 Research
 Open Access
 Published:
Error bounds of regularized gap functions for weak vector variational inequality problems
Journal of Inequalities and Applications volume 2014, Article number: 331 (2014)
Abstract
In this paper, by the nonlinear scalarization method, a global error bound of a weak vector variational inequality is established via a regularized gap function. The result extends some existing results in the literature.
MSC:49K40, 90C31.
1 Introduction
Throughout this paper, let K be a closed convex subset of an Euclidean space {R}^{n} and F:{R}^{n}\to B({R}^{n},{R}^{m}) be a continuously differentiable mapping. We consider a weak vector variational inequality (WVVI) of finding {x}^{\ast}\in K such that
where C\subseteq {R}^{m} is a closed convex and pointed cone with nonempty interior intC. (WVVI) was firstly introduced by Giannessi [1]. It has been shown to have many applications in vector optimization problems and traffic equilibrium problems (e.g., [2, 3]).
Error bounds are to depict the distance from a feasible solution to the solution set, and have played an important role not only in sensitivity analysis but also in convergence analysis of iterative algorithms. Recently, kinds of error bounds have been presented for weak vector variational inequalities in [4–7]. By using a scalarization approach of Konnov [8], Li and Mastroeni [5] established the error bounds for two kinds of (WVVIs) with setvalued mappings. By a regularized gap function and a Dgap function for a weak vector variational inequality, Charitha and Dutta [4] obtained the error bounds of (WVVI), respectively. Recently, in virtue of the regularized gap functions, Sun and Chai [6] studied some error bounds for generalized (WVVIs). By using the image space analysis, Xu and Li [7] got a gap function for (WVVI). Then, they established an error bound for (WVVI) without the convexity of the constraint set. These papers have a common characteristic: the solution set of (WVVI) is a singleton [6, 7]. Even though the solution set of (WVVI) is not a singleton [4, 5], the solution set of the corresponding variational inequality (VI) is a singleton, when their results reduce to (VI).
In this paper, by the nonlinear scalarization method, we study a global error bound of (WVVI). This paper is organized as follows. In Section 2, we establish a global error bound of (VI) via the generalized gap functions. In Section 3, we discuss a global error bound of (WVVI) by the nonlinear scalarization method.
2 A global error bound of (VI)
Let h:{R}^{n}\to R\cup \{+\mathrm{\infty}\} be a proper lower semicontinuous function, and let S=\{x\in {R}^{n}h(x)\le 0\}. h has a global error bound if there exists \tau >0 such that
where h{(x)}_{+}:=max\{h(x),0\} and d(x,S):=inf\{\parallel xs\parallel s\in S\} if S is nonempty and d(x,S)=+\mathrm{\infty} if S is empty. f:{R}^{n}\to {R}^{n} is said to be coercive on K if
f:{R}^{n}\to {R}^{n} is said to be strongly monotone on {R}^{n} with the modulus \lambda >0 if
In this section, we establish a global error bound of (VI) of finding x\in K such that
where f:{R}^{n}\to {R}^{n} is a continuously differentiable mapping.
To study the error bound of (VI), we need to construct a class of merit functions which were made to reformulate (VI) as an optimization problem; see [9–16]. One of such functions is a generalized regularized gap function [17] defined by
where \phi :{R}^{n}\times {R}^{n}\to R is a realvalued function with the following properties:
(P1) φ is continuously differentiable on {R}^{n}\times {R}^{n}.
(P2) \phi (x,y)\ge 0, \mathrm{\forall}x,y\in {R}^{n} and the equality holds if and only if x=y.
(P3) \phi (x,\cdot ) is uniformly strongly convex on {R}^{n} with the modulus \beta >0 in the sense that
where {\mathrm{\nabla}}_{2}\phi denotes the partial derivative of φ with respect to the second variable.
(P4) {\mathrm{\nabla}}_{2}\phi (\cdot ,y) is uniformly Lipschitz continuous on {R}^{n} with the modulus α, i.e., for all x\in {R}^{n},
Now we recall some properties of φ in (1).
Proposition 2.1 The following statements hold for each x,y\in K:

(i)
\u3008{\mathrm{\nabla}}_{2}\phi (x,y),u\u3009\le \alpha \parallel xy\parallel \parallel u\parallel, \mathrm{\forall}u\in span(Kx).

(ii)
\beta {\parallel xy\parallel}^{2}\le \phi (x,y)\le (\alpha \beta ){\parallel xy\parallel}^{2}.

(iii)
\phi (x,y)\u3008{\mathrm{\nabla}}_{2}\phi (x,y),yx\u3009\ge (\alpha \beta ){\parallel xy\parallel}^{2}.

(iv)
{\mathrm{\nabla}}_{2}\phi (x,y)=0 if and only if x=y.
Proof Parts (i)(iii) are taken from [[14], Lemma 2.1] and part (iv) is from [[17], Lemma 2.1]. □
Remark 2.1 In light of (ii) in Proposition 2.1, it holds true that \alpha \ge 2\beta.
Then we list some basic properties of the generalized regularized gap function {f}_{\gamma}.
Proposition 2.2 The following conclusions are valid for (VI).

(i)
For every x\in {R}^{n}, there exists a unique vector {y}_{\gamma}^{\phi}(x)\in K at which the infimum in (1) is attained, i.e.,
{f}_{\gamma}(x)=\u3008f(x),{y}_{\gamma}^{\phi}(x)x\u3009\gamma \phi (x,{y}_{\gamma}^{\phi}(x)). 
(ii)
{f}_{\gamma} is a gap function of (VI).

(iii)
x={y}_{\gamma}^{\phi}(x) if and only if x is a solution of (VI).

(iv)
{f}_{\gamma} is continuously differentiable on {R}^{n} with
\mathrm{\nabla}{f}_{\gamma}(x)=\mathrm{\nabla}f(x)({y}_{\gamma}^{\phi}(x)x)+f(x)\gamma {\mathrm{\nabla}}_{1}\phi (x,{y}_{\gamma}^{\phi}(x)). 
(v)
{y}_{\gamma}^{\phi} and {f}_{\gamma} are both locally Lipschitz on {R}^{n}.

(vi)
If f is coercive on K, then (VI) has a nonempty compact solution set.

(vii)
{f}_{\gamma}(x)\ge \beta \gamma {\parallel {y}_{\gamma}^{\phi}(x)x\parallel}^{2}, \mathrm{\forall}x\in K.
Proof Parts (i)(iv) are from [16], part (v) from [[18], Lemma 3.1] and part (vi) from [[11], Proposition 2.2.7].
It follows from (ii) and (iii) that we only need to prove (vii) for x\in K\setminus S. Since {y}_{\gamma}^{\phi}(x) is the minimizer of the function
the firstorder optimality condition implies that
Letting y=x, we get
i.e.,
It follows from (P3) that the mapping G is strongly convex on {R}^{n} with the modulus \beta >0, i.e., \mathrm{\forall}x,{y}_{1},{y}_{2}\in {R}^{n}
Letting {y}_{1}=x and {y}_{2}={y}_{\gamma}^{\phi}(x), by {f}_{\gamma}(x)=G(x,{y}_{\gamma}^{\phi}(x)), we obtain
Thus, one has {f}_{\gamma}(x)\ge \beta \gamma {\parallel {y}_{\gamma}^{\phi}(x)x\parallel}^{2}. □
Theorem 2.1 Let f be coercive on K and \gamma (\alpha 2\beta )<\mu. Assume that φ satisfies
(P5) \u3008{\mathrm{\nabla}}_{1}\phi (x,{y}_{\gamma}^{\phi}(x))+{\mathrm{\nabla}}_{2}\phi (x,{y}_{\gamma}^{\phi}(x)),{y}_{\gamma}^{\phi}(x)x\u3009\ge 0, \mathrm{\forall}x\in K.
Suppose further that the following condition holds:
where S is the solution set of (VI). Then \sqrt{{f}_{\gamma}} has a global error bound with the modulus
Proof It follows from (vi) of Proposition 2.2 that S is a nonempty compact set of K. If x\in S, then the assertion obviously holds. Let x\in K\setminus S. Then {f}_{\gamma}(x)>0. For brevity, we denote w:={y}_{\gamma}^{\phi}(x)x and d:=\frac{w}{\parallel w\parallel}. It follows from [[19], Theorem 2.5] that we only need to prove
It follows from (iv) of Proposition 2.2 that
By (P5) and (2), we have
It follows from (iii) of Proposition 2.1 that
Thus,
In light of (4) and (vii) of Proposition 2.2, we have
If \mu <\gamma (\alpha \beta ), then it follows from \gamma (\alpha 2\beta )<\mu that
If \mu \ge \gamma (\alpha \beta ), then
Hence, (3) follows from (5) and (6). The proof is complete. □
Now we use two examples to show that (2) cannot be dropped and that Theorem 2.1 is applicable, respectively.
Example 2.1 Consider K=R, \phi (x,y)=\frac{1}{2}{xy}^{2}, \gamma =\frac{1}{2} and f(x)={x}^{3}. Then we can easily get that \alpha =2\beta =1, {y}_{\gamma}^{\phi}(x)=x2f(x), {f}_{\gamma}(x)={x}^{6} and S=\{0\}. It is clear that f is coercive on K and \mu =0. Thus, (2) does not hold. Moreover, it is obvious that \sqrt{{f}_{\gamma}} does not have a global error bound.
Example 2.2 Consider K=[0,+\mathrm{\infty}), \phi (x,y)=\frac{1}{2}{xy}^{2}, \gamma =1 and f(x)=x. Then we can easily get that \alpha =2\beta =1, \mathrm{\nabla}f(x)=1, {y}_{\gamma}^{\phi}(x)=0, {f}_{\gamma}(x)=\frac{1}{2}{x}^{2} and S=\{0\}. It is clear that f is coercive on K and (2) holds. Thus, it follows from Theorem 2.1 that \sqrt{{f}_{\gamma}} has a global error bound.
By [[21], Proposition 2.3(ii)] Huang and Ng [[14], Theorem 2.1] have obtained the following conclusion. Now we give a slightly different proof by Theorem 2.1.
Corollary 2.1 Let f be strongly monotone on {R}^{n} with the modulus \lambda >0 and \gamma (\alpha 2\beta )<\lambda. Assume that φ satisfies (P5). Then \sqrt{{f}_{\gamma}} has a global error bound with the modulus
Proof Let x\in K\setminus S and w={y}_{\gamma}^{\phi}(x)x. Since f is continuously differentiable, then
where \frac{o(t)}{t}\to 0 as t\to 0. Since f is strongly monotone with the modulus λ, one has
which implies that
Thus, (2) holds. Moreover, the strong monotonicity of f implies the coerciveness of f (cf. [[22], Remark 2.1]). Thus, by Theorem 2.1, we get that \sqrt{{f}_{\gamma}} has a global error bound. □
3 A global error bound of (WVVI)
In this section, by the nonlinear scalarization method and by Theorem 2.1, we discuss a global error bound of (WVVI). The dual cone of C is defined by {C}^{\ast}:=\{\xi \in {R}^{m}:\u3008\xi ,z\u3009\ge 0,\mathrm{\forall}z\in C\}. For each \xi \in {R}^{m}, \parallel \xi \parallel :=sup\{\u3008\xi ,z\u3009:\parallel z\parallel \le 1\}, where \u3008\xi ,z\u3009 denotes the value of ξ at z. Let e\in intC and {B}_{e}^{\ast}:=\{\xi \in {C}^{\ast}:\u3008\xi ,e\u3009=1\}. It is well known that {B}_{e}^{\ast} is a compact convex base of {C}^{\ast}.
Lemma 3.1 [20]
where {S}_{\xi}:=\{x\in K:\u3008{\sum}_{i=1}^{m}{\xi}_{i}{F}_{i}({x}^{\ast}),x{x}^{\ast}\u3009\ge 0,\mathrm{\forall}y\in K\} and S is the solution set of (WVVI).
Recall the generalized regularized gap function for (WVVI) which is defined by
where {f}_{\gamma}(x,\xi )={max}_{y\in K}\{\u3008{\sum}_{i=1}^{m}{\xi}_{i}{F}_{i}(x),xy\u3009\gamma \phi (x,y)\}. When \varphi (x,y)=\frac{1}{2}{\parallel xy\parallel}^{2}, the generalized regularized gap function reduces to the regularized gap function which was defined in [4].
Theorem 3.1 Let \gamma (\alpha 2\beta )<{min}_{\xi \in {B}_{e}^{\ast}}{\mu}_{\xi}. Assume that φ satisfies (P5). For each \xi \in {B}_{e}^{\ast}, suppose that \xi \circ F is coercive on K, and that the following condition holds:
Then \sqrt{{\varphi}_{\gamma}} has a global error bound with the modulus
Proof It follows from (vi) of Proposition 2.2 that {S}_{\xi} is a nonempty compact set of K for each \xi \in {B}_{e}^{\ast}. If x\in S, then the assertion obviously holds. Let x\in K\setminus S. Then {\varphi}_{\gamma}(x)>0 and there exists {\xi}_{0}\in {B}_{e}^{\ast} such that {f}_{\gamma}(x,{\xi}_{0})={\varphi}_{\gamma}(x). It follows from Theorem 2.1 that
where {\tau}_{\xi}=max\{\frac{2\sqrt{\beta \gamma}}{{\mu}_{\xi}+2\gamma \beta \gamma \alpha},\frac{2\sqrt{\beta \gamma}}{\beta \gamma}\}. Thus, by Lemma 3.1, one has
Hence, \sqrt{{\varphi}_{\gamma}} has a global error bound with the modulus {max}_{\xi \in {B}_{e}^{\ast}}{\tau}_{\xi}. □
Remark 3.1 If {F}_{i} is strongly monotone with the modulus {\lambda}_{i} for i=1,2,\dots ,m and C={R}_{+}^{m}, it follows from [[21], Proposition 2.3] that
Moreover, the strong monotonicity of {F}_{i} implies the coerciveness of {F}_{i} (cf. [[22], Remark 2.1]) and that (VI) has a unique solution (cf. [[11], Theorem 2.3.3]). Thus, by Theorem 3.1, we get that \sqrt{{\varphi}_{\gamma}} has a global error bound. Hence, our results extend those of [[4], Theorem 2.9].
References
Giannessi F: Theorems of the alternative, quadratic programs and complementarity problems. In Variational Inequalities and Complementarity Problems. Edited by: Cottle RW, Giannessi F, Lions JL. Wiley, New York; 1980:151–186.
Chen GY, Goh CJ, Yang XQ: Vector network equilibrium problems and nonlinear scalarization methods. Math. Methods Oper. Res. 1999, 49: 239–253.
Li SJ, Teo KL, Yang XQ: A remark on a standard and linear vector network equilibrium problem with capacity constraints. Eur. J. Oper. Res. 2008, 184: 13–23. 10.1016/j.ejor.2005.11.059
Charitha C, Dutta J: Regularized gap functions and error bounds for vector variational inequalities. Pac. J. Optim. 2010, 6: 497–510.
Li J, Mastroeni G: Vector variational inequalities involving setvalued mappings via scalarization with applications to error bounds for gap functions. J. Optim. Theory Appl. 2010, 145: 355–372. 10.1007/s1095700996251
Sun XK, Chai Y: Gap functions and error bounds for generalized vector variational inequalities. Optim. Lett. 2014, 8: 1663–1673. 10.1007/s1159001306857
Xu YD, Li SJ: Gap functions and error bounds for weak vector variational inequalities. Optimization 2014, 63: 1339–1352. 10.1080/02331934.2012.721115
Konnov IV: A scalarization approach for vector variational inequalities with applications. J. Glob. Optim. 2005, 32: 517–527. 10.1007/s108980032688x
Auchmuty G: Variational principles for variational inequalities. Numer. Funct. Anal. Optim. 1989, 10: 863–874. 10.1080/01630568908816335
Auslender A: Optimization, Méthodes Numériques. Masson, Paris; 1976.
Facchinei F, Pang JS: FiniteDimensional Variational Inequalities and Complementarity Problems. Springer, Berlin; 2003.
Fukushima M: Equivalent differentiable optimization problems and descent methods for asymmetric variational inequality problems. Math. Program. 1992, 53: 99–110. 10.1007/BF01585696
Harker PT, Pang JS: Finitedimensional variational inequality and nonlinear complementarity problems: a survey of theory, algorithms, and applications. Math. Program. 1990, 48: 161–220. 10.1007/BF01582255
Huang LR, Ng KF: Equivalent optimization formulations and error bounds for variational inequality problems. J. Optim. Theory Appl. 2005, 125: 299–314. 10.1007/s1095700418397
Ng KF, Tan LL: Error bounds of regularized gap functions for nonsmooth variational inequality problems. Math. Program. 2007, 110: 405–429. 10.1007/s1010700600072
Wu JH, Florian M, Marcotte P: A general descent framework for the monotone variational inequality problem. Math. Program. 1993, 61: 281–300. 10.1007/BF01582152
Yamashita N, Taji K, Fukushima M: Unconstrained optimization reformulations of variational inequality problems. J. Optim. Theory Appl. 1997, 92: 439–456. 10.1023/A:1022660704427
Li G, Tang C, Wei Z: Error bound results for generalized Dgap functions of nonsmooth variational inequality problems. J. Comput. Appl. Math. 2010, 233: 2795–2806. 10.1016/j.cam.2009.11.025
Ng KF, Zheng XY: Error bounds for lower semicontinuous functions in normed spaces. SIAM J. Optim. 2001, 12: 1–17. 10.1137/S1052623499358884
Lee GM, Kim DS, Lee BS, Yen ND: Vector variational inequality as a tool for studying vector optimization problems. Nonlinear Anal. 1998, 34: 745–765. 10.1016/S0362546X(97)005786
Jiang HY, Qi LQ: Local uniqueness and convergence of iterative methods for nonsmooth variational inequalities. J. Math. Anal. Appl. 1995, 196: 314–331. 10.1006/jmaa.1995.1412
Li G, Ng KF: Error bounds of generalized Dgap functions for nonsmooth and nonmonotone variational inequality problems. SIAM J. Optim. 2009, 20: 667–690. 10.1137/070696283
Acknowledgements
This research was supported by the Natural Science Foundation of Shaanxi Province, China (Grant number: 2014JQ1023).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The author declares that he has no competing interests.
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.
About this article
Cite this article
Li, M. Error bounds of regularized gap functions for weak vector variational inequality problems. J Inequal Appl 2014, 331 (2014). https://doi.org/10.1186/1029242X2014331
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/1029242X2014331
Keywords
 error bound
 regularized gap function
 weak vector variational inequality