Skip to main content

On the Hölder continuity of solution maps to parametric generalized vector quasi-equilibrium problems via nonlinear scalarization

Abstract

In this paper, by using a nonlinear scalarization technique, we obtain sufficient conditions for Hölder continuity of the solution mapping for a parametric generalized vector quasi-equilibrium problem with set-valued mappings. The results are different from the recent ones in the literature.

1 Introduction

The generalized vector quasi-equilibrium problem is a unified model of several problems, namely generalized vector quasi-variational inequalities, vector quasi-optimization problems, traffic network problems, fixed point and coincidence point problems, etc. (see, for example, [1, 2] and the references therein). It is well known that the stability analysis of a solution mapping for equilibrium problems is an important topic in optimization theory and applications. Stability may be understood as lower or upper semicontinuity, continuity, and Lipschitz or Hölder continuity. There have been many papers to discuss the stability of solution mapping for equilibrium problems when they are perturbed by parameters (also known the parametric (generalized) equilibrium problems). Last decade, many authors intensively studied the sufficient conditions of upper (lower) semicontinuity of various solution mappings for parametric (generalized) equilibrium problems, see [310]. Let us begin now, Yen [11] obtained the Hölder continuity of the unique solution of a classic perturbed variational inequality by the metric projection method. Mansour and Riahi [12] proved the Hölder continuity of the unique solution for a parametric equilibrium problem under the concepts of strong monotonicity and Hölder continuity. Bianchi and Pini [13] introduced the concept of strong pseudomonotonicity and got the Hölder continuity of the unique solution of a parametric equilibrium problem. Anh and Khanh [14] generalized the main results of [13] to two classes of perturbed generalized equilibrium problems with set-valued mappings. Anh and Khanh [15] further discussed the uniqueness and Hölder continuity of the solutions for perturbed equilibrium problems with set-valued mappings. Anh and Khanh [16] extended the results of [15] to the case of perturbed quasi-equilibrium problems with set-valued mappings and obtained the Hölder continuity of the unique solutions. Li et al. [17] introduced an assumption, which is weaker than the corresponding ones of [13, 14], and established the Hölder continuity of the set-valued solution mappings for two classes of parametric generalized vector quasi-equilibrium problems in general metric spaces. Li et al. [18] extended the results of [17] to perturbed generalized vector quasi-equilibrium problems.

Among many approaches for dealing with the lower semicontinuity, continuity and Hölder continuity of the solution mapping for a parametric vector equilibrium problem in general metric spaces, the scalarization method is of considerable interest. The classical scalarization method using linear functionals has been already used for studying the lower semicontinuity of the solution mapping [1921] and the Hölder continuity [22] of the solution mapping to parametric vector equilibrium problems. Wang et al. [23] established the lower semicontinuity and upper semicontinuity of the solution set to a parametric generalized strong vector equilibrium problem by using a scalarization method and a density result. Recently, by using this method, Peng [24] established the sufficient conditions for the Hölder continuity of the solution mapping to a parametric generalized vector quasi-equilibrium problem with set-valued mappings.

On the other hand, a useful approach for analyzing a vector optimization problem is to reduce it to a scalar optimization problem. Nonlinear scalarization functions play an important role in this reduction in the context of nonconvex vector optimization problems. The nonlinear scalarization function ξ q , commonly known as the Gerstewitz function in the theory of vector optimization [25, 26], has been also used to study the lower semicontinuity of the set-valued solution mapping to a parametric vector variational inequality [27]. Using this method, Bianchi and Pini [28] obtained the Hölder continuity of the single-valued solution mapping to a parametric vector equilibrium problem. Recently, Chen and Li [29] studied Hölder continuity of the solution mapping for both set-valued and single-valued cases to parametric vector equilibrium problems. The key role in their paper is a globally Lipschitz property of the Gerstewitz function. Very recently, by using the idea in [29], Chen [30] obtained Hölder continuity of the unique solution to a parametric vector quasi-equilibrium problem based on nonlinear scalarization approach under three different kinds of monotonicity hypotheses. It is natural to raise and give an answer to the following question.

Question Can one establish the Hölder continuity of a solution mapping to the parametric generalized vector quasi-equilibrium problem with set-valued mappings by using a nonlinear scalarization method?

Motivated and inspired by Peng [24] and Chen [30] and research going on in this direction, in this paper we aim to give positive answers to the above question. We first establish the sufficient conditions which guarantee the Hölder continuity of a solution mapping to the parametric generalized vector quasi-equilibrium problem with set-valued mappings by using a nonlinear scalarization method. We further study several kinds of the monotonicity conditions to obtain the Hölder continuity of the solution mapping. The main results of this paper are different from the corresponding results in Peng [24] and Chen [30]. These results improve the corresponding ones in recent literature.

The structure of the paper is as follows. Section 2 presents the parametric generalized vector quasi-equilibrium problem and materials used in the rest of this paper. We establish, in Section 3, a sufficient condition for the Hölder continuity of the solution mapping to a parametric generalized vector quasi-equilibrium problem.

2 Preliminaries

Throughout the paper, unless otherwise specified, we denote by and d(,) the norm and the metric on a normed space and a metric space, respectively. A closed ball with center 0X and radius δ>0 is denoted by B(0,δ). We always consider X, Λ, M as metric spaces, and Y as a linear normed space with its topological dual space Y . For any y Y , we define y :=sup{ y ,y:y=1}, where y ,y denotes the value of y at y. Let CY be a pointed, closed and convex cone with intC, where intC stands for the interior of C. Let

C := { y Y : y , y 0 , y C }

be the dual cone of C. Since intC, the dual cone C of C has a weak* compact base. Let eintC. Then

B e := { y C : y , e = 1 }

is a weak*-compact base of C . Clearly, C q is a weak-compact base of C , that is, C q is convex and weak-compact such that 0 C q and C = t 0 t C q .

Let qintC, the nonlinear scalarization function [25, 26] ξ q :YR is defined by

ξ q =min{tR:ytqC}.

It is well known that ξ q is a continuous, positively homogeneous, subadditive and convex function on Y, and it is monotone (that is, y 2 y 1 C ξ q ( y 1 ) ξ q ( y 2 )) and strictly monotone (that is, y 2 y 1 intC ξ q ( y 1 )< ξ q ( y 2 )) (see [25, 26]). In case, Y= R l , C= R + l and q=(1,1,,1)int R + l , the nonlinear scalarization function can be expressed in the following equivalent form [[25], Corollary 1.46]:

ξ q (y)= max 1 i l { y i },y=( y 1 , y 2 ,, y l ) R l .
(1)

Lemma 2.1 [[25], Proposition 1.43]

For any fixed qintC, yY and rR,

  1. (i)

    ξ q <ryrqintC (that is, ξ q (y)ryrqintC);

  2. (ii)

    ξ q (y)ryrqC;

  3. (iii)

    ξ q (y)=ryrqC, where ∂C denotes the boundary of C;

  4. (iv)

    ξ q (rq)=r.

The property (i) of Lemma 2.1 plays an essential role in scalarization. From the definition of ξ q , property (iv) in Lemma 2.1 could be strengthened as

ξ q (y+rq)= ξ q (y)+r,yY,rR.
(2)

For any qintC, the set C q defined by

C q := { y C : y , q = 1 }

is a weak-compact set of Y (see [[19], Lemma 5.1]). The following equivalent form of ξ q can be deduced from [[31], Corollary 2.1] or [[32], Proposition 2.2] ([[25], Proposition 1.53]).

Proposition 2.2 [[30], Proposition 2.2]

Let qintC. Then, for yY,

ξ q (y)= max y C q y , y .

Proposition 2.3 [[30], Proposition 2.3]

ξ q is Lipschitz on Y, and its Lipschitz constant is

L:= sup y C q y [ 1 q ,+).

The following example can be found in [[30], Example 2.1].

Example 2.4

  1. (i)

    If Y=R and C= R + , then the Lipschitz constant of ξ q is L= 1 q (q>0). Indeed, | ξ q (x) ξ q (y)|= 1 q |xy| for all x,yR.

  2. (ii)

    If Y= R 2 and C={( y 1 , y 2 ) R 2 : 1 4 y 1 y 2 2 y 1 }. Take q=(2,3)intC, then

    C q := { ( y 1 , y 2 ) R : 2 y 1 + 3 y 2 = 1 , y 1 [ 0.1 , 2 ] } ,

and the Lipschitz constant is L= sup y C q y =(2,1)= 5 . Hence,

| ξ q ( y ) ξ q ( y ) | 5 y y ,y, y R 2 .

Now we recall some basic definitions and their properties which will be used in the sequel.

Definition 2.5 (Classical notion)

Let l0 and α>0. A set-valued mapping G:Λ 2 X is said to be lα-Hölder continuous at λ 0 on a neighborhood N( λ 0 ) of λ 0 if and only if

G( λ 1 )G( λ 2 )+l B X ( 0 , d α ( λ 1 , λ 2 ) ) , λ 1 , λ 2 N( λ 0 ).
(3)

When X is a normed space, we say that the vector-valued mapping g:ΛX is lα-Hölder continuous at λ 0 on a neighborhood N( λ 0 ) of λ 0 iff

g ( λ 1 ) g ( λ 2 ) l d α ( λ 1 , λ 2 ), λ 1 , λ 2 N( λ 0 ).
(4)

Definition 2.6 Let l 1 , l 2 0 and α 1 , α 2 >0. A set-valued mapping G:X×Λ 2 X is said to be ( l 1 α 1 , l 2 α 2 )-Hölder continuous at x 0 , λ 0 on neighborhoods N( x 0 ) and N( λ 0 ) of x 0 and λ 0 if and only if

G( x 1 , λ 1 )G( x 2 , λ 2 )+ ( l 1 d X α 1 ( x 1 , x 2 ) + l 2 d Λ α 2 ( λ 1 , λ 2 ) ) B X (0,1)
(5)

for all x 1 , x 2 N( x 0 ), λ 1 , λ 2 N( λ 0 ).

3 Main results

By using a nonlinear scalarization technique, we present the sufficient conditions for Hölder continuity of the solution mapping for a parametric generalized vector quasi-equilibrium problem.

Let N( λ 0 )Λ and N( μ 0 )M be neighborhoods of λ 0 and μ 0 , respectively, and let K:X×Λ 2 X and F:X×X×M 2 Y be set-valued mappings. For each λN( λ 0 ) and μN( μ 0 ), we consider the following parametric generalized vector quasi-equilibrium problem (PGVQEP):

Find x 0 K( x 0 ,λ) such that

F( x 0 ,y,μ)Y(intC),yK( x 0 ,λ).
(6)

For each λN( λ 0 ) and μN( μ 0 ), let

E(λ):= { x X | x K ( x , λ ) } .

The weak solution set of (6) is denoted by

S W (λ,μ):= { x E ( λ ) : F ( x , y , μ ) Y ( int C ) , y K ( x , λ ) } .

For each λN( λ 0 ), μN( μ 0 ) and fixed qintC, the ξ q -solution set of (6) is denoted by

S( ξ q ,λ,μ):= { x E ( λ ) : inf z F ( x , y , μ ) ξ q ( z ) 0 , y K ( x , λ ) } .

We first establish the following lemmas which will be used in the sequel.

Lemma 3.1 For each λN( λ 0 ), μN( μ 0 ) and fixed qintC,

S W (λ,μ)=S( ξ q ,λ,μ).

Proof Let λN( λ 0 ), μN( μ 0 ) and fixed qintC. For any x S W (λ,μ), we have

xE(λ)andF(x,y,μ)Y(intC),yK(x,λ).

Therefore, for each yK(x,λ) and each zF(x,y,μ), we have

zintC=0qintC.

By Lemma 2.1(i), we conclude that ξ q (z)0. Since z is arbitrary, we have

inf z F ( x , y , μ ) ξ q (z)0for all yK(x,λ),

which gives that S W (λ,μ)S( ξ q ,λ,μ).

On the other hand, for each xS( ξ q ,λ,μ), we have that

xE(λ)and inf z F ( x , y , μ ) ξ q (z)0,yK(x,λ).
(7)

Thus, for each yK(x,λ) and each zF(x,y,μ), we have that ξ q (z)0. By Lemma 2.1(i), we can obtain zintC. Therefore, we have zY(intC), which implies that

xE(λ)andF(x,y,μ)Y(intC),yK(x,λ).

Hence, S( ξ q ,λ,μ) S W (λ,μ). The proof is completed. □

Lemma 3.2 Suppose that N( λ 0 ) and N( μ 0 ) are the given neighborhoods of λ 0 and μ 0 , respectively.

  1. (a)

    If for each x,yE(N( λ 0 )), F(x,y,) is m 1 γ 1 -Hölder continuous at μ 0 M, then for any fixed qintC, the function

    ψ ξ q (x,y,)= inf z F ( x , y , ) ξ q (z)

    is L m 1 γ 1 -Hölder continuous at μ 0 .

  2. (b)

    If for each xE(N( λ 0 )) and μN(E( μ 0 )), F(x,,μ) is m 2 γ 2 -Hölder continuous on E(N( λ 0 )), then for any fixed qintC, the function

    ψ ξ q (x,,μ)= inf z F ( x , , μ ) ξ q (z)

    is L m 2 γ 2 -Hölder continuous on E(N( λ 0 )).

Proof (a) Let x,yE(N( λ 0 )). The m 1 γ 1 -Hölder continuity of F(x,y,) implies that there exists a neighborhood N( μ 0 ) of μ 0 such that for all μ 1 , μ 2 N( μ 0 ),

F(x,y, μ 1 )F(x,y, μ 2 )+ m 1 d M γ 1 ( μ 1 , μ 2 ) B Y .

So, for any z 1 F(x,y, μ 1 ), there exist z 2 F(x,y, μ 2 ) and e B Y such that

z 1 = z 2 + m 1 d M γ 1 ( μ 1 , μ 2 )e.

By using Proposition 2.3, we obtain

| ξ q ( z 1 ) ξ q ( z 2 ) | L z 1 z 2 = L m 1 d M γ 1 ( μ 1 , μ 2 ) e L m 1 d M γ 1 ( μ 1 , μ 2 ) ,
(8)

which gives that

L m 1 d γ 1 ( μ 1 , μ 2 ) ξ q ( z 1 ) ξ q ( z 2 ).

Since z 1 is arbitrary and ξ q ( z 2 ) inf z F ( x , y , μ 2 ) ξ q (z), we have

L m 1 d M γ 1 ( μ 1 , μ 2 ) inf z F ( x , y , μ 1 ) ξ q (z) inf z F ( x , y , μ 2 ) ξ q (z).

Applying the symmetry between μ 1 and μ 2 , we arrive at

L m 1 d M γ 1 ( μ 1 , μ 2 ) inf z F ( x , y , μ 2 ) ξ q (z) inf z F ( x , y , μ 1 ) ξ q (z).

It follows from the last two inequalities that

| ψ ξ q ( x , y , μ 1 ) ψ ξ q ( x , y , μ 2 ) | L m 1 d M γ 1 ( μ 1 , μ 2 ), μ 1 , μ 2 N( μ 0 ).

Therefore, we conclude that ψ ξ q (x,y,)= inf z F ( x , y , ) ξ q (z) is L m 1 γ 1 -Hölder continuous at  μ 0 .

  1. (b)

    It follows by a similar argument as in part (a). The proof is completed. □

Now, by using the nonlinear scalarization technique, we propose some sufficient conditions for Hölder continuity of the solution mapping for (PGVQEP).

Theorem 3.3 For each fixed qintC, let S( ξ q ,λ,μ) be nonempty in a neighborhood N( λ 0 )×N( μ 0 ) of ( λ 0 , μ 0 )Λ×M. Assume that the following conditions hold.

  1. (i)

    K(,) is ( l 1 α 1 , l 2 α 2 )-Hölder continuous on E(N( λ 0 ))×N( λ 0 );

  2. (ii)

    For each x,yE(N( λ 0 )), F(x,y,) is m 1 γ 1 -Hölder continuous at μ 0 M;

  3. (iii)

    For each xE(N( λ 0 )) and μN( μ 0 ), F(x,,μ) is m 2 γ 2 -Hölder continuous on E(N( λ 0 ));

  4. (iv)

    F(,,μ) is hβ-Hölder strongly monotone with respect to ξ q , that is, there exist constants h>0, β>0 such that for every x,yE(N( λ 0 )), xy,

    h d X β (x,y)d ( inf z F ( x , y , μ ) ξ q ( z ) , R + ) +d ( inf z F ( y , x , μ ) ξ q ( z ) , R + ) ;
  5. (v)

    β= α 1 γ 2 , h>2 m 2 L l 1 γ 1 , where L:= sup λ C q λ[ 1 q ,+) is the Lipschitz constant of ξ q on Y.

Then, for every (λ,μ)N( λ 0 )×N( μ 0 ), the solution x(λ,μ) of (PVQGEP) is unique, and x(λ,μ) as a function of λ and μ satisfies the Hölder condition: for all ( λ 1 , μ 1 ),( λ 2 , μ 2 )N( λ 0 )×N( μ 0 ),

d X ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) ( 2 m 2 L l 2 γ 2 h 2 m 2 L l 1 γ 1 ) 1 β d Λ α 2 γ 2 / β ( λ 1 , λ 2 ) + ( m 1 L h 2 m 2 L l 1 γ 1 ) 1 β d M γ 1 / β ( μ 1 , μ 2 ) ,

where x( λ i , μ i ) S W ( λ i , μ i ), i=1,2.

Proof Let ( λ 1 , μ 1 ),( λ 2 , μ 2 )N( λ 0 )×N( μ 0 ). The proof is divided into the following three steps based on the fact that

d X ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) d X ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) + d X ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) ,

where x( λ i , μ i ) S W ( λ i , μ i ), i=1,2.

Step 1: We prove that

d 1 := d X ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) ( m 1 L h 2 m 2 L l 1 γ 1 ) 1 β d M γ 1 / β ( μ 1 , μ 2 )
(9)

for all x( λ 1 , μ 1 ) S W ( λ 1 , μ 1 ) and x( λ 1 , μ 2 ) S W ( λ 1 , μ 2 ).

If x( λ 1 , μ 1 )=x( λ 1 , μ 2 ), then we are done. So, we assume that x( λ 1 , μ 1 )x( λ 1 , μ 2 ). Since x( λ 1 , μ 1 )K(x( λ 1 , μ 1 ), λ 1 ) and x( λ 1 , μ 2 )K(x( λ 1 , μ 2 ), λ 1 ), by the l 1 α 1 -Hölder continuity of K(, λ 1 ), there exist x 1 K(x( λ 1 , μ 1 ), λ 1 ) and x 2 K(x( λ 1 , μ 2 ), λ 1 ) such that

d X ( x ( λ 1 , μ 1 ) , x 2 ) l 1 d X α 1 ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) = l 1 d 1 α 1
(10)

and

d X ( x ( λ 1 , μ 2 ) , x 1 ) l 1 d X α 1 ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) = l 1 d 1 α 1 .
(11)

Since x( λ 1 , μ 1 ) S W ( λ 1 , μ 1 ) and x( λ 1 , μ 2 ) S W ( λ 1 , μ 2 ), by Lemma 3.1, we obtain

ψ ξ q ( x ( λ 1 , μ 1 ) , x 1 , μ 1 ) := inf z F ( x ( λ 1 , μ 1 ) , x 1 , μ 1 ) ξ q (z)0
(12)

and

ψ ξ q ( x ( λ 1 , μ 2 ) , x 2 , μ 2 ) := inf z F ( x ( λ 1 , μ 2 ) , x 2 , μ 2 ) ξ q (z)0.
(13)

By virtue of (iv), we have

h d 1 β = h d X β ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) d ( ψ ξ q ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) , μ 1 ) , R + ) + d ( ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 1 , μ 1 ) , μ 1 ) , R + ) .

By combining (12) and (13) with the last inequality, we have

h d 1 β | ψ ξ q ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) , μ 1 ) ψ ξ q ( x ( λ 1 , μ 1 ) , x 1 , μ 1 ) | + | ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 1 , μ 1 ) , μ 1 ) ψ ξ q ( x ( λ 1 , μ 2 ) , x 2 , μ 2 ) | | ψ ξ q ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) , μ 1 ) ψ ξ q ( x ( λ 1 , μ 1 ) , x 1 , μ 1 ) | + | ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 1 , μ 1 ) , μ 1 ) ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 1 , μ 1 ) , μ 2 ) | + | ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 1 , μ 1 ) , μ 2 ) ψ ξ q ( x ( λ 1 , μ 2 ) , x 2 , μ 2 ) | L m 2 d X γ 2 ( x ( λ 1 , μ 2 ) , x 1 ) + L m 1 d M γ 1 ( μ 1 , μ 2 ) + L m 2 d X γ 2 ( x ( λ 1 , μ 1 ) , x 2 ) L m 2 l 1 γ 2 d X α 1 γ 2 ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) + L m 1 d M γ 1 ( μ 1 , μ 2 ) + L m 2 l 1 γ 2 d X α 1 γ 2 ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) = 2 L m 2 l 1 γ 2 d X α 1 γ 2 ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) + L m 1 d M γ 1 ( μ 1 , μ 2 ) .
(14)

Whence, assumption (iv) implies that

d X ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) ( L m 1 h L m 2 l 1 γ 2 ) 1 β d M γ 1 / β ( μ 1 , μ 2 ).

Step 2: We prove that

d 2 := d X ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) ( 2 L m 2 l 2 γ 2 h 2 L m 2 l 1 γ 1 ) 1 β d Λ α 2 γ 2 / β ( λ 1 , λ 2 )
(15)

for all x( λ 1 , μ 2 ) S W ( λ 1 , μ 2 ) and x( λ 2 , μ 2 ) S W ( λ 2 , μ 2 ).

If x( λ 1 , μ 2 )=x( λ 2 , μ 2 ), then we are done. So, we assume that x( λ 1 , μ 2 )x( λ 2 , μ 2 ). Since x( λ 1 , μ 2 )K(x( λ 1 , μ 2 ), λ 1 ) and x( λ 2 , μ 2 )K(x( λ 2 , μ 2 ), λ 2 ), by the l 2 α 2 -Hölder continuity of K(x( λ 1 , μ 2 ),) and K(x( λ 2 , μ 2 ),), there exist x 1 K(x( λ 2 , μ 2 ), λ 1 ) and x 2 K(x( λ 1 , μ 2 ), λ 2 ) such that

d X ( x ( λ 1 , μ 2 ) , x 2 ) l 2 d Λ α 2 ( λ 1 , λ 2 )
(16)

and

d X ( x ( λ 2 , μ 2 ) , x 1 ) l 2 d Λ α 2 ( λ 1 , λ 2 ).
(17)

Again, by the Hölder continuity of K(,), there exist x 1 K(x( λ 1 , μ 2 ), λ 1 ) and x 2 K(x( λ 2 , μ 2 ), λ 2 ) such that

d X ( x 1 , x 1 ) l 1 d X α 1 ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) = l 1 d 2 α 1
(18)

and

d X ( x 2 , x 2 ) l 1 d X α 1 ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) = l 1 d 2 α 1 .
(19)

Since x( λ 1 , μ 2 ) S W ( λ 1 , μ 2 ) and x( λ 2 , μ 2 ) S W ( λ 2 , μ 2 ), by Lemma 3.1, we obtain the following:

ψ ξ q ( x ( λ 1 , μ 2 ) , x 1 , μ 2 ) := inf z F ( x ( λ 1 , μ 2 ) , x 1 , μ 2 ) ξ q (z)0
(20)

and

ψ ξ q ( x ( λ 2 , μ 2 ) , x 2 , μ 2 ) := inf z F ( x ( λ 2 , μ 2 ) , x 2 , μ 2 ) ξ q (z)0.
(21)

By virtue of (iv), we have

h d 2 β = h d X β ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) d ( ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) , μ 2 ) , R + ) + d ( ψ ξ q ( x ( λ 2 , μ 2 ) , x ( λ 1 , μ 2 ) , μ 2 ) , R + ) .

By combining (20) and (21) with the last inequality, we have

h d 2 β | ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) , μ 2 ) ψ ξ q ( x ( λ 1 , μ 2 ) , x 1 , μ 2 ) | + | ψ ξ q ( x ( λ 2 , μ 2 ) , x ( λ 1 , μ 2 ) , μ 2 ) ψ ξ q ( x ( λ 2 , μ 2 ) , x 2 , μ 2 ) | | ψ ξ q ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) , μ 2 ) ψ ξ q ( x ( λ 1 , μ 2 ) , x 1 , μ 2 ) | + | ψ ξ q ( x ( λ 1 , μ 2 ) , x 1 , μ 2 ) ψ ξ q ( x ( λ 1 , μ 2 ) , x 1 , μ 2 ) | + | ψ ξ q ( x ( λ 2 , μ 2 ) , x ( λ 1 , μ 2 ) , μ 2 ) ψ ξ q ( x ( λ 2 , μ 2 ) , x 2 , μ 2 ) | + | ψ ξ q ( x ( λ 2 , μ 2 ) , x 2 , μ 2 ) ψ ξ q ( x ( λ 2 , μ 2 ) , x 2 , μ 2 ) | L m 2 d X γ 2 ( x ( λ 2 , μ 2 ) , x 1 ) + L m 2 d X γ 2 ( x 1 , x 1 ) + L m 2 d X γ 2 ( x ( λ 1 , μ 2 ) , x 2 , ) + L m 2 d X γ 2 ( x 2 , x 2 ) .
(22)

By virtue of (16), (17), (18) and (19), we get

h d X β ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) L m 2 l 2 γ 2 d Λ α 2 γ 2 ( λ 1 , λ 2 ) + L m 2 l 1 γ 2 d X α 1 γ 2 ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) + L m 2 l 2 γ 2 d Λ α 2 γ 2 ( λ 1 , λ 2 ) + L m 2 l 1 γ 2 d X α 1 γ 2 ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) = 2 L m 2 l 2 γ 2 d Λ α 2 γ 2 ( λ 1 , λ 2 ) + 2 L m 2 l 1 γ 2 d X α 1 γ 2 ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) .
(23)

Whence, condition (v) implies that

d X β ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) ( 2 L m 2 l 2 γ 2 h 2 L m 2 l 1 γ 2 ) 1 β d Λ α 2 γ 2 ( λ 1 , λ 2 ).

Step 3: Let x( λ 1 , μ 1 ) S W ( λ 1 , μ 1 ) and x( λ 2 , μ 2 ) S W ( λ 2 , μ 2 ). It follows from (9) and (15) that

d ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) d ( x ( λ 1 , μ 1 ) , x ( λ 1 , μ 2 ) ) + d ( x ( λ 1 , μ 2 ) , x ( λ 2 , μ 2 ) ) ( m 1 L h 2 m 2 L l 1 γ 1 ) 1 β d M γ 1 / β ( μ 1 , μ 2 ) + ( 2 L m 2 l 2 γ 2 h 2 L m 2 l 1 γ 1 ) 1 β d Λ α 2 γ 2 / β ( λ 1 , λ 2 ) .

Thus,

ρ ( S W ( λ 1 , μ 1 ) , S W ( λ 2 , μ 2 ) ) = sup x ( λ 1 , μ 1 ) S W ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) S W ( λ 2 , μ 2 ) d X ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) ( m 1 L h 2 m 2 L l 1 γ 1 ) 1 β d M γ 1 / β ( μ 1 , μ 2 ) + ( 2 L m 2 l 2 γ 2 h 2 L m 2 l 1 γ 1 ) 1 β d Λ α 2 γ 2 / β ( λ 1 , λ 2 ) .

Taking λ 2 = λ 1 and μ 2 = μ 1 , we see that the diameter of S( λ 1 , μ 1 ) is 0, that is, this set is a singleton {x( λ 1 , μ 1 )}. This implies that the (PGVQEP) has a unique solution in a neighborhood of ( λ 0 , μ 0 ). The proof is completed. □

Definition 3.4 Let F:X×X×M 2 Y be a set-valued mapping. A set-valued mapping F(,,μ) 2 Y is said to be

  1. (A)

    hβ-Hölder strongly monotone with respect to ξ q if there exist qintC and h>0, β>0 such that for every x,yE(N(λ)) with xy,

    inf z F ( x , y , μ ) ξ q (z)+ inf z F ( y , x , μ ) ξ q (z)+h d X β (x,y)0;
  2. (B)

    hβ-Hölder strongly pseudomonotone with respect to qintC and h>0, β>0 such that for every x,yE(N( λ 0 )) with xy,

    zintC,zF(x,y,μ) z +h d X β (x,y)qC, z F(y,x,μ).
  3. (C)

    quasi-monotone on E(N( λ 0 )) if x,yE(N( λ 0 )) with xy,

    zintC,zF(x,y,μ) z intC, z F(y,x,μ).

The following proposition provides the relation among monotonicity conditions defined above.

Proposition 3.5

  1. (i)

    (A) (iv).

  2. (ii)

    (B) and (C) (iv).

Proof (i) From the definition of (A), we have

h d X β ( x , y ) inf z F ( x , y , μ ) ξ q ( z ) inf z F ( y , x , μ ) ξ q ( z ) d ( inf z F ( x , y , μ ) ξ q ( z ) , R + ) + d ( inf z F ( y , x , μ ) ξ q ( z ) , R + ) .

(ii) Assume that F satisfies definitions (B) and (C). We consider two cases.

Case 1. zintC, zF(x,y,μ), then there exists z F(y,x,μ) such that z +h d X β (x,y)qC. From Lemma 2.1, we have

ξ q ( z ) +h d X β (x,y)= ξ q ( z + h d X β ( x , y ) q ) 0,

which implies that inf z F ( y , x , μ ) ξ q (z) ξ q ( z )h d X β (x,y). Hence,

h d X β (x,y) inf z F ( y , x , μ ) ξ q (z)d ( inf z F ( x , y , μ ) ξ q ( z ) , R + ) +d ( inf z F ( y , x , μ ) ξ q ( z ) , R + ) .

Case 2. zintC, zF(x,y,μ), then there exists z F(y,x,μ) such that zintC. By a similar argument as in the previous case, we have the desired result. □

Remark 3.6 The converse of Proposition 3.5 does not hold in general, even in the special case X=Y=R and C= R + . See, for example, Examples 1.1 and 1.2 in [15]. Therefore, Theorem 3.3 still holds when condition (iv) is replaced by condition (A) or conditions (B) and (C). We can immediately obtain the following two theorems.

Theorem 3.7 Theorem  3.3 still holds when condition (iv) is replaced by condition (A).

Theorem 3.8 Theorem  3.3 still holds when condition (iv) is replaced by conditions (B) and (C).

Let f:X×X×MY be a vector-valued mapping. Then (PGVQEP) becomes the following parametric vector quasi-equilibrium problem (PVQEP):

Find x 0 K( x 0 ,λ) such that

f( x 0 ,y,μ)intC,yK( x 0 ,λ).
(24)

Remark 3.9 In the case of a vector-valued mapping, condition (iv) in Theorem 3.3 and condition (ii′′) coincide. Also, condition (A) and conditions (B) and (C) are the same as conditions (ii) and (ii′) in [30], respectively. It is obvious that Theorems 3.3, 3.7 and 3.8 extend Theorems 3.3, 3.1 and 3.2 in [30], respectively, in the case that the vector-valued mapping f(,,) is extended to a set-valued one.

4 Applications

Since the parametric generalized vector quasi-equilibrium problem (PGVQEP) contains as special cases many optimization-related problems, including quasi-variational inequalities, traffic equilibrium problems, quasi-optimization problems, fixed point and coincidence point problems, complementarity problems, vector optimization, Nash equilibria, etc., we can derive from Theorem 3.3 a direct consequence for such special cases. We discuss now only some applications of our results.

4.1 Quasi-variational inequalities

In this section, we assume that X is a normed space. Let K:X×ΛX and T:X×M B (X,Y) be set-valued mappings, where B (X,Y) denotes the space of all bounded linear mappings of X into Y. Setting F(x,y,μ)=T(x,μ),yx:= t T ( x , μ ) t,yx in (6), we obtain parametric generalized vector quasi-variational inequalities (PGVQVI) in the case of set-valued mappings as follows:

Find  x 0 K( x 0 ,λ) such that  T ( x 0 , μ ) , y x 0 YintC,yK( x 0 ,λ).
(25)

For each λN( λ 0 ) and μN( μ 0 ), let

E(λ):= { x X : x K ( x , λ ) } .

The solution set of (25) is denoted by

S Q V I V (λ,μ):= { x E ( λ ) : T ( x , μ ) , y x Y int C , y K ( x , λ ) } .

For each λN( λ 0 ), μN( μ 0 ) and fixed qintC, the ξ q -solution set of (25) is

S Q V I V ( ξ q ,λ,μ):= { x E ( λ ) : inf z T ( x , μ ) , y x ξ q ( z ) 0 , y K ( x , λ ) } .

Theorem 4.1 Assume that for each fixed qintC, S Q V I V ( ξ q ,λ,μ) is nonempty in a neighborhood N( λ 0 )×N( μ 0 ) of the considered point ( λ 0 , μ 0 )Λ×M. Assume further that the following conditions hold.

  1. (i')

    K(,) is ( l 1 α 1 , l 2 α 2 )-Hölder continuous on E(N( λ 0 ))×N( λ 0 );

  2. (ii')

    For each xE(N( λ 0 )), T(x,) is m 3 γ 3 -Hölder continuous at μ 0 M;

  3. (iii')

    T(,) is bounded in xE(N( λ 0 )), and E(N( λ 0 )) is bounded;

  4. (iv')

    T(,μ) is hβ-Hölder strongly monotone with respect to ξ q , i.e., there exist constants h>0, β>0 such that for every x,yE(N( λ 0 )): xy,

    h x y β d ( inf z T ( x , μ ) , y x ξ q ( z ) , R + ) +d ( inf z T ( y , μ ) , x y ξ q ( z ) , R + ) ;
  5. (v')

    β= α 1 , h>2ML l 1 γ 1 , where L:= sup λ C q λ[ 1 q ,+) is the Lipschitz constant of ξ q on Y.

Then, for every (λ,μ)N( λ 0 )×N( μ 0 ), the solution of (PGVQVI) is unique, x(λ,μ), and this function satisfies the Hölder condition: for all ( λ 1 , μ 1 ),( λ 2 , μ 2 )N( λ 0 )×N( μ 0 ),

d X ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) ( 2 M L l 2 h 2 M L l 1 γ 3 ) 1 β d λ α 2 / β ( λ 1 , λ 2 ) + ( N m 3 L h 2 M L l 1 γ 3 ) 1 β d M γ 3 / β ( μ 1 , μ 2 ) ,

where x( λ i , μ i ) S Q V I ( λ i , μ i ), i=1,2.

Proof We verify that all the assumptions of Theorem 3.3 are fulfilled. First, (i′), (iv′) and (v′) are the same as (i), (iv) and (v) in Theorem 3.3. We need only to verify conditions (ii) and (iii). Taking M, M ˜ >0 such that

T ( x , μ ) M,(x,μ)E ( N ( λ 0 ) ) ×N( μ 0 )

and

xy M ˜ ,x,yE ( N ( λ 0 ) ) .

We put m 1 = M ˜ m 3 and γ 1 = γ 3 . For any fixed x,yE(N( λ 0 )), by assumption (ii′), we have

T(x, μ 1 )T(x, μ 2 )+ m 3 d γ 3 ( μ 1 , μ 2 ) B B ( X , Y ) , μ 1 , μ 2 N( μ 0 ).

Then

T ( x , μ 1 ) , y x T ( x , μ 2 ) + m 3 d γ 3 ( μ 1 , μ 2 ) B B ( X , Y ) , y x = T ( x , μ 2 ) , y x + m 3 d γ 3 ( μ 1 , μ 2 ) B B ( X , Y ) , y x = T ( x , μ 2 ) , y x + m 3 d γ 3 ( μ 1 , μ 2 ) B B ( X , Y ) , y x = T ( x , μ 2 ) , y x + m 3 d γ 3 ( μ 1 , μ 2 ) g B B ( X , Y ) g , y x T ( x , μ 2 ) , y x + m 3 d γ 3 ( μ 1 , μ 2 ) M ˜ B Y .

Hence

T ( x , μ 1 ) , y x T ( x , μ 2 ) , y x + m 1 d γ 1 ( μ 1 , μ 2 ) M ˜ B Y .

Also, we put m 2 =M and γ 2 =1. We need to show that

T ( x , μ ) , y 1 x T ( x , μ ) , y 2 x + M ˜ y 1 y 2 B Y .

For each fixed xE(N( λ 0 )) and μN( μ 0 ),

T ( x , μ ) , y 1 x = t T ( x , μ ) t , y 1 x = t T ( x , μ ) t , y 1 x + y 2 y 2 = t T ( x , μ ) t , y 2 x + t T ( x , μ ) t , y 1 y 2 T ( x , μ ) , y x + M y 1 y 2 B Y .

Hence, condition (iii) is verified, and so we obtain the result. □

For (PGVQVI), if we put Y=R, C=[0,+), then (25) becomes the following parametric generalized quasi-variational inequality problem in the case of scalar-valued one:

Find  x 0 K( x 0 ,λ) such that t,y x 0 0,yK( x 0 ,λ),tT( x 0 ,μ).
(26)

For each λN( λ 0 ) and μN( μ 0 ), let

E(λ):= { x X : x K ( x , λ ) } .

The solution set of (26) is denoted by

S Q V I S (λ,μ):= { x E ( λ ) : t , y x 0 , y K ( x , λ ) , t T ( x , μ ) } .

For each λN( λ 0 ), μN( μ 0 ) and fixed 1intC, the ξ q -solution set of (25) is

S Q V I S ( ξ 1 ,λ,μ):= { x E ( λ ) : inf z T ( x , μ ) , y x ξ 1 ( z ) 0 , y K ( x , λ ) } .

It follows from Lemma 2.1 that S Q V I S ( ξ 1 ,λ,μ) coincides with S Q V I S (λ,μ).

Corollary 4.2 Assume that S Q V I S (λ,μ) is nonempty in a neighborhood N( λ 0 )×N( μ 0 ) of the considered point ( λ 0 , μ 0 )Λ×M. Assume further that conditions (i′)-(iii′) and (v′) in Corollary  4.1 hold. Replace (iv′) by (iv′′).

(iv′′) T(,μ) is hβ-Hölder strongly monotone, i.e., there exist constants h>0, β>0, such that for every x,yE(N( λ 0 )): xy,

uv,xyh x y β ,uT(x),vT(y).

Then, for every (λ,μ)N( λ 0 )×N( μ 0 ), the solution of (PGVQVI) is unique, x(λ,μ), and this function satisfies the Hölder condition: for all ( λ 1 , μ 1 ),( λ 2 , μ 2 )N( λ 0 )×N( μ 0 ),

d X ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) ( 2 M l 2 h 2 M l 1 γ 3 ) 1 β d λ α 2 / β ( λ 1 , λ 2 )+ ( N m 3 h 2 M l 1 γ 3 ) 1 β d M γ 3 / β ( μ 1 , μ 2 ),

where x( λ i , μ i ) S Q V I S ( λ i , μ i ), i=1,2.

Proof It is not hard to show that (iv′′) implies (iv′). Indeed, for any x,yE(N( λ 0 )) with xy,

h x y β u v , x y = u , x y + v , y x sup u T ( x ) u , x y + sup v T ( y ) v , y x = sup u T ( x ) u , y x + sup v T ( y ) v , x y = inf u T ( x ) u , y x inf v T ( y ) v , x y d ( inf u T ( x ) u , y x , R + ) + d ( inf v T ( y ) v , x y , R + ) .

Therefore, (iv′) is satisfied. □

Remark 4.3 Corollary 4.2 extends Corollary 3.1 in [33] since the mapping T is a multivalued mapping.

4.2 Traffic equilibrium problems

The foundation of the study of traffic network problems goes back to Wardrop [34], who stated the basic equilibrium principle in 1952. Over the past decades, a large number of efforts have been devoted to the study of traffic assignment models, with emphasis on efficiency and optimality, in order to improve practicability, reduce gas emissions and contribute to the welfare of the community. The variational inequality approach to such problems begins with the seminal work of Smith [35] who proved that the user-optimized equilibrium can be expressed in terms of a variational inequality. Thus, the possibility of exploiting the powerful tools of variational analysis has led to dealing with a large variety of models, reaching valuable theoretical results and providing applications in practical situations. In this paper, we are concerned with a class of equilibrium problems which can be studied in the framework of quasi-variational inequalities, see [36, 37].

Let a set N of nodes, a set L of links, a set W:=( W 1 ,, W l ) of origin-destination pairs (O/D pairs for short) be given. Assume that there are r j 1 paths connecting the pairs W j , j=1,,l, whose set is denoted by P j . Set m:= r 1 ++ r l ; i.e., there are in whole m paths in the traffic network. Let F:=( F 1 ,, F m ) stand for the path flow vector. Assume that the travel cost of the path R s , s=1,,m, is a set T s (F) R + . So, we have a multifunction T: R + m R + m with T(F):=( T 1 (F),, T m (F)). Let the capacity restriction be

FA:= { F R + m : F s Γ s , s = 1 , , m } ,

where Γ s are given real numbers. Extending the Wardrop definition to the case of multivalued costs, we propose the following definition.

A path flow vector H is said to be a weak equilibrium flow vector if

W j , R q P j , R s P j ,  there exists  t T ( H )  such that t q < t s H q = Γ q  or  H s = 0 ,
(27)

where j=1,,l and q,s{1,,m} are among r j indices corresponding to P j .

A path flow vector H is said to be a strong equilibrium flow vector if

W j , R q P j , R s P j , for all tT(H) such that  t q < t s H q = Γ q  or  H s =0.
(28)

Suppose that the travel demand ρ j of the O/D pair W j , j=1,,l, depends on the weak (or strong) equilibrium problem flow H. So, considering all the O/D pairs, we have a mapping ρ: R + m R + l . We use the Kronecker notation

ϕ j s ={ 1 if  s P j , 0 if  s P j .

Then the matrix

ϕ={ ϕ j s },j=1,,l,s=1,,m,

is called an O/D pair/path incidence matrix. The path flow vectors meeting the travel demands are called the feasible path flow vectors and form the constraint set, for a given weak (or strong) equilibrium flow H,

K(H,λ):= { F A : ϕ F = ρ ( H , λ ) } .

Assume further that the path costs are also perturbed, i.e., depend on a perturbation parameter μ of a metric space M: T s (F,μ), s=1,,m.

Our traffic equilibrium problem is equivalent to a quasi-variational inequality as follows (see [38]).

Lemma 4.4 A path vector flow HK(H,λ) is a weak equilibrium flow if and only if it is a solution of the following quasi-variational inequality:

Find  H K ( H , λ )  such that there exists  t T ( H , λ )  satisfying  t , F H 0 , F K ( H , λ ) .

Lemma 4.5 A path vector flow HK(H,λ) is a strong equilibrium flow if and only if it is a solution of the following quasi-variational inequality:

Find  H K ( H , λ )  such that for all  t T ( H , λ )  it satisfies  t , F H 0 , F K ( H , λ ) .

Corollary 4.6 Assume that solutions of the traffic network equilibrium problem exist and all the assumptions of Corollary  4.2 are satisfied. Then, in a neighborhood of ( λ 0 , μ 0 ), the solution is unique and satisfies the same Hölder condition as in Corollary  4.2.

4.3 Quasi-optimization problem

For the normed linear space Y and pointed, closed and convex cone C with nonempty interior, we denote the ordering induced by C as follows:

x y iff y x C ; x < y iff y x int C .

The orderings ≥ and > are defined similarly. Let g:X×MY be a vector-valued mapping. For each (λ,μ)Λ×M, consider the problem of parametric quasi-optimization problem (PQOP) finding x 0 K( x 0 ,λ) such that

g( x 0 ,μ)= min y K ( x 0 , λ ) g(y,μ).
(29)

Since the constraint set depends on the minimizer x 0 , this is a quasi-optimization problem. Setting f(x,y,μ)=g(y,μ)g(x,μ), (PVQEP) becomes a special case of (PQOP).

The following results are derived from Theorem 3.8 (Theorem 3.3 cannot be applied since f(x,y,μ)+f(y,x,μ)=0, x,yA and μM).

Theorem 4.7 For (PQOP), assume that the solution exists in a neighborhood N( λ 0 )×N( μ 0 ) of the considered point ( λ 0 , μ 0 )Λ×M. Assume further that the following conditions hold.

  1. (i)

    K(,) is ( l 1 α 1 , l 2 α 2 )-Hölder continuous on E(N( λ 0 ))×N( λ 0 );

  2. (ii)

    For each x,yE(N( λ 0 )), F(x,y,) is m 1 γ 1 -Hölder continuous at μ 0 M;

  3. (iii)

    For each xE(N( λ 0 )) and μN( μ 0 ), F(x,,μ) is m 2 γ 2 -Hölder continuous on E(N( λ 0 ));

  4. (iv)

    F(,,μ) is hβ-Hölder strongly monotone with respect to ξ q , i.e., there exist constants h>0, β>0 such that for every x,yE(N( λ 0 )): xy,

    h d X β (x,y)d ( inf z F ( x , y , μ ) ξ q ( z ) , R + ) +d ( inf z F ( y , x , μ ) ξ q ( z ) , R + ) ;
  5. (v)

    β= α 1 γ 2 , h>2 m 2 L l 1 γ 1 , where L:= sup λ C q λ[ 1 q ,+) is the Lipschitz constant of ξ q on Y.

Then, for every (λ,μ)N( λ 0 )×N( μ 0 ), the solution of (PVQGEP) is unique, x(λ,μ), and this function satisfies the Hölder condition:

for all ( λ 1 , μ 1 ),( λ 2 , μ 2 )N( λ 0 )×N( μ 0 ),

d X ( x ( λ 1 , μ 1 ) , x ( λ 2 , μ 2 ) ) ( 2 m 2 L l 2 γ 2 h 2 m 2 L l 1 γ 1 ) 1 β d Λ α 2 γ 2 / β ( λ 1 , λ 2 ) + ( m 1 L h 2 m 2 L l 1 γ 1 ) 1 β d M γ 1 / β ( μ 1 , μ 2 ) ,

where x( λ i , μ i ) S W ( λ i , μ i ), i=1,2.

5 Conclusions

In this paper, by using a nonlinear scalarization technique, we obtain sufficient conditions for Hölder continuity of the solution mapping for a parametric generalized vector quasi-equilibrium problem in the case where the mapping F is a general set-valued one. As applications, we derived this Hölder continuity for some quasi-variational inequalities, traffic network problems and quasi-optimization problems.

References

  1. 1.

    Ansari QH: Vector equilibrium problems and vector variational inequalities. In Vector Variational Inequalities and Vector Equilibria Mathematical Theories. Edited by: Giannessi F. Kluwer Academic, Dordrecht; 2000:1–16.

    Google Scholar 

  2. 2.

    Ansari QH, Yao JC: An existence result for the generalized vector equilibrium problem. Appl. Math. Lett. 1999, 12: 53–56.

    MathSciNet  Article  MATH  Google Scholar 

  3. 3.

    Chuong TD, Yao JC: Isolated and proper efficiencies in semi-infinite vector optimization problems. J. Optim. Theory Appl. 2014, 162: 447–462. 10.1007/s10957-013-0425-2

    MathSciNet  Article  MATH  Google Scholar 

  4. 4.

    Xue XW, Li SJ, Liao CM, Yao JC: Sensitivity analysis of parametric vector set-valued optimization problems via coderivatives. Taiwan. J. Math. 2011, 15: 2533–2554.

    MathSciNet  MATH  Google Scholar 

  5. 5.

    Chuong TD, Yao JC, Yen ND: Further results on the lower semicontinuity of efficient point multifunctions. Pac. J. Optim. 2010, 6: 405–422.

    MathSciNet  MATH  Google Scholar 

  6. 6.

    Gong XH, Kimura K, Yao JC: Sensitivity analysis of strong vector equilibrium problems. J. Nonlinear Convex Anal. 2008, 9: 83–94.

    MathSciNet  MATH  Google Scholar 

  7. 7.

    Kimura K, Yao JC: Semicontinuity of solution mappings of parametric generalized vector equilibrium problems. J. Optim. Theory Appl. 2008, 138: 429–443. 10.1007/s10957-008-9386-2

    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 quasiequilibrium problems. J. Glob. Optim. 2008, 41: 187–202. 10.1007/s10898-007-9210-9

    MathSciNet  Article  MATH  Google Scholar 

  10. 10.

    Wangkeeree R, Wangkeeree R, Preechasilp P: Continuity of the solution mappings of parametric generalized vector equilibrium problems. Appl. Math. Lett. 2014, 29: 42–45.

    MathSciNet  Article  MATH  Google Scholar 

  11. 11.

    Yen ND: Hölder continuity of solutions to parametric variational inequalities. Appl. Math. Optim. 1995, 31: 245–255. 10.1007/BF01215992

    MathSciNet  Article  MATH  Google Scholar 

  12. 12.

    Mansour MA, Riahi H: Sensitivity analysis for abstract equilibrium problems. J. Math. Anal. Appl. 2005, 306: 684–691. 10.1016/j.jmaa.2004.10.011

    MathSciNet  Article  MATH  Google Scholar 

  13. 13.

    Bianchi M, Pini R: A note on stability for parametric equilibrium problems. Oper. Res. Lett. 2003, 31: 445–450. 10.1016/S0167-6377(03)00051-8

    MathSciNet  Article  MATH  Google Scholar 

  14. 14.

    Anh LQ, Khanh PQ: On the Hölder continuity of solutions to multivalued vector equilibrium problems. J. Math. Anal. Appl. 2006, 321: 308–315. 10.1016/j.jmaa.2005.08.018

    MathSciNet  Article  MATH  Google Scholar 

  15. 15.

    Anh LQ, Khanh PQ: Uniqueness and Hölder continuity of solutions to multivalued vector equilibrium problems in metric spaces. J. Glob. Optim. 2007, 37: 349–465.

    MathSciNet  Article  Google Scholar 

  16. 16.

    Anh LQ, Khanh PQ: Sensitivity analysis for multivalued quasiequilibrium problems in metric spaces: Hölder continuity of solutions. J. Glob. Optim. 2008, 42: 515–531. 10.1007/s10898-007-9268-4

    MathSciNet  Article  MATH  Google Scholar 

  17. 17.

    Li SJ, Li XB, Teo KL: The Hölder continuity of solutions to generalized vector equilibrium problems. Eur. J. Oper. Res. 2011, 199: 334–338.

    MathSciNet  Article  MATH  Google Scholar 

  18. 18.

    Li SJ, Chen CR, Li XB, Teo KL: Hölder continuity and upper estimates of solution to vector quasiequilibrium problems. Eur. J. Oper. Res. 2011, 210: 148–157. 10.1016/j.ejor.2010.10.005

    MathSciNet  Article  MATH  Google Scholar 

  19. 19.

    Chen CR, Li SJ, Teo KL: Solution semicontinuity of parametric generalized vector equilibrium problems. J. Glob. Optim. 2009, 45: 309–318. 10.1007/s10898-008-9376-9

    MathSciNet  Article  MATH  Google Scholar 

  20. 20.

    Gong XH: Continuity of the solution set to parametric weak vector equilibrium problems. J. Glob. Optim. 2008, 139: 3–46.

    MathSciNet  Google Scholar 

  21. 21.

    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 

  22. 22.

    Li SJ, Li XB: Hölder continuity of solutions to parametric weak generalized Ky Fan inequality. J. Optim. Theory Appl. 2011, 149: 540–553. 10.1007/s10957-011-9803-9

    MathSciNet  Article  MATH  Google Scholar 

  23. 23.

    Wang QL, Lin Z, Li XX: Semicontinuity of the solution set to a parametric generalized strong vector equilibrium problem. Positivity 2014. 10.1007/s11117-014-0273-9

    Google Scholar 

  24. 24.

    Peng ZY: Hölder continuity of solutions to parametric generalized vector quasiequilibrium problems. Abstr. Appl. Anal. 2012. 10.1155/2012/236413

    Google Scholar 

  25. 25.

    Chen GY, Huang XX, Yang XQ: Vector Optimization: Set-Valued and Variational Analysis. Springer, Berlin; 2005.

    Google Scholar 

  26. 26.

    Luc DT: Theory of Vector Optimization. Springer, Berlin; 1989.

    Google Scholar 

  27. 27.

    Chen CR, Li SJ: Semicontinuity of the solution set map to a set-valued weak vector variational inequality. J. Ind. Manag. Optim. 2007, 3: 519–528.

    MathSciNet  Article  MATH  Google Scholar 

  28. 28.

    Bianchi M, Pini R: Sensitivity of parametric vector equilibria. Optimization 2006, 55: 221–230. 10.1080/02331930600662732

    MathSciNet  Article  MATH  Google Scholar 

  29. 29.

    Chen CR, Li MH: Hölder continuity of solutions to parametric vector equilibrium problem with nonlinear scalarization. Numer. Funct. Anal. Optim. 2014, 35: 685–707. 10.1080/01630563.2013.818549

    MathSciNet  Article  MATH  Google Scholar 

  30. 30.

    Chen CR: Hölder continuity of the unique solution to parametric vector quasiequilibrium problems via nonlinear scalarization. Positivity 2013, 17: 133–150. 10.1007/s11117-011-0153-5

    MathSciNet  Article  MATH  Google Scholar 

  31. 31.

    Chen GY, Yang XQ, Yu H: A nonlinear scalarization function and generalized quasi-vector equilibrium problems. J. Glob. Optim. 2005, 32: 451–466. 10.1007/s10898-003-2683-2

    MathSciNet  Article  MATH  Google Scholar 

  32. 32.

    Li SJ, Yang XQ, Chen GY: Nonconvex vector optimization of set-valued mappings. J. Math. Anal. Appl. 2003, 283: 337–350. 10.1016/S0022-247X(02)00410-9

    MathSciNet  Article  MATH  Google Scholar 

  33. 33.

    Anh LQ, Khanh PQ: Hölder continnuity of the unique solution to quasequilibrium problems in metric spaces. J. Optim. Theory Appl. 2009, 141: 37–54. 10.1007/s10957-008-9508-x

    MathSciNet  Article  MATH  Google Scholar 

  34. 34.

    Wardrop JG: Some theoretical aspects of road traffic research. Proc. Inst. Civil Eng., Part II 1952.

    Google Scholar 

  35. 35.

    Smith MJ: The existence, uniqueness and stability of traffic equilibrium. Transp. Res., Part B, Methodol. 1979, 13: 295–304.

    Article  Google Scholar 

  36. 36.

    De Luca M, Maugeri A: Quasi-variational inequality and applications to equilibrium problems with elastic demands. In Nonsmooth Optimization and Related Topic. Edited by: Clarke FM, Demyanov VF, Giannessi F. Plenum, New York; 1989:61–77.

    Google Scholar 

  37. 37.

    De Luca M: Existence of solutions for a time-dependent quasi-variational inequality. Rend. Circ. Mat. Palermo Suppl. 1997, 48: 101–106.

    MathSciNet  MATH  Google Scholar 

  38. 38.

    Khanh PQ, Luu LM: On the existence of solution to vector quasivariational inequalities and quasicomplementarity with applications to traffic equilibria. J. Optim. Theory Appl. 2004, 123: 1533–1548.

    MathSciNet  Article  MATH  Google Scholar 

Download references

Acknowledgements

The authors were partially supported by the Thailand Research Fund and Naresuan University, Grant No. RSA5780003. The authors would like to thank the referees for their remarks and suggestions, which helped to improve the paper.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Rabian Wangkeeree.

Additional information

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

Both 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

Wangkeeree, R., Preechasilp, P. On the Hölder continuity of solution maps to parametric generalized vector quasi-equilibrium problems via nonlinear scalarization. J Inequal Appl 2014, 425 (2014). https://doi.org/10.1186/1029-242X-2014-425

Download citation

Keywords

  • parametric generalized vector quasi-equilibrium problem
  • solution mapping
  • Hölder continuity
  • nonlinear scalarization