Volume 2009, Article ID 364615,13pages doi:10.1155/2009/364615
Research Article
General System of
A
-Monotone Nonlinear
Variational Inclusions Problems with Applications
Jian-Wen Peng
1and Lai-Jun Zhao
21College of Mathematics and Computer Science, Chongqing Normal University, Chongqing 400047, China 2Management School, Shanghai University, Shanghai 200444, China
Correspondence should be addressed to Lai-Jun Zhao,zhao laijun@163.com
Received 25 September 2009; Accepted 2 November 2009
Recommended by Vy Khoi Le
We introduce and study a new system of nonlinear variational inclusions involving a combination ofA-Monotone operators and relaxed cocoercive mappings. By using the resolvent technique of theA-monotone operators, we prove the existence and uniqueness of solution and the convergence of a new multistep iterative algorithm for this system of variational inclusions. The results in this paper unify, extend, and improve some known results in literature.
Copyrightq2009 J.-W. Peng and L.-J. Zhao. 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.
1. Introduction
A-monotonicity is defined in terms of relaxed monotone mappings—a more general notion than the monotonicity or strong monotonocity—which gives a significant edge over the H-monotonocity. Very recently, Verma 6 studied the solvability of a system of variational inclusions involving a combination ofA-monotone and relaxed cocoercive mappings using resolvent operator techniques ofA-monotone mappings. Since relaxed cocoercive mapping is a generalization of strong monotone mappings, the main result in6is more general than the corresponding results in1,2.
Inspired and motivated by recent works in 1, 2, 6, the purpose of this paper is to introduce a new mathematical model, which is called a general system of A-monotone nonlinear variational inclusion problems, that is, a family of A-monotone nonlinear variational inclusion problems defined on a product set. This new mathematical model contains the system of inclusions in1,2,6, the variational inclusions in7,8, and some variational inequalities in literature as special cases. By using the resolvent technique for the A-monotone operators, we prove the existence and uniqueness of solution for this system of variational inclusions. We also prove the convergence of a multistep iterative algorithm approximating the solution for this system of variational inclusions. The result in this paper unifies, extends, and improves some results in1,2,6–8and the references therein.
2. Preliminaries
We suppose thatHis a real Hilbert space with norm and inner product denoted by · and ·,·, respectively, 2H denotes the family of all the nonempty subsets ofH. IfM :H → 2H
be a set-valued operator, then we denote the effective domainDMofMas follows:
DM {x∈ H:Mx/∅}. 2.1
Now we recall some definitions needed later.
Definition 2.1see2,6,7. LetA:H → Hbe a single-valued operator and letM:H → 2H be a set-valued operator.Mis said to be
im-relaxed monotone, if there exists a constantm >0 such that
x−y, u−v≥ −mu−v2, ∀u, v∈DM, x∈Mu, y∈Mv, 2.2
iiA-monotone with a constantmif
aMism-relaxed monotone,
bAλMis maximal monotone forλ >0 i.e.,AλMH H, for allλ >0.
It follow from3, Lemma 7.11we know that ifXis a reflexive Banach space withX∗ its dual, andA:X → X∗bem-strongly monotone andf :X → Ris a locally Lipschitz such that∂fisα-relaxed monotone, then∂fisA-monotone with a constantm−α.
Definition 2.3see1,7,8. LetA, T :H → H, be two single-valued operators.T is said to be
imonotone if
Tu−Tv, u−v ≥0, ∀u, v∈ H; 2.3
iistrictly monotone ifTis monotone and
Tu−Tv, u−v0, iffuv; 2.4
iiiγ-strongly monotone if there exists a constantγ >0 such that
Tu−Tv, u−v ≥γu−v2, ∀u, v∈ H; 2.5
ivs-Lipschitz continuous if there exists a constants >0 such that
Tu−Tv ≤su−v, ∀u, v∈ H; 2.6
vr-strongly monotone with respect toAif there exists a constantγ >0 such that
Tu−Tv, Au−Av ≥ru−v2, u, v∈ H. 2.7
Definition 2.4see2. LetA:H → Hbe aγ-strongly monotone operator and letM:H → 2Hbe anA-monotone operator. Then the resolvent operatorRAM,λ:H → His defined by
RA
M,λx AλM−1x, ∀x∈ H. 2.8
We also need the following result obtained by Verma2.
Lemma 2.5. LetA : H → Hbe aγ-strongly monotone operator and letM : H → 2H be an A-monotone operator. Then, the resolvent operatorRAM,λ : H → His Lipschitz continuous with constant1/γ−mλfor0< λ < γ/m, that is,
RA
M,λx−RAM,λ
y≤ γ−1mλx−y, ∀x, y∈H. 2.9
Definition 2.6. LetH1,H2, . . . ,Hpbe Hilbert spaces and · 1denote the norm ofH1, also let
A1:H1 → H1andN1:pj1Hj → H1be two single-valued mappings:
iN1is said to beξ-Lipschitz continuous in the first argument if there exists a constant
ξ >0 such that
N1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp1 ≤ξx1−y11,
∀x1, y1∈ H1, xj∈ Hj
j2,3, . . . , p; 2.10
iiN1is said to be monotone with respect toA1in the first argument if
N1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp, A1x1−A1
y1
≥0, ∀x1, y1∈ H1, xj∈ Hj j2,3, . . . , p;
2.11
iiiN1is said to beβ-strongly monotone with respect toA1in the first argument if there
exists a constantβ >0 such that
N1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp, A1x1−A1
y1
≥βx1−y121,
∀x1, y1∈ H1, xj∈ Hjj2,3, . . . , p;
2.12
ivN1is said to beγ-cocoercive with respect toA1in the first argument if there exists
a constantγ >0 such that
N1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp, A1x1−A1
y1
≥γN1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp21, ∀x1, y1∈ H1, xj∈ Hj j2,3, . . . , p;
2.13
vN1is said to beγ-relaxed cocoercive with respect toA1in the first argument if there
exists a constantγ >0 such that
N1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp, A1x1−A1
y1
≥ −γN1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp21, ∀x1, y1∈ H1, xj∈ Hj j 2,3, . . . , p;
2.14
viN1is said to beγ, r-relaxed cocoercive with respect toA1in the first argument if
there exists a constantγ >0 such that
N1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp, A1x1−A1
y1
≥ −γN1
x1, x2, . . . , xp−N1
y1, x2, . . . , xp21rx1−y121,
∀x1, y1∈ H1, xj∈ Hj j2,3, . . . , p.
In a similar way, we can define the Lipschitz continuity and the strong monotonicity
monotonicity, relaxed cocoercivity cocoercivity ofNi : pj1Hj → Hi with respect to Ai:Hi → Hiin theith argumenti2,3, . . . , p.
3. A System of Set-Valued Variational Inclusions
In this section, we will introduce a new system of nonlinear variational inclusions in Hilbert spaces. In what follows, unless other specified, for eachi 1,2, . . . , p, we always suppose thatHiis a Hilbert space with norm denoted by · i,Ai : Hi → Hi,Fi : pj1Hj → Hi are single-valued mappings, andMi : Hi → 2Hi is a nonlinear mapping. We consider the following problem of findingx1, x2, . . . , xp∈pi1Hisuch that for eachi1,2, . . . , p,
0∈Fix1, x2, . . . , xp
Mixi. 3.1
Below are some special cases of3.1.
Ifp2, then3.1becomes the following problem of findingx1, x2∈ H1× H2such
that
0∈F1x1, x2 M1x1,
0∈F2x1, x2 M2x1.
3.2
However, 3.2 is called a system of set-valued variational inclusions introduced and researched by Fang and Huang1,9and Verma2,6.
Ifp 1, then3.1becomes the following variational inclusion with anA-monotone operator, which is to findx1∈ H1such that
0∈F1x1 M1x1, 3.3
problem3.3is introduced and studied by Fang and Huang8. It is easy to see that the mathematical model2studied by Verma7is a variant of3.3.
4. Existence of Solutions and Convergence of an Iterative Algorithm
In this section, we will prove existence and uniqueness of solution for 3.1. For our main results, we give a characterization of the solution of3.1as follows.
Lemma 4.1. Fori 1,2, . . . , p, letAi : Hi → Hi be a strictly monotone operator and letMi : Hi → 2Hi be anAi-monotone operator. Thenx1, x2, . . . , xp∈pi1Hiis a solution of 3.1if and
only if for eachi1,2, . . . , p,
xiRAMii,λiAixi−λiFix1, x2, . . . , xp, 4.1
Proof. It holds thatx1, x2, . . . , xp∈pi1Hiis a solution of3.1
⇐⇒θi∈Fix1, x2, . . . , xpMixi, i1,2, . . . , p,
⇐⇒Aixi−λiFix1, x2, . . . , xp∈AiλiMixi, i1,2, . . . , p,
⇐⇒xiRAMii,λiAixi−λiFix1, x2, . . . , xp
, i1,2, . . . , p.
4.2
LetΓ {1,2, . . . , p}.
Theorem 4.2. For i 1,2, . . . , p, let Ai : Hi → Hi be γi-strongly monotone and let τi -Lipschitz continuous, Mi : Hi → 2Hi be an Ai-monotone operator with a constant mi, let
Fi :pj1Hj → Hibe a single-valued mapping such thatFi isθi, ri-relaxed cocoercive monotone
with respect toAiandsi-Lipschitz continuous in theith argument,Fiislij-Lipschitz continuous in thejth arguments for eachj∈Γ, j /i. Suppose that there exist constantsλi>0i1,2, . . . , psuch that
1 γ1−m1λ1
τ2
1θ21−2λ1r12λ1θ1s21λ1 2s2
1
p
k2
lk1λk γk−mkλk <1,
1 γ2−m2λ2
τ2
2θ22−2λ2r22λ2θ2s22λ2 2s2
2
k∈Γ, k /2
lk2λk γk−mkλk <1,
. . . ,
1 γp−mpλp
τ2
pθp2−2λprp2λpθps2pλp2s2p p−1
k1
lk,pλk γk−mkλk <1.
4.3
Then,3.1admits a unique solution.
Proof. For i 1,2, . . . , p and for any given λi > 0, define a single-valued mapping Ti,λi :
p
j1Hj → Hiby
Ti,λi
x1, x2, . . . , xpRMAii,λiAixi−λiFix1, x2, . . . , xp, 4.4
For anyx1, x2, . . . , xp,y1, y2, . . . , yp∈pi1Hi, it follows from4.4andLemma 2.5
that fori1,2, . . . , p,
Ti,λ
i
x1, x2, . . . , xp−Ti,λi
y1, y2, . . . , ypi
RAi
Mi,λi
Aixi−λiFix1, x2, . . . , xp−RAMii,λiAiyi−λiFiy1, y2, . . . , ypi
≤ γ 1 i−miλi
Aixi−Ai
yi−λiFix1, x2, . . . , xp−Fiy1, y2, . . . , ypi
≤ γ 1 i−miλi
Aixi−Ai
yi
−λiFix1, x2, . . . , xi−1, xi, xi1, . . . , xp−Fix1, x2, . . . , xi−1, yi, xi1, . . . , xpi
λi
γi−miλi
⎛ ⎝
j∈Γ,j /i
Fi
x1, x2, . . . , xj−1, xj, xj1, . . . , xp
−Fix1, x2, . . . , xj−1, yj, xj1, . . . , xpi
⎞ ⎠.
4.5
Fori 1,2, . . . , p, sinceAi isτi-Lipschitz continuous, Fi isθi, ri-relaxed cocoercive with respected toAiandsi-Lipschitz continuous in theith argument, we have
Aixi−Ai
yi
−λiFix1, x2, . . . , xi−1, xi, xi1, . . . , xp−Fix1, x2, . . . , xi−1, yi, xi1, . . . , xp2i
≤Aixi−Aiyi2i
−2λiFix1, x2, . . . , xi−1, xi, xi1, . . . , xp
−Fix1, x2, . . . , xi−1, yi, xi1, . . . , xp, Aixi−Aiyi
λi2Fix1, x2, . . . , xi−1, xi, xi1, . . . , xp−Fix1, x2, . . . , xi−1, yi, xi1, . . . , xp2i
≤τ2
ixi−yi2i −2λirixi−yi2i 2λiθis2ixi−yi2i λi2s2ixi−yi2i ≤τ2
i −2λiri2λiθis2i λi2s2i
xi−yi2i.
4.6
Fori1,2, . . . , p, sinceFiislij-Lipschitz continuous in thejth argumentsj∈Γ, j /i, we have
Fi
x1, x2, . . . , xj−1, xj, xj1, . . . , xp−Fix1, x2, . . . , xj−1, yj, xj1, . . . , xpi ≤lijxj−yjj,
It follows from4.5–4.7that for eachi1,2, . . . , p,
Ti,λi
x1, x2, . . . , xp
−Ti,λi
y1, y2, . . . , ypi
≤ γ 1 i−miλi
τ2
i −2λiri2λiθis2i λi2s2ixi−yii λi γi−miλi
⎛ ⎝
j∈Γ,j /i
lijxj−yjj
⎞ ⎠.
4.8
Hence,
p
i1
Ti,λi
x1, x2, . . . , xp−Ti,λi
y1, y2, . . . , ypi
≤ p i1
⎡ ⎣ 1
γi−miλi
τ2
i −2λiri2λiθis2i λi2s2ixi−yiiγ λi i−miλi
⎛ ⎝
j∈Γ,j /i
lijxj−yjj
⎞ ⎠ ⎤ ⎦ 1 γ1−m1λ1
τ2
1θ21−2λ1r12λ1θ1s21λ1 2s2
1
p
k2
lk1λk γk−mkλk
x1−y1
1
⎛
⎝ 1
γ2−m2λ2
τ2
2θ22−2λ2r22λ2θ2s22λ2 2s2
2
k∈Γ,k /2
lk2λk γk−mkλk
⎞
⎠x2−y2
2
· · ·
1 γp−mpλp
τ2
pθ2p−2λprp2λpθps2pλp2s2p p−1
k1
lk,pλk γk−mkλk
xp−yp
p
≤ξ
p
k1
xk−yk
k
,
4.9
where
ξmax
1 γ1−m1λ1
τ2
1θ21−2λ1r12λ1θ1s21λ1 2s2
1
p
k2
lk1λk
γk−mkλk,
1 γ2−m2λ2
τ2
2θ22−2λ2r22λ2θ2s22λ2 2s2
2
k∈Γ,k /2
lk2λk γk−mkλk,
. . . ,
1 γp−mpλp
τ2
pθ2p−2λprp2λpθps2pλp2s2p p−1
k1
lk,pλk γk−mkλk
.
Define · Γ on pi1Hi by x1, x2, . . . , xpΓ x11 x22 · · · xpp, for all x1, x2, . . . , xp ∈ pi1Hi. It is easy to see that
p
i1Hi is a Banach space. For any given
λi>0i∈Γ, defineWΓ,λ1,λ2,...,λp :
p i1Hi →
p i1Hiby
WΓ,λ1,λ2,...,λp
x1, x2, . . . , xp
T1,λ1x1, x2, . . . , xp, T2,λ2x1, x2, . . . , xp, . . . , Tp,λp
x1, x2, . . . , xp,
4.11
for allx1, x2, . . . , xp∈pi1Hi.
By4.3, we know that 0< ξ <1, it follows from4.9that
WΓ,λ1,λ2,...,λp
x1, x2, . . . , xp−WΓ,λ1,λ2,...,λp
x1, x2, . . . , xpΓ
≤ξx1, x2, . . . , xp−y1, y2, . . . , ypΓ.
4.12
This shows that WΓ,λ1,λ2,...,λp is a contraction operator. Hence, there exists a unique
x1, x2, . . . , xp∈pi1Hi, such that
WΓ,λ1,λ2,...,λp
x1, x2, . . . , xp
x1, x2, . . . , xp
, 4.13
that is, fori1,2, . . . , p,
xiRAMii,λiAixi−λiFix1, x2, . . . , xp. 4.14
ByLemma 4.1,x1, x2, . . . , xpis the unique solution of3.1. This completes this proof.
Corollary 4.3. For i 1,2, . . . , p, let Hi : Hi → Hi beγi-strongly monotone andτi-Lipschitz continuous, letMi :Hi → 2Hi be anHi-monotone operator, letFi :pj1Hj → Hi be a
single-valued mapping such thatFiisri-strongly monotone with respect toHiandsi-Lipschitz continuous in theith argument,Fiislij-Lipschitz continuous in thejth arguments for eachj ∈Γ, j /i. Suppose that there exist constantsλi>0 i1,2, . . . , psuch that
1 γ1
τ2
1−2λ1r1λ1 2s2
1
p
k2
lk1λk γk <1,
1 γ2
τ2
2 −2λ2r2λ22s22
k∈Γ,k /2
lk2λk
γk <1, ..
.
1 γp
τ2
p−2λprpλp2s2p p−1
k1
lk,pλk γk <1.
4.15
Remark 4.4. Theorem 4.2andCorollary 4.3unify, extend, and generalize the main results in
1,2,6–8.
5. Iterative Algorithm and Convergence
In this section, we will construct some multistep iterative algorithm for approximating the unique solution of3.1and discuss the convergence analysis of these Algorithms.
Lemma 5.1see8,9. Let{cn}and{kn}be two real sequences of nonnegative numbers that satisfy the following conditions:
10≤kn<1, n0,1,2, . . .andlim sup n kn<1,
2cn1≤kncn, n0,1,2, . . . ,
thencnconverges to 0 asn → ∞.
Algorithm 5.2. Fori 1,2, . . . , p, letAi, Mi, Fibe the same as inTheorem 4.2. For any given
x0
1, x02, . . . , x0p∈
p
j1Hj, define a multistep iterative sequence{x1n, xn2, . . . , xnp}by
xn1
i αnxni 1−αn
RAi Mi,λi
Aixni
−λiFi
xn
1, xn2, . . . , xnp
, 5.1
where
0≤αn <1, lim sup
n αn<1. 5.2
Theorem 5.3. For i 1,2, . . . , p, let Ai, Mi, Fi be the same as in Theorem 4.2. Assume that all the conditions of theorem 4.1 hold. Then {x1n, x2n, . . . , xpn} generated byAlgorithm 5.2 converges strongly to the unique solutionx1, x2, . . . , xpof 3.1.
Proof. ByTheorem 4.2, problem3.1admits a unique solutionx1, x2, . . . , xp, it follows from
Lemma 4.1that for eachi1,2, . . . , p,
xiRAMii,λiAixi−λiFix1, x2, . . . , xp
. 5.3
It follows from4.3,5.1and5.3that for eachi1,2, . . . , p,
xn1
i −xii αn
xn i −xi
1−αn
RAi Mi,λi
Aixni−λiFi
xn
1, xn2, . . . , xnp
−RAi Mi,λi
≤αnxni −xii 1−αnγ 1 i−miλi
×Aixni−Aixi−λi
Fi
xn
1, xn2, . . . , xni−1, xni, xni1, . . . , xnp
−Fi
xn
1, xn2, . . . , xni−1, xi, xin1, . . . , xnpi
γ λi
i−miλi
⎛ ⎝
j∈Γ,j /i
Fi
xn
1, xn2, . . . , xnj−1, xjn, xjn1, . . . , xnp
−Fi
xn
1, xn2, . . . , xnj−1, xj, xnj1, . . . , xnpi
⎞ ⎠.
5.4
Fori 1,2, . . . , p, sinceAi isτi-Lipschitz continuous, Fi isθi, ri-relaxed cocoercive with respected toAi, andsi-Lipschitz is continuous in theith argument, we have
Aixni
−Aixi
−λi
Fi
xn
1, xn2, . . . , xni−1, xni, xin1, . . . , xnp
−Fi
xn
1, xn2, . . . , xni−1, xi, xin1, . . . , xnp 2
i
≤τ2
i −2λiri2λiθis2i λi2s2i
xn i −xi2.
5.5
Fori1,2, . . . , p, sinceFiislij-Lipschitz continuous in thejth argumentsj∈Γ, j /i, we have
Fi
xn
1, xn2, . . . , xnj−1, xnj, xnj1, . . . , xnp
−Fi
xn
1, x2n, . . . , xjn−1, xj, xnj1, . . . , xpni ≤lijxnj −xjj.
5.6
It follows from5.4–5.6that fori1,2, . . . , p,
xn1
i −xii≤αnxni −xii 1−αn 1 γi−miλi
τ2
i −2λiri2λiθis2i λi2s2ixni −xii
1−αn λi γi−miλi
⎛ ⎝
j∈Γ,j /i
lijxnj −xjj
⎞ ⎠.
Hence,
p
i1
xn1
i −xii≤ p
i1
⎡ ⎣αnxn
i −xii 1−αnγ 1 i−miλi
τ2
i −2λiri2λiθis2i λi2s2ixni −xii
1−αn λi
γi−miλi
⎛ ⎝
j∈Γ,j /i
lijxnj −xjj
⎞ ⎠ ⎤ ⎦
≤αn
p
i1
xn i −xii
1−αnξ
p
i1
xn i −xii
ξ 1−ξαn
p
i1
xn i −xii
,
5.8
where
ξmax
1 γ1−m1λ1
τ2
1θ21−2λ1r12λ1θ1s21λ1 2s2
1
p
k2
lk1λk γk−mkλk,
1 γ2−m2λ2
τ2
2θ22−2λ2r22λ2θ2s22λ22s22
k∈Γ,k /2
lk2λk
γk−mkλk,
. . . ,
1 γp−mpλp
τ2
pθp2−2λprp2λpθps2pλp2s2p p−1
k1
lk,pλk γk−mkλk
.
5.9
It follows from hypothesis4.3that 0< ξ <1.
Letanpi1xni −xii, ξnξ1−ξαn. Then,5.8can be rewritten asan1≤ξnan, n 0,1,2, . . . .By5.2, we know that lim supnξn<1, it follows fromLemma 5.1that
an p
i1
xn
i −xiiconverges to 0 as n−→ ∞. 5.10
Therefore,{xn1, xn2, . . . , xnp}converges to the unique solutionx1, x2, . . . , xpof3.1. This completes the proof.
Acknowledgment
This study was supported by grants from National Natural Science Foundation of China
References
1 Y. P. Fang and N. J. Huang, “H-monotone operators and system of variational inclusions,”
Communications on Applied Nonlinear Analysis, vol. 11, no. 1, pp. 93–101, 2004.
2 R. U. Verma, “A-monotonicity and applications to nonlinear variational inclusion problems,”Journal of Applied Mathematics and Stochastic Analysis, vol. 17, no. 2, pp. 193–195, 2004.
3 Z. Naniewicz and P. D. Panagiotopoulos, Mathematical Theory of Hemivariational Inequalities and Applications, vol. 188 ofMonographs and Textbooks in Pure and Applied Mathematics, Marcel Dekker, New York, NY, USA, 1995.
4 R. U. Verma, “Nonlinear variational and constrained hemivariational inequalities involving relaxed operators,”Zeitschrift f ¨ur Angewandte Mathematik und Mechanik, vol. 77, no. 5, pp. 387–391, 1997. 5 P. D. Panagiotopoulos,Hemivariational Inequalities and Their Applications in Mechanics and Engineering,
Springer, New York, NY, USA, 1993.
6 R. U. Verma, “General system of A-monotone nonlinear variational inclusion problems with applications,”Journal of Optimization Theory and Applications, vol. 131, no. 1, pp. 151–157, 2006. 7 R. U. Verma, “General nonlinear variational inclusion problems involvingA-monotone mappings,”
Applied Mathematics Letters, vol. 19, no. 9, pp. 960–963, 2006.
8 Y.-P. Fang and N.-J. Huang, “H-monotone operator and resolvent operator technique for variational inclusions,”Applied Mathematics and Computation, vol. 145, no. 2-3, pp. 795–803, 2003.