Available online at www.sciencedirect.

com

Applied Mathematics and Computation 200 (2008) 547556

www.elsevier.com/locate/amc

Existence of weakly ecient solutions in nonsmooth

vector optimization
Lucelina Batista Santos a,*, Marko Rojas-Medar b, Gabriel Ruiz-Garzon c,
Antonio Ruan-Lizana d
a

Departamento de Matematica, Universidade Federal do Parana, 81531-990 Curitiba, PR, Brazil

Facultad de Ciencias, Universidad del Bo-Bo, Campus Fernando May, Casilla 447, Chillan, Chile
c
Departamento de Estadstica e Investigacion Operativa, Universidad de Cadiz, 11403 Jerez de la Frontera, Spain
d
Departamento de Estadstica e Investigacion Operativa, Facultad de Matematica, Universidad de Sevilla, 41012 Sevilla, Spain
b

Abstract
In this paper we study the existence of weakly ecient solutions for some nonsmooth and nonconvex vector optimization
problems. We consider problems whose objective functions are dened between innite and nite-dimensional Banach
spaces. Our results are stated under hypotheses of generalized convexity and make use of variational-like inequalities.
2007 Elsevier Inc. All rights reserved.
Keywords: Multiobjective optimization; Nonsmooth analysis; Invexity; Variational-like inequalities

1. Introduction
The connection between variational inequalities and optimization problems is well known (e.g. [1,6,11,12])
and have been extensively investigated in the recent years by several authors. One of the main works in this
direction was done by Giannessi [8], where many existence results for optimization problems were obtained by
using variational inequalities.
For multiobjective optimization problems, Giannessi proved in [9] that there exists an equivalence between
ecient solutions of dierentiable convex optimization problems and the solutions of a variational inequality
of Minty type. He also established similar results for ecient solutions. On the other hand, using subdierentials, Lee showed in [13] that analogous results are true for nonsmooth convex problems dened between
nite-dimensional spaces.
For some nonconvex dierentiable vector problems dened between innite-dimensional Banach spaces,
Chen and Craven [4] proved the equivalence of weakly ecient solutions and the solutions of a certain

Corresponding author.
E-mail addresses: [email protected] (L.B. Santos), [email protected] (M. Rojas-Medar), [email protected] (G. Ruiz-Garzon),
[email protected] (A. Ruan-Lizana).
0096-3003/$ - see front matter 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.amc.2007.11.027

548

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

variational-like inequality. Using this characterization they proved an existence result for weakly ecient
solutions.
In this work we consider the following two problems.
1. An innite-dimensional problem:
Minimize

f x

subject to

x 2 K;

P1

where X and Y are two Banach spaces, f : X ! Y is a given function and K is a nonempty subset of X.
2. A nite-dimensional problem:
Minimize
subject to

f x : f1 x; . . . ; fp x
x 2 X;

P2

where fi : Rn ! R i 1; . . . ; p are given functions and X is a nonempty subset of Rn .

For both problems, by minimize we mean nd the weakly ecient solution of the problem Our objective is to solve problem (P1) without assuming hypotheses of dierentiability, which extends early results by
Chen and Craven [4], and to solve problem (P2) under assumptions of generalized convexity, which extends
early results by Lee [13].
This paper is organized as follows: In Section 2 we x the notation and recall some facts from nonsmooth
analysis. In Section 3 we consider the problem (P1) and establish our existence result. In Section 4 we consider
the problem (P2).
2. Preliminaries
Let X and Y be two real Banach spaces. We will
T denote by k  k the norm in Y. Let K be a nonempty subset of X and P  Y a pointed convex cone (i.e. P P f0g) such that int P 6 ;. Also, let f : X ! Y be a
given function. We consider the problem (P1) given in the previous section. The notion of optimality (or
equilibria) that we consider here is the weak eciency. We say that x0 2 K is a weakly ecient solution of
(P1) if
f x  f x0 62 int P ;

8x 2 K:

In particular, for the problem (P2), the denition of weakly ecient solution is done by taking P Rp in the
previous denition, that is, x0 2 X is a weakly ecient solution of (P2) if does not exist x 2 X such that
fi x < fi x0 ;

8 i 1; . . . ; p:

Now, we recall some notions and results from nonsmooth analysis. Let / be a locally Lipschitz function from
a Banach space X into R. The Clarke generalized directional derivative of /, at a point x 2 X , and in the direction d 2 X , denoted by /0 x; d, is given by:
/0 x; d lim sup
y!x
t#0

/y td  /y
t

and the Clarke generalized gradient of / at x is given by

o/x fx 2 X  : /0 x; d P hx ; di; 8d 2 X g;
where X  denotes the topological dual of X and h; i is the canonical bilinear form pairing X  and X.
The next proposition establish some properties of the generalized directional derivative and the generalized
gradient of Clarke.
Proposition 1. Let f : X ! R be a locally Lipschitz function with Lipschitz constant k. Then the following
assertions are true:

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

549

1. The function v 7! f 0 x; v is finite, sublinear and satisfies

jf 0 x; vj 6 kkvk:
2. For each x 2 X; of x is a w -compact and nonempty subset of X  . Furthermore, knk 6 k, for all n 2 of x,
where
knk sup hn; xi;

n 2 X :

x2X
kxk61

3. For each v 2 X , we have

f 0 x; v maxfhn; vi : n 2 of xg:
4. One has n 2 of x if and only if f 0 x; v P hn; vi for each v 2 X .
5. The function x; v 7! f 0 x; v is upper semicontinuous.
A function f : X ! R is called regular (or Clarke regular) at x 2 X, if
(i) For each v 2 X , there exists the usual directional derivative of f at x, in the direction v, denoted by
f 0 x; v.
(ii) For all v 2 X ; f 0 x; v f 0 x; v.
Furthermore, if f is regular at x 2 X, for each x 2 X, then we say that f is regular in X. The following result
holds (see [5]).
Proposition 2. These following assertions are true:
 Pn 
Pn
(a) If fi are regular at x 2 X, then o
i1 fi x
i1 ofi x. (The inclusion  is true, without the regularity
assumption).
(b) If f is convex and Lipschitz near x 2 X then f is regular at x.
(c) If f is continuously differentiable at x 2 X, then of x ff 0 xg, where f 0 is the usual derivative.
We note that if C is a nonempty subset of X, then the distance function d C : E ! R, dened by
d C x inffkx  ck : c 2 Cg, is not dierentiable, but it is (globally) Lipschitz, with Lipschitz constant equal
to 1.
Let C be a nonempty subset of X and x 2 C. We say that v 2 X is a tangent vector to C at x if d 0 x; v 0.
We denote by T C x the set of tangent vectors to C at x and T C x is called tangent cone of C at x. The normal
cone of C at x is dened by
N C x fn 2 X  : hn; vi 6 0; 8 v 2 T C xg:
It can be proved that, for each x 2 C; T C x is a convex cone, closed in X, and N C x is a convex cone,
w -closed in X  .
The following Proposition establishes a necessary condition for optimality.
Proposition 3. Let f : X ! R be a locally Lipschitz function and let x be a minimum of f in C  X. Then
0 2 of x N C x :

Note that the condition (1) is equivalent to

f 0 x ; v P 0;

8v 2 T C x :

For more details on nonsmooth analysis, we refer the reader to the book by Clarke [5].
Next, we recall the denition of strongly compactly Lipschitz function, which is very important for the
analysis of the innite-dimensional problem (P1).

550

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

Denition 4 [19]. A function h : X ! Y is said to be strongly compactly Lipschitz at x 2 X if there exists a

multifunction, that is, a point-to-set-map, R : X ! CompY , where CompY denotes the set of all compact
subsets of Y, and there exists a function r : X  X ! R satisfying
(i) lim x!x rx; d 0.
d!0

(ii) Exists a P 0 such that

t1 hx td  hx 2 Rd kdkrx; tBY
for all x 2 x aBY and t 2 0; a, where BY denotes the closed unit ball around the origin of Y.
(iii) R0 f0g and R is an upper semicontinuous multifunction.
Moreover, we say that h is a strongly compactly Lipschitz function if it is strongly compactly Lipschitz at all
x 2 X .
If Y has nite dimension, then h is strongly compactly Lipschitz at x if and only if it is locally Lipschitz near
x. If h is strongly compactly Lipschitz, then u  hx hu ; hxi is locally Lipschitz, for all u 2 Y  . This fact
is very important because it allows us to extend some results of the nonsmooth analysis to functions dened
between innite-dimensional spaces. For more details about strongly compactly Lipschitz functions, we refer
the reader to [19].
3. The innite-dimensional problem (P1)
We recall some denitions of generalized convexity for functions dened between Banach spaces. Given a
cone P  Y , the dual cone of P is dened by
P  fn 2 Y  : hn; xi P 0; 8x 2 P g:
It can be proved that, if P is a convex cone, with int P 6 ;, then hn; vi > 0, for all v 2 int P (see [10]). We will
need the following concepts of generalized convexity:
Denition 5 [15]. We say that a locally Lipschitz function h : K  X ! R is invex on K, with respect to g, if
for any x; y 2 K, there exists a vector gx; y 2 T K y such that
hx  hy  h0 y; gx; y P 0:
Denition 6 [2]. Let X and Y be two Banach spaces and suppose that P  Y is a convex cone. We say that
f : K  X ! Y is P-invex on K, with respect to g, if x  f : K ! R is invex on K with respect to g, for each
x 2 P  .
Certainly, if f : K  X ! R is dierentiable and convex on K, then f is invex and gx; y x  y.
Note that when K is open, or more generally, if y 2 int K, we have T K y X and, in this case, the above
denition coincides with that given by Weir and Jeyakumar [20]. Moreover, when Y is innite-dimensional
and f is dierentiable, the Denition 6 is the same to that given by Santos et al. in [17]. Recall that the set
K is called invex with respect to g if the vector y agx; y is in K for each x; y 2 K and a 2 0; 1.
In the sequence, we suppose that X and Y are two Banach spaces, K is a nonempty subset of X and P  Y is
a convex cone such that int P 6 ;.
We consider the following vector variational-like inequality.
VI Find x0 2 K such that, for each x 2 K, there exists x 2 P  n f0g such that
0

x  f x0 ; gx; x0 P 0:
Under suitable hypotheses, each solution of VI is a weakly ecient solution of (P1). In eect we have.
Theorem 7. Let K be an invex set with respect to g and f : K  X ! Y be a P-invex function with respect to g.
Then each solution of VI is a weakly efficient solution of (P1).

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

551

Proof. Suppose that x0 is solution of VI which is not a weakly ecient solution of (P1). Then there exists
x 2 K such that
f x  f x0 2 int P :

3


On the other hand, there exists x 2 P n f0g such that x  f x0 ; gx; x0 P 0. Since f is P-invex, it follows
from (3) that
0

x  f x  x  f x0 P x  f x0 ; gx; x0 P 0:

On the other hand, it follows from (3) that

x f x  f x0 x  f x  x  f x0 < 0;
which contradicts (4). h
The following Lemma will be useful in the proof of our main result of this section.
Lemma 8 (KKM-Fan Theorem [7]). Let X be a topological vector space, E  X be a nonempty set and
F : E  X a multifunction such that for each x 2 E, the set F x is closed and nonempty. Moreover, suppose that
there exists some x 2 E such that F x is compact. If for each finite subset fx1 ; . . . ; xn g of E one has
[n
cofx1 ; . . . ; xn g  i1 F xi ;
where cofx1 ; . . . ; xn g is the convex hull of fx1 ; . . . ; xn g, then
\
F x 6 ;:
x2E
Now we establish our existence result for (P1).
Theorem 9. Let X be a reflexive Banach space and K a closed, convex and bounded subset of X. Let f : K ! Y be
a strongly compactly Lipschitz function, P-invex respect to g. Suppose that for each x 2 K and each
x 2 P  n f0g, the sets
0

Ux; x : fy 2 K : x  f x; gy; x < 0g

are convex. Furthermore, assume that g is continuous and gx; x 0 for each x 2 K. Then problem (P1) has a
weakly efficient solution.
Proof. For y 2 K and x 2 P  n f0g we dene
0

F y; x : fx 2 K : x  f x; gy; x P 0g:
By Theorem 7, it is sucient to prove that the variational-like inequality VI has a solution, that is,
\ [
F y; x 6 ;:
y2K
x 2P  nf0g
To this end, we will prove that
\ \
F y; x 6 ;:
y2K
x 2P  nf0g
Consider the multifunction G : K  X , dened by
\
Gy x 2P  nf0g F y; x :
T
We will prove that y2K Gy 6 ; by using of Lemma 8. Suppose that the space X is equipped with the weak
topology. Note that Gy is nonempty for all y 2 K. In fact, gy; y 0 and thus y 2 F y; x for all
x 2 P  n f0g. Hence y 2 Gy. Furthermore, Gy is closed. In fact, let y 2 K and xk  Gy be a sequence
such that xk ! x. Then x  f xk ; gy; xk P 0 for all x 2 P  n f0g. Fixed x 2 P  n f0g, we have
0

lim sup x  f xk ; gy; xk P 0:

k!1

552

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556
0

But the function x  f ;  is upper semicontinuous (see Proposition 1). Hence, it follows from (5) that
0

x  f x; gy; x P 0;
that is, x 2 Gy and thus Gy is closed. But Gy is convex and hence, weakly closed.
On the other hand, K is weakly compact, because it is a convex, closed and bounded subset of X and X is a
reexive space. So Gy is weakly compact.
S
Now, take x1 ; . . . ; xn 2 K. We should prove that cofx1 ; . . . ; xn g  ni1 Gxi .
Pn
Suppose
thatS this is false. Then there exist ai P 0, i 1; . . . ; n, such that
i1 ai 1 and
Pn
x : i1 ai xi 62 ni1 Gxi . Hence, for each i, there exists xi 2 P  n f0g such that x 62 F xi ; xi , that is,
xi  f 0 x; gxi ; x < 0.
Next we construct a x 2 P  n f0g such that xi 2 Xx; x for all i, which will contradict the convexity of
Xx; x since gx; x 0.
Dene for i; j 2 f1; . . . ; ng; m 2 N,
( 
xj
if xj  f 0 x; gxi ; x < 0;
m
xi;j
0
1 
x if xj  f x; gxi ; x P 0;
m j
n
X
m
m
xi;j :
xi
j1
m
xi

Clearly
are linear, and for each m; xi
tinuous. In fact, dene
0

J 1 fj : xj  f x; gxi ; x < 0g;

J 2 fj :

xj

M max

j1;...;n

 f x; gxi ; x P 0g;

are not all zeros, for i 1; . . . ; n. Furthermore, xi

are con-

a1 #J 1 ;
a2 #J 2 ;

kxj k;

where #J is the number of elements of J.

Then, for all u 2 Y , we have




X
1 X   X 
1 X 
1

m

jxi uj 
x u
x u 6
kx kkuk
kx kkuk a1 a2 Mkuk
 j2J j
m j2J 2 j  j2J 1 j
m j2J 2 j
m
1
m

that is, xi 2 Y  . Furthermore, xi u P 0, for each u 2 P (because xi

of functionals in P  ).
On the other hand,
"
#0
X
1 X 
m
0

xi  f xi ; gxi ; x
xj  f
x  f x; gxi ; x
m j2J 2 j
j2J 1
6

is a nonnegative linear combination

X
1 X 
xj  f 0 x; gxi ; x
x  f 0 x; gxi ; x
m j2J 2 j
j2J 1

and hence, by taking m suciently big, say, m P mi, we have

X
m
xj  f 0 x; gxi ; x < 0; 8m P mi:
xi  f 0 xi ; gxi ; x 6
j2J 1

Pn M
Take M max16i6n mi and dene xM i1 xi . Then we have xM is in P  n f0g and
X M
xi  f 0 x; gxi ; x < 0
xM  f 0 x; gxi ; x 6
M
and so xi 2 Ux; xM for all i 1; . . . ; n. But Ux; xM
T is convex and thus x 2 Ux; x . This contradicts
gx; x 0. Therefore, it follows from Lemma 8, that y2K Gy 6 ;. The proof is complete. h

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

553

Remark 10. We notice that if K is compact and f is continuous, then (P1) has a weakly ecient solution (see
Sawaragi et al. [18]). Thus, if we apply Theorem 9 to the problem (P2), we do not obtain any new result,
because the hypotheses considered there imply that K is compact and f is continuous (see Brezis [3]).
4. The nite-dimensional problem (P2)
In this Section, we will consider a variational-like inequality of (weak) Minty type and we establish a characterization of weakly eciency for (P2), in terms of the solutions of that inequality. Using this result and a
xed point theorem for multifunctions, we establish our existence result for (P2).
Our variational-like inequality is:
WMVLI Find y 2 X such that, for each x 2 X and each ni 2 ofi x; i 1; . . . ; p,
nT1 gx; y; . . . ; nTp gx; y 62 int Rp :
In [16], Santos et al. proved that, under certain hypotheses, the solutions of WMVLI are weakly ecient
solutions of (P2).
Proposition 11 [16]. Let X be a nonempty set of Rn , invex with respect to g, and fi : Rn ! R; i 1; . . . ; p are
invex functions, locally Lipschitz with respect to g. Suppose that g is skew (that is, gx; y gy; x; 8x; y 2 X ).
Then y 2 X is a weakly efficient solution of (P2) if and only if y 2 X is a solution of WMVLI.
Now, we recall a xed-point theorem of FanBrowder, whose proof can be found in Park [14].
Lemma 12 [14]. Let X be a nonempty convex subset of a Hausdorff topological vector space E and let K be a
nonempty compact subset of X. Suppose that A; B : X  X are multifunctions satisfying the following conditions:
1.
2.
3.
4.
5.

Ax  Bx for all x 2 X ,
Bx is a convex set for all x 2 X ,
Ax 6 ; for all x 2 K,
A1 y fx 2 X ; y 2 Axg is an open set for each y 2 X ,
for eachTfinite subset N of X, there exists a compact, convex and nonempty subset LN of X such that LN N
and Ax LN 6 ; for all x 2 LN n K.

Then there is a x 2 Bx (that is, x is a fixed point of B).

Next, we will use Proposition 11 and Lemma 12 to establish our result about the existence of weakly ecient solution for nonsmooth invex vectorial problem, under a weaker compactness hypothesis on the feasible
set X.
Theorem 13. Let X be a nonempty and invex subset of Rn with respect to g, and fi : Rn ! R; i 1; . . . ; p, invex
and locally Lipschtz functions with respect to the same g. Assume that g is skew and such that g; y is convex and
continuous, for each y 2 X . Moreover, suppose that there exists a compact, convex and nonempty set LN  X ,
such that LN N and for all x 2 LN n K, there is z 2 LN such that there exist ni 2 ofi z; i 1; . . . ; p satisfying
nT1 gz; x; . . . ; nTp gz; x 2 int Rp :
Then problem (P2) has a weakly efficient solution.
Proof. For sake of readability, we give a proof using a concise notation. We denote by of x the set
of1 x  . . . ofp x; x 2 X . Let s s1 ; . . . ; sp , where si 2 Rn ; i 1; . . . ; p. We denote by sT gx; y the vector
sT1 gx; y; . . . ; sTp gx; y 2 Rp :
Let A; B : X  X be multifunctions dened by:
Ax : fz 2 X : 9t 2 of z; tT gz; x 2 int Rp g;
Bx : fz 2 X : 8t 2 of x; tT gz; x 2 int Rp g:

554

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

We will prove (by using of Lemma 12) that exists y 2 K such that Ay ;, that is, y is solution of (WMVLI).
By Proposition 11, this is sucient to prove our result.
First we prove that the multifunctions A and B satisfy the conditions (1), (2), (4) and (5) of Lemma 12 and
that B does not have a xed point. So, Lemma 12 will imply the existence of y 2 K such that Ay ;.
We show that the condition (1) of Lemma 12 holds. Take x 2 X and z 2 Ax. Then, exists t n1 ; . . . ; np 2
of z such that
nT1 gz; x; . . . ; nTp z; x 2 int Rp :

Let s b
n1; . . . ; b
n p 2 of x. By using of the invexity of functions fi and the skewness of g, we have that, for
each i 1; . . . ; p,
b
n i gz; x 6 fi0 x; gz; x 6 fi z  fi x fi x  fi z 6 nTi gx; z nTi gz; x:

Follows from (6) and (7),

b
n T1 gz; x; . . . ; b
n Tp z; x 2 int Rp ;
and so z 2 Bx.
Now, we will see that the second condition of Lemma 12 holds: Let x 2 X ; z1 ; z2 2 Bx and k 2 0; 1. Then,
for each s n1 ; . . . ; np 2 of x, we have
nT1 gz1 ; x; . . . ; nTp gz1 ; x; nT1 gz2 ; x; . . . ; nTp gz2 ; x 2 int Rp :

For each j 1; . . . ; p, we consider nj nj1 ; . . . ; njn , where nk

j 2 R; gx; y g1 x; y; . . . ; gn x; y;
gk x; y 2 R. Then, from the convexity of gk and (8), we obtain
nTj gkz1 1  kz2 ; x

n
X

nk
j gk kz1 1  kz2 ; x 6

k1

knTj gz1 ; x

n
X

nk
j kgk z1 ; x 1  kgk z2 ; x

k1

1 

knTj gz2 ; x

< 0;

for each j 1; . . . ; p. Hence kz1 1  kz2 2 Bx.

The fourth condition is proved as follows. For all z 2 X , we will show that the set A1 zc is closed. To do
this, consider a sequence xn  A1 zc such that xn converges to x. Then xn 62 A1 z; 8n 2 N. Let
t n1 ; . . . ; np 2 of z be such that
nT1 gz; xn ; . . . ; nTp gz; xn 62 int Rp :

9
c

Because g; z is continuous and skew, we have that gz;  is also continuous and skew and, since int Rp is
a closed set, by taking n ! 1 in (9) we obtain
nT1 gz; x; . . . ; nTp gz; x 62 int Rp ;
c

and thus x 2 A1 z .

From the hypotheses, condition (5) of Lemma 12 also holds.
However, B does not have a xed point, because otherwise it would exists some x 2 X such that s 2 of x
and sT gx; x 0 2 int Rp , which is an absurd. Thus, from Lemma 12, it follows that exists y 2 K such that
Ay ;. h
We have the following consequence of Theorem 13.
Corollary 14. Let X be a nonempty subset of Rn , invex with respect to g, and suppose that g is skew and such that
g; y is convex and continuous. Moreover, assume that
K fx 2 X : f10 z0 ; gz0 ; x; . . . ; fp0 z0 ; gz0 ; x 62 int Rp g;
is compact for some z0 2 X . Then (P2) has a weakly efficient solution.

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

Proof. Let N be a nonempty and nite subset of X. Dene LN coN

convex hull of A). Then, for each x 2 LN n K we have
f10 z0 ; gz0 ; x; . . . ; fp0 z0 ; gz0 ; x 2 int Rp :

555

K (where co denotes the closed

10

Take z z0 2 K  LN and ni 2 ofi z. We have

nTi gz0 ; x 6 fi0 z0 ; gz0 ; x;

i 1; . . . ; p;

11

and from (10) and (11), we obtain

nT1 gz; x; . . . ; nTp gz; x 2 int Rp :
Thus, by Theorem 13, (P2) has a weakly ecient solution.

5. Conclusions
In this paper, we obtain an existence theorem for weakly ecient solutions of vector optimization problems
dened between innite dimensional Banach spaces whose objective function is invex and strongly compactly
Lipschitz. We characterize the solutions of this problem in terms of the solutions of a variational-like inequality and, by applying this characterization and the KKM-Fan Theorem, we establish our result. The approach
that we used here is similar to the one employed by Chen and Craven [4], where the authors considered a differentiable problem.
Also, we consider the vector problem dened between nite-dimensional spaces, and for this problem we
also obtain a result on the existence of weakly ecient solution. Our approach is analogous to the one used
by Lee [13], where it was considered the non-dierentiable convex case. We use a characterization of the
weakly eciency in terms of the solutions of the (weak) Minty type inequalities and a xed point theorem
for multifunctions to prove our result.
The results that we present in this paper generalize those obtained by Chen and Craven [4] and Lee [13].
References
[1] C. Baiocchi, A. Capelo, Variational and Quasivariational Inequalities: Applications to the Free Boundary Problems, John Wiley and
Sons Inc., New York, 1984.
[2] A.J.V. Brandao, M.A. Rojas-Medar, G.N. Silva, Optimality conditions for Pareto nonsmooth nonconvex programming in Banach
spaces, J. Optim. Theory Appl. 103 (1) (1999) 6573.
[3] H. Brezis, Analyse fonctionnelle: Theorie et Applications, Masson, Paris, 1983.
[4] G.Y. Chen, B.D. Craven, Existence and continuity of solutions for vector optimization, J. Optim. Theory Appl. 81 (1994) 459468.
[5] F.H. Clarke, Optimization and Nonsmooth Analysis, Wiley-Interscience, 1983.
[6] G. Duvault, J.L. Lions, Les inequations en Mecanique et en Physique, Dunod, Paris, 1972.
[7] K. Fan, A generalization of Tichonos xed point theorem, Mathematische Annalen 142 (1961) 305310.
[8] F. Giannessi, Theorems of the alternative, quadratic problems, and complementarity problems, in: R.W. Cottle, F. Giannessi, J.L.
Lions (Eds.), Variational Inequalities and Complementarity Problems (Proceedings of the International School, Erice, 1978), John
Wiley, Chichester, England, 1980, pp. 151186.
[9] F. Giannessi, On Minty variational principle, in: New Trends in Mathematical Programming, Kluwer, 1997, pp. 9399.
[10] I.V. Girsanov, Lectures on Mathematical Theory of Extremum Problems, Lecture Notes in Economics and Mathematical Systems,
vol. 67, Springer-Verlag, 1972.
[11] R. Glowinsky, J.L. Lions, R. Tremolieres, Numerical Analysis of Variational Inequalities, North-Holland, 1976.
[12] P.T. Harker, J.-S. Pang, Finite-dimensional variational inequality and nonlinear complementarity problems: a survey of theory,
algorithms and applications, Math. Program. 48 (1990) 161220.
[13] G.M. Lee, On relations between vector variational inequality and vector optimization problem, in: X. Yang et al. (Eds.), Progress in
Optimization, Kluwer, 2000.
[14] S. Park, Some coincidence theorems on acyclic multifunctions and applications to KKM Theory, in: Fixed Point Theory and
Applications, World Scientic Publishing, River Edge, NJ, 1992, pp. 248277.
[15] T.D. Phuong, P.H. Sach, N.D. Yen, Strict lower semicontinuity of the level sets and invexity of a locally Lipschitz function, J. Optim.
Theory Appl. 87 (1995) 579594.
[16] L.B. Santos, M. Rojas-Medar, A. Ruan-Lizana, Some relations between variational-like inequalities and ecient solutions of certain
nonsmooth optimization problems, Int. J. Math. Sci. 2006 (2006), 16 (Article ID 26808), doi:10.1155/IJMMS/2006/26808.

556

L.B. Santos et al. / Applied Mathematics and Computation 200 (2008) 547556

[17] L.B. Santos, R. Osuna-Gomez, M.A. Rojas-Medar, A. Ruan-Lizana, Preinvex functions and weak ecient solutions for some
vectorial optimization problems in Banach spaces, Comput. Math. Appl. 48 (2004) 885895.
[18] Y. Sawaragi, H. Nakayama, T. Tanino, Theory of Multiobjective Optimization, Academic Press, New York, 1985.
[19] L. Thibault, On generalized dierentials and subdierentials of Lipschitz vector valued functions, Nonlinear Anal. 6 (1980)
10371053.
[20] T. Weir, V. Jeyakumar, A class of nonconvex functions and mathematical programming, Bull. Austr. Math. Soc. 38 (1988) 177189.