Research Article Finite-Step Relaxed Hybrid Steepest-Descent Methods for Variational Inequalities

Σχετικά έγγραφα
COMMON RANDOM FIXED POINT THEOREMS IN SYMMETRIC SPACES

Data Dependence of New Iterative Schemes

1. For each of the following power series, find the interval of convergence and the radius of convergence:

On Certain Subclass of λ-bazilevič Functions of Type α + iµ

On Inclusion Relation of Absolute Summability

n r f ( n-r ) () x g () r () x (1.1) = Σ g() x = Σ n f < -n+ r> g () r -n + r dx r dx n + ( -n,m) dx -n n+1 1 -n -1 + ( -n,n+1)

A study on generalized absolute summability factors for a triangular matrix

On Generating Relations of Some Triple. Hypergeometric Functions

Introduction of Numerical Analysis #03 TAGAMI, Daisuke (IMI, Kyushu University)

Homework for 1/27 Due 2/5

Congruence Classes of Invertible Matrices of Order 3 over F 2

Last Lecture. Biostatistics Statistical Inference Lecture 19 Likelihood Ratio Test. Example of Hypothesis Testing.

Supplemental Material: Scaling Up Sparse Support Vector Machines by Simultaneous Feature and Sample Reduction

Μια εισαγωγή στα Μαθηματικά για Οικονομολόγους

Biorthogonal Wavelets and Filter Banks via PFFS. Multiresolution Analysis (MRA) subspaces V j, and wavelet subspaces W j. f X n f, τ n φ τ n φ.


Every set of first-order formulas is equivalent to an independent set

IIT JEE (2013) (Trigonomtery 1) Solutions

Other Test Constructions: Likelihood Ratio & Bayes Tests

Binet Type Formula For The Sequence of Tetranacci Numbers by Alternate Methods

MINIMAL CLOSED SETS AND MAXIMAL CLOSED SETS

Solutions: Homework 3

The Heisenberg Uncertainty Principle

Solve the difference equation

2. Let H 1 and H 2 be Hilbert spaces and let T : H 1 H 2 be a bounded linear operator. Prove that [T (H 1 )] = N (T ). (6p)

Στα επόμενα θεωρούμε ότι όλα συμβαίνουν σε ένα χώρο πιθανότητας ( Ω,,P) Modes of convergence: Οι τρόποι σύγκλισης μιας ακολουθίας τ.μ.

Ψηφιακή Επεξεργασία Εικόνας

ORDINAL ARITHMETIC JULIAN J. SCHLÖDER

Example Sheet 3 Solutions

A Note on Intuitionistic Fuzzy. Equivalence Relation

Uniform Convergence of Fourier Series Michael Taylor

Uniform Estimates for Distributions of the Sum of i.i.d. Random Variables with Fat Tail in the Threshold Case

Homework 4.1 Solutions Math 5110/6830

SUPPLEMENT TO ROBUSTNESS, INFINITESIMAL NEIGHBORHOODS, AND MOMENT RESTRICTIONS (Econometrica, Vol. 81, No. 3, May 2013, )

SUPERPOSITION, MEASUREMENT, NORMALIZATION, EXPECTATION VALUES. Reading: QM course packet Ch 5 up to 5.6

C.S. 430 Assignment 6, Sample Solutions

Statistical Inference I Locally most powerful tests

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

Nowhere-zero flows Let be a digraph, Abelian group. A Γ-circulation in is a mapping : such that, where, and : tail in X, head in

Lecture 17: Minimum Variance Unbiased (MVUB) Estimators

SOME PROPERTIES OF FUZZY REAL NUMBERS

2 Composition. Invertible Mappings

SCITECH Volume 13, Issue 2 RESEARCH ORGANISATION Published online: March 29, 2018

Fractional Colorings and Zykov Products of graphs

LAD Estimation for Time Series Models With Finite and Infinite Variance

ANOTHER EXTENSION OF VAN DER CORPUT S INEQUALITY. Gabriel STAN 1

Degenerate Perturbation Theory

Supplement to A theoretical framework for Bayesian nonparametric regression: random series and rates of contraction

A New Class of Analytic p-valent Functions with Negative Coefficients and Fractional Calculus Operators

Math 446 Homework 3 Solutions. (1). (i): Reverse triangle inequality for metrics: Let (X, d) be a metric space and let x, y, z X.

Coefficient Inequalities for a New Subclass of K-uniformly Convex Functions

Econ 2110: Fall 2008 Suggested Solutions to Problem Set 8 questions or comments to Dan Fetter 1

Inverse trigonometric functions & General Solution of Trigonometric Equations

Math221: HW# 1 solutions

Homomorphism of Intuitionistic Fuzzy Groups

Presentation of complex number in Cartesian and polar coordinate system

FREE VIBRATION OF A SINGLE-DEGREE-OF-FREEDOM SYSTEM Revision B

The Simply Typed Lambda Calculus

EE512: Error Control Coding

Bessel function for complex variable

α β

The Neutrix Product of the Distributions r. x λ

4.6 Autoregressive Moving Average Model ARMA(1,1)

Lecture 2: Dirac notation and a review of linear algebra Read Sakurai chapter 1, Baym chatper 3

Homomorphism in Intuitionistic Fuzzy Automata

Απόκριση σε Μοναδιαία Ωστική Δύναμη (Unit Impulse) Απόκριση σε Δυνάμεις Αυθαίρετα Μεταβαλλόμενες με το Χρόνο. Απόστολος Σ.

Factorial. Notations. Specific values. Traditional name. Traditional notation. Mathematica StandardForm notation. Specialized values

THE SECOND ISOMORPHISM THEOREM ON ORDERED SET UNDER ANTIORDERS. Daniel A. Romano

Supplementary Materials: Trading Computation for Communication: Distributed Stochastic Dual Coordinate Ascent

Approximation of distance between locations on earth given by latitude and longitude

Tridiagonal matrices. Gérard MEURANT. October, 2008

k A = [k, k]( )[a 1, a 2 ] = [ka 1,ka 2 ] 4For the division of two intervals of confidence in R +

Matrices and Determinants

INTEGRATION OF THE NORMAL DISTRIBUTION CURVE

1. Introduction and Preliminaries.

Proof of Lemmas Lemma 1 Consider ξ nt = r

Intuitionistic Fuzzy Ideals of Near Rings

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

ST5224: Advanced Statistical Theory II

12. Radon-Nikodym Theorem

On a four-dimensional hyperbolic manifold with finite volume

Fourier Series. MATH 211, Calculus II. J. Robert Buchanan. Spring Department of Mathematics

Research Article Existence of Positive Solutions for m-point Boundary Value Problems on Time Scales

Problem Set 3: Solutions

derivation of the Laplacian from rectangular to spherical coordinates

Certain Sequences Involving Product of k-bessel Function

ESTIMATES FOR WAVELET COEFFICIENTS ON SOME CLASSES OF FUNCTIONS

1. Introduction. Main Result. 1. It is well known from Donsker and Varadhan [1], the LDP for occupation measures. I(ξ k A) (1.

6.1. Dirac Equation. Hamiltonian. Dirac Eq.

1. Matrix Algebra and Linear Economic Models

Solution Series 9. i=1 x i and i=1 x i.

ΔΙΕΡΕΥΝΗΣΗ ΤΗΣ ΣΕΞΟΥΑΛΙΚΗΣ ΔΡΑΣΤΗΡΙΟΤΗΤΑΣ ΤΩΝ ΓΥΝΑΙΚΩΝ ΚΑΤΑ ΤΗ ΔΙΑΡΚΕΙΑ ΤΗΣ ΕΓΚΥΜΟΣΥΝΗΣ ΤΕΧΝΟΛΟΓΙΚΟ ΠΑΝΕΠΙΣΤΗΜΙΟ ΚΥΠΡΟΥ ΣΧΟΛΗ ΕΠΙΣΤΗΜΩΝ ΥΓΕΙΑΣ

Partial Differential Equations in Biology The boundary element method. March 26, 2013

DIRECT PRODUCT AND WREATH PRODUCT OF TRANSFORMATION SEMIGROUPS

F19MC2 Solutions 9 Complex Analysis

Concrete Mathematics Exercises from 30 September 2016

Finite Field Problems: Solutions

3.4 SUM AND DIFFERENCE FORMULAS. NOTE: cos(α+β) cos α + cos β cos(α-β) cos α -cos β

5. Choice under Uncertainty

Fourier Series. constant. The ;east value of T>0 is called the period of f(x). f(x) is well defined and single valued periodic function

Transcript:

Hidawi Publishig Corporatio Joural of Iequalities ad Applicatios Volume 2008, Article ID 598632, 13 pages doi:10.1155/2008/598632 Research Article Fiite-Step Relaxed Hybrid Steepest-Descet Methods for Variatioal Iequalities Ye-Cherg Li Departmet of Occupatioal Safety ad Health, Geeral Educatio Ceter, Chia Medical Uiversity, Taichug 404, Taiwa Correspodece should be addressed to Ye-Cherg Li, ycli@mail.cmu.edu.tw Received 22 August 2007; Revised 2 Jauary 2008; Accepted 13 March 2008 Recommeded by Jog Kim The classical variatioal iequality problem with a Lipschitzia ad strogly mootoe operator o a oempty closed covex subset i a real Hilbert space was studied. A ew fiite-step relaxed hybrid steepest-descet method for this class of variatioal iequalities was itroduced. Strog covergece of this method was established uder suitable assumptios imposed o the algorithm parameters. Copyright q 2008 Ye-Cherg Li. This is a ope access article distributed uder the Creative Commos Attributio Licese, which permits urestricted use, distributio, ad reproductio i ay medium, provided the origial work is properly cited. 1. Itroductio Let H be a real Hilbert space with ier product, ad orm.letc be a oempty closed covex subset of H, ad let F : C H be a operator. The classical variatioal iequality problem: fid u C such that VI F, C F u ),v u 0, v C, 1.1 was iitially studied by Kiderlehrer ad Stampacchia. It is also kow that the VI F, C is equivalet to the fixed-poit equatio u P C u μf u )), 1.2 where P C is the earest poit projectio from H oto C, thatis,p C x argmi y C x y for each x H ad where μ>0 is a arbitrarily fixed costat. If F is strogly mootoe ad Lipschitzia o C ad μ > 0 is small eough, the the mappig determied by the right-had side of this equatio is a cotractio. Hece the Baach cotractio priciple guaratees that the Picard iterates coverge i orm to the uique solutio of the VI F, C. Such a method is called the projectio method. However, Zeg ad Yao 2 poit out that the fixed-poit

2 Joural of Iequalities ad Applicatios equatio ivolves the projectio P C which may ot be easy to compute due to the complexity of the covex set C. To reduce the complexity problem probably caused by the projectio P C,a class of hybrid steepest-descet methods for solvig VI F, C has bee itroduced ad studied recetly by may authors see, e.g., 3, 4. Zeg ad Yao 2 have established the method of two-step relaxed hybrid steepest-descet for variatioal iequalities. A atural arisig problem is whether there exists a geeral relaxed hybrid steepest-descet algorithm that is more tha two steps for fidig approximate solutios of VI F, C or ot. Motivated ad ispired by the recet research work i this directio, we itroduce the followig fiite step relaxed hybrid steepest-descet algorithm for fidig approximate solutios of VI F, C ad aim to uify the covergece results of this kid of methods. Algorithm 1.1. Let {α k } 0, 1, {λ k } 0, 1, fork 1, 2,...,m, ad take fixed umbers 0, 2η/κ 2, k 1, 2,...,m. Startig with arbitrarily chose iitial poits u 0 H, compute the sequeces {u k } such that u 1 u 3 α u α 2 u α 3 u 1 α )[ 2 Tu λ 1 t F T )], 1 α 2 )[ 3 Tu λ 2 1 t 2 F Tu 3 )], 1 α 3 )[ 4 Tu λ 3 1 t 3 F Tu 4 )], 1.3 u m α m u. 1 α m )[ Tu λ m 1 t m F Tu )]. We will prove a strog covergece result for Algorithm 1.1 uder suitable restrictios imposed o the parameters. 2. Prelimiaries The followig lemmas will be used for provig the mai result of the paper i ext sectio. Lemma 2.1 see 5. Let {s } be a sequece of oegative real umbers satisfyig the iequality where {α }, {τ }, ad {γ } satisfy the followig coditios: s 1 1 α ) s α τ γ, 0, 2.1 i {α } 0, 1, 0 α, or equivaletly, 0 1 α 0; ii lim sup τ 0; iii {γ } 0,, 0 γ <. The lim s 0. Lemma 2.2 see 6. Demiclosedess priciple: assume that T is a oexpasive self-mappig o a oempty closed covex subset C of a Hilbert space H. If T has a fixed poit, the I T is demiclosed; that is, wheever {x } is a sequece i C weakly covergig to some x C ad the sequece { I T x } strogly coverges to some y H, it follows that I T x y. Here I is the idetity operator of H.

Ye-Cherg Li 3 The followig lemma is a immediate cosequece of a ier product. Lemma 2.3. I a real Hilbert space H, there holds the iequality x y 2 x 2 2 y, x y, x, y H. 2.2 Lemma 2.4. Let C be a oempty closed covex subset of H. For ay x, y H ad z C, the followig statemets hold: i P C x x, z P C x 0; ii P C x P C y 2 x y 2 P C x x y P C y 2. 3. Covergece theorem Let H be a real Hilbert space ad let C be a oempty closed covex subset of H. Let F : C H be a operator such that for some costats κ, η > 0,F is κ-lipschitzia ad η-strogly mootoe o C;thatis,F satisfies the coditios Fx Fy κ x y, x, y C, Fx Fy, x y η x y 2, x, y C, 3.1 respectively. Sice F is η-strogly mootoe, the variatioal iequality problem VI F, C has a uique solutio u C see, e.g., 7. Assume that T : H H is a oexpasive mappig with the fixed poits set Fix T C. Note that obviously Fix P C C. For ay give umbers λ 0, 1 ad μ 0, 2η/κ 2,we defie the mappig T λ μ : H H by T λ μx : Tx λμf Tx, x H. Lemma 3.1 see 3. Let T λ μ be a cotractio provided that 0 <λ<1 ad 0 <μ<2η/κ 2. Ideed, where τ 1 1 μ 2η μκ 2 0, 1. T λ μx T λ μy 1 λτ x y, x, y H, 3.2 We ow state ad prove the mai result of this paper. Theorem 3.2. Let H be a real Hilbert space ad let C be a oempty closed covex subset of H. Let F : C H be a operator such that for some costats κ, η > 0, Fis κ-lipschitzia ad η- strogly mootoe o C. Assume that T : H H is a oexpasive mappig with the fixed poits set Fix T C, the real sequeces {α k }, {λ k },fork 1, 2,...,m,iAlgorithm 1.1 satisfy the followig coditios: i 1 α k α k 1 <, fork 1, 2,...,m; ii lim α 0 ad lim α k 1, for k 2, 3,...,m;

4 Joural of Iequalities ad Applicatios iii lim λ iv λ 0, lim λ /λ 1 1, 1 λ ; max{λ k : k 2, 3,...,m}, for all 1. The the sequeces {u k } geerated by Algorithm 1.1 coverge strogly to u which is the uique solutio of the VI F, C. Proof. Sice F is η-strogly mootoe, by 7,theVI F, C has the uique solutio u C.Next we divide the rest of the proof ito several steps. Step 1. Let {u k } is bouded for each k 1, 2,...,m. Ideed, let us deote that Tt λu Tu λtf Tu,thewehave u 1 u α α u u α u α u 1 α λ 1 u 1 α u 1 α ) T λ u 1 )[ T λ 1 u T λ 1 u T λ 1u u ] u 1 α )[ 1 λ 1 τ ) u λ 1 t F u ) ], 3.3 where τ 1 1 2η κ 2 0, 1. Moreover, we also have u α 2 u α 2 u α 2 u α 2 u 1 α 2 λ 2 1 t 2 u 1 α 2 u 3 u )[ T λ 2 1 t 2 u 3 T λ 2 1 t 2 u T λ 2 1u u ] t 2 u 2 )[ 2 1 α 1 λ 1 τ 2 ) 3 u u 2 λ ) 2 λ u 2 ) 1 α u 3 u 1 α 2 1 t 2 F u ) ] 1 t 2 F u ), 3.4 where τ 2 1 1 t 2 2η t 2 κ 2 0, 1, adfork 2, 3,...,m 1, u k u α k u α k α k α k u u u 1 α k λ k 1 u 1 α k u k )[ T λ k u 1 u k T λ k 1 u T λ k 1u u ] u 1 α k )[ k 1 λ 1 τ k ) u k u λ k ) k λ u 1 α k ) u k u 1 α k 1 t k F u ) ] 1 t k F u ), 3.5

Ye-Cherg Li 5 where τ k 1 1 2η κ 2 0, 1, ad u m u α m u α m α m u u u 1 α m λ m 1u t m u u 1 α m )[ T λ m 1u t m T λ m 1u T λ m 1u u ] t m t m u 1 α m )[ m 1 λ 1 τ m ) u u λ m 1 t m F u ) ] u λ m 1 t m F u ), where τ m 1 1 t m 2η t m κ 2 0, 1. Thus we obtai m 1 u u α m 1 u u m 1 ) 1 α u m α m 1 u u u m 1 λ 1 t m 1 F u ) u m 1 )[ 1 α u u m λ 1 t m F u ) ] λ m 1 ) { m } max λ t m t m 1 ) F u ), u 1 α m 1 1,λ m 1 1 3.6 1 t m 1 F u ) ] u k u u u 1 α k ) { j } ) m max λ t j F u ), k j m 1 j k 3.7 for k 2, 3,...,m 1. I particular, u u u 1 α 2 ) { j } ) m max λ t j F u ). 2 j m Hece, substitutig 3.8 i 3.3 ad by coditio iv,weobtai u 1 u α u u 1 α )[ 1 λ 1 τ ) 2 u λ 1 t F u ) ] α u α u u ) [ 1 α 1 λ 1 τ ) u u m max 2 j m {λ j 1 } u 1 α ) [ 1 λ 1 τ ) u 1 t j F u ) λ u { j max λ 1 j m 1 1 t F ] u ) } m t j F ] u ). j 1 3.8 3.9 By iductio, it is easy to see that u u M, 0, 3.10

6 Joural of Iequalities ad Applicatios where M max{ u 0 u, m j 1 t j /τ F u }. Ideed, for 0, from 3.9 we obtai u 1 u α 0 u 0 u 1 α ) [ 0 1 λ 1 τ ) u 0 u max 1 j m λ j 1 m ) t j F ] u ) j 1 α 0 M 1 α )[ 0 1 λ 1 τ ) M λ 1 τ M ] M. 3.11 Suppose that u u 1 u α u M, for 1. We wat to claim that u 1 u M. Ideed, u u 1 α ) [ 1 λ 1 τ ) u u λ 1 m t j F ] u ) j 1 α M 1 α )[ 1 λ 1 τ ) M λ 1 τ M ] M. 3.12 Therefore, we have u u M, for all 0, ad u m u M λ m 1 τ M τ M, for all 0. I this case, from 3.8, it follows that u k u { j M max λ 1} τ M 1 τ ) M, 0, k 2, 3,...,m 1. 3.13 k j m Step 2. Let u 1 Tu 0,. Ideed by Step 1, {u k } is bouded for 1 k m ad so are {Tu k } ad {F Tu k } for 1 k m. Thus from the coditios that lim α 0, lim α k 1, for k 2, 3,...,mad lim λ 0, we have, for k 2,...,m, u k u α k u 1 α k ) k Tu λ k 1 t k F Tu k )) u 1 α k ) u 1 α k ) k Tu λ k 1 t k F 3.14 Tu k )) ad so u 1 α k 1 Tu α u ) u k ) 1 α Tu k 1 α ) 2 Tu λ 1 t F T k λ 1 t k k F Tu )) Tu ) ) 0 α u Tu ) ) 2 1 α Tu Tu ) 1 λ 1 t F T )) α u Tu ) 1 α T Tu ) 1 α λ 1 t 2 ) F Tu α u Tu u λ 1 t F T ) 0 as. 3.15

Ye-Cherg Li 7 Step 3. Let u u m u m 1 1 u 0, as. Ideed, we observe that m α u α m 1 u 1 1 α m ) λ T m 1 α m u u α m α m 1 1 u t m u 1 1 α m 1 λ m u t m 1 1 α m ) T λ m 1u t m T λ m 1u t m 1 1 α m ) λ T m 1u t m 1 1 α m ) λ 1 T m u t m 1 α m u u 1 α m α m 1 u m ) m 1 1 α 1 λ 1 τ m ) u u 1 α m α m 1 Tu 1 1 α m ) m λ 1 1 α m ) m 1 λ t m F Tu ) 1 1 1 α m ) m λ 1 τ m ) u u 1 1 α m ) m λ 1 1 α m ) m 1 λ t m ) F Tu 1 α m α m 1 u 1 Tu ) 1 u α m u 1 1 α m ) m λ 1 1 α m ) m 1 λ t m F Tu ) 1 α m u Tu ), 1 1 1 3.16 ad, for 2 k m 1, k u u k 1 k α u α k 1 u 1 1 α k α k u u α k α k 1 1 λ k 1 u k u 1 1 α k ) λ 1 T k 1 α k ) T λ k u k 1 1 u k T λ k 1 u k 1 1 α k ) λ T k 1u k 1 1 α k ) λ 1 T k u k 1 α k u u 1 α k α k 1 u k ) k 1 1 α 1 λ 1 τ k ) k u u k 1 α k α k k 1 Tu 1 1 α k ) k λ 1 1 α k ) k 1 λ F Tu k ) 1 α k u u α k α k u Tu k ) 1 1 1 1 α k ) k 1 λ 1 τ k ) u k u k 1 1 α k ) k λ 1 1 α k ) k 1 λ F Tu k ) 1, 3.17 1

8 Joural of Iequalities ad Applicatios m u u m u u α m α m u u m 1 1 1 1 Tu ) 1 1 α m ) m λ 1 1 α m ) m 1 λ t m F Tu 1), u m 1 u u 1 α m ) m λ 1 2 u 1 1 1 α m 1 ) m 1 λ 1 α m 1 α m... 1 α m 1 1 1 1 1 α m 1 α m 1 u Tu m ) 1 1 ) m λ t m FTu ) m 1 λ t m 1 m FTu 1 α m u Tu ), 1 1 1 u u m 1 1 α k ) k λ 1 k 2 1 1 α k 1 1 1 ) k λ t k k FTu 1 3.18 m 1 k α α k u Tu k ) m α α m u Tu ). k 2 1 Hece it follows from the above iequalities 3.17 3.19 that u 1 u α u α 1 u 1 1 α λ 1 1 1 1 α ) λ 1 T 1 α u u α α u ) 1 α T λ 1 1 1 1 1 T λ 1 1 1 1 3.19 1 α α α u λ 1 1 1 α 1 u α α 1 1 λ 1 u 1 1 α ) 1 λ 1 τ ) 1 α 2 1 Tu 1 1 α ) λ 1 1 α ) 1 λ F T ) 1 α u u α α k u Tu k ) 1 1 1 1 α ) 1 λ 1 τ ) 1 1 1 α ) λ 1 1 α ) 1 λ F T 1). 3.20

Ye-Cherg Li 9 Let us substitute 3.19 ito 3.20,thewehave u 1 u α 1 α ) 1 λ 1 τ )) u u 1 m 1 α k k 1 α k u Tu k ) 1 1 1 m 1 1 α k ) k λ 1 1 α k ) k 1 λ k 1 k FTu 1 1 α m ) m λ 1 1 α m ) m 1 λ t m FTu 1 α m α m 1 u 1 Tu ) 1 1 1 α ) λ 1 τ ) u u 1 1 α ) λ 1 τ ν δ, 3.21 where m 1 δ k α α k u Tu k ) m α α m u Tu ), k 1 1 1 1 1 1 1 1 m ) m ν 1 α 1 α λ 1 τ λ 1 1 α m ) m 1 λ t m FTu 1 m 1 1 α k ) k λ 1 1 α k ) k 1 λ ) t k k FTu 1. k 1 3.22 We put ξ sup { } { u : 0 sup Tu k } : 0, k 1, 2,...,m sup { FTu k : 0, k 1, 2,...,m }, M ) m u F u ) ξ. k 1 3.23 The δ 2M m k 1 α k α k 1 0, as, ad m ν t )M k k 1 1 α 1 ) λ 1 τ m 1 α k ) k λ 1 1 α k ) k 1 λ ) 1 α λ 1 ) 1 α ) 1 λ. k 2 3.24

10 Joural of Iequalities ad Applicatios From ii iv,weobtaiν 0as. Furthermore, from i, 1 δ <. By Lemma 2.1, we deduce that u 0as. Step 4. Let u 1 u Tu 0as.FromSteps2 ad 3,wehave u Tu u 1 u u 1 Tu 0 3.25 as. Step 5. Let lim sup F u,tu k of {Tu } such that u 0, for k 2, 3,...,m.Let{Tu i } be a subsequece lim sup F u ),Tu u lim F u ),Tu i u. 3.26 i Without loss of geerality, we assume that Tu i ũ weakly for some ũ H. ByStep 4, we derive u i ũ weakly. But by Lemma 2.2 ad Step 4, wehaveũ Fix T C. Sice u is the uique solutio of the VI F, C, we obtai lim sup F u ),Tu u F u ), ũ u 0. 3.27 From the proof of Step 2, Tu k Tu u 0, as, 3.28 for k 2, 3,...,m.The lim sup F u ),Tu k u [ lim sup F u ),Tu k F u ),Tu k lim sup lim sup 0, Tu F u ),Tu u ] Tu F u ),Tu u lim sup F u ),Tu u 3.29 for k 2, 3,...,m.

Ye-Cherg Li 11 Step 6. Let u u i orm ad so does {u k } for k 2, 3,...,m. Ideed usig Lemma 2.3 ad 3.7 we get u 1 u 2 α u α u α u α α u u u ) 1 α u 2 1 α u 2 1 α u 2 1 α λ 1 ) T λ 1 ) T λ 1 )[ T λ 1 u ) 2 u 2 T λ 1 T λ 1 u ) T λ 1u u ) 2 u 2 2 T λ 1 u 2 1 α ) 1 λ 1 τ ) 2 u 2 2 λ 1 F u ),T λ 1 t F T ) u α u u 2 [ 1 α ) 1 λ 1 τ ) 2 u u 1 α 2 2 1 α ) λ 1 t F u ),T λ 1 t F T ) u α 1 α ) 1 λ 1 τ )) u u 2 2 1 α 1 α ) 2 ) 1 α 1 λ ) 1 α 2 1 τ ) 2 max 2 j m {λ j 1 } m ) 2 1 λ 1 τ ) 2 2 λ 1 F u ),T u λ 1 t F T ) 1 1 α ) λ 1 α 1 τ ) u ) 1 α 2 u 2 2 1 α ) 2 1 λ 1 τ ) 2 2λ 1 t F u ),T u λ 1 t F T ) 1 1 α ) λ 1 τ ) u u u,t λ 1 ) m max u ] { j } ) m F λ t j u ) 1 t j ) F u ) u ) 2 ) m 2 max 2 j m {λ j 1 } t j F u ) 2 ) 2 ) 1 α 1 λ 1 τ ) 2 max 2 j m ) 2 ) m 2 max 2 j m {λ j 1 } t j M 2 u 2 1 α ) λ 1 τ u { j } ) m λ t j 1 M 2 ] 2

12 Joural of Iequalities ad Applicatios [ 2t F u ),T u λ 1 t F T ) τ 1 α ) 2 1 α 2 ) 1 λ 1 τ ) 2 max2 j m {λ j 1 } m t j ) M 2 τ λ 1 2 ) 2 1 α 1 λ 1 τ ) 2 { j }) 2 max2 j m λ m 1 t j ) 2 M 2 ] τ λ 1 1 1 α ) λ 1 τ ) u u 2 1 α ) λ 1 τ [ 2t F u ),T u λ 1 t F T ) τ 1 α ) 2 2 ) 1 α τ 1 λ 1 ) ) 2 ) 2 m 2 1 α τ 1 λ 1 τ ) 2 max 2 j m {λ j 1 } t j M ]. 2 From ii, iii, adstep 5, wehavelim α lim α k by Lemma 2.4, we coclude that lim sup lim λ k 1, for k 2,...,m, lim sup F u,t 2t F u ),T lim sup 0 0 0. u λ 1 t F T τ 1 α 2 ) 2 1 α 1 λ 1 τ ) 2 max2 j m {λ j 1 }) m 2 τ 1 α τ ) F u ),T Cosequetly from Lemma 2.1, weobtai u u 0, for k 2, 3,...,m,that u k ) u lim sup 1 τ ) 2 ) m t j M 2 3.30 0, for k 1, 2,...,m ad } is bouded; u 0, ad {F T ) 2 1 α 2 1 λ 1 τ ) 2 m t j ) M 2 τ t j ) 2 M 2 ) 2 ) 2 λ 1 τ 1 α ) F u ), F T ) 3.31 u 0 ad hece it follows from u k u 0, for k 2, 3,...,m. Ackowledgmet This research was partially supported by Grat o. NSC95-2115-M-039-001- from the Natioal Sciece Coucil of Taiwa. Refereces D. Kiderlehrer ad G. Stampacchia, A Itroductio to Variatioal Iequalities ad Their Applicatios, vol. 88 of Pure ad Applied Mathematics, Academic Press, New York, NY, USA, 1980.

Ye-Cherg Li 13 2 L. C. Zeg ad J.-C. Yao, Two step relaxed hybrid steepest-descet methods for variatioal iequalities, to appear i Applied Mathematics ad Mechaics. 3 I. Yamada, The hybrid steepest-descet method for the variatioal iequality problem over the itersectio of fixed poit sets of oexpasive mappigs, i Iheretly Parallel Algorithms i Feasibility ad Optimizatio ad Their Applicatios, D. Butariu, Y. Cesor, ad S. Reich, Eds., vol. 8 of Studies i Computatioal Mathematics, pp. 473 504, North-Hollad, Amsterdam, The Netherlads, 2001. 4 L. C. Zeg, N. C. Wog, ad J.-C. Yao, Covergece aalysis of modified hybrid steepest-descet methods with variable parameters for variatioal iequalities, Joural of Optimizatio Theory ad Applicatios, vol. 132, o. 1, pp. 51 69, 2007. 5 H.-K. Xu, Iterative algorithms for oliear operators, Joural of the Lodo Mathematical Society, vol. 66, o. 1, pp. 240 256, 2002. 6 K. Goebel ad W. A. Kirk, Topics i Metric Fixed Poit Theory, vol.28ofcambridge Studies i Advaced Mathematics, Cambridge Uiversity Press, Cambridge, UK, 1990. 7 J.-C. Yao, Variatioal iequalities with geeralized mootoe operators, Mathematics of Operatios Research, vol. 19, o. 3, pp. 691 705, 1994.