Skip to main content

Second-order composed contingent epiderivatives and set-valued vector equilibrium problems

Abstract

In this paper, we introduce the concept of a second-order composed contingent epiderivative for set-valued maps and discuss some of its properties. Then, by virtue of the second-order composed contingent epiderivative, we establish second-order sufficient optimality conditions and necessary optimality conditions for the weakly efficient solution of set-valued vector equilibrium problems with unconstraints and set-valued vector equilibrium problems with constraints, respectively.

MSC:90C46, 91B50.

1 Introduction

The vector equilibrium problem, which contains vector optimization problems, vector variational inequality problems and vector complementarity problems as special case, has been studied (see [114]). But so far, most papers focused mainly on the existence of solutions and the properties of the solutions, there are a few papers which deal with the optimality conditions. Giannessi et al. [15] turned the vector variational inequalities with constraints into another vector variational inequalities without constraints. They gave the sufficient conditions for the efficient solution and the weakly efficient solution of the vector variational inequalities in finite dimensional spaces. Morgan and Romaniello [16] gave the scalarization and Kuhn-Tucker-like conditions for weak vector generalized quasivariational inequalities in Hilbert space by using the subdifferential of the function. Gong [17] presented the necessary and sufficient conditions for the weakly efficient solution, the Henig efficient solution and the superefficient solution for vector equilibrium problems with constraints under the condition of cone-convexity. Qiu [18] presented the necessary and sufficient conditions for globally efficient solutions of vector equilibrium problems under generalized cone-subconvexlikeness. Gong and Xiong [19] weakened the convexity assumption in [17] and obtained the necessary and sufficient conditions for weakly efficient solutions of vector equilibrium problems. Under the nearly cone-subconvexlikeness, Long et al. [20] obtained the necessary and sufficient conditions for the Henig efficient solution and the superefficient solution to the vector equilibrium problems with constraints. By using the concept of Fréchet differentiability of mapping, Wei and Gong [21] obtained the Kuhn-Tucker optimality conditions for weakly efficient solutions, Henig efficient solutions, superefficient solutions and globally efficient solutions to the vector equilibrium problems with constraints. Ma and Gong [22] obtained the first-order necessary and sufficient conditions for the weakly efficient solution, the Henig efficient solution, and the globally proper efficient solution to the vector equilibrium problems with constraints.

It is well known that the second-order tangent sets and higher-order tangent sets introduced in [23], in general, are not cones and convex sets, there are some difficulties in studying second-order and higher-order optimality conditions for general set-valued optimization problems. Until now, there are some papers to deal with higher-order optimality conditions by virtue of the higher-order derivatives or epiderivatives introduced by the higher-order tangent sets (see [2432]). However, as far as we know the second-order optimality conditions of the solutions remain unstudied in set-valued vector equilibrium problems.

Motivated by the work reported in [22, 24, 2631, 33], we introduce a new second-order derivative called second-order composed contingent epiderivative for set-valued maps and obtain some of its properties. By virtue of the second-order composed contingent epiderivative, we obtain second-order sufficient optimality conditions and necessary optimality conditions for the weakly efficient solution of set-valued vector equilibrium problems.

The rest of the paper is organized as follows. In Section 2, we recall some notions. In Section 3, we introduce second-order composed contingent epiderivatives for set-valued maps and discuss some of its properties. In Section 4, we establish second-order necessary and sufficient optimality conditions for weakly efficient solutions to set-valued vector equilibrium problems.

2 Preliminaries and notations

Throughout this paper, let X, Y, and Z be three real normed spaces, Y and Z be the topological dual spaces of Y and Z, respectively. 0 X , 0 Y and 0 Z denote the origins of X, Y, and Z, respectively. Let CY and DZ be closed convex pointed cones in Y and Z, respectively. Let M be a nonempty subset in Y. The cone hull of M is defined by cone(M)={ty|t0,yM}. Let C be the dual cone of coneC, defined by

C = { y Y : y ( c ) 0 ,  for all  c C } .

Let E be a nonempty subset of X, G:E 2 Z be a set-valued map. The domain, the graph and the epigraph of G are defined, respectively, by

dom ( G ) = { x E | G ( x ) } , graph ( G ) = { ( x , z ) X × Z | x E , z G ( x ) } , epi ( G ) = { ( x , z ) X × Z | x E , z G ( x ) + C } .

Denote

G(E)= x E G(x)and(G z 0 )(x)=G(x){ z 0 }.

Definition 2.1 (see [33, 34])

A set-valued map W:X 2 Y is said to be

  1. (i)

    strictly positive homogeneous if

    W(αx)=αW(x),α>0,xX;
  2. (ii)

    subadditive if

    W( x 1 )+W( x 2 )W( x 1 + x 2 )+C.

Definition 2.2 (see [34, 35])

Let EX be a nonempty convex set and G:E 2 Z be a set-valued map with G(x), for all xE. G is said to be D-convex on E, if for any x 1 , x 2 E and λ(0,1),

λG( x 1 )+(1λ)G( x 2 )G ( λ x 1 + ( 1 λ ) x 2 ) +D.

Let X be a normed space supplied with a distance d and K be a subset of X. We denote by d(x,K)= inf y K d(x,y) the distance from x to K, where we set d(x,)=+.

Definition 2.3 (see [23, 36])

Let K be a nonempty subset of X and xK, uX. The contingent cone of K at x is

T(K,x):={vX| t n 0, v n v, such that x+ t n v n K,nN}.

Proposition 2.1 (see [36])

Let KX and xK. The following statements are equivalent:

  1. (i)

    uT(K,x);

  2. (ii)

    there exist sequences { λ n } with λ n + and { x n } with x n K and x n x such that λ n ( x n x)v.

Proposition 2.2 (see [23])

Let KX and xK. Then T(K,x) is a closed cone.

Proposition 2.3 (see [37])

Let KX be a convex set, xK, and uT(K,x). Then

T ( T ( K , x ) , u ) =clcone ( cone ( K x ) u ) .

Let E be a nonempty subset of X, F:E×E 2 Y be a set-valued bifunction, F( x 1 , x 2 ), for all x 1 , x 2 E. We suppose that 0 Y F(x,x), for any xE.

Let x 0 E be given. F x 0 :E 2 Y is the set-valued map defined by

graph( F x 0 )= { ( x , y ) E × Y : y F ( x 0 , x ) } .

The set

epi( F x 0 )= { ( x , y ) E × Y : y F ( x 0 , x ) + C }

is called the epigraph of F x 0 . Denote

F x 0 (E)=F( x 0 ,E)= { y F ( x 0 , x ) : x E } .

Let G:E 2 Z be a set-valued map with G(x), for all xE.

In this paper, we consider the set-valued vector equilibrium problem with unconstraints (USVVEP): find x 0 E such that

F( x 0 ,x)( A 0 )=,for all xE,

where A 0 =A{ 0 Y }, A is a convex cone in Y.

We also consider the set-valued vector equilibrium problem with constraints (CSVVEP): find x 0 K such that

F( x 0 ,x)( A 0 )=,for all xK,

where A 0 =A{ 0 Y }, A is a convex cone in Y, and K:={xE:G(x)(D)}.

Definition 2.4 Let intC.

  1. (i)

    A vector x 0 E is called a weakly efficient solution of (USTVEP) if

    F( x 0 ,E)(intC)=.
  2. (ii)

    A vector x 0 K is called a weakly efficient solution of (CSTVEP) if

    F( x 0 ,K)(intC)=.

3 Second-order composed contingent epiderivatives

Let EX. Let F:E 2 Y be a set-valued map, y 0 F( x 0 ), and (u,v)X×Y. We first recall the definition of the generalized second-order composed contingent epiderivative introduced by Li et al. [24].

Definition 3.1 (see [24])

The generalized second-order composed contingent epiderivative D g ′′ F( x 0 , y 0 ,u,v) of F at ( x 0 , y 0 ) in the directive (u,v) is the set-valued map from X to Y defined by

D g ′′ F( x 0 , y 0 ,u,v)(x)= Min C { y Y | ( x , y ) T ( T ( epi ( F ) , ( x 0 , y 0 ) ) , ( u , v ) ) } .

Now we introduce the following second-order composed contingent epiderivatives of set-valued maps, and then we investigate some of its properties.

Definition 3.2 Let ( x 0 , y 0 )graph(F), (u,v)X×Y. The second-order composed contingent epiderivative D F + ( x 0 , y 0 ,u,v) of F at ( x 0 , y 0 ) in the directive (u,v) is the set-valued map from X to Y defined by

D F + ( x 0 , y 0 ,u,v)(x)= { y Y | ( x , y ) T ( T ( epi ( F ) , ( x 0 , y 0 ) ) , ( u , v ) ) } ,xX.

Proposition 3.1 Let ( x ˆ , y ˆ )graph(F), ( u ˆ , v ˆ )X×Y, and M:=dom D F + ( x ˆ , y ˆ , u ˆ , v ˆ ). Then

D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x)+C= D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x),xM.
(1)

Proof Let xM, y D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x)+C. Then there exist (x, y ¯ )T(T(epiF,( x ˆ , y ˆ )),( u ˆ , v ˆ )), and cC, such that (x,y)=(x, y ¯ +c). Since (x, y ¯ )T(T(epiF,( x ˆ , y ˆ )),( u ˆ , v ˆ )), there exist sequences ( x n , y n )(x, y ¯ ) and t n 0 + , such that

( u ˆ , v ˆ )+ t n ( x n , y n )T ( epi F , ( x ˆ , y ˆ ) ) ,nN.

Moreover, nN, there exist sequences ( x n k , y n k )( u ˆ , v ˆ )+ t n ( x n , y n ) and t n k 0 + , such that ( x ˆ , y ˆ )+ t n k ( x n k , y n k )epiF, kN. Then we have

y ˆ + t n k y n k F ( x ˆ + t n k x n k ) +C,n,kN.
(2)

Since cC, combine with (2), we have y ˆ + t n k ( y n k + t n c)= y ˆ + t n k y n k + t n k t n cF( x ˆ + t n k x n k )+C, n,kN. That is ( x ˆ , y ˆ )+ t n k ( x n k , y n k + t n c)epi F + , n,kN. Since ( x n k , y n k )( u ˆ , v ˆ )+ t n ( x n , y n ), we have ( x n k , y n k + t n c)( u ˆ , v ˆ )+ t n ( x n , y n +c) as k+. Thus,

( u ˆ , v ˆ )+ t n ( x n , y n +c)T ( epi F , ( x ˆ , y ˆ ) ) ,nN.

Simultaneously, ( x n , y n +c)(x, y ¯ +c), since ( x n , y n )(x, y ¯ ) as n+. Together with (x,y)=( x ¯ , y ¯ +c), we have (x,y)=T(T(epiF,( x ˆ , y ˆ )),( u ˆ , v ˆ )), which implies

y D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x).

So

D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x)+C D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x).

Naturally, D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x) D F + ( x ˆ , y ˆ , u ˆ , v ˆ )(x)+C. Thus (1) holds, and this completes the proof. □

By definitions and Proposition 3.1, we can conclude that the following result holds.

Proposition 3.2 Let F:E 2 Y , ( x 0 , y 0 )graph(F), (u,v)T(epi(F),( x 0 , y 0 )), and xX. Then D g ′′ F( x 0 , y 0 ,u,v)(x)+C D F + ( x 0 , y 0 ,u,v)(x).

Remark 3.1 The inclusion relation

D F + ( x 0 , y 0 ,u,v)(x) D g ′′ F( x 0 , y 0 ,u,v)(x)+C

may not hold.

Now we give the following example to explain Remark 3.1.

Example 3.1 Let C= R + 2 and F(x)={( y 1 , y 2 ) R 2 | y 1 x 2 , y 2 R}, x R + . Let ( x 0 , y 0 )=(0,(0,0)), and (u,v)=(1,(0,1)). Then

T ( T ( epi ( F ) , ( x 0 , y 0 ) ) , ( u , v ) ) = { ( x , ( y 1 , y 2 ) ) R × R 2 | x R , y 1 0 , y 2 R } .

Therefore, for any xR, we have

D g ′′ F( x 0 , y 0 ,u,v)(x)=

and

D F + ( x 0 , y 0 ,u,v)(x)= { ( y 1 , y 2 ) | y 1 0 , y 2 R } .

And then, for any xR, we have

D F + ( x 0 , y 0 ,u,v)(x) D g ′′ F( x 0 , y 0 ,u,v)(x)+C.

Now we discuss some crucial properties of the second-order composed contingent epiderivative.

Proposition 3.3 Let ( x 0 , y 0 )graphF, (u,v)T(epiF,( x 0 , y 0 )) with vC and EX be convex. If F is C-convex on E, then for all xE,

F(x){ y 0 }+C D F + ( x 0 , y 0 ,u,v)(x x 0 u).

Proof Since F is C-convex on E, epiF is a convex set. So it follows from Proposition 2.3 that

T ( T ( epi F , ( x 0 , y 0 ) ) , ( u , v ) ) =clcone ( cone ( epi F { ( x 0 , y 0 ) } ) { ( u , v ) } ) .
(3)

Since for every cC, xS and yF(x), one has

( x x 0 u , y y 0 + c ) = ( x x 0 u , y + c + v y 0 v ) { x } × ( F ( x ) + C ) { ( x 0 , y 0 ) } { ( u , v ) } .

Then it follows from (3) that

(x x 0 u,y y 0 )T ( T ( epi F , ( x 0 , y 0 ) ) , ( u , v ) ) .

Thus, by the definition of the second-order composed contingent epiderivative, we have

y y 0 +c D F + ( x 0 , y 0 ,u,v)(x x 0 u),

and then

F(x){ y 0 }+C D F + ( x 0 , y 0 ,u,v)(x x 0 u),xE.

The proof is complete. □

Proposition 3.4 Let ( x 0 , y 0 )graph(F), (u,v)T(epiF,( x 0 , y 0 )). Then

  1. (i)

    D F + ( x 0 , y 0 ,u,v) is strictly positive homogeneous.

    Moreover, if F is C-convex on a nonempty convex set E, then

  2. (ii)

    D F + ( x 0 , y 0 ,u,v) is subadditive.

Proof (i) Let α>0 and xX.

If y D F + ( x 0 , y 0 ,u,v)(x), then there exist sequences { h n } with h n 0 + and {( x n , y n )} with ( x n , y n )T(epiF,( x 0 , y 0 )) such that

( x n , y n ) ( u , v ) h n (x,y),

and then

( x n , y n ) ( u , v ) 1 α h n (αx,αy).

So (αx,αy)T(T(epiF,( x 0 , y 0 )),u,v), and then we can obtain

αy D F + ( x 0 , y 0 ,u,v)(αx).

Thus

α D F + ( x 0 , y 0 ,u,v)(x) D F + ( x 0 , y 0 ,u,v)(x)(αx).
(4)

The proof of

D F + ( x 0 , y 0 ,u,v)(αx)α D F + ( x 0 , y 0 ,u,v)(x)(x)

follows along the lines of (4). So D F + ( x 0 , y 0 ,u,v) is strictly positive homogeneous.

  1. (ii)

    Let x 1 , x 2 X, y 1 D F + ( x 0 , y 0 ,u,v)( x 1 ), y 2 D F + ( x 0 , y 0 ,u,v)( x 2 ). Then one has

    ( x 1 , y 1 )T ( T ( epi F , ( x 0 , y 0 ) ) , ( u , v ) ) ,( x 2 , y 2 )T ( T ( epi F , ( x 0 , y 0 ) ) , ( u , v ) ) .

Since F is C-convex on S, epiF is convex, and then T(T(epiF,( x 0 , y 0 )),(u,v)) is a close and convex cone. Thus we have

( x 1 + x 2 , y 1 + y 2 )T ( T ( epi F , ( x 0 , y 0 ) ) , ( u , v ) ) ,

and then

y 1 + y 2 D F + ( x 0 , y 0 ,u,v)( x 1 + x 2 ).

Thus

D F + ( x 0 , y 0 ,u,v)( x 1 )+ D F + ( x 0 , y 0 ,u,v)( x 2 ) D F + ( x 0 , y 0 ,u,v)( x 1 + x 2 )+C,

and the proof is complete. □

By the proof of Proposition 3.4, we can conclude that the following result holds.

Proposition 3.5 Let ( x 0 , y 0 )graph(F), (u,v)T(epiF,( x 0 , y 0 )), M:=dom D F + ( x 0 , y 0 ,u,v). If F is C-convex on a nonempty convex set E, then D F + ( x 0 , y 0 ,u,v)(M) is a convex cone.

4 Second-order optimality conditions of weakly efficient solutions

Throughout this section, let x 0 K, y 0 = 0 Y F x 0 ( x 0 ), intC, and intD. Firstly, we recall a definition and a result in [25].

Let KX and x 0 K. The interior tangent cone of K at x 0 defined as

IT(K, x 0 )= { u X | δ > 0  such that  x 0 + t u K , t ( 0 , δ ] , u B X ( u , δ ) } ,

where B X (u,δ) stands for the closed ball centered at uX and of radius δ.

Lemma 4.1 (see [25])

If KX is convex, x 0 K, and intK, then

IT(intK, x 0 )=intcone(K x 0 ).

Theorem 4.1 Let x 0 be a weakly efficient solution of the problem (USVVEP). Then, for every (u,v)T(epi F x 0 ,( x 0 , y 0 )) with vC, we have

[ D ( F x 0 ) + ( x 0 , y 0 , u , v ) ( x ) + line { v } ] (intC)=,
(5)

for every xdom D ( F x 0 ) + ( x 0 , y 0 ,u,v).

Proof Suppose to the contrary that there exists an xdom D ( F x 0 ) + ( x 0 , y 0 ,u,v) such that (5) does not hold. Then there exist λR and

y D ( F x 0 ) + ( x 0 , y 0 ,u,v)(x)
(6)

such that

y :=y+λvintC.
(7)

Let us consider two possible cases for λ.

Case 1: If λ>0, then it follows from Proposition 3.1 and (6) that y D ( F x 0 ) + ( x 0 , y 0 ,u,v)(x). So

( x , y ) T ( T ( epi F x 0 , ( x 0 , y 0 ) ) , ( u , v ) ) .
(8)

By definition, there exist sequences λ n + and ( u n , v n )T(epiF,( x 0 , y 0 )) such that ( u n , v n )(u,v) and

λ n ( ( u n , v n ) ( u , v ) ) ( x , y ) ,as n+.
(9)

It follows from (9), (7), and vC that there exists N 1 N such that

v n intC,n> N 1 .
(10)

Since ( u n , v n )T(epi F x 0 ,( x 0 , y 0 )), for every nN, there exist a sequence λ n k + as k+ and a sequence ( x n k , y n k )epiF, such that ( x n k , y n k )( x 0 , y 0 ) and

λ n k ( ( x n k , y n k ) ( x 0 , y 0 ) ) ( u n , v n ),as k+.
(11)

It follows from (10) that there exists N 1 (n)N such that

λ n k ( y n k y 0 ) intC,k> N 1 (n),n> N 1 ,

which implies

y n k y 0 intC,k> N 1 (n),n> N 1 .
(12)

Since ( x n k , y n k )epi F x 0 , there exists y ¯ n k F x 0 ( x n k ) such that y n k { y ¯ n k }+C. Then, by (12), we have

y ¯ n k y 0 intC,k> N 1 (n),n> N 1 .

Therefore

F ( x 0 , x n k ) (intC),

which contradicts that ( x 0 , y 0 ) is a weakly efficient solution of the problem (USVVEP).

Case 2: If λ0, then it follows from Proposition vC and (7) that yintC and

(x,y)T ( T ( epi F x 0 , ( x 0 , y 0 ) ) , ( u , v ) ) .

By a similar proof method to case 1, there exist consequences x ¯ n k and N 2 , N 2 (n)N such that

F ( x 0 , x ¯ n k ) (intC),k> N 2 (n),n> N 2 ,

which contradicts that x 0 is a weakly efficient solution of the problem (USVVEP). Thus (5) holds, and the proof is complete. □

Remark 4.1 In Theorem 4.1, we cannot use vintC instead of vC. Since ( x 0 , y 0 ) is a weakly efficient solution of the problem (USVVEP),

F x 0 (x)(intC)=,xE.
(13)

It follows from (u,v)T(epi( F x 0 ),( x 0 , y 0 )), and y 0 =0 that there exist sequences { λ n } with λ n + and ( x n , y n )epi( F x 0 ) with ( x n , y n )( x 0 ,0) such that

λ n ( ( x n , y n ) ( x 0 , 0 ) ) (u,v),as n+.

So, it follows from vintC that there exists NN such that

λ n y n intC,n>N,

which implies

y n intC,n>N.
(14)

Since ( x n , y n )epi( F x 0 ), there exists y ¯ n F x 0 ( x n ) such that y n { y ¯ n }+C. Then, combined with (14), we have

y ¯ n intC,n>N.

Therefore

F x 0 ( x n )(intC), x n E,n>max{N, N 1 },

which contradicts (13).

Next, we give an example to illustrate Theorem 4.1.

Example 4.1 Let F(x)={( y 1 , y 2 ) R 2 | y 1 R, y 2 x 2 }, x R + , ( x 0 , y 0 )=(0,(0,0)), and C= R + 2 . Then T(epi(F),( x 0 , y 0 ))={(u,( v 1 , v 2 ))R× R 2 |x R + , v 1 R, v 2 0}. Take (u,v)=(1,(1,0)). Then

T ( T ( epi ( F ) , ( x 0 , y 0 ) ) , ( u , v ) ) = { ( x , ( y 1 , y 2 ) ) R × R 2 | x R , y 1 R , y 2 0 } .

Therefore, for any xR, we have

D F + ( x 0 , y 0 ,u,v)(x)+line{v}= { ( y 1 , y 2 ) | y 1 R , y 2 0 } .

And then, for any xR, we have

[ D F + ( x 0 , y 0 , u , v ) ( x ) + line { v } ] (intC)=,

which shows that Theorem 4.1 holds.

Theorem 4.2 Let (u,v)T(epi F x 0 ( x 0 , y 0 )) with vC and EX be convex. If F x 0 is C-convex on E, and for all xE,

D ( F x 0 ) + ( x 0 , y 0 ,u,v)(x x 0 u)(intC)=,
(15)

then x 0 is a weakly efficient solution of the problem (USVVEP).

Proof It follows from Proposition 3.3 that

( F x 0 )(x) D ( F x 0 ) + ( x 0 , y 0 ,u,v)(x x 0 u),xE.

Then, from (15), we have

( F x 0 )(x)(intC)=,xE.

So x 0 is a weakly efficient solution of (USVVEP), and the proof is complete. □

Theorem 4.3 Let (u,v,w)T(epi( F x 0 ,G), x 0 , y 0 , z 0 ) with vC and wD. If x 0 is a weakly efficient solution of (CSVVEP), then for any z 0 G( x 0 )(D),

D ( F x 0 , G ) + ( x 0 , y 0 , z 0 , u , v , w + z 0 ) ( x ) cone ( int ( C × D ) + line { ( v , w + z 0 ) } ) { ( 0 Y , 0 Z ) } = ,
(16)

for all xΩ:=dom[ D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w+ z 0 )].

Proof To prove the result by contradiction, suppose that there exists an xΩ such that (16) does not hold, that is, there exists a (y,z)Y×Z such that

(y,z) D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w+ z 0 )(x)

and

(y,z)cone ( int ( C × D ) + line { ( v , w + z 0 ) } ) { ( 0 Y , 0 Z ) } .
(17)

Then, by the definition of second-order composed contingent epiderivatives, there exist sequences λ n + and ( u n , v n , w n )T(epi( F x 0 ,G),( x 0 , y 0 , z 0 )) such that ( u n , v n , w n )(u,v,w+ z 0 ) and

λ n ( ( u n , v n , w n ) ( u , v , w + z 0 ) ) (x,y,z),as n+.
(18)

It follows from (17) that there exist μ>0, νR, cintC, and dintD such that

y=μ(c+νv),z=μ ( d + ν ( w + z 0 ) ) .
(19)

Let us consider two possible cases for ν.

Case 1: If ν0, then, from (19), vC, and w, z 0 D, we have yintC and zintD. Thus, by (18), there exists N 1 N such that

λ n ( v n v)intC, λ n ( w n ( w + z 0 ) ) intD,n> N 1 .

Thus, it follows from vC and wD that

v n intC, w n intD,n> N 1 .
(20)

Since ( u n , v n , w n )T(epi( F x 0 ,G),( x 0 , y 0 , z 0 )), for every nN, there exist a sequence { λ n k } with λ n k + as k+ and a sequence ( x n k , y n k , z n k )epi( F x 0 ,G), such that ( x n k , y n k , z n k )( x 0 , y 0 , z 0 ) and

λ n k ( ( x n k , y n k , z n k ) ( x 0 , y 0 , z 0 ) ) ( u n , v n , w n ),as k+.
(21)

It follows from (20) and (21) that there exists N 1 (n)N such that

λ n k ( y n k y 0 ) intC, λ n k ( z n k z 0 ) intD,k> N 1 (n),n> N 1 ,

which implies

y n k y 0 intC, z n k z 0 intD,k> N 1 (n),n> N 1 .
(22)

Since ( x n k , y n k , z n k )epi( F x 0 ,G), there exist y ¯ n k F x 0 ( x n k ), z ¯ n k G( x n k ), cC and dD such that y n k = y ¯ n k +c and z n k = z ¯ n k +d. Then, by (22) and y 0 =0, we have

y ¯ n k intC,k> N 1 (n), z ¯ n k intD,n> N 1 .

So

F x 0 ( x n k ) intC, x n k K,k> N 1 (n),n> N 1 ,

which contradicts that x 0 is a weakly efficient solution of (CSVVEP).

Case 2: If ν>0, then, from (19), we get y=μν( 1 ν c+v) and z=μν( 1 ν d+(w+ z 0 )). So it follows from cintC and dintD that

yintcone ( C + { v } ) ,z=intcone ( D + { w + z 0 } ) .
(23)

Then, by Lemma 4.1 and (23), we get yIT(intC,v) and z=IT(intD,w+ z 0 ). Therefore, there exists δ>0 such that

v+δ y intC, y B Y (y,δ),
(24)
w+ z 0 +δ z intD, y B Z (z,δ).
(25)

For this δ, it follows from (18) that there exists N 2 N such that

δ λ n >1, λ n ( v n v) B Y (y,δ), λ n ( w n ( w + z 0 ) ) B Z (z,δ),n> N 2 .

Then, by (24) and (25), we have

v n ( 1 1 δ λ n ) vintC, w n ( 1 1 δ λ n ) (w+ z 0 )intD,n> N 2 .

Thus, from vC, w, z 0 D, and δ λ n >1, n> N 2 , we have

v n intC, w n intD,n> N 2 .

By a similar proof method to case 1, there exists N 2 (n)N such that

F x 0 ( x n k ) intC, x n k K,k> N 2 (n),n> N 2 ,

which contradicts that x 0 is a weakly efficient solution of (CSVVEP). Thus (16) holds and the proof is complete. □

Theorem 4.4 Let (u,v,w)T(epi( F x 0 ,G), x 0 , y 0 , z 0 ) with vC and wD. If x 0 is a weakly efficient solution of (CSVVEP), then for any z 0 G( x 0 )(D),

D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w+ z 0 )(x) ( int ( C × D ) + { ( 0 , z 0 ) } ) =,
(26)

for all xΩ:=dom[ D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w+ z 0 )].

Proof To prove the result by contradiction, suppose that there exists an xΩ such that (26) does not hold, that is, there exists (y,z)Y×Z such that

( y , z ) D ( F x 0 , G ) + ( x 0 , y 0 , z 0 , u , v , w + z 0 ) ( x ) , ( y , z ) ( int ( C × D ) + { ( 0 , z 0 ) } ) .
(27)

Then, by the definition of second-order composed contingent epiderivatives, there exist sequences λ n + and ( u n , v n , w n )T(epi( F x 0 ,G),( x 0 , y 0 , z 0 )) such that ( u n , v n , w n )(u,v,w+ z 0 ) and

λ n ( ( u n , v n , w n ) ( u , v , w + z 0 ) ) (x,y,z),as n+.
(28)

It follows from (27) that there exist cintC and dintD such that

y=c,z=(d+ z 0 ).
(29)

Thus, by (28), there exists N 1 N such that

λ n >1, λ n ( v n v)intC, λ n ( w n ( w + z 0 ) ) (intD+ z 0 ),n> N 1 .

Thus, it follows from vC and w, z 0 D that

v n intC, w n intD,n> N 1 .
(30)

Since ( u n , v n , w n )T(epi( F x 0 ,G),( x 0 , y 0 , z 0 )), for every nN, there exist a sequence { λ n k } with λ n k + as k+ and a sequence ( x n k , y n k , z n k )epi( F x 0 ,G), such that ( x n k , y n k , z n k )( x 0 , y 0 , z 0 ) and

λ n k ( ( x n k , y n k , z n k ) ( x 0 , y 0 , z 0 ) ) ( u n , v n , w n ),as k+.
(31)

It follows from (30) and (31) that there exists N 1 (n)N such that

λ n k ( y n k y 0 ) intC, λ n k ( z n k z 0 ) intD,k> N 1 (n),n> N 1 ,

which implies

y n k y 0 intC, z n k z 0 intD,k> N 1 (n),n> N 1 .
(32)

Since ( x n k , y n k , z n k )epi( F x 0 ,G), there exist y ¯ n k F x 0 ( x n k ), z ¯ n k G( x n k ), cC, and dD such that y n k = y ¯ n k +c and z n k = z ¯ n k +d. Then, by (32) and y 0 =0, we have

y ¯ n k intC,k> N 1 (n), z ¯ n k intD,n> N 1 .

So

F x 0 ( x n k ) intC, x n k K,k> N 1 (n),n> N 1 ,

which contradicts that x 0 is a weakly efficient solution of (CSVVEP). Thus (26) holds and the proof is complete. □

Theorem 4.5 Let EX be a nonempty convex set, z 0 G( x 0 )(D) and (u,v,w)X×(C)×(D). Suppose that the following conditions are satisfied:

  1. (i)

    ( F x 0 ,G) is C×D-convex on E.

  2. (ii)

    x 0 is a weakly efficient solution of (CSVVEP).

Then there exist ϕ C and ψ D , not both zero functionals, such that

inf { ( y , z ) A ϕ ( y ) + ψ ( z ) } =0andψ( z 0 )=0,

where A:= x Ω D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w+ z 0 )(x) and Ω:=dom[ D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w+ z 0 )].

Proof Define M=A+( 0 Y , z 0 ). By Proposition 3.5, we see that M is a convex set. By Theorem 4.4, we get

M ( int ( C × D ) ) =.

By the separation theorem of convex sets, there exist ϕ Y and ψ Z , not both zero functionals, such that

ϕ(y)+ψ(z)ϕ( y ¯ )+ψ( z ¯ ),for all (y,z)M,( y ¯ , z ¯ )int(C×D).
(33)

Since intC{ 0 Y } and intD{ 0 Z } are cones, by (33), we have

ϕ( y ¯ )ψ( z ¯ ),for all ( y ¯ , z ¯ )intC×intD,
(34)

and

ϕ(y)+ψ(z)0,for all (y,z)M.
(35)

From (34), we find that ψ is bounded below on intD. Then ψ(z)0, for all zintD. Naturally ψ D .

By a similar line of proof to ψ D , we can obtain ϕ C .

It follows from Proposition 3.5 that ( 0 Y , 0 Z )A, and then, from z 0 D, ψ D , and (35), we obtain

inf { ( y , z ) A ϕ ( y ) + ψ ( z ) } =0andψ( z 0 )=0.

The proof is complete. □

Theorem 4.6 Let EX be a nonempty convex set, (u,v,w)T(epi( F x 0 ,G), x 0 , y 0 , z 0 ) with vC and wD and z 0 G( x 0 )(D). Suppose that the following conditions are satisfied:

  1. (i)

    ( F x 0 ,G) is C×D-convex on E;

  2. (ii)

    there exist ϕ C {0} and ψ D such that

    inf { ( y , z ) V ϕ ( y ) + ψ ( z ) } =0andψ( z 0 )=0,

where U:= x Ω D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w)(x) and V:=dom[ D ( F x 0 , G ) + ( x 0 , y 0 , z 0 ,u,v,w)].

Then x 0 is a weakly efficient solution of (CSVVEP).

Proof To prove the result by contradiction, suppose that x 0 is not a weakly efficient solution of (CSVVEP). Then there exist x K and y F( x 0 , x ) such that y intC. Since x K, there exists z G( x )(D). It follows from assumption (i) and Proposition 3.3 that we have

( y y 0 , z z 0 ) U,

and then, from assumption (ii), we obtain

ϕ ( y y 0 ) +ψ ( z z 0 ) 0.
(36)

Since y y 0 = y intC, ϕ C {0}, ϕ( y y 0 )<0. It follows from z G( x )(D), ψ D , and ψ( z 0 )=0 that ψ( z z 0 )0, thus

ϕ ( y y 0 ) +ψ ( z z 0 ) <0,

which contradicts (36). So x 0 is a weakly efficient solution of (CSVVEP), and this completes the proof. □

5 Conclusions

In this paper, we propose a new concept of a second-order derivative for set-valued maps, which is called the second-order composed contingent epiderivative, and we investigate some of its properties. Simultaneously, by virtue of the derivative, we obtain second-order sufficient optimality conditions and necessary optimality conditions for set-valued equilibrium problems.

References

  1. 1.

    Giannessi F: Theorem of alternative, quadratic programs, and complementarity problem. In Variational Inequality and Complementarity Problem. Edited by: Cottle RW, Giannessi F, Lions JL. Wiley, Chichester; 1980:151-186.

    Google Scholar 

  2. 2.

    Chen GY, Yang XQ: Vector complementarity problem and its equivalence with weak minimal element in ordered spaces. J. Math. Anal. Appl. 1990, 153: 136-158. 10.1016/0022-247X(90)90270-P

    MathSciNet  Article  MATH  Google Scholar 

  3. 3.

    Ansari QH, Oettli W, Schager D: A generalization of vectorial equilibria. Math. Methods Oper. Res. 1997, 46: 147-152. 10.1007/BF01217687

    MathSciNet  Article  MATH  Google Scholar 

  4. 4.

    Bianchi M, Hadjisavvas N, Schaible S: Vector equilibrium problems with generalized monotone bifunctions. J. Optim. Theory Appl. 1997, 92: 527-542. 10.1023/A:1022603406244

    MathSciNet  Article  MATH  Google Scholar 

  5. 5.

    Fu J: Simultaneous vector variational inequalities and vector implicit complementarity problems. J. Optim. Theory Appl. 1997, 93: 141-151. 10.1023/A:1022653918733

    MathSciNet  Article  MATH  Google Scholar 

  6. 6.

    Gong XH: Efficiency and Henig efficiency for vector equilibrium problems. J. Optim. Theory Appl. 2001, 108: 139-154. 10.1023/A:1026418122905

    MathSciNet  Article  MATH  Google Scholar 

  7. 7.

    Ansari QH, Konnov IV, Yao JC: Characterizations of solutions for vector equilibrium problems. J. Optim. Theory Appl. 2002, 113: 435-447. 10.1023/A:1015366419163

    MathSciNet  Article  MATH  Google Scholar 

  8. 8.

    Kimura K, Yao JC: Semicontinuity of solution mappings of parametric generalized strong vector equilibrium problems. J. Ind. Manag. Optim. 2008, 4: 167-181.

    MathSciNet  Article  MATH  Google Scholar 

  9. 9.

    Kimura K, Yao JC: Sensitivity analysis of solution mappings of parametric vector quasi-equilibrium problems. J. Glob. Optim. 2008, 41: 187-202. 10.1007/s10898-007-9210-9

    MathSciNet  Article  MATH  Google Scholar 

  10. 10.

    Kimura K, Yao JC: Sensitivity analysis of solution mappings of parametric generalized quasi vector equilibrium problems. Taiwan. J. Math. 2008, 12: 2233-2268.

    MathSciNet  MATH  Google Scholar 

  11. 11.

    Kimura K, Yao JC: Sensitivity analysis of vector equilibrium problems. Taiwan. J. Math. 2008, 12: 649-669.

    MathSciNet  MATH  Google Scholar 

  12. 12.

    Gong XH: Connectedness of the solution sets and scalarization for vector equilibrium problems. J. Optim. Theory Appl. 2007, 133: 151-161. 10.1007/s10957-007-9196-y

    MathSciNet  Article  MATH  Google Scholar 

  13. 13.

    Gong XH, Yao JC: Connectedness of the set of efficient solutions for generalized systems. J. Optim. Theory Appl. 2008, 138: 189-196. 10.1007/s10957-008-9378-2

    MathSciNet  Article  MATH  Google Scholar 

  14. 14.

    Gong XH, Yao JC: Lower semicontinuity of the set of efficient solutions for generalized systems. J. Optim. Theory Appl. 2008, 138: 197-205. 10.1007/s10957-008-9379-1

    MathSciNet  Article  MATH  Google Scholar 

  15. 15.

    Giannessi F, Mastroeni G, Pellegrini L: On the theory of vector optimization and variational inequalities. Image space analysis and separation. In Vector Variational Inequalities and Vector Equilibria: Mathematical Theories. Kluwer Academic, Dordrecht; 2000:153-215.

    Google Scholar 

  16. 16.

    Morgan J, Romaniello M: Scalarization and Kuhn-Tucker-like conditions for weak vector generalized quasivariational inequalities. J. Optim. Theory Appl. 2006, 130: 309-316. 10.1007/s10957-006-9104-x

    MathSciNet  Article  MATH  Google Scholar 

  17. 17.

    Gong XH: Optimality conditions for vector equilibrium problems. J. Math. Anal. Appl. 2008, 342: 1455-1466. 10.1016/j.jmaa.2008.01.026

    MathSciNet  Article  MATH  Google Scholar 

  18. 18.

    Qiu Q: Optimality conditions of globally efficient solution for vector equilibrium problems with generalized convexity. J. Inequal. Appl. 2009. Article ID 898213, 2009:

    Google Scholar 

  19. 19.

    Gong XH, Xiong SQ: The optimality conditions of the convex-like vector equilibrium problems. J. Nanchang Univ., Nat. Sci. Ed. 2009 (in Chinese), 33: 409-414.

    Google Scholar 

  20. 20.

    Long XJ, Huang YQ, Peng ZY: Optimality conditions for the Henig efficient solution of vector equilibrium problems with constraints. Optim. Lett. 2011,5(4):717-728. 10.1007/s11590-010-0241-7

    MathSciNet  Article  MATH  Google Scholar 

  21. 21.

    Wei ZF, Gong XH: Kuhn-Tucker optimality conditions for vector equilibrium problems. J. Inequal. Appl. 2010. Article ID 842715, 2010:

    Google Scholar 

  22. 22.

    Ma BC, Gong XH: Optimality conditions for vector equilibrium problems in normed spaces. Optimization 2011, 60: 1441-1455. 10.1080/02331931003657709

    MathSciNet  Article  MATH  Google Scholar 

  23. 23.

    Aubin JP, Frankowska H: Set-Valued Analysis. Birkhäuser, Boston; 1990.

    Google Scholar 

  24. 24.

    Li SJ, Zhu SK, Teo KL: New generalized second-order contingent epiderivatives and set-valued optimization problems. J. Optim. Theory Appl. 2012,152(3):587-604. 10.1007/s10957-011-9915-2

    MathSciNet  Article  MATH  Google Scholar 

  25. 25.

    Jiménez B, Novo V: Second-order necessary conditions in set constrained differentiable vector optimization. Math. Methods Oper. Res. 2003, 58: 299-317. 10.1007/s001860300283

    MathSciNet  Article  MATH  Google Scholar 

  26. 26.

    Li SJ, Chen CR: Higher-order optimality conditions for Henig efficient solutions in set-valued optimization. J. Math. Anal. Appl. 2006, 323: 1184-1200. 10.1016/j.jmaa.2005.11.035

    MathSciNet  Article  MATH  Google Scholar 

  27. 27.

    Li SJ, Teo KL, Yang XQ: Higher-order optimality conditions for set-valued optimization. J. Optim. Theory Appl. 2008,137(3):533-553. 10.1007/s10957-007-9345-3

    MathSciNet  Article  MATH  Google Scholar 

  28. 28.

    Li SJ, Teo KL, Yang XQ: Higher-order Mond-Weir duality for set-valued optimization. J. Comput. Appl. Math. 2008,217(2):339-349. 10.1016/j.cam.2007.02.011

    MathSciNet  Article  MATH  Google Scholar 

  29. 29.

    Chen CR, Li SJ, Teo KL: Higher order weak epiderivatives and applications to duality and optimality conditions. Comput. Math. Appl. 2009, 57: 1389-1399. 10.1016/j.camwa.2009.01.012

    MathSciNet  Article  MATH  Google Scholar 

  30. 30.

    Wang QL, Li SJ: Generalized higher-order optimality conditions for set-valued optimization under Henig efficiency. Numer. Funct. Anal. Optim. 2009,30(7-8):849-869. 10.1080/01630560903139540

    MathSciNet  Article  MATH  Google Scholar 

  31. 31.

    Wang QL, Li SJ, Teo KL: Higher-order optimality conditions for weakly efficient solutions in nonconvex set-valued optimization. Optim. Lett. 2010, 4: 425-437. 10.1007/s11590-009-0170-5

    MathSciNet  Article  MATH  Google Scholar 

  32. 32.

    Jahn J, Khan AA, Zeilinger P: Second-order optimality conditions in set optimization. J. Optim. Theory Appl. 2005, 125: 331-347. 10.1007/s10957-004-1841-0

    MathSciNet  Article  MATH  Google Scholar 

  33. 33.

    Chen GY, Jahn J: Optimality conditions for set-valued optimization problems. Math. Methods Oper. Res. 1998, 48: 187-200. 10.1007/s001860050021

    MathSciNet  Article  MATH  Google Scholar 

  34. 34.

    Jahn J: Vector Optimization-Theory, Applications and Extensions. Springer, Berlin; 2004.

    Google Scholar 

  35. 35.

    Corley HW: Existence and Lagrangian duality for maximization of set-valued functions. J. Optim. Theory Appl. 1987, 54: 489-501. 10.1007/BF00940198

    MathSciNet  Article  MATH  Google Scholar 

  36. 36.

    Aubin JP, Ekeland I: Applied Nonlinear Analysis. Wiley, New York; 1984.

    Google Scholar 

  37. 37.

    Jiménez B, Novo V: Optimality conditions in differentiable vector optimization via second-order tangent sets. Appl. Math. Optim. 2004, 49: 123-144. 10.1007/s00245-003-0782-6

    MathSciNet  Article  MATH  Google Scholar 

Download references

Acknowledgements

The authors would like to thank anonymous referees for their valuable comments and suggestions, which helped to improve the paper. This research was partially supported by Science and Technology Research Project of Chongqing Municipal Education Commission (KJ130414), Chongqing Natural Science Foundation Project of CQ CSTC(cstc2012jjA00038) and the National Natural Science Foundation of China (Nos: 11171362, 11271389, 11201509, and 11301571).

Author information

Affiliations

Authors

Corresponding author

Correspondence to Qilin Wang.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

All authors read and approved the final manuscript.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wang, Q., Lin, Z., Zeng, J. et al. Second-order composed contingent epiderivatives and set-valued vector equilibrium problems. J Inequal Appl 2014, 406 (2014). https://doi.org/10.1186/1029-242X-2014-406

Download citation

Keywords

  • set-valued vector equilibrium problems
  • second-order composed contingent epiderivatives
  • weakly efficient solutions
  • second-order optimality conditions