010) 147: 507515

In [7], by virtue of a density result and a scalarization technique, Gong and Yao discussed the lower semicontinuity of the vector solutions to a parametric GKFI. By using the idea of Cheng and Zhu [11], Gong [8] discussed the continuity of the solution mapping to a parametric weak GKFI in topological vector spaces. Chen and Li [9] discussed the continuity of the solution mappings of vector solutions, weak vector solutions and various proper vector solutions to a parametric GKFI in a real locally convex Hausdorff topological vector space without the uniform compactness assumption, which improved the corresponding results of [7, 8]. In [10], Chen et al. investigated a parametric GKFI involving set-valued mappings. They established the lower semicontinuity and continuity of the solution mapping by using a new proof, which is different from the ones of [7, 8]. In [12], Li et al. obtained the sufcient conditions for the lower semicontinuity and continuity of the solution mapping to a parametric GKFI with set-valued mappings, which is different from the model considered in [10]. In the papers [710, 12], where the semicontinuity has been discussed by virtue of the scalarization technique, the f -solution set (see Denition 2.3) is a singleton for a xed linear continuous functional f in a dual cone because of the assumption of strict C -mappings (see Denition 2.5) for the vector-valued or set-valued mappings. However, in general, the f -solution set should be a general set, but not a singleton. So, in the present paper, we aim at weakening this assumption and studying the lower semicontinuity of the weak vector solutions and global vector solutions to a parametric GKFI under the case where the f -solution set is a general set. The rest of the paper is organized as follows. In Sect. 2, we introduce a parametric GKFI, and recall some concepts and their properties. In Sects. 3 and 4, based on a scalarization method, we discuss the lower semicontinuity of the weak vector solutions and global vector solutions to the parametric GKFI, respectively.

2 Preliminaries Throughout this paper, let X , Y and Z be Banach spaces and C be a pointed, closed and convex cone in Y with nonempty topological interior int C . Let Y be the topological dual space of Y and C := {f Y |f (y) 0, y C } be the dual cone of C . Denote the quasi-interior of C by C , i.e., C := {f Y |f (y) > 0, y C \{0}}. Let A be a nonempty subset of X and F : A A Y be a vector-valued mapping. Consider the following generalized Ky Fan inequality Find x A such that F (x, y) K, y A, where K {0} is a convex cone in Y . When the subset A and the mapping F are perturbed by a parameter , which varies over a subset of Z , we consider the following parametric generalized Ky Fan inequality Find x A() such that F (x, y, ) K, y A(), where A : 2X \{} is a set-valued mapping, F : B B is a vector-valued mapping with A( ) = A() B . XXZY

J Optim Theory Appl (2010) 147: 507515

509

Denition 2.1 [13] A vector x A() is called a weak vector solution to the parametric GKFI, iff F (x, y, ) int C, y A().

The set of weak vector solutions to the parametric GKFI is denoted by S(). Denition 2.2 [14] A vector x A() is called a global vector solution to the parametric GKFI, iff there exists a point, convex cone H Y with C \{0} int H , such that F (x, A, ) ((H )\{0}) = , where F (x, A(), ) = y A() F (x, y, ). The set of global vector solutions to the parametric GKFI is denoted by SG (). Throughout this paper, we always assume S() = and SG () = for all . This paper aims at investigating the lower semicontinuity of the solution mappings S() and SG (). Now we recall some basic denitions and their properties. Given any . Denition 2.3 [13] Let f C \{0}. A vector x A() is called an f -solution to the parametric GKFI, iff f (F (x, y, )) 0, y A().

The set of the f -solutions to the parametric GKFI is denoted by Vf (). Denition 2.4 [15] Let f C . A vector x A() is called a positive proper solution to the parametric GKFI, iff f (F (x, y, )) 0, y A().

P (). The set of the positive proper solutions to the parametric GKFI is denoted by Vf

Denition 2.5 (i) The mapping F : A A Y is a C -mapping on A A, iff F (x, y) + F (y, x) C , for all x, y A. When C is contained in the non-negative orthant, then F is called C -antitone (or C -monotone nonincreasing); if C equals the nonnegative orthant, F is antitone (or monotone nonincreasing). When C is contained in the non-positive orthant, then F is called C -isotone (or C -monotone nondecreasing); if C equals the non-positive orthant, F is isotone (or monotone nondecreasing). (ii) The mapping F : A A Y is a strict C -mapping on A A, iff F is a C mapping on A A and for all x, y A with x = y , F (x, y) + F (y, x) int C . When C is contained in the non-negative orthant, then F is called strictly C antitone (or strictly C -monotone nonincreasing); if C equals the non-negative orthant, F is strictly antitone (or strictly monotone nonincreasing). When C is

510

J Optim Theory Appl (2010) 147: 507515

contained in the non-positive orthant, then F is called strictly C -isotone (or strictly C -monotone nondecreasing); if C equals the non-positive orthant, F is strictly isotone (or strictly monotone nondecreasing). (iii) A set D Y is called a C -convex set, iff its conic extension D + C is a convex set in Y . Lemma 2.1 [14] Let F (x, A(), ) be a C -convex set for each A(). (i) If int C = , S() = f C \{0} Vf (); P (). (ii) If C has a base, SG () = f C Vf , ) denotes the open ball with center The notion B( . F : 2X be a set-valued mapping, and given Denition 2.6 [16] , iff for any open set V (i) F is called lower semicontinuous (l.s.c., in short) at = , there exists > 0, such that for every B(, ), satisfying V F () V F () = . , iff for any open set V (ii) F is called upper semicontinuous (u.s.c., in short) at V , there exists > 0, such that for every B(, ), F () satisfying F () V. We say F is l.s.c. (resp. u.s.c.) on , iff it is l.s.c. (resp. u.s.c.) at each said to be continuous on , iff it is both l.s.c. and u.s.c. on . Proposition 2.1 [17, 18] if and only if for any sequence {n } with n and any (i) F is l.s.c. at , there exists xn F (n ), such that xn x x F () . (ii) If F has compact values (i.e., F () is a compact set for each ), then F if and only if for any sequences {n } with n and {xn } is u.s.c. at and a subsequence {xnk } of {xn }, such that with xn F (n ), there exist x F () xnk x . We shall also deal with the property of lower semicontinuity proposed by Berge (Theorem 2, p. 114, [19]), which plays an important role in the proof of lower semicontinuity of the solution mappings S() and SG (). Lemma 2.2 [19] The union = i I i of a family of l.s.c. set-valued mappings i from a topological space X into a topological space Y is also an l.s.c. set-valued mapping from X into Y , where I is an index set. . F is and radius > 0. Let and x

3 Lower Semicontinuity to Weak Vector Solutions In this section, we discuss the lower semicontinuity of the weak vector solutions to the parametric GKFI.

J Optim Theory Appl (2010) 147: 507515

511

Lemma 3.1 Let f C \{0}. Suppose that the following conditions be satised: (i) A() is continuous with compact convex values on ; (ii) For each , F (, , ) is continuous on A() A() ; (iii) For each , x A()\Vf (), there exists y Vf (), such that F (x, y, ) + F (y, x, ) + B(0, d(x, y)) C. Then, Vf () is l.s.c. on .

Proof Suppose to the contrary that there exists 0 , such that Vf () be not l.s.c. at 0 . Then, there exist a sequence {n } with n 0 and x0 Vf (0 ), such that for any xn Vf (n ), xn x0 . Since x0 A(0 ) and A() is l.s.c. at 0 , then there exists x n A(n ), such that x n x0 . Obviously, x n A(n )\Vf (n ). By (iii), there exists yn Vf (n ), such that F (x n , yn , n ) + F (yn , x n , n ) + B(0, d(x n , yn )) C. (1)

For yn A(n ), from the upper semicontinuity and compactness of A() at 0 , it follows that there exist y0 A(0 ) and a subsequence {ynk } of {yn }, such that ynk y0 . In particular, for (1), we have F (x nk , ynk , nk ) + F (ynk , x nk , nk ) + B(0, d(x nk , ynk )) C. Taking the limit as nk +, we have F (x0 , y0 , 0 ) + F (y0 , x0 , 0 ) + B(0, d(x0 , y0 )) C. Noting that x0 Vf (0 ) and y0 A(0 ), we can obtain f (F (x0 , y0 , 0 )) 0. (3) (2)

On the other hand, for ynk Vf (nk ) and x nk A(nk ), we get f (F (ynk , x nk , nk )) 0. Moreover, it follows from the continuity of f and F that f (F (y0 , x0 , 0 )) 0. By (3), (4) and the linearity of f , we get f (F (x0 , y0 , 0 ) + F (y0 , x0 , 0 )) 0. (5) (4)

If x0 = y0 , by (2), we can obtain F (x0 , y0 , 0 ) + F (y0 , x0 , 0 ) int C . Thus, we have f (F (x0 , y0 , 0 ) + F (y0 , x0 , 0 )) < 0, which contradicts (5). Therefore, we have x0 = y0 . This is impossible by the contradiction assumption. Thus, our result holds and the proof is complete.

512

J Optim Theory Appl (2010) 147: 507515

Remark 3.1 In [710, 12], under the condition of strict C -mappings, the continuity of the f -solutions to the parametric GKFI is obtained. However, this condition is so strict that the f -solution set is conned to be a singleton. In our paper, we introduce the assumption (iii) of Lemma 3.1 to weaken this condition. In the case, the f -solution set may be a general set, but not a singleton. Now we give the following example to show that.
2, Example 3.1 Let X = R , Y = R 2 , C = R+ = [1, 2], A() = [1, 1] and 1 F (x, y, ) = (1, x). For any given , let f ((x, y)) = y C \{0}. It follows from a direct computation that Vf () = [0, 1]. Obviously, the f -solution set to the parametric GKFI is not a singleton, but a general set. The assumption (iii) in Lemma 3.1 can be checked as follows: For any x A\Vf () = [1, 0), there exists y = 0 Vf (), such that

F (x, y, ) + F (y, x, ) + B(0, d(x, y)) C. However, the condition that F is a strict C -mapping in Lemma 4.2 of [8] is violated. Indeed, for x A\Vf () = [1, 0), there exists y Vf () with y = x , such that F (x, y, ) + F (y, x, ) = (2, 0) C \{0} where C is the boundary of C . Obviously, F (x, y, ) + F (y, x, ) int C , which implies that F (, , ) is not a strict C -mapping on A() A(). Theorem 3.1 For each f C \{0}, suppose that the following conditions be satised: (i) (ii) (iii) (iv) A() is continuous with compact convex values on ; For each , F (, , ) is continuous on A() A() ; For each and x A(), F (x, , ) is C -convex on A(); For each and x A()\Vf (), there exists y Vf (), such that F (x, y, ) + F (y, x, ) + B(0, d(x, y)) C. Then, S() is l.s.c. on .

Proof For each and x A(), since F (x, , ) is C -convex on A(), then F (x, A(), ) + C is a convex set. Thus, it follows from Lemma 2.1 that for each , S() =
f C \{0}

Vf ().

By Lemma 3.1, for each f C \{0}, Vf () is l.s.c. on . Therefore, in view of Lemma 2.2, we have S() is l.s.c. on . The proof is complete. Now, we give an example to illustrate that our result extends that of [8].

J Optim Theory Appl (2010) 147: 507515

513

2, Example 3.2 Let X = R , Y = R 2 , C = R+ = [1, 2], A() = [1, 0] and F (x, y, ) = (1, x). For any f C \{0}, it follows from a direct computation that, if Vf () = , then 0 Vf () . Clearly, conditions (i)(iii) of Theorem 3.1 are satised. For any x A()\Vf (), there exists y = 0 Vf (), such that

F (x, y, ) + F (y, x, ) + B(0, d(x, y)) C. Thus, the condition (iv) of Theorem 3.1 is also satised. By Theorem 3.1, S() is lower semicontinuous on . However, for x =y = 0, F (x, y, ) + F (y, x, ) = (2, 0) C \{0}, where C is the boundary of C . Obviously, F (x, y, ) + F (y, x, ) int C , i.e., F (, , ) is not a strict C -mapping on A() A(). Thus, Theorem 4.1 in [8] is not applicable.

4 Lower Semicontinuity to Global Vector Solutions In this section, we discuss the lower semicontinuity of the global vector solutions to the parametric GKFI. Lemma 4.1 Let f C . Suppose that the following conditions be satised: (i) A() is continuous with compact convex values on ; (ii) For each , F (, , ) is continuous on A() A() ; P (), there exists y V P (), such that (iii) For each and x A()\Vf f F (x, y, ) + F (y, x, ) C \{0}; (iv) For each and x = y , F (x, y, ) + F (y, x, ) = 0. .

P () is l.s.c. on Then, Vf

Proof In a way similar to the proof of Lemma 3.1, with suitable modications, we can obtain that the conclusion of Lemma 4.1 holds. Theorem 4.1 Suppose that C = , and the following conditions be satised: (i) (ii) (iii) (iv) For each and x A(), F (x, A(), ) is a C -convex set; A() is continuous with compact convex values on ; For each , F (, , ) is continuous on A() A() ; P (), there exists y V P (), such For each f C , and x A()\Vf f that F (x, y, ) + F (y, x, ) C \{0}; (v) For each and x = y , F (x, y, ) + F (y, x, ) = 0. .

Then, SG () is l.s.c. on

514

J Optim Theory Appl (2010) 147: 507515

Proof For each and x A(), since F (x, A(), ) is a C -convex set, then it follows from Lemma 2.1 that for each , SG () =
f C P () is l.s.c. on By Lemma 4.1, for each f C , Vf . Therefore, in view of Lemma 2.2, we have SG () is l.s.c. on . The proof is complete. P Vf ().

Remark 4.1 In Theorem 5.1 of [9], Chen and Li dealt with the continuity of the global vector solutions by the continuity of the of positive proper solutions under the condition of strict C -mappings. In our paper, we weaken this condition and improve the corresponding results of [9]. We give the following example to illustrate the case.
2, = [1, 2], A() = [1, 1] and Example 4.1 Let X = R , Y = R 2 , C = R+ F (x, y, ) = ((x y), 1). It is easy to verify that all assumptions of Theorem 4.1 are satised. By Theorem 4.1, SG () is lower semicontinuous on . However, the condition that F is a strict C -mapping is violated. Indeed, for any x, y A(), we have F (x, y, ) + F (y, x, ) = (0, 2) C \{0}, where C is the boundary of C , which implies that F (, , ) is not a strict C -mapping on A() A(). Then, Theorem 5.1 in [9] is not applicable.

5 Conclusion In this paper, by using a scalarization method, we have obtained the lower semicontinuity of the weak vector solutions and global vector solutions to a parametric GKFI under new assumptions, which are weaker than strict C -mappings. These results extend and improve the corresponding ones obtained in [710, 12]. We have also showed some examples to illustrate the case.

References
1. Fan, K.: Extensions of two xed point theorems of F.E. Browder. Math. Z. 112, 234240 (1969) 2. Anh, L.Q., Khanh, P.Q.: Semicontinuity of the solution set of parametric multivalued vector quasiequilibrium problems. J. Math. Anal. Appl. 294, 699711 (2004) 3. Anh, L.Q., Khanh, P.Q.: On the stability of the solution sets of general multivalued vector quasiequilibrium problems. J. Optim. Theory Appl. 135, 271284 (2007) 4. Huang, N.J., Li, J., Thompson, H.B.: Stability for parametric implict vector equilibrium problems. Math. Comput. Model. 43, 12671274 (2006) 5. Hou, S.H., Gong, X.H., Yang, X.M.: Existence and stability of solutions for generalized Ky Fan inequality problems with trifunctions. J. Optim. Theory Appl. (2010). DOI:10.1007/s10957010-9656-7 6. Kimura, K., Yao, J.C.: Sensitivity analysis of solution mappings of parametric vector quasiequilibrium problems. J. Glob. Optim. 41, 187202 (2008) 7. Gong, X.H., Yao, J.C.: Lower semicontinuity of the set of the efcient solutions for generalized systems. J. Optim. Theory Appl. 138, 197205 (2008) 8. Gong, X.H.: Continuity of the solution set to parametric weak vector equilibrium problems. J. Optim. Theory Appl. 139, 3546 (2008)

J Optim Theory Appl (2010) 147: 507515

515

9. Chen, C.R., Li, S.J.: On the solution continuity of parametric generalized systems. Pac. J. Optim. 6, 141151 (2010) 10. Chen, C.R., Li, S.J., Teo, K.L.: Solution semicontinuity of parametric generalized vector equilibrium problems. J. Glob. Optim. 45, 309318 (2009) 11. Cheng, Y.H., Zhu, D.L.: Global stability results for the weak vector variational inequality. J. Glob. Optim. 32, 543550 (2005) 12. Li, S.J., Liu, H.M., Chen, C.R.: Lower semicontinuity of parametric generalized weak vector equilibrium problems. Bull. Aust. Math. Soc. 81, 8595 (2010) 13. Gong, X.H.: Efciency and Henig efciency for vector equilibrium problems. J. Optim. Theory Appl. 108, 139154 (2001) 14. Gong, X.H.: Connectedness of the solution sets and scalarization for vector equilibrium problems. J. Optim. Theory Appl. 133, 151161 (2007) 15. Gong, X.H., Yao, J.C.: Connectedness of the set of efcient solutions for generalized systems. J. Optim. Theory Appl. 138, 189196 (2008) 16. Kien, B.T.: On the lower semicontinuity of optimal solution sets. Optimization 54, 123130 (2005) 17. Aubin, J.P., Ekeland, I.: Applied Nonlinear Analysis. Wiley, New York (1984) 18. Ferro, F.: A minimax theorem for vector-valued functions. J. Optim. Theory Appl. 60, 1931 (1989) 19. Berge, C.: Topological Spaces. Oliver and Boyd, London (1963)

Você também pode gostar