• 検索結果がありません。

Well-Posedness for a Class of

N/A
N/A
Protected

Academic year: 2022

シェア "Well-Posedness for a Class of"

Copied!
22
0
0

読み込み中.... (全文を見る)

全文

(1)

Journal of Applied Mathematics Volume 2012, Article ID 712306,21pages doi:10.1155/2012/712306

Research Article

Well-Posedness for a Class of

Strongly Mixed Variational-Hemivariational Inequalities with Perturbations

Lu-Chuan Ceng,

1, 2

Ngai-Ching Wong,

3

and Jen-Chih Yao

4

1Department of Mathematics, Shanghai Normal University, Shanghai 200234, China

2Scientific Computing Key Laboratory, Shanghai Universities, Shanghai 200234, China

3Department of Applied Mathematics, National Sun Yat-Sen University, Kaohsiung 80424, Taiwan

4Center for General Education, Kaohsiung Medical University, Kaohsiung 80708, Taiwan

Correspondence should be addressed to Ngai-Ching Wong,wong@math.nsysu.edu.tw Received 1 August 2011; Accepted 19 November 2011

Academic Editor: Ya Ping Fang

Copyrightq2012 Lu-Chuan Ceng et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The concept of well-posedness for a minimization problem is extended to develop the concept of well-posedness for a class of strongly mixed variational-hemivariational inequalities with perturbations which includes as a special case the class of variational-hemivariational inequalities with perturbations. We establish some metric characterizations for the well-posed strongly mixed variational-hemivariational inequality and give some conditions under which the strongly mixed variational-hemivariational inequality is strongly well-posed in the generalized sense. On the other hand, it is also proven that under some mild conditions there holds the equivalence between the well posedness for a strongly mixed variational-hemivariational inequality and the well-posedness for the corresponding inclusion problem.

1. Introduction

It is well known that the classical notion of well-posedness for the minimization problem MPis due to Tykhonov1, which has been known as the Tykhonov well-posedness. Let V be a Banach space andf :VR∪ {∞}be a real-valued functional onV. The problem MP, that is, minx∈Vfx, is said to be well posed if there exists a unique minimizer and every minimizing sequence converges to the unique minimizer. Furthermore, the notion of generalized Tykhonov well-posedness is also introduced for the problemMP, which means the existence of minimizers and the convergence of some subsequence of every minimizing sequence toward a minimizer. Clearly, the concept of well-posedness is inspired by numerical methods producing optimizing sequences for optimization problems and plays a crucial

(2)

role in the optimization theory. Therefore, various concepts of well-posedness have been introduced and studied for optimization problems. For more details, we refer to2–8and the references therein.

On the other hand, the concept of well-posedness has been extended to other related problems, such as variational inequalities 5, 9–14, saddle-point problem 15, inclusion problems10,11, and fixed-point problems10,11. An initial notion of well-posedness for variational inequalities is due to Lucchetti and Patrone5. They introduced the notion of well-posedness for variational inequalities and proved some related results by means of Eke- land’s variational principle. Since then, many authors have been devoted to generating the concept of well-posedness from the minimization problem to various variational inequalities.

In2, Crespi et al. gave the notions of well-posedness for a vector optimization problem and a vector variational inequality of the differential type, explored their basic properties, and investigated their links. Lignola13 introduced two concepts of well-posedness and L-well-posedness for quasivariational inequalities and investigated some equivalent char- acterizations of these two concepts. Recently, Fang et al. 11generalized the concepts of well-posedness and α-well-posedness to a generalized mixed variational inequality which includes as a special case the classical variational inequality and discussed its links with the well-posedness of corresponding inclusion problem and the well-posedness of corresponding fixed-point problem. They also derived some conditions under which the mixed variational inequality is well posed. For further results on the well-posedness for variational inequalities and equilibrium problems, we refer to5,8,11,13,16–18and the references therein.

In 1983, in order to formulate variational principles involving energy functions with no convexity and no smoothness, Panagiotopoulos19first introduced the hemivariational inequality which is an important and useful generalization of variational inequality and investigated it by using the mathematical notion of the generalized gradient of Clarke for nonconvex and nondifferentiable functions20. The hemivariational inequalities have been proved very efficient to describe a variety of mechanical problems, for instance, uni- lateral contact problems in nonlinear elasticity, problems describing the adhesive and fric- tional effects, and nonconvex semipermeability problems see, for instance, 19, 21, 22.

Therefore, in recent years all kinds of hemivariational inequalities have been studied by many authors14,21,23–29, and the study of hemivariational inequalities has emerged as a new and interesting branch of applied mathematics. However, there are very few researchers extending the well-posedness to hemivariational inequalities. In 1995, Goeleven and Mentagui14first introduced the notion of well-posedness for hemivariational inequalities and established some basic results concerning the well-posed hemivariational inequality.

Very recently, Xiao and Huang30generalized the well-posedness of minimization problems to a class of variational-hemivariational inequalities with perturbations, which in- cludes as special cases the classical hemivariational inequalities and variational inequalities.

Under appropriate conditions, they derived some metric characterizations for the well- posed variational-hemivariational inequality and presented some conditions under which the variational-hemivariational inequality is strongly well posed in the generalized sense.

Meantime, they also proved that the well-posedness for a variational-hemivariational ine- quality is equivalent to the well-posedness for the corresponding inclusion problem.

In this paper, we extend the notion of well-posedness for minimization problems to a class of strongly mixed variational-hemivariational inequalities with perturbations, which includes as a special case the class of variational-hemivariational inequalities with perturba- tions in30. Under very mild conditions, we establish some metric characterizations for the well-posed strongly mixed variational-hemivariational inequality and give some conditions

(3)

under which the strongly mixed variational-hemivariational inequality is strongly well- posed in the generalized sense. On the other hand, it is also proven that the well-posed- ness for a strongly mixed variational-hemivariational inequality is equivalent to the well-posed- ness for the corresponding inclusion problem.

2. Preliminaries

Throughout this paper, unless stated otherwise, we always suppose thatV is a real reflexive Banach space, where its dual space is denoted by V and the generalized duality pairing betweenV and V is denoted by·,·. We denote the norms of Banach spacesV and V by · V and · V, respectively. In what follows, letN : V×VV,A, T : VV and g : VV be four mappings, G : VR∪ {∞}be a proper, convex, and lower semicontinuous functional, andfVbe some given element. Denote by domGthe efficient domain of functional, that is,

domG: {u∈V :Gu<∞}. 2.1

Consider the following strongly mixed variational-hemivariational inequality: finduV such that

SMVHVI : N

Agu, Tu

f, vgu J

u, vgu

GvG gu

≥0, ∀v∈V, 2.2 whereJu, vdenotes the generalized directional derivative in the sense of Clarke of a locally Lipschitz functionalJ:VR atuin the directionvsee20given by

Ju, v: lim sup

wu λ↓0

JwλvJw

λ . 2.3

In particular, ifNu, v uv, for allu, vVandg Ithe identity mapping ofV, then the problem2.2reduces to the following variational-hemivariational inequality of findinguVsuch that

VHVI : AuTu, vuJu, v−u GvGu

f, vu

, ∀v∈V, 2.4

whereTis perturbation, which was first introduced and studied by Xiao and Huang30.

LetΩ be an open bounded subset of R3 which is occupied by a linear elastic body and Γ the boundary of the Ω which is assumed to be appropriately regular C0,1, i.e., a Lipschitzian boundary, is sufficient. We denote byS {Si}the stress vector onΓ, which can be decomposed into a normal componentSNand a tangential componentSTwith respect to Γ, that is,

SN σijnjni, STi σijnjσijninj

ni, 2.5

where σij} is an appropriately defined stress tensor and n {ni} is the outward unit normal vector on Γ. Analogously, uN and uT denote the normal and the tangential

(4)

components of the displacement vector uwith respect to Γ. As pointed out in30, the re- action-displacement law presents in compression ideal locking effectthe infinite branchEF, that is, alwaysuNa, whereasuN> ais impossible. Specifically,

ifuN < athen−SNβu N, ifuN athen− ∞<−SNβa, ifuN > athen SN ∅,

2.6

whereβis a multivalued function defined as follows. Suppose thatβ :RR is a function such thatβLlocR, that is, a function essentially bounded on any bounded interval ofR.

For anyρ >0 andξR, we defineβρξ ess inf1−ξ|≤ρβξ1andβρξ ess sup1−ξ|≤ρβξ1. By the monotonicity of the functionsβρandβρwith respect toρ, we infer that the limits as ρ → 0exist, that is,

βξ lim

ρ→0βρξ, βξ lim

ρ→0βρξ. 2.7

Then,

βξ

βξ, βξ

. 2.8

Furthermore, a locally Lipschitz functionjNcan be determined up to an additive constant by

jNξ ξ

0

βξ11 2.9

such that∂jNξ βξ for eachξR when the limitsβξ±exist, where∂jNis the Clarke’s generalized gradient of locally Lipschitz functionjNwhich will be specified in what follows.

Now, letK {uN |uNa},NKthe normal cone toKatuN, andIKthe indicator of the setK. Then2.6can be written as

−SNβu N NKuN ∂jNuN ∂IKuN, 2.10

where ∂IK is the subgradient of the convex functional IK in the sense of convex analysis, which will also be specified in what follows. By the definitions of the Clarke’s generalized gradient of locally Lipschitz function and the subgradient of the convex functional, 2.10 gives rise to the following variational-hemivariational inequality

uNR : SN, vuNjNuN, vuN IKv−IKuN≥0, ∀v∈R, 2.11 which is a special case of the variational-hemivariational inequality VHVI. Beyond question, the problem2.11is a special case of the strongly mixed variational-hemivariational inequal- ity SMVHVI as well. More special cases of the SMVHVI are stated as follows.

(5)

iIfG δKandJu

Ωjx, udΩ, whereδKdenotes the indicator functional of a nonempty, convex subsetKof a function spaceV defined onΩandj:Ω×R → R is a locally Lipschitz continuous function, then the SMVHVI reduces to the following strongly mixed variational-hemivariational inequality:

SMVHVI :

N

Agu, Tu

f, vgu J

u, vgu

≥0, ∀v∈K. 2.12

Remark that the SMVHVI 2.12with NAgu, Tu Agu Tu and g I is equivalent to the VHVI which was considered by Goeleven and Mentagui in14.

iiIfG 0, then the SMVHVI2.2withNAgu, Tu Agu Tureduces to the strongly mixed hemivariational inequality of findinguV such that

SMHVI :

Agu Tuf, vgu J

u, vgu

≥0, ∀v∈V. 2.13

Remark that the SMHVI2.13withT 0 andg Iis equivalent to the hemivar- iational inequalityHVIstudied intensively by many authorssee, e.g.,21,22.

iiiIfJ 0, then the SMVHVI2.2withNAgu, Tu Agu Tu reduces to the strongly mixed variational inequality of findinguVsuch that

SMVI :

Agu Tuf, vgu

GvG gu

≥0, ∀v∈V. 2.14

Remark that the SMVI2.14withT 0 andg I is equivalent to the mixed var- iational inequalitysee, e.g.,11,31and the references therein.

ivIfT 0, J 0, g I andG δK, then the SMVHVI2.2withNAgu, Tu Agu Tureduces to the classical variational inequality:

VI :

Auf, vu

≥0, ∀v∈K. 2.15

vIfN 0, J 0, g I, andf 0, then the SMVHVI2.2 reduces to the global minimization problem:

MP : min

u∈V Gu. 2.16

Let ∂Gu : V → 2V\ {∅}and ∂Ju : V → 2V \ {∅}denote the subgradient of convex functionalGin the sense of convex analysissee32and the Clarke’s generalized gradient of locally Lipschitz functionalJsee20, respectively. That is,

∂Gu {uV:GvGuu, vu,∀v∈V},

∂Ju {ω∈V:Ju, v≥ ω, v,∀v∈V}. 2.17

(6)

Remark 2.1see33. The Clarke’s generalized gradient of a locally Lipschitz functionalJ : VR at a pointuis given by

∂Ju ∂Ju,·0. 2.18

About the subgradient in the sense of convex analysis, the Clarke’s generalized di- rectional derivative, and the Clarke’s generalized gradient, we have the following basic propertiessee, e.g.,20,30,32,33.

Proposition 2.2. LetVbe a Banach space andG:VR∪{∞}be a convex and proper functional.

Then we have the following properties of∂G:

i∂Guis convex and weak-closed;

iiifGis continuous atu∈domG, then∂Guis nonempty, convex, bounded, and weak- compact;

iiiifGis Gateaux differentiable atu∈domG, then∂Gu {DGu}, whereDGuis the Gateaux derivative ofGatu.

Proposition 2.3. LetV be a Banach space andG1, G2:VR∪ {∞}be two convex functionals.

If there is a pointu0 ∈ domG1∩domG2 at whichG1 is continuous, then the following equation holds:

∂G1G2u ∂G1u ∂G2u, ∀u∈V. 2.19

Proposition 2.4. LetV be a Banach space,u, vV, andJa locally Lipschitz functional defined on V. Then

ithe functionvJu, vis finite, positively homogeneous, subadditive, and then convex onV;

iiJu, v is upper semicontinuous as a function of u, v, as a function of v alone, is Lipschitz continuous onV;

iiiJu−v −Ju, v;

iv∂Juis a nonempty, convex, bounded, weak-compact subset ofV; vfor everyvV, one has

Ju, v max

ξ, v:ξ∂Ju

. 2.20

Now we recall some important definitions and useful results.

Definition 2.5see34. LetV be a real Banach space with its dualVandT be an operator fromVto its dual spaceV.Tis said to be monotone if

Tu−Tv, uv ≥0, ∀u, v∈V. 2.21

(7)

Definition 2.6see34. A mapping T : VV is said to be hemicontinuous if for any u, vV, the functiont→ Tutvu, vufrom0,1intoR is continuous at 0.

It is clear that the continuity implies the hemicontinuity, but the converse is not true in general.

Theorem 2.7see35. LetCV be nonempty, closed, and convex,CV nonempty, closed, convex, and bounded,ϕ:VR∪ {∞}proper, convex, and lower semicontinuous, andyCbe arbitrary. Assume that, for eachxC, there existsxx∈Csuch that

xx, x−y

ϕ y

ϕx. 2.22

Then, there existsyCsuch that y, xy

ϕ y

ϕx, ∀x∈C. 2.23

Definition 2.8see36. LetSbe a nonempty subset ofV. The measure, sayμ, of noncom- pactness for the setSis defined by

μS: inf

>0 :Sn

i 1

Si, diamSi< , i 1,2, . . . , n

, 2.24

where diamSimeans the diameter of setSi.

Definition 2.9 see36. LetA, B be nonempty subsets of V. The Hausdorffmetric H·,· betweenAandBis defined by

HA, B: max{eA, B, eB, A}, 2.25

whereeA, B: supa∈Ada, Bwithda, B: infb∈B a−b V.

Let{An}be a sequence of nonempty subsets ofV. We say thatAn converges toAin the sense of Hausdorffmetric ifHAn, A → 0. It is easy to see thateAn, A → 0 if and only ifdan, A → 0 for all sectionanAn. For more details on this topic, we refer the reader to 36.

3. Well-Posedness of the SMVHVI with Metric Characterizations

In this section, we generalize the concept of well-posedness to the strongly mixed varia- tional-hemivariational inequality SMVHVI with perturbations, establish its metric charac- terizations, and derive some conditions under which the strongly mixed variational-hemi- variational inequality is strongly well-posed in the generalized sense in Euclidean spaceRn.

(8)

Definition 3.1. A sequence{un} ⊂Vis said to be an approximating sequence for the SMVHVI if there exists a nonnegative sequence{n}withn → 0 asn → ∞such that

N

Agun, Tun

f, vgun J

un, vgun

GvG gun

≥ −nvgun

V, ∀v∈V. 3.1

Definition 3.2. The SMVHVI is said to be stronglyresp., weaklywell posed if the SMVHVI has a unique solution in V and every approximating sequence converges stronglyresp., weaklyto the unique solution.

Remark 3.3. Strong well-posedness implies weak well-posedness, but the converse is not true in general.

Definition 3.4. The SMVHVI is said to be stronglyresp., weaklywell posed in the general- ized sense if the SMVHVI has a nonempty solution set S in V and every approximating sequence has a subsequence which converges stronglyresp., weaklyto some point of the solution setS.

Remark 3.5. Strong well-posedness in the generalized sense implies weak well-posedness in the generalized sense, but the converse is not true in general.

Definition 3.6. LetN:V×VVandA:VVbe two mappings. Then

iAis said to be monotone with respect to the first argument ofNif there holds NAu, wNAv, w, u−v ≥0, ∀u, v∈V, wV; 3.2

iiAis said to be continuous with respect to the first argument ofNif for eachwV the mappingvNAv, wfromVintoVis continuous;

iiiA is said to be hemicontinuous with respect to the first argument ofN if for all u, vV andwV, the functiont → NAutvu, w, v−ufrom0,1 intoR is continuous at 0.

For any >0, we define the following two sets:

Ω

uV : N

Agu, Tu

f, vgu J

u, vgu

GvG gu

≥ −vgu

V, ∀v∈V , Ψ

uV :

NAv, Tuf, vgu J

u, vgu

GvG gu

≥ −vgu

V, ∀v∈V .

3.3

Lemma 3.7. Suppose thatA: VV is both monotone and hemicontinuous with respect to the first argument ofN, G:VR∪ {∞}is a proper, convex, and lower semicontinuous functional.

ThenΩ Ψfor all >0.

(9)

Proof. Letu ∈ Ω. Then, by the monotonicity of the mappingAwith respect to the first argument ofN, we have for allvV

0≤ N

Agu, Tu

f, vgu J

u, vgu

GvG gu

vgu

V

NAv, Tuf, vgu J

u, vgu

GvG gu

vgu

V.

3.4 This implies thatu∈Ψ. Thus, we get the inclusionΩ⊂Ψ.

Next let us show thatΨ⊂Ω. Indeed, for anyu∈Ψ, we have NAv, Tuf, vgu

J

u, vgu

GvG gu

≥ −vgu

V, ∀v∈V.

3.5 For anywV andt∈0,1, puttingv tw 1−tgu gu twguin3.5, we obtain

t

wgu

VN

A

tw 1−tgu , Tu

f, t

wgu J

u, t

wgu G

tw 1−tgu

G gu

. 3.6

Since the Clarke’s generalized directional derivativeJu, vis positively homogeneous with respect tovandGis convex, it follows that

N A

tw 1−tgu , Tu

f, wgu J

u, wgu

GwG gu

≥ −wgu

V. 3.7

Taking the limit for3.7ast → 0, we obtain from the hemicontinuity of the mappingA with respect to the first argument ofNthat

N

Agu, Tu

f, wgu J

u, wgu

GwG gu

≥ −wgu

V. 3.8

By the arbitrariness ofwV, we conclude thatu∈ Ω, which implies thatΨ ⊂ Ω.

This completes the proof.

Lemma 3.8. Suppose thatT : VV is continuous with respect to the second argument ofN, g :VV is continuous, andG:VR∪ {∞}is a proper, convex, and lower semicontinuous functional. ThenΨis closed inV for all >0.

Proof. Let{un} ⊂Ψbe a sequence such thatunuinV. Then NAv, Tunf, vgun

J

un, vgun

GvG gun

≥ −vgun

V,

∀v∈V.

3.9

(10)

SinceT : VV is continuous with respect to the second argument of N,g : VV is continuous,G : VR∪ {∞} is lower semicontinuous, and the Clarke’s generalized directional derivativeJu, vis upper semicontinuous with respect tou, v, we deduce that gungu, NAv, TunNAv, Tu, and

nlim→ ∞

NAv, Tun, v−gun

NAv, Tu, vgu ,

lim sup

n→ ∞ J

un, vgun

J

u, vgu ,

lim sup

n→ ∞G gun

≤ −G gu

.

3.10

Taking the lim sup for3.9asn → ∞, we obtain from3.10that NAv, Tuf, vgu

J

u, vgu

GvG gu

≥ −vgu

V, ∀v∈V, 3.11

which implies thatu∈Ψ. Therefore,Ψis closed inV. This completes the proof.

Corollary 3.9. Suppose thatA:VVis both monotone and hemicontinuous with respect to the first argument ofNandT :VVis continuous with respect to the second argument ofN. Let g :VV be continuous andG:VR∪ {∞}be a proper, convex, and lower semicontinuous functional. Then, for all >0,Ω Ψis closed inV.

Theorem 3.10. Suppose thatA:VVis both monotone and hemicontinuous with respect to the first argument ofNandT :VVis continuous with respect to the second argument ofN. Let g : VV be continuous andG : VR∪ {∞}a proper, convex, and lower semicontinuous functional. Then, the SMVHVI is strongly well posed if and only if

Ω/∅, ∀ >0, diamΩ−→0 as−→0. 3.12

Proof. “Necessity”. Suppose that the SMVHVI is strongly well posed. Then the SMVHVI has a unique solution which lies inΩand soΩ/∅for all >0. If diamΩ0 as → 0, then there exist a constantl >0, a nonnegative sequence{n}withn → 0 andun, vn∈Ωn such that

unvn V > l, ∀n≥1. 3.13

Sinceun, vn ∈Ωn, it is known that{un}and{vn}are both approximating sequences for the SMVHVI. From the strong well-posedness of the SMVHVI, it follows that both{un}and{vn} converge strongly to the unique solution of the SMVHVI, which is a contradiction to3.13.

(11)

“Sufficiency”. Let {un} ⊂ V be an approximating sequence for the SMVHVI. Then there exists a nonnegative sequence{n}withn → 0 such that

N

Agun, Tun

f, vgun J

un, vgun

GvG gun

≥ −nvgun

V,

∀v∈V, 3.14 which implies that un ∈ Ωn. By condition 3.12, {un} is a Cauchy sequence and so {un} converges strongly to some point uV. Since the mapping A is monotone with respect to the first argument ofN, the mappingT is continuous with respect to the second argument ofN,g is continuous, the Clarke’s generalized directional derivative Ju, vis upper semicontinuous with respect tou, v, andGis lower semicontinuous, it follows from 3.14that

NAv, Tuf, vgu J

u, vgu

GvG gu

≥lim sup

n→ ∞

NAv, Tunf, vgun J

un, vgun

GvG

gun

≥lim sup

n→ ∞

N

Agun, Tun

f, vgun J

un, vgun

GvG

gun

≥lim sup

n→ ∞

nvgun

V

0, ∀v∈V.

3.15 Furthermore, sinceAis also hemicontinuous with respect to the first argument ofNandGis convex, by the argument similar to that inLemma 3.7we can readily prove that

N

Agu, Tu

f, vgu J

u, vgu

GvG gu

≥0, ∀v∈V, 3.16 which implies thatusolves the SMVHVI.

To complete the proof ofTheorem 3.10, we need only to prove that the SMVHVI has a unique solution. Assume by contradiction that the SMVHVI has two distinct solutionsu1 andu2. Then it is easy to see thatu1, u2∈Ωfor all >0 and

0< u1u2 V ≤diamΩ−→0, 3.17

which is a contradiction. Therefore, the SMVHVI has a unique solution. This completes the proof.

Corollary 3.11 see30, Theorem 3.1. Suppose thatA : VV is a monotone and hemi- continuous mapping,T :VVis a continuous mapping, andG :VR∪ {∞}is a proper, convex, and lower semicontinuous functional. Then, the VHVI is strongly well posed if and only if

Ω/∅, ∀ >0, diamΩ−→0 as−→0. 3.18

(12)

Proof. InTheorem 3.10, putNu, v u v, for allu, vV and g I the identity mapping ofV. Then from the monotonicity and hemicontinuity ofAit follows thatA:VVis both monotone and hemicontinuous with respect to the first argument ofN. Moreover, from the continuity ofTit follows thatT :VVis continuous with respect to the second argument ofN. Thus, utilizingTheorem 3.10, we obtain the desired result.

Theorem 3.12. Suppose thatA:VVis both monotone and hemicontinuous with respect to the first argument ofNandT :VVis continuous with respect to the second argument ofN. Let g :VV be continuous andG:VR∪ {∞}be a proper, convex, and lower semicontinuous functional. Then, the SMVHVI is strongly well posed in the generalized sense if and only if

Ω/∅, ∀ >0, μΩ−→0 as−→0. 3.19

Proof. “Necessity”. Suppose that the SMVHVI is strongly well posed in the generalized sense.

Then the solution set of the SMVHVI is nonempty andS⊂Ωfor any >0. Furthermore, the solution set of the SMVHVI also is compact. In fact, for any sequence{un} ⊂S, it follows fromS ⊂ Ωfor any > 0 that{un} ⊂ Sis an approximating sequence for the SMVHVI.

Since the SMVHVI is strongly well posed in the generalized sense,{un}has a subsequence which converges strongly to some point of the solution setS. Thus, the solution setSof the SMVHVI is compact. Now let us show thatμΩ → 0 as → 0. FromS ⊂Ωfor any >0, we get

HΩ, S max{eΩ, S, eS,Ω} eΩ, S. 3.20

Taking into account the compactness of the solution setS, we obtain from3.20that

μΩ≤2HΩ, S 2eΩ, S. 3.21

In order to prove thatμΩ → 0 as → 0, it is sufficient to show thateΩ, S → 0 as → 0. Assume by contradiction thateΩ, S 0 as → 0. Then there exist a constant l >0, a sequence{n} ⊂Rwithn → 0 andun∈Ωnsuch that

un/SB0, l, 3.22

whereB0, lis the closed ball centered at 0 with radiusl. Since{un}is an approximating sequence for the SMVHVI and the SMVHVI is strongly well posed in the generalized sense, there exists a subsequence{unk}which converges strongly to some pointuSwhich is a contradiction to3.22. ThenμΩ → 0 as → 0.

“Sufficiency”. Assume that condition3.19holds. ByCorollary 3.9, we conclude that Ωis nonempty and closed for all >0. Observe that

S

>0

Ω. 3.23

(13)

SinceμΩ → 0 as → 0, by applying the theorem36, page 412, it can be easily found thatSis nonempty and compact with

eΩ, S HΩ, S−→0 as−→0. 3.24

Let{un} ⊂Vbe an approximating sequence for the SMVHVI. Then there exists a nonnegative sequence{n}withn → 0 such that

N

Agun, Tun

f, vgun J

un, vgun

GvG gun

≥ −nvgun

V,

∀v∈V, 3.25

and soun∈Ωnby the definition ofΩn. It follows from3.24that

dun, SeΩ, S−→0. 3.26

Since the solution setSis compact, there existsunSsuch that

unun V dun, S−→0. 3.27 Again from the compactness of the solution setS,{un}has a subsequence{unk}converging strongly to someuS. It follows from3.27that

unku Vunkunk V unku V −→0, 3.28 which implies that{unk}converges strongly to u. Therefore, the SMVHVI is strongly well- posed in the generalized sense. This completes the proof.

Corollary 3.13 see 30, Theorem 3.2. Suppose that A : VV is a monotone and hemicontinuous mapping,T : VV is a continuous mapping, andG : VR∪ {∞}is a proper, convex, and lower semicontinuous functional. Then, the VHVI is strongly well posed in the generalized sense if and only if

Ω/∅, ∀ >0, μΩ−→0 as−→0. 3.29 The following theorem gives some conditions under which the strongly mixed variational-hemi- variational inequality is strongly well posed in the generalized sense in Euclidean spaceRn.

Theorem 3.14. Suppose thatA: RnRnis both monotone and hemicontinuous with respect to the first argument ofN and T : RnRn is continuous with respect to the second argument of N. Letg : RnRn be continuous and G : RnR∪ {∞} be a proper, convex, and lower semicontinuous functional. If there exists some > 0 such thatΩ is nonempty and bounded.

Then the strongly mixed variational-hemivariational inequality SMVHVI is strongly well posed in the generalized sense.

(14)

Proof. Suppose that{un}is an approximating sequence for the SMVHVI. Then there exists a nonnegative sequence{n}withn → 0 asn → ∞such that

N

Agun, Tun

f, vgun J

un, vgun

GvG gun

≥ −nvgunRn,

∀v∈Rn. 3.30 Let0 >0 be such thatΩ0is nonempty and bounded. Then there existsn0such thatun ∈ Ω0for alln > n0. This implies that{un}is bounded by the boundedness ofΩ0. Thus, there exists a subsequence{unk} such that unku ask → ∞. Since the mappingA is monotone with respect to the first argument ofN, the mappingTis continuous with respect to the second argument ofN,gis continuous, the Clarke’s generalized directional derivative Ju, vis upper semicontinuous with respect to u, v, and G is lower semicontinuous, it follows from3.30that

NAv, Tuf, vgu J

u, vgu

GvG gu

≥lim sup

k→ ∞

NAv, Tunkf, vgunk J

unk, vgunk

GvG

gunk

≥lim sup

k→ ∞

N

Agunk, Tunk

f, vgunk J

unk, vgunk

GvG

gunk

≥lim sup

k→ ∞

nkvgunkRn 0, ∀v∈Rn.

3.31 Meantime, sinceAis also hemicontinuous with respect to the first argument ofNandGis convex, by the argument similar to that inLemma 3.7we can readily prove that

N

Agu, Tu

f, vgu J

u, vgu

GvG gu

≥0, ∀v∈Rn, 3.32 which implies thatusolves the SMVHVI. Therefore, the SMVHVI is strongly well-posed in the generalized sense. This completes the proof.

Corollary 3.15see30, Theorem 3.3. Suppose thatA : RnRn is a monotone and hemi- continuous mapping, T : RnRn is a continuous mapping, and G : RnR∪ {∞}is a proper, convex, and lower semicontinuous functional. If there exists some > 0 such thatΩis nonempty and bounded. Then the variational-hemivariational inequality VHVI is strongly well posed in the generalized sense.

4. Well-Posedness of Inclusion Problem

In this section, we first recall the concept of well-posedness for inclusion problems and then investigate the relations between the well-posedness for the strongly mixed variational- hemivariational inequality and the well-posedness for the corresponding inclusion problem.

(15)

In what follows we always assume thatFis a set-valued mapping from real reflexive Banach spaceV to its dual spaceV. The inclusion problem associated with mappingFis defined by

IPF: findxV such that 0∈Fx. 4.1

Definition 4.1see18,37. A sequence{un} ⊂ V is called an approximating sequence for the inclusion problem IPFifd0, Fun → 0 or, equivalently, there exists a sequencewnFunsuch that wn V → 0 asn → ∞.

Definition 4.2 see 18, 37. We say that the inclusion problem IPF is strongly resp., weaklywell posed if it has a unique solution and every approximating sequence converges stronglyresp., weaklyto the unique solution of IPF.

Definition 4.3 see 18, 37. We say that the inclusion problem IPF is strongly resp., weaklywell posed in the generalized sense if the solution setSof the IPFis nonempty and every approximating sequence has a subsequence which converges stronglyresp., weakly to some point of the solution setSfor the IPF.

The following two theorems establish the relations between the strongresp., weak well-posedness for the strongly mixed variational-hemivariational inequality and the strong resp., weakwell-posedness for the corresponding inclusion problem.

Theorem 4.4. LetN : V ×VV,A, T : VV, andg : VV be four mappings, J : VR a locally Lipschitz functional, and G : VR∪ {∞} a proper, convex, and lower semicontinuous functional. Then the strongly mixed variational-hemivariational inequality SMVHVI is strongly (resp., weakly) well posed if and only if the corresponding inclusion problem IPNAg, T−f∂J∂Ggis strongly (resp., weakly) well posed.

Theorem 4.5. LetN : V ×VV,A, T : VV, andg : VV be four mappings, J : VR a locally Lipschitz functional, andG : VR∪ {∞} a proper, convex, and lower semicontinuous functional. Then the strongly mixed variational-hemivariational inequality SMVHVI is strongly (resp., weakly) well posed in the generalized sense if and only if the corresponding inclusion problem IPNAg, Tf∂J∂Ggis strongly (resp., weakly) well posed in the generalized sense.

Lemma 4.6. Let N : V×VV, A, T : VV, and g : VV be four mappings, J : VR a locally Lipschitz functional, andG : VR∪ {∞} a proper, convex, and lower semicontinuous functional. ThenuV is a solution of the SMVHVI if and only ifuis a solution of the corresponding inclusion problem IPNAg, Tf∂J∂Ggof findinguV such that

0∈N A

gu , Tu

f∂Ju ∂G gu

. 4.2

Proof. “Sufficiency”. Assume thatuis a solution of the inclusion problem IPNAg, T− f∂J∂Gg. Then there existw1∂Juandw2∂Ggusuch that

N A

gu , Tu

fw1w2 0. 4.3

(16)

By multiplyingvguat both sides of the above equation4.3, we obtain from the de- finitions of the Clarke’s generalized gradient for locally Lipschitz functional and the sub- gradient for convex functional that

0

N A

gu , Tu

fw1w2, vgu

N

A gu

, Tu

f, vgu J

u, vgu

GvG gu

, ∀v∈V, 4.4

which implies thatuis a solution of the SMVHVI.

“Necessity”. Suppose thatuis a solution of the SMVHVI. Then, N

A gu

, Tu

f, vgu J

u, vgu

GvG gu

≥0, ∀v∈V. 4.5

From the fact that J

u, vgu

max

w, vgu

:w∂Ju

, 4.6

we deduce that there exists awu, v∂Jusuch that N

A gu

, Tu

f, vgu

w

gu, v

, vgu

GvG gu

≥0, ∀v∈V.

4.7 In terms of Proposition 2.4 iv, ∂Ju is a nonempty, convex, bounded, weak-compact subset of V. Note that V is a real reflexive Banach space. Hence, ∂Ju is a nonempty, convex, bounded, weak-compact subset inV. Thus∂Juis a nonempty, closed, convex, and bounded subset inVwhich implies that{NAgu, Tu−fw:w∂Ju}is nonempty, closed, convex, and bounded inV. SinceG:VR∪ {∞}is a proper, convex, and lower semicontinuous functional, it follows fromTheorem 2.7withϕu Guand4.7that there existswu∂Jusuch that

N A

gu , Tu

f, vgu

wu, vgu

GvG gu

≥0, ∀v∈V. 4.8

For the sake of simplicity we writew wu, and hence from4.8we have GvG

gu

−N A

gu , Tu

fw, vgu

, ∀v∈V, 4.9

which implies that−NAgu, Tu fw∂Ggu. Consequently, it follows fromw

∂Juthat

0∈N A

gu , Tu

f∂Ju ∂G gu

, 4.10

which implies thatuis a solution of the inclusion problem IPNAg, T−f∂J∂Gg.

This completes the proof.

(17)

Proof ofTheorem 4.4. “Necessity”. Assume that the SMVHVI is stronglyresp., weaklywell posed. Then there is a unique solution u for the SMVHVI. ByLemma 4.6, u also is the unique solution for the inclusion problem IPNAg, T−f∂J ∂Gg. Let{un}be an approximating sequence for the IPNAg, T−f∂J∂Gg. Then there exists a sequence wnNAgun, Tunf∂Jun ∂Ggunsuch that wn V → 0 asn → ∞. And so there existξn∂Junandηn∂Ggunsuch that

wn N A

gun , Tun

nηn. 4.11

From the definitions of the Clarke’s generalized gradient for locally Lipschitz functional and the subgradient for convex functional, we obtain by multiplyingvgunat both sides of the above equation4.11that

N A

gun , Tun

f, vgun J

un, vgun

GvG gun

N

A gun

, Tun

f, vgun

ξn, vgun

ηn, vgun wn, vgun

≥ − wn Vvgun

V, ∀v∈V.

4.12

Lettingn wn V, we obtain that{un}is an approximating sequence for the SMVHVI from 4.12with wn V → 0 asn → ∞. Therefore, it follows from the strongresp., weakwell- posedness of the SMVHVI that{un}converges stronglyresp., weaklyto the unique solution u. Thus, the inclusion problem IPNAg, T−f∂J∂Ggis stronglyresp., weakly well posed.

“Sufficiency”. Suppose that the inclusion problem IPNAg, T−f∂J∂Ggis stronglyresp., weaklywell posed. Then the IPNAg, T−f∂J∂Gghas a unique solutionu, which implies thatuis the unique solution for the SMVHVI byLemma 4.6. Let {un}be an approximating sequence for the SMVHVI. Then there exists a sequence{n}with n → 0 asn → ∞such that

N A

gun , Tun

−f, v−gun J

un, vgun

Gv−G gun

≥ −nvgun

V,

∀v∈V.

4.13

By the same argument as in the proof ofLemma 4.6, there exists awun, v∂Junsuch that N

A gun

, Tun

f, vgun

w

gun, v

, vgun

GvG gun

≥ −nvgun

V, ∀v∈V, 4.14

(18)

and{NAgun, Tunfw :w∂Jun}is nonempty, closed, convex, and bounded in V. Then, it follows from4.14andTheorem 2.7withϕu Gu n u−gun V, which is proper, convex, and lower semicontinuous, that there existswun∂Junsuch that

N A

gun , Tun

f, vgunwun, v−gunGvG gun

≥ −nvgun

V, ∀v∈V. 4.15

For the sake of simplicity we writewn wun, and hence from4.15we have G

gun

Gv N

A gun

, Tun

fwn, vgun

nvgun

V, ∀v∈V.

4.16

Define functionalGn:VR∪ {∞}as follows:

Gnv Gv Pnv nQnv, 4.17

wherePnvandQnvare two functionals onV defined by Pnv

N A

gun , Tun

fwn, vgun

, Qnv vgun

V. 4.18

Clearly,Gnis proper, convex, and lower semicontinuous andv gunis a global minimizer ofGnonV. Thus, 0∈∂Gngun. Since the functionalsPnandQnare continuous onV and Gis proper, convex, and lower semicontinuous, it follows fromProposition 2.3that

∂Gnv ∂Gv N A

gun , Tun

fwnn∂Qnv. 4.19

It is easy to calculate that

∂Qnv

vV: v V 1,

v, vgun vgun

V

, 4.20

and so there exists aξn∂Qngunwith ξn V 1 such that 0∈∂G

gun N

A gun

, Tun

fwnnξn. 4.21

Lettingunnξn, we have un V → 0 asn → 0. Moreover, sincewn∂Jun, it follows from4.21that

unN A

gun , Tun

f∂Jun ∂G gun

, 4.22

which implies that{un}is an approximating sequence for the IPNAg, T−f∂J∂Gg.

Since the inclusion problem IPNAg, T−f ∂J ∂Gg is strongly resp., weakly

(19)

well posed,{un}converges stronglyresp., weaklyto the unique solutionu. Therefore, the strongly mixed variational-hemivariational inequality SMVHVI is strongly resp., weakly well posed. This completes the proof.

Proof ofTheorem 4.5. The proof is similar to that inTheorem 4.4and so we omit it here.

Corollary 4.7see30, Theorem 4.1. LetAandT be two mappings from Banach spaceV to its dualV,J:VR be a locally Lipschitz functional, andG:VR∪{∞}be a proper, convex, and lower semicontinuous functional. Then the variational-hemivariational inequality VHVI is strongly (resp., weakly) well posed if and only if the corresponding inclusion problem IPATf∂J∂G is strongly (resp., weakly) well posed.

Proof. InTheorem 4.4, putg Ithe identity mapping ofV andNu, v uv, for all u, vV. Then, in terms ofTheorem 4.4we derive the desired result.

Corollary 4.830, Theorem 4.2. LetAandT be two mappings from Banach spaceV to its dual V,J :VR be a locally Lipschitz functional, andG:VR∪ {∞}be a proper, convex, and lower semicontinuous functional. Then the variational-hemivariational inequality VHVI is strongly (resp., weakly) well posed in the generalized sense if and only if the corresponding inclusion problem IPATf∂J∂Gis strongly (resp., weakly) well posed in the generalized sense.

5. Concluding Remarks

In this paper, we introduce some concepts of well-posedness for a class of strongly mixed variational-hemivariational inequalities with perturbations, which includes as a special case the class of variational-hemivariational inequalities in30. We establish some metric char- acterizations for the well-posed strongly mixed variational-hemivariational inequality and give some conditions under which the strongly mixed variational-hemivariational inequality is strongly well posed in the generalized sense inRn. On the other hand, we first recall the concept of well-posedness for inclusion problems and then investigate the relations between the strongresp., weakwell-posedness for a strongly mixed variational-hemivariational ine- quality and the strongresp., weakwell-posedness for the corresponding inclusion problem.

It is well known that there are many other concepts of well-posedness for optimization problems, variational inequalities, and Nash equilibrium problems, such asα-well-posedness 17, well-posedness by perturbations12, and Levitin-Polyak well-posedness38. How- ever, we wonder whether the concepts mentioned as above can be extended to the strong- ly mixed variational-hemivariational inequality. Beyond question, this is an interesting prob- lem.

Acknowledgments

This paper was partially supported by the National Science Foundation of China11071169, Innovation Program of Shanghai Municipal Education Commission09ZZ133, and Leading Academic Discipline Project of Shanghai Normal UniversityDZL707 to L.-C. Ceng. This paper was partially supported by the Taiwan NSC Grant 99-2115-M-110-007-MY3to N.-C.

Wang. This paper was partially supported by the Taiwan NSC Grant 99-2221-E-037-007-MY3 to J.-C. Yao.

参照

関連したドキュメント

Kilbas; Conditions of the existence of a classical solution of a Cauchy type problem for the diffusion equation with the Riemann-Liouville partial derivative, Differential Equations,

In 2, the regularity of weak solutions to the characteristic BVP 1.2-1.3 was studied, under the assumption that the problem is strongly L 2 -well posed, namely, that a unique L

Turmetov; On solvability of a boundary value problem for a nonhomogeneous biharmonic equation with a boundary operator of a fractional order, Acta Mathematica Scientia.. Bjorstad;

Merle; Global wellposedness, scattering and blow up for the energy critical, focusing, nonlinear Schr¨ odinger equation in the radial case, Invent.. Strauss; Time decay for

Analogs of this theorem were proved by Roitberg for nonregular elliptic boundary- value problems and for general elliptic systems of differential equations, the mod- ified scale of

We give another global upper bound for Jensen’s discrete inequal- ity which is better than already existing ones.. For instance, we determine a new converses for generalized A–G and

In addition, under the above assumptions, we show, as in the uniform norm, that a function in L 1 (K, ν) has a strongly unique best approximant if and only if the best

In Section 2, we establish conditions under which (1.2) is well-posed using stable families of generators of semigroups and Kato’s stability conditions [8, 11]; our work also