By using a linear scalarization method, we establish sufficient
conditions for the Hölder continuity of the solution mappings to a parametric generalized
vector quasiequilibrium problem with set-valued mappings. These results extend the
recent ones in the recent literature, (e.g., Li et al. (2009), Li et al. (2011)). Furthermore, two examples are given
to illustrate the obtained result.
1. Introduction
The vector equilibrium problem has been attracting great interest because it provides a unified model for several important problems such as vector variational inequalities, vector complementarity problems, vector optimization problems, vector min-max inequality, and vector saddle point problems. Many different types of vector equilibrium problems have been intensively studied for the past years; see, for example, [1–3] and the references therein.
It is important to derive results for parametric vector equilibrium problems concerning the properties of the solution mapping when the problems data vary. Among many desirable properties of vector equilibrium problems, the stability analysis of solutions is an essential topic in vector optimization theory and applications. In general, stability may be understood as lower (upper) semicontinuity, continuity, Lipschitz and Hölder continuity and so on. Recently, semicontinuity, especially lower semicontinuity, of solution mappings to parametric vector variational inequalities and parametric vector equilibrium problems has been intensively studied in the literature; see [4–12]. On the other hand, Hölder continuity of solutions to parametric vector equilibrium problems has also been discussed recently; see [13–22], although there are less works in the literature devoted to this property than to semicontinuity. There have been many papers devoted to discussing the local uniqueness and Hölder continuity of the solutions to parametric variational inequalities and parametric equilibrium problems; see [14–20] and the references therein. Yên [14] obtained Hölder continuity of the unique solution of a classic perturbed variational inequality by the metric projection method. Ait Mansour and Riahi [15] proved Hölder continuity of the unique solution for a parametric vector equilibrium problem under the concepts of strong monotonicity. Bianchi and Pini [16] introduced the concept of strong pseudomonotonicity and got the Hölder continuity of the unique solution of a parametric vector equilibrium problem. Bianchi and Pini [17] extend the results of [16] to vector equilibrium problems. Anh and Khanh [18] generalized the main results of [16] to the vector case and obtained Hölder continuity of the unique solutions for two classes of perturbed generalized vector equilibrium problems. Anh and Khanh [19] further discussed uniqueness and Hölder continuity of the solutions for perturbed generalized vector equilibrium problems, which improved remarkably the results in [16, 18]. Anh and Khanh [20] extended the results of [19] to the case of perturbed generalized vector quasiequilibrium problems and obtained Hölder continuity of the unique solutions.
For general perturbed vector quasiequilibrium problems, it is well known that a solution mapping is, in general, a set-valued one, but not a single-valued one. Naturally, there is a need to study Hölder continuous properties of the set-valued solution mappings. Under the Hausdorff distance and the strong quasimonotonicity, Lee et al. [21] first showed that the set-valued solution mapping for a parametric vector variational inequality is Hölder continuous. Recently, by virtue of the strong quasimonotonicity, Ait Mansour and Aussel [22] discussed Hölder continuity of set-valued solution mappings for parametric generalized variational inequalities. Li et al. [23] introduced an assumption, which is weaker than the corresponding ones of [16, 18], and established the Hölder continuity of the set-valued solution mappings for two classes of parametric generalized vector quasiequilibrium problems in general metric spaces. Li et al. [24] extended the results of [23] to perturbed generalized vector quasiequilibrium problems. Later, S. J. Li and X. B. Li [25] use a scalarization technique to obtain the Hölder continuity of the set-valued solution mappings for a parametric vector equilibrium problem in general metric spaces.
Motivated by the work reported in [21, 23, 25], this paper aims at establishing sufficient conditions for Hölder continuity of the solution sets for a class of parametric generalized vector quasiequilibrium problem (PGVQEP, in short) with set-valued mapping, by using a linear scalarization method. The main results in this paper are different from corresponding results in [23, 24] and overcome the drawback, which requires the knowledge of detailed values of the solution mapping in a neighborhood of the point under consideration. Our main results also extend and improve the corresponding ones in [25].
The rest of the paper is organized as follows. In Section 2, we introduce the (PGVQEP) and define the solution and ξ-solution to the (PGVQEP). Then, we recall some notions and definitions which are needed in the sequel. In Section 3, we discuss Hölder continuity of the solution mapping for the (PGVQEP) and compare our main results with the corresponding ones in the recent literature. We also give two examples to illustrate that our main results are applicable.
2. Preliminaries
Throughout this paper, if not other specified, ∥·∥ and d(·,·) denote the norm and metric in any metric space, respectively. Let B(0,δ) denote the closed ball with radius δ≥0 and center 0 in any metric linear spaces. Let X,Λ,M, Y be metric linear spaces. Let Y* be the topological dual space of Y. Let C⊂Y be a pointed, closed, and convex cone with intC≠∅, where intC denotes the interior of C. Let C*:={f∈Y*:f(y)≥0,forally∈C} be the dual cone of C. Since intC≠∅, the dual cone C* of C has a weak* compact base. Letting e∈intC be given, then Be*:={ξ∈C*:∥ξ∥=1} is a weak* compact base of C*.
Let N(λ0)⊂Λ and N(μ0)⊂M be neighborhoods of considered points λ0 and μ0, respectively. Let K:X×Λ⇉X be a set-valued mapping, and let F:X×X×M⇉Y be a set-valued mapping. For each λ∈N(λ0) and μ∈N(μ0), consider the following parameterized generalized vector quasiequilibrium problem of finding x0∈K(x0,λ) such thatF(x0,y,μ)⊂Y∖-intC,∀y∈K(x0,λ).
For each λ∈N(λ0) and μ∈N(μ0), letE(λ)∶={x∈X∣x∈K(x,λ)}.
Let S(λ,μ) be the solution set of (PGVQEP), that is,S(λ,μ)∶={x∈E(λ)∣F(x,y,μ)⊂Y∖-intC,∀y∈K(x,λ)}.
For each ξ∈C*∖{0}, each λ∈N(λ0) and μ∈N(μ0), let Sξ(λ,μ) denote the set of ξ-solution set to (PGVQEP), that is,Sξ(λ,μ)∶={x∈E(λ):infz∈F(x,y,μ)f(z)≥0,∀y∈K(x,λ)}.
Special Case
When K(x,λ)=K(λ), that is, K does not depend on x, the (PGVQEP) reduces to the parametric generalized vector equilibrium problem (PGVEP) considered by Li et al. [23].
If F:X×X×M→ℝ, the (PGVQEP) collapses to the quasiequilibrium problem (QEP) considered by Anh and Khanh [26].
If K(x,λ)=K(λ) and F is a vector-valued mapping, that is, F:X×X×M→Y, the (PGVQEP) reduce to the parametric Ky Fan inequality (PKI) considered by S. J. Li and X. B. Li [25].
Now we recall some basic definitions and their properties which are needed in this paper.
Definition 2.1 (classical notion).
A set-valued mapping G:M⇉X is said to be ℓ·α-Hölder continuous at μ0 if there is a neighborhood U(μ0) of μ0 such that, for all μ1,μ1∈U(μ0),
G(μ1)⊆G(μ2)+lB(0,dα(μ1,μ2)),
where ℓ≥0 and α>0.
Definition 2.2.
A set-valued mapping G:X×Λ⇉Y is said to be (ℓ1·α1,ℓ2·α2)-Hölder continuous at (x0,λ0) if and only if there exists neighborhoods N(x0) of x0 and N(μ0) of μ0 such that, forallx1,x2∈N(x0),forallλ1,λ2∈N(λ0),
G(x1,λ1)⊆G(x2,λ2)+(l1dα1(x1,x2)+l2dα2(λ1,λ2))B(0,1),
where ℓ1,ℓ2≥0 and α1,α2>0.
Definition 2.3 (see [25]).
A set-valued mapping G:M⇉Y is said to be (ℓ·α)-Hölder continuous with respect to e∈intC at μ0 if and only if there exists neighborhoods N(μ0) of μ0 such that, for all μ1,μ2∈N(μ0),
G(μ1)⊆G(μ2)+ldα(μ1,μ2)[-e,e],
where ℓ≥0, α>0 and [-e,e]={x:x∈e-C,x∈-e+C}.
Definition 2.4.
Let F:X×X×Λ⇉Y be a set-valued mapping with nonempty values; F(x,·,μ) is called C-like convex on A(λ) if and only if for any x1,x2∈X and any t∈[0,1], there exists x3∈X such that
tF(x,x1,λ)+(1-t)F(x,x2,λ)⊂F(x,x3,λ)+C.
Remark 2.5.
If for each μ∈N(μ0) and each x∈E(N(λ0)), F(x,·,μ) is C-like convex on E(N(λ0)), then F(x,E(N(λ0)),μ)+C is a convex set.
3. Main Results
In this section, we mainly discuss the Hölder continuity of the solution mappings to (PGVQEP).
Lemma 3.1.
Suppose that N(λ0),N(μ0) are the given neighborhoods of λ0, μ0, respectively.
If for each x,y∈E(N(λ0)), F(x,y,·) is m1·γ1-Hölder continuous with respect to e∈intC at μ0∈M, then for any ξ∈Be*, the function φξ(x,y,·)=infz∈F(x,y,·)ξ(z) is m1·γ1-Hölder continuous at μ0.
If for each x∈E(N(λ0)) and μ∈N(E(μ0)), F(x,·,μ) is m2·γ2-Hölder continuous with respect to e∈intC on E(N(λ0)), then for each ξ∈Be*, φξ(x,·,μ)=infz∈F(x,·,μ)ξ(z) is also m2·γ2-Hölder continuous on E(N(λ0)).
Proof.
(a) By assumption, there exists a neighborhood N(μ0) of μ0, such that for all μ1,μ2∈N(μ0),forallx,y∈E(N(λ0)):x≠y,
F(x,y,μ1)⊂F(x,y,μ2)+m1dγ1(μ1,μ2)[-e,e].
So, for any z1∈F(x,y,μ1), there exist z2∈F(x,y,μ2) and e0∈[-e,e] such that
z1=z2+m1dγ1(μ1,μ2)e0.
Then, by the linearity of ξ, we have
ξ(z1)-ξ(z2)=m1dγ1(μ1,μ2)ξ(e0).
It follows from ξ(e)=1,e0∈[-e,e], and the structure of [-e,e] that
ξ(e0)≥-1.
Therefore, (3.3) and (3.4) together yield that
-m1dγ1(μ1,μ2)≤ξ(z1)-ξ(z2).
Since z1 is arbitrary and ξ(z2)≥infz∈F(x,y,μ2)ξ(z), we have
-m1dγ1(μ1,μ2)≤infz∈F(x,y,μ1)ξ(z)-infz∈F(x,y,μ2)ξ(z).
Due to the symmetry between μ1 and μ2, the same estimate is also valid, that is,
-m1dγ1(μ1,μ2)≤infz∈F(x,y,μ2)ξ(z)-infz∈F(x,y,μ1)ξ(z).
Thus, it follows (3.6) and (3.7) that
|infz∈F(x,y,μ1)ξ(z)-infz∈F(x,y,μ2)ξ(z)|=|φξ(x,y,μ1)-φξ(x,y,μ2)|≤m1dγ1(μ1,μ2)
and the proof is completed.
(b) As the proof of (b) is similar to (a), we omit it. Then the proof is completed.
Lemma 3.2.
If for each μ∈N(μ0) and each x∈E(N(λ0)), F(x,·,μ) is C-like convex on E(N(λ0)), that is, F(x,E(N(λ0)),μ)+C is a convex set, then
S(λ,μ)=∪ξ∈C*∖0Sξ(λ,μ)=∪ξ∈Be*Sξ(λ,μ).
Proof.
In a similar way to the proof of Lemma 3.1 in [8], with suitable modifications, we can obtain the conclusion.
Theorem 3.3.
Assume that for each ξ∈Be*, the ξ-solution set for (PGVQEP) exists in a neighborhood N(λ0)×N(μ0) of the considered point (λ0,μ0)∈Λ×M. Assume further that the following conditions hold.
K(·,·) is (ℓ1·α1,ℓ2·α2)-Hölder continuous in E(N(λ0))×N(μ0).
For each x,y∈E(N(λ0)), F(x,y,·) is m1·γ1-Hölder continuous with respect to e∈intC at μ0∈M.
For each x∈E(N(λ0)) and μ∈N(E(μ0)), F(x,·,μ) is m2·γ2-Hölder continuous with respect to e∈intC on E(N(λ0)).
forallξ∈Be*,μ∈N(μ0),forallx,y∈E(N(λ0)):x≠y, there exists two constants h>0 and β>0 such that
hdβ(x,y)≤d(infz∈F(x,y,μ)ξ(z),R+)+d(infz∈F(y,x,μ)ξ(z),R+).
α1γ2=β and h>2m2ℓ1γ2.
Then, for any ξ¯∈Be*, there exists open neighborhoods N(ξ¯) of ξ¯, Nξ¯(λ0) of λ0 and Nξ¯(μ0) of μ0, such that the ξ-solution set Sξ(·,·) on N(ξ¯)×Nξ¯(λ0)×Nξ¯(μ0) satisfies the following Hölder condition: forallξ∈N(ξ¯),forall(λ1,μ1),(λ2,μ2)∈Nξ¯(λ0)×Nξ¯(μ0),
d(xξ(λ1,μ1),xξ(λ2,μ2))≤(m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2)+(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2),
where xξ(λi,μi)∈Sξ(λi,μi),i=1,2.
Proof.
Let (λ1,μ1),(λ2,μ2)∈Nξ¯(λ0)×Nξ¯(μ0) be arbitrarily given. For all ξ∈Be*, x,y∈X, and μ∈M, we set φξ(x,y,·)∶=infz∈F(x,y,·)ξ(z) for the sake of convenient statement in the sequel. We prove that (3.11) holds by the following three steps.Step 1.
We first show that, for all xξ(λ1,μ1)∈Sξ(λ1,μ1), for all xξ(λ1,μ2)∈Sξ(λ1,μ2),
d(xξ(λ1,μ1),xξ(λ1,μ2))≤(m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2).
Obviously, if xξ(λ1,μ1)=xξ(λ1,μ2), we have that (3.12) holds. So we suppose xξ(λ1,μ1)≠xξ(λ1,μ2). Since xξ(λ1,μ1)∈K(xξ(λ1,μ1),λ1),xξ(λ1,μ2)∈K(xξ(λ1,μ2),λ1), and by the Hölder continuity of K(·,λ1), there exists x1∈K(xξ(λ1,μ1),λ1) and x2∈K(xξ(λ1,μ2),λ1) such that
d(xξ(λ1,μ1),x2)≤l1dα1(xξ(λ1,μ1),xξ(λ1,μ2)),d(xξ(λ1,μ2),x1)≤l1dα1(xξ(λ1,μ1),xξ(λ1,μ2)).
Since xξ(λ1,μ1),xξ(λ1,μ2) are ξ-solutions to (PGVQEP) at parameters (λ1,μ1),(λ1,μ2), respectively, we obtain
φξ(xξ(λ1,μ1),x1,μ1)≥0,φξ(xξ(λ1,μ2),x2,μ2)≥0.
By virtue of (iv), we get that
hdβ(xξ(λ1,μ1),xξ(λ1,μ2))≤d(ϕξ(xξ(λ1,μ2),xξ(λ1,μ1),μ1),R+)+d(ϕξ(xξ(λ1,μ1),xξ(λ1,μ2),μ1),R+),
which together with (3.14) yields that
hdβ(xξ(λ1,μ1),xξ(λ1,μ2))≤|ϕξ(xξ(λ1,μ2),xξ(λ1,μ1),μ1)-ϕξ(xξ(λ1,μ2),x2,μ2)|+|ϕξ(xξ(λ1,μ1),xξ(λ1,μ2),μ1)-ϕξ(xξ(λ1,μ1),x1,μ1)|≤|ϕξ(xξ(λ1,μ2),xξ(λ1,μ1),μ1)-ϕξ(xξ(λ1,μ2),xξ(λ1,μ1),μ2)|+|ϕξ(xξ(λ1,μ2),xξ(λ1,μ1),μ2)-ϕξ(xξ(λ1,μ2),x2,μ2)|+|ϕξ(xξ(λ1,μ1),xξ(λ1,μ2),μ1)-ϕξ(xξ(λ1,μ1),x1,μ1)|.
Then, from Lemma 3.1, (3.13), we have
hdβ(xξ(λ1,μ1),xξ(λ1,μ2))≤m1dγ1(μ1,μ2)+m2dγ2(xξ(λ1,μ2),x1)+m2dγ2(xξ(λ1,μ1),x2)≤m1dγ1(μ1,μ2)+2m2lγ2dα1γ2(xξ(λ1,μ1),xξ(λ1,μ2)).
The assumption (v) yields that
d(xξ(λ1,μ1),xξ(λ1,μ2))≤(m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2).
Hence, we have that (3.12) holds.
Step 2.
Now we show that, for all xξ(λ1,μ2)∈Sξ(λ1,μ2), for all xξ(λ2,μ2)∈Sξ(λ2,μ2),
d(xξ(λ1,μ2),xξ(λ2,μ2))≤(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2).
Obviously, we only need to prove that (3.19) holds when xξ(λ1,μ2)≠xξ(λ2,μ2). By virtue of assumption (i), there exists x1′∈K(xξ(λ2,μ2),λ1) and x2′∈K(xξ(λ1,μ2),λ2) such that
d(xξ(λ2,μ2),x1′)≤l2dα2(λ1,λ2),d(xξ(λ1,μ2),x2′)≤l2dα2(λ1,λ2).
By the Hölder continuity of K(·,·), there exists x1′′∈K(xξ(λ1,μ2),λ1) and x2′′∈K(xξ(λ2,μ2),λ2) such that
d(x1′,x1′′)≤l1dα1(xξ(λ1,μ2),xξ(λ2,μ2)),d(x2′,x2′′)≤l1dα1(xξ(λ1,μ2),xξ(λ2,μ2)).
From the definition of ξ-solution for (PGVQEP), we have
ϕξ(xξ(λ1,μ2),x1′′,μ2)≥0,ϕξ(xξ(λ2,μ2),x2′′,μ2)≥0.
From assumptions (ii)–(iv), (3.22), and Lemma 3.1, we have
hdβ(xξ(λ1,μ2),xξ(λ2,μ2))≤d(ϕξ(xξ(λ1,μ2),xξ(λ2,μ2),μ2),R+)+d(ϕξ(xξ(λ2,μ2),xξ(λ1,μ2),μ2),R+)≤d(ϕξ(xξ(λ1,μ2),x(λ2,μ2),μ2)-ϕξ(xξ(λ1,μ2),x1′′,μ2))+d(ϕξ(xξ(λ2,μ2),xξ(λ1,μ2),μ2)-ϕξ(xξ(λ2,μ2),x2′′,μ2))≤d(ϕξ(xξ(λ1,μ2),x(λ2,μ2),μ2)-ϕξ(xξ(λ1,μ2),x1′,μ2))+d(ϕξ(xξ(λ1,μ2),x1′,μ2)-ϕξ(xξ(λ1,μ2),x1′′,μ2))+d(ϕξ(xξ(λ2,μ2),xξ(λ1,μ2),μ2)-ϕξ(xξ(λ2,μ2),x2′,μ2))+d(ϕξ(xξ(λ2,μ2),x2′,μ2)-ϕξ(xξ(λ2,μ2),x2′′,μ2))≤m2dγ2(x(λ2,μ2),x1′)+m2dγ2(x1′,x1′′)+m2dγ2(x(λ1,μ2),x2′)+m2dγ2(x2′,x2′′).
By virtue of (3.20)–(3.21) and (3.23), we can get
hdβ(xξ(λ1,μ2),xξ(λ2,μ2))≤m2l2γ2dα2γ2(λ1,λ2)+m2l1γ2dα1γ2(xξ(λ1,μ2),xξ(λ2,μ2))+m2l2γ2dα2γ2(λ1,λ2)+m2l1γ2dα1γ2(xξ(λ1,μ2),xξ(λ2,μ2)).
Therefore, it follows from (v) that
d(xξ(λ1,μ2),xξ(λ2,μ2))≤(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2)
and the conclusion (3.19) holds.
Step 3.
Finally, by the arbitrariness of xξ(λ1,μ1)∈Sξ(λ1,μ1),xξ(λ1,μ2)∈Sξ(λ1,μ2),xξ(λ2,μ2)∈Sξ(λ2,μ2), (3.12) and (3.19), we can easily get that
d(xξ(λ1,μ1),xξ(λ2,μ2))≤d(xξ(λ1,μ1),xξ(λ1,μ2))+d(xξ(λ1,μ2),xξ(λ2,μ2))≤(m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2)+(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2)
and the conclusion (3.11) holds. This completes the proof.
Remark 3.4.
Theorem 3.3 generalizes Lemma 3.3 in S. J. Li and X. B. Li [25] from vector-valued version to set valued version. Moreover, the assumption (H4) of Lemma 3.3 in [25] is removed.
Now, we give an example to illustrate that Theorem 3.3 is applicable under the case that the mapping F is set valued.
Example 3.5.
Let X=Y=R,Λ=M=[0,1],C=ℝ+ and e=3/2∈intC. Let K:X×M⇉Y be defined by
K(x,λ)=[λ2+x16,1]
and F:X×X×M⇉Y a set-valued mapping defined by
F(x,y,λ)=[(1+λ)(x+12)(y-x),28-2x3/2].
Then, E(λ)=[λ2/15,1]. Consider that λ0=0.5 and N(λ0)=Λ. Direct computation shows that E(Λ)=E(N(λ0))=[0,1].
It can be checked that K(·,·) is ((1/16)·1,(3/2)·1)-Hölder continuous in E(N(λ0))×N(μ0); for all x,y∈E(N(λ0)), F(x,y,·) is 62.1-Hölder continuous with respect to e=3/2∈intC at λ0∈M; for each x∈E(N(λ0)) and λ∈N(E(λ0)), F(x,·,λ) is 3.1-Hölder continuous with respect to e∈intC on E(N(λ0)). Here ℓ1=1/16,α1=1,ℓ2=3/2,α2=1,m1=62,γ1=1,m2=3,γ2=1. Take β=1 and h=1/2, for any ξ∈Be* and for all x,y∈E(N(λ0)):x≠y, we have
hdβ(x,y)≤d(infz∈F(x,y,μ)ξ(z),R+)+d(infz∈F(y,x,μ)ξ(z),R+)
and also have α1γ2=β and h>2m2ℓ1γ2=3/8. Hence, all assumptions of Theorem 3.3 hold, and thus it is valid.
Theorem 3.6.
Assume that for each ξ∈Be*, the ξ-solution set for (PGVQEP) exists in a neighborhood N(λ0)×N(μ0) of the considered point (λ0,μ0)∈Λ×M. Assume further that the following conditions hold:
K(·,·) is (ℓ1·α1,ℓ2·α2)-Hölder continuous in E(N(λ0))×N(μ0);
for each x,y∈E(N(λ0)), F(x,y,·) is m1·γ1-Hölder continuous with respect to e∈intC at μ0∈M;
for each x∈E(N(λ0)) and μ∈N(E(μ0)), F(x,·,μ) is m2·γ2-Hölder continuous with respect to e∈intC on E(N(λ0));
forallξ∈Be*,μ∈N(μ0),forallx,y∈E(N(λ0))(x≠y), there exist two constants h>0 and β>0 such that
hdβ(x,y)≤d(infz∈F(x,y,μ)ξ(z),R+)+d(infz∈F(y,x,μ)ξ(z),R+);
forallx∈E(N(λ0)),forallμ∈N(μ0), F(x,·,μ) is C-like convex on E(N(λ0));
α1γ2=β and h>2m2ℓ1γ2.
Then there exist neighborhoods Ñ(λ0) of λ0 and Ñ(μ0) of μ0, such that the solution set S(·,·) on Ñ(λ0)×Ñ(μ0) is nonempty and satisfies the following Hölder continuous condition, for all (λ1,μ1),(λ2,μ2)∈Ñ(λ0)×Ñ(μ0):
S(λ1,μ1)⊂S(λ2,μ2)+((m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2)+(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2))B(0,1).
Proof.
Since the system of {N′(ξ¯)}ξ¯∈Be*, which are given by Theorem 3.3, is an open covering of the weak* compact set Be*, there exist a finite number of points (ξi)(i=1,2,…,n.) from Be* such that
Be*⊂⋃i=1nN′(ξi).
Hence, let Ñ(λ0)=⋂i=1nN′ξi(λ0) and Ñ(μ0)=⋂i=1nN′ξi(μ0). Then Ñ(λ0) and Ñ(μ0) are desired neighborhoods of λ0 and μ0, respectively. Indeed, let (λ,μ)∈Ñ(λ0)×Ñ(μ0) be given arbitrarily. For any ξ∈Be*, by virtue of (3.32), there exists i0∈{1,2,…,n} such that ξ∈N′(ξi0). From the construction of the neighborhoods Ñ(λ0) and Ñ(μ0), one has
(λ,μ)∈Nξi0′(λ0)×Nξi0′(μ0).
Then, from the assumption of existence for ξ-solution set and Lemma 3.2, S(λ,μ)=⋃ξ∈Be*Sξ(λ,μ) is nonempty.
Now, we show that (3.31) holds. Indeed, taking any (λ1,μ1),(λ2,μ2)∈Ñ(λ0)×Ñ(μ0), we need to show that for any x1∈S(λ1,μ1), there exists x2∈S(λ2,μ2) satisfying
d(x1,x2)≤(m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2)+(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2).
Since x1∈S(λ1,μ1)=⋃ξ∈Be*Sξ(λ1,μ1), there exists ξ̂∈Be* such that
x1=xξ̂(λ1,μ1)∈Sξ̂(λ1,μ1).
It follows from (3.32) that there exists i0∈{1,2,…,n} such that ξ̂∈N′(ξi0). Thus, by the construction of the neighborhoods Ñ(λ0) and Ñ(μ0), we have
(λ1,μ1),(λ2,μ2)∈Nξi0(λ0)×Nξi0(μ0).
Obviously, thanks to Theorem 3.3, we have
d(xξ̂(λ1,μ1),xξ̂(λ2,μ2))≤(m1h-2m2l1γ2)1/βdγ1/β(μ1,μ2)+(2m2l2γ2h-2m2l1γ2)1/βdα2γ2/β(λ1,λ2).
Let x2=xξ̂(λ2,μ2). Then, (3.34) holds, and the proof is complete.
Remark 3.7.
Theorem 3.6 generalizes, and improves the corresponding results of S. J. Li and X. B. Li [25] in the following three aspects.
The vector-valued mapping F(x,y,μ) is extended to set-valued, and the parametric vector equilibrium problem is extended to the parametric vector quasiequilibrium problem.
The assumption (H4) of Theorem 3.1 in [25] is removed.
The C-convexity of F(x,·,μ) (see Theorem 3.1 in [25]) is extended to C-convexlikeness.
In addition, it is easy to see that the assumption (iv) of Theorem 3.6 is different form the assumption (H1) of Theorem 3.1 in S. J. Li and X. B. Li [25].
Moreover, we also can see that the obtained result extends the ones of [23]. Now, we give the following example to illustrate the case.
Example 3.8.
Let X=Y=ℝ,Λ=M=[0,1],C=ℝ+, and e=2/2∈intC. Let K:X×M⇉Y be defined by K(x,λ)=[λ2,1], and let F:X×X×M⇉Y be a set-valued mapping defined by
F(x,y,λ)=[(34+2λ)(y+3)(x-y),20-|x|1/2].
Consider that λ0=0.5 and N(λ0)=Λ. Then, E(λ)=[λ2,1] and E(Λ)=E(N(λ0))=[0,1].
Obviously, K(·,·) is (0.1,2·1)-Hölder continuous in E(N(λ0))×N(μ0); for all x,y∈E(N(λ0)), F(x,y,·) is 72·1-Hölder continuous with respect to e=2/2∈intC at λ0∈M; for each x∈E(N(λ0)) and λ∈N(E(λ0)), F(x,·,λ) is 95·1-Hölder continuous with respect to e∈intC on E(N(λ0)). Here ℓ1=0,α1=1,ℓ2=2,α2=1,m1=72,γ1=1,m2=95,γ2=1. Take β=1 and h=3/4, for any ξ∈Be* and for all x,y∈E(N(λ0))(x≠y), we havehdβ(x,y)≤d(infz∈F(x,y,μ)ξ(z),R+)+d(infz∈F(y,x,μ)ξ(z),R+)
and also have α1γ2=β and h=3/4>2m2ℓ1γ2. Therefore, all assumptions of Theorem 3.3 hold, and thus it is applicable.
However, the assumption (ii) of Theorem 3.1 (or (ii′) of Theorem 4.1) in [23] does not hold. In fact, for any λ∈Λ, for any h>0 and β>0, there exists y0=0∈E(N(λ0))∖Sξ(λ,μ) such thatF(y0,x¯,λ)+hB(0,dβ(x¯,y0))=[-(34+2λ)(x¯+3)x¯,20-|x¯|1/2]+hB(0,dβ(0,x¯))⊈-R+
for all x¯∈Sξ(λ,μ). Thus, Theorems 3.1 and 4.1 in Li et al. [23] are not applicable.
Acknowledgments
The author would like to thank the anonymous referees for valuable comments and suggestions, which helped to improve the paper. This work was supported by the Natural Science Foundation of China (no. 10831009. 11001287), the Natural Science Foundation Project of ChongQing (no. CSTC, 2010BB9254. 2011AC6104), and the Research Grant of Chongqing Key Laboratory of Operations and System Engineering.
GiannessiF.200038Dordrecht, The NetherlandsKluwer Academic Publishersxiv+523Nonconvex Optimization and Its Applications1789109GiannessiF.MaugeriA.PardalosP. M.2001Dordrecht, The NetherlandsKluwer Acad. Publ.ChenG.-Y.HuangX.YangX.2005541Berlin, GermanySpringerx+306Lecture Notes in Economics and Mathematical Systems2164220ChengY. H.ZhuD. L.Global stability results for the weak vector variational inequality200532454355010.1007/s10898-004-2692-92191187ZBL1097.49006AnhL. Q.KhanhP. Q.Semicontinuity of the solution set of parametric multivalued vector quasiequilibrium problems20042942699711206135210.1016/j.jmaa.2004.03.014ZBL1048.49004AnhL. Q.KhanhP. Q.On the stability of the solution sets of general multivalued vector quasiequilibrium problems20071352271284234653510.1007/s10957-007-9250-9ZBL1146.90516HuangN. J.LiJ.ThompsonH. B.Stability for parametric implicit vector equilibrium problems20064311-1212671274222855010.1016/j.mcm.2005.06.010ZBL1187.90286ChenC. R.LiS. J.TeoK. L.Solution semicontinuity of parametric generalized vector equilibrium problems200945230931810.1007/s10898-008-9376-92539163ZBL1213.54028ChenC. R.LiS. J.On the solution continuity of parametric generalized systems2010611411512604521ZBL1190.49032GongX. H.YaoJ. C.Lower semicontinuity of the set of efficient solutions for generalized systems20081382197205241499310.1007/s10957-008-9379-1GongX. H.Continuity of the solution set to parametric weak vector equilibrium problems200813913546243859310.1007/s10957-008-9429-8ZBL1189.90195KimuraK.YaoJ.-C.Sensitivity analysis of solution mappings of parametric vector quasi-equilibrium problems200841218720210.1007/s10898-007-9210-92398931ZBL1165.49009AnhL. Q.KhanhP. Q.Sensitivity analysis for weak and strong vector quasiequilibrium problems2009372-32372532568019ZBL1202.90247YênN. D.Hölder continuity of solutions to a parametric variational inequality1995313245255131625910.1007/BF01215992Ait MansourM.RiahiH.Sensitivity analysis for abstract equilibrium problems20053062684691213634210.1016/j.jmaa.2004.10.011ZBL1068.49005BianchiM.PiniR.A note on stability for parametric equilibrium problems2003316445450200381810.1016/S0167-6377(03)00051-8ZBL1112.90082BianchiM.PiniR.Sensitivity for parametric vector equilibria200655322123010.1080/023319306006627322238411ZBL1149.90156AnhL. Q.KhanhP. Q.On the Hölder continuity of solutions to parametric multivalued vector equilibrium problems20063211308315223656010.1016/j.jmaa.2005.08.018ZBL1104.90041AnhL. Q.KhanhP. Q.Uniqueness and Hölder continuity of the solution to multivalued equilibrium problems in metric spaces200737344946510.1007/s10898-006-9062-82293665ZBL1156.90025AnhL. Q.KhanhP. Q.Sensitivity analysis for multivalued quasiequilibrium problems in metric spaces: Hölder continuity of solutions200842451553110.1007/s10898-007-9268-42448995ZBL1188.90274LeeG. M.KimD. S.LeeB. S.YenN. D.Vector variational inequality as a tool for studying vector optimization problems199834574576510.1016/S0362-546X(97)00578-61634819ZBL0956.49007Ait MansourM.AusselD.Quasimonotone variational inequalities and quasiconvex programming: quantitative stability2006236116262265758ZBL1124.49023LiS. J.LiX. B.TeoK. L.The Hölder continuity of solutions to generalized vector equilibrium problems20091992334338253327910.1016/j.ejor.2008.12.024ZBL1176.90643LiS. J.ChenC. R.LiX. B.TeoK. L.Hölder continuity and upper estimates of solutions to vector quasiequilibrium problems20112102148157274687710.1016/j.ejor.2010.10.005LiS. J.LiX. B.Hölder continuity of solutions to parametric weak generalized Ky Fan inequality20111493540553279549110.1007/s10957-011-9803-9AnhL. Q.KhanhP. Q.Hölder continuity of the unique solution to quasiequilibrium problems in metric spaces200914113754249591710.1007/s10957-008-9508-xZBL1176.90584