nach oben

Journal of Inequalities and Applications

Erschienen in:

Open Access 01.12.2013 | Research

Strong convergence of a general iterative algorithm in Hilbert spaces

verfasst von: Songtao Lv

Erschienen in: Journal of Inequalities and Applications | Ausgabe 1/2013

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config

KI-gestützte Suche

Patentsuche

Aus

Abstract

In this paper, the problem of approximating a common element in the common fixed point set of an infinite family of nonexpansive mappings, in the solution set of a variational inequality involving an inverse-strongly monotone mapping and in the solution set of an equilibrium problem is investigated based on a general iterative algorithm. Strong convergence of the iterative algorithm is obtained in the framework of Hilbert spaces. The results obtained in this paper improve the corresponding results announced by many authors.

AMS Subject Classification:47H09, 47J05, 47J25.

Competing interests

The author declares that they have no competing interests.

1 Introduction and preliminaries

Let H be a real Hilbert space, whose inner product and norm are denoted by

〈 \cdot, \cdot 〉

and

∥ \cdot ∥

respectively. Let C be a nonempty, closed and convex subset of H and

T : C \to C

be a mapping. In this paper, we use

F (T)

to denote the set of fixed points of T. Recall that T is said to be a κ-contraction iff there exists a constant

κ \in (0, 1)

such that

∥ T x - T y ∥ \leq κ ∥ x - y ∥, \forall x, y \in C .

T is said to be a nonexpansive mapping iff

∥ T x - T y ∥ \leq ∥ x - y ∥, \forall x, y \in C .

Let

B : C \to H

be a mapping. Recall that B is said to be an α-inverse-strongly monotone iff there exits a positive constant α such that

〈 B x - B y, x - y 〉 \geq α {∥ B x - B y ∥}^{2}, \forall x, y \in C .

The classical variational inequality is to find

u \in C

such that

〈 B u, v - u 〉 \geq 0, \forall v \in C .

(1.1)

In this paper, we use

V I (C, B)

to denote the solution set of the variational inequality.

Let

P_{C}

be the metric projection from H onto C. It is also known that

P_{C}

satisfies

〈 x - y, P_{C} x - P_{C} y 〉 \geq {∥ P_{C} x - P_{C} y ∥}^{2}, \forall x, y \in H .

Moreover,

P_{C} x

is characterized by the properties

P_{C} x \in C

and

〈 x - P_{C} x, P_{C} x - y 〉 \geq 0

for all

y \in C

. One can see that the variational inequality is equivalent to a fixed point problem. The element

u \in C

is a solution of the variational inequality if and only if u is a fixed point of the mapping

P_{C} (I - λ B)

, where

λ > 0

is a constant and I is the identity mapping. This alternative equivalent formulation has played a significant role in the studies of the variational inequality and related optimization problems.

Recall that an operator A is strongly positive on H iff there exists a constant

\bar{γ} > 0

with the property

〈 A x, x 〉 \geq \bar{γ} {∥ x ∥}^{2}, \forall x \in H .

Recall that a set-valued mapping

S : H \to 2^{H}

is said to be monotone if for all

x, y \in H

f \in S x

and

g \in S y

imply

〈 x - y, f - g 〉 \geq 0

. A monotone mapping

S : H \to 2^{H}

is maximal if the graph of

Graph (S)

of S is not properly contained in the graph of any other monotone mapping. It is known that a monotone mapping S is maximal iff for

(x, f) \in H \times H

〈 x - y, f - g 〉 \geq 0

for every

(y, g) \in Graph (S)

implies

f \in S x

. Let B be a monotone map of C into H and let

N_{C} v

be the normal cone to C at

v \in C

, i.e.,

N_{C} v = {w \in H : 〈 v - u, w 〉 \geq 0, \forall u \in C}

and define

S v = {\begin{matrix} B v + N_{C} v, & v \in C, \\ \emptyset, & v \notin C . \end{matrix}

Then S is maximal monotone and

0 \in S v

iff

v \in V I (C, B)

; see [1] and the references therein.

Let F be a bifunction of

C \times C

into ℝ, where ℝ is the set of real numbers. The equilibrium problem for

F : C \times C \to R

is to find

x \in C

such that

F (x, y) \geq 0, \forall y \in C .

(1.2)

The set of solutions of the problem (1.2) is denoted by

E P (F)

. Numerous problems in physics, optimization and economics reduce to finding a solution of (1.2). Recently, many iterative algorithms have been studied to solve the equilibrium problem (1.2); see, for instance, [2‐19].

For solving the equilibrium problem (1.2), let us assume that F satisfies the following conditions:

(A1)

F (x, x) = 0

for all

x \in C

;

(A2) F is monotone, i.e.,

F (x, y) + F (y, x) \leq 0

for all

x, y \in C

;

(A3) for each

x, y, z \in C

\underset{t ↓ 0}{lim sup} F (t z + (1 - t) x, y) \leq F (x, y);

(A4) for each

x \in C

y \mapsto F (x, y)

is convex and lower semicontinuous.

In 2007, Takahashi and Takahashi [17] proved the following result.

Theorem TT Let C be a nonempty closed convex subset of H. Let F be a bifunction from

C \times C

to R satisfying (A1)-(A4) and let T be a nonexpansive mapping of C into H such that

F (S) \cap E P (F) \neq \emptyset

. Let f be a contraction of H into itself and let

{x_{n}}

and

{u_{n}}

be sequences generated by

x_{1} \in H

and

{\begin{matrix} F (y_{n}, u) + \frac{1}{r_{n}} 〈 u - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall u \in C, \\ x_{n + 1} = α_{n} f (x_{n}) + (1 - α_{n}) T y_{n}, n \geq 0, \end{matrix}

where

{α_{n}} \subset [0, 1]

and

{r_{n}} \subset (0, \infty)

satisfy

{lim}_{n \to \infty} α_{n} = 0

\sum_{n = 1}^{\infty} α_{n} = \infty

\sum_{n = 1}^{\infty} | α_{n + 1} - α_{n} | < \infty

\sum_{n = 1}^{\infty} | r_{n + 1} - r_{n} | < \infty

, and

{lim inf}_{n \to \infty} r_{n} > 0

. Then

{x_{n}}

and

{y_{n}}

strongly converge to some point z, where

z = P_{C} F (T) \cap E P (T) f (z)

Recently, Plubtieng and Punpaeng [19] further improved the above results by involving a strongly positive self-adjoint operator. To be more precise, they proved the following results.

Theorem PP Let H be a real Hilbert space, let F be a bifunction from

H \times H \to R

satisfying (A1)-(A4) and let T be a nonexpansive mapping on H such that

F (T) \cap E P (F) \neq \emptyset

. Let f be a contraction of H into itself with

α \in (0, 1)

and let A be a strongly positive bounded linear operator on H with the coefficient

\bar{γ} > 0

and

0 < γ < \frac{\bar{γ}}{α}

. Let

{x_{n}}

be a sequence generated by

x_{1} \in H

and

{\begin{matrix} F (y_{n}, u) + \frac{1}{r_{n}} 〈 u - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall u \in C, \\ x_{n + 1} = α_{n} γ f (x_{n}) + (I - α_{n} A) T y_{n}, n \geq 1, \end{matrix}

where

{α_{n}} \subset [0, 1]

and

{r_{n}} \subset (0, \infty)

satisfy

{lim}_{n \to \infty} α_{n} = 0

\sum_{n = 1}^{\infty} α_{n} = \infty

\sum_{n = 1}^{\infty} | α_{n + 1} - α_{n} | < \infty

\sum_{n = 1}^{\infty} | r_{n + 1} - r_{n} | < \infty

, and

{lim inf}_{n \to \infty} r_{n} > 0

. Then

{x_{n}}

and

{y_{n}}

strongly converge to some point z, where

z = P_{F (T) \cap E P (T)} (I - A + γ f) (z)

In 2008, Su, Shang and Qin [2] considered the variational inequality (1.1), and the equilibrium problem (1.2) based on a composite iterative algorithm and proved the following theorem.

Theorem SSQ Let C be a nonempty closed convex subset of H. Let F be a bifunction from

C \times C

to R satisfying (A1)-(A4). Let A be α-inverse-strongly monotone and let T be a nonexpansive mapping of C into H such that

F (S) \cap E P (F) \cap V I (C, A) \neq \emptyset

. Let f be a contraction of H into itself and let

{x_{n}}

and

{u_{n}}

be sequences generated by

x_{1} \in H

and

{\begin{matrix} F (y_{n}, u) + \frac{1}{r_{n}} 〈 u - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall u \in C, \\ x_{n + 1} = α_{n} f (x_{n}) + (1 - α_{n}) T P_{C} (I - λ_{n} A) y_{n}, n \geq 0, \end{matrix}

where

{λ_{n}} \subset [a, b]

, where

0 < a < b < 2 α

{α_{n}} \subset [0, 1]

and

{r_{n}} \subset (0, \infty)

satisfy

{lim}_{n \to \infty} α_{n} = 0

\sum_{n = 1}^{\infty} α_{n} = \infty

\sum_{n = 1}^{\infty} | α_{n + 1} - α_{n} | < \infty

\sum_{n = 1}^{\infty} | r_{n + 1} - r_{n} | < \infty

\sum_{n = 1}^{\infty} | λ_{n + 1} - λ_{n} | < \infty

, and

{lim inf}_{n \to \infty} r_{n} > 0

. Then

{x_{n}}

and

{y_{n}}

strongly converge to some point z, where

z = P_{C} F (T) \cap E P (T) f (z)

The above results only involve a single mapping, we will consider an infinite family of mappings in this paper. To be more precise, we study the mapping

W_{n}

defined by

\begin{array}{l} U_{n, n + 1} = I, \\ U_{n, n} = γ_{n} T_{n} U_{n, n + 1} + (1 - γ_{n}) I, \\ U_{n, n - 1} = γ_{n - 1} T_{n - 1} U_{n, n} + (1 - γ_{n - 1}) I, \\ ⋮ \\ U_{n, k} = γ_{k} T_{k} U_{n, k + 1} + (1 - γ_{k}) I, \\ u_{n, k - 1} = γ_{k - 1} T_{k - 1} U_{n, k} + (1 - γ_{k - 1}) I, \\ ⋮ \\ U_{n, 2} = γ_{2} T_{2} U_{u, 3} + (1 - γ_{2}) I, \\ W_{n} = U_{n, 1} = γ_{1} T_{1} U_{n, 2} + (1 - γ_{1}) I, \end{array}

(1.3)

where

{γ_{1}}, {γ_{2}}, \dots

are real numbers such that

0 \leq γ_{n} \leq 1

T_{1}, T_{2}, \dots

are an infinite family of mappings of C into itself.

Considering

W_{n}

, we have the following lemmas which are important in proving our main results.

Lemma 1.1 [20]

Let C be a nonempty closed convex subset of a strictly convex Banach space E. Let

T_{1}, T_{2}, \dots

be nonexpansive mappings of C into itself such that

⋂_{n = 1}^{\infty} F (T_{n})

is nonempty, and let

γ_{1}, γ_{2}, \dots

be real numbers such that

0 < γ_{n} \leq b < 1

for any

n \geq 1

. Then, for every

x \in C

and

k \in N

, the limit

{lim}_{n \to \infty} U_{n, k} x

exists.

Using Lemma 1.1, one can define the mapping W of C into itself as follows:

W x = lim_{n \to \infty} W_{n} x = lim_{n \to \infty} U_{n, 1} x, \forall x \in C .

Such a W is called the W-mapping generated by

T_{1}, T_{2}, \dots

and

γ_{1}, γ_{2}, \dots

. Throughout this paper, we will assume that

0 < γ_{n} \leq b < 1

, where b is some constant.

Lemma 1.2 [20]

Let C be a nonempty closed convex subset of a strictly convex Banach space E. Let

T_{1}, T_{2}, \dots

be nonexpansive mappings of C into itself such that

⋂_{n = 1}^{\infty} F (T_{n})

is nonempty, and let

γ_{1}, γ_{2}, \dots

be real numbers such that

0 < γ_{n} \leq b < 1

for any

n \geq 1

. Then

F (W) = ⋂_{n = 1}^{\infty} F (T_{n})

In this paper, based on a general iterative algorithm, we study the problem of approximating a common element in the common fixed point set of an infinite family of nonexpansive mappings, in the solution set of a variational inequality involving an inverse-strongly monotone mapping and in the solution set of an equilibrium problem. Strong convergence of the iterative algorithm is obtained in the framework of Hilbert spaces.

In order to obtain the strong convergence, we need the following tools.

Lemma 1.3 In Hilbert spaces, the following inequality holds:

{∥ x + y ∥}^{2} \leq {∥ x ∥}^{2} + 2 〈 y, x + y 〉, \forall x, y \in H .

Lemma 1.4 [21]

Assume that

{α_{n}}

is a sequence of nonnegative real numbers such that

α_{n + 1} \leq (1 - γ_{n}) α_{n} + δ_{n},

where

{γ_{n}}

is a sequence in

(0, 1)

and

{δ_{n}}

is a sequence such that

(i)

\sum_{n = 1}^{\infty} γ_{n} = \infty

;

(ii)

{lim sup}_{n \to \infty} δ_{n} / γ_{n} \leq 0

\sum_{n = 1}^{\infty} | δ_{n} | < \infty

Then

{lim}_{n \to \infty} α_{n} = 0

Lemma 1.5 [22]

Assume B is a strong positive linear bounded operator on a Hilbert space H with the coefficient

\bar{γ} > 0

and

0 < ρ \leq {∥ B ∥}^{- 1}

. Then

∥ I - ρ B ∥ \leq 1 - ρ \bar{γ}

Lemma 1.6 [22]

Let H be a Hilbert space. Let B be a strongly positive linear bounded self-adjoint operator with the constant

\bar{γ} > 0

and f be a contraction with the constant κ. Assume that

0 < γ < \bar{γ} / κ

. Let T be a nonexpansive mapping with a fixed point

x_{t} \in H

of the contraction

x \mapsto t γ f (x) + (I - t B) T x

. Then

{x_{t}}

converges strongly as

t \to 0

to a fixed point

\bar{x}

of T, which solves the variational inequality

〈 (A - γ f) \bar{x}, z - \bar{x} 〉 \leq 0, \forall z \in F (T) .

Equivalently, we have

P_{F (T)} (I - A + γ f) \bar{x} = \bar{x}

Lemma 1.7 [23, 24]

Let C be a nonempty closed convex subset of H and let B be a bifunction of

C \times C

into ℝ satisfying (A1)-(A4). Let

r > 0

and

x \in H

. Then there exists

z \in C

such that

F (z, y) + \frac{1}{r} 〈 y - z, z - x 〉 \geq 0, \forall y \in C .

Define a mapping

T_{r} : H \to C

as follows:

T_{r} (x) = {z \in C : F (z, y) + \frac{1}{r} 〈 y - z, z - x 〉 \geq 0, \forall y \in C} .

Then the following hold:

(1)

T_{r}

is single-valued;

(2)

T_{r}

is firmly nonexpansive, i.e., for any

x, y \in H

{∥ T_{r} x - T_{r} y ∥}^{2} \leq 〈 T_{r} x - T_{r} y, x - y 〉;

(3)

F (T_{r}) = E P (F)

;

(4)

E P (F)

is closed and convex.

Lemma 1.8 [25]

Let

{x_{n}}

and

{y_{n}}

be bounded sequences in a Banach space X and let

β_{n}

be a sequence in

[0, 1]

with

0 < {lim inf}_{n \to \infty} β_{n} \leq {lim sup}_{n \to \infty} β_{n} < 1

. Suppose

x_{n + 1} = (1 - β_{n}) y_{n} + β_{n} x_{n}

for all integers

n \geq 0

and

\underset{n \to \infty}{lim sup} (∥ y_{n + 1} - y_{n} ∥ - ∥ x_{n + 1} - x_{n} ∥) \leq 0 .

Then

{lim}_{n \to \infty} ∥ y_{n} - x_{n} ∥ = 0

Lemma 1.9 [26, 27]

Let K be a nonempty closed convex subset of a Hilbert space H,

{T_{i} : C \to C}

be a family of infinitely nonexpansive mappings with

⋂_{i = 1}^{\infty} F (T_{i})

{γ_{n}}

be a real sequence such that

0 < γ_{n} \leq b < 1

for each

n \geq 1

. If C is any bounded subset of K, then

{lim}_{n \to \infty} {sup}_{x \in C} ∥ W x - W_{n} x ∥ = 0

2 Main results

Theorem 2.1 Let C be a nonempty closed convex subset of a Hilbert space H. Let F be a bifunction from

C \times C

to ℝ which satisfies (A1)-(A4). Let

{T_{n}}_{n = 1}^{\infty}

be an infinite family of nonexpansive mappings of C into C. Let

B : C \to H

be an α-inverse-strongly monotone mapping. Let A be a strongly positive linear bounded self-adjoint operator on H with the coefficient

\bar{γ} > 0

. Assume that

0 < γ < \bar{γ} / κ

and

F : = ⋂_{i = 1}^{\infty} F (T_{i}) \cap E P (F) \cap V I (C, B) \neq \emptyset

. Let

f : C \to H

be a κ-contraction. Let

{x_{n}}

be a sequence generated in the following iterative process:

{\begin{matrix} x_{1} \in H, \\ F (y_{n}, z) + \frac{1}{r_{n}} 〈 z - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall z \in C, \\ x_{n + 1} = α_{n} x_{n} + (1 - α_{n}) β_{n} γ f (y_{n}) + (1 - α_{n}) (I - β_{n} A) W_{n} P_{C} (I - s_{n} B) y_{n}, n \geq 1, \end{matrix}

where

W_{n}

is generated in (1.3),

{α_{n}}

{β_{n}}

are real number sequences in

(0, 1)

{r_{n}}

and

{s_{n}}

are positive real number sequences. Assume that the following restrictions are satisfied:

(a)

0 < {lim inf}_{n \to \infty} α_{n} \leq {lim sup}_{n \to \infty} α_{n} < 1

;

(b)

{lim}_{n \to \infty} β_{n} = 0

\sum_{n = 1}^{\infty} β_{n} = \infty

;

(c)

{lim}_{n \to \infty} | r_{n + 1} - r_{n} | = 0

{lim}_{n \to \infty} | s_{n + 1} - s_{n} | = 0

;

(d)

{lim inf}_{n \to \infty} r_{n} > 0

{s_{n}} \subset [s, s^{'}]

for some s,

s^{'}

with

0 < s < s^{'} < 2 α

Then

{x_{n}}

converges strongly to

q \in F

, where

q = P_{F} (γ f + (I - A)) (q)

, which solves the following variational inequality:

〈 γ f (q) - A q, p - q 〉 \leq 0, \forall p \in F .

Proof We divide the proof into five steps.

Step 1. Show that the sequence

{x_{n}}

is bounded.

Notice that

I - s_{n} B

is nonexpansive. Indeed, we see from the restriction (d) that

\begin{matrix} {∥ (I - s_{n} B) x - (I - s_{n} B) y ∥}^{2} \\ = {∥ (x - y) - s_{n} (B x - B y) ∥}^{2} \\ = {∥ x - y ∥}^{2} - 2 s_{n} 〈 x - y, B x - B y 〉 + s_{n}^{2} {∥ B x - B y ∥}^{2} \\ \leq {∥ x - y ∥}^{2} - s_{n} (2 α - s_{n}) {∥ B x - B y ∥}^{2} \\ \leq {∥ x - y ∥}^{2}, \end{matrix}

which implies the mapping

I - s_{n} B

is nonexpansive. Fix

p \in F

. Since

z_{n} = T_{r_{n}} x_{n}

, we have

∥ y_{n} - p ∥ = ∥ T_{r_{n}} x_{n} - T_{r_{n}} p ∥ \leq ∥ x_{n} - p ∥ .

Put

ζ_{n} = β_{n} γ f (y_{n}) + (I - β_{n} A) W_{n} ρ_{n},

where

ρ_{n} = P_{C} (I - s_{n} B) y_{n} .

It follows that

∥ ρ_{n} - p ∥ \leq ∥ y_{n} - p ∥ \leq ∥ x_{n} - p ∥ .

Since

β_{n} \to 0

n \to \infty

, we may assume, with no loss of generality, that

β_{n} < {∥ A ∥}^{- 1}

for all n. It follows that

\begin{array}{rcl} ∥ ζ_{n} - p ∥ & = & ∥ β_{n} (γ f (y_{n}) - A p) + (I - β_{n} A) (W_{n} ρ_{n} - p) ∥ \\ \leq & β_{n} ∥ γ f (y_{n}) - A p ∥ + ∥ I - β_{n} A ∥ ∥ W_{n} ρ_{n} - p ∥ \\ \leq & β_{n} [γ ∥ f (y_{n}) - f (p) ∥ + ∥ γ f (p) - A p ∥] + (1 - β_{n} \bar{γ}) ∥ ρ_{n} - p ∥ \\ \leq & β_{n} [γ ∥ f (y_{n}) - f (p) ∥ + ∥ γ f (p) - A p ∥] + (1 - β_{n} \bar{γ}) ∥ x_{n} - p ∥ \\ \leq & [1 - (\bar{γ} - γ κ) β_{n}] ∥ x_{n} - p ∥ + β_{n} ∥ γ f (p) - A p ∥, \end{array}

which yields

\begin{array}{rcl} ∥ x_{n + 1} - p ∥ & = & ∥ α_{n} (x_{n} - p) + (1 - α_{n}) (ζ_{n} - p) ∥ \\ \leq & α_{n} ∥ x_{n} - p ∥ + (1 - α_{n}) ∥ ζ_{n} - p ∥ \\ \leq & α_{n} ∥ x_{n} - p ∥ + (1 - α_{n}) [1 - (\bar{γ} - γ α) β_{n}] ∥ x_{n} - p ∥ \\ + (1 - α_{n}) β_{n} ∥ γ f (p) - A p ∥ . \end{array}

This in turn implies that

∥ x_{n} - p ∥ \leq max {∥ x_{1} - p ∥, \frac{∥ γ f (p) - A p ∥}{\bar{γ} - γ κ}} .

This completes the proof that the sequence

{x_{n}}

is bounded. This completes the proof of Step 1.

Step 2. Show that

{lim}_{n \to \infty} ∥ x_{n + 1} - x_{n} ∥ = 0

In view of

y_{n} = T_{r_{n}} x_{n}

and

y_{n + 1} = T_{r_{n + 1}} x_{n + 1}

, we see that

F (y_{n}, z) + \frac{1}{r_{n}} 〈 z - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall z \in C,

(2.1)

and

F (y_{n + 1}, z) + \frac{1}{r_{n + 1}} 〈 z - y_{n + 1}, y_{n + 1} - x_{n + 1} 〉 \geq 0, \forall z \in C .

(2.2)

Putting

z = y_{n + 1}

in (2.1) and

z = y_{n}

in (2.2), we find that

F (y_{n}, y_{n + 1}) + \frac{1}{r_{n}} 〈 y_{n + 1} - y_{n}, y_{n} - x_{n} 〉 \geq 0

and

F (y_{n + 1}, y_{n}) + \frac{1}{r_{n + 1}} 〈 y_{n} - y_{n + 1}, y_{n + 1} - x_{n + 1} 〉 \geq 0 .

It follows from (A2) that

〈 y_{n + 1} - y_{n}, \frac{y_{n} - x_{n}}{r_{n}} - \frac{y_{n + 1} - x_{n + 1}}{r_{n + 1}} 〉 \geq 0 .

That is,

〈 y_{n + 1} - y_{n}, y_{n} - y_{n + 1} + y_{n + 1} - x_{n} - \frac{r_{n}}{r_{n + 1}} (y_{n + 1} - x_{n + 1}) 〉 \geq 0 .

Without loss of generality, let us assume that there exists a real number m such that

r_{n} > m > 0

for all n. It follows that

{∥ y_{n + 1} - y_{n} ∥}^{2} \leq ∥ y_{n + 1} - y_{n} ∥ (∥ x_{n + 1} - x_{n} ∥ + | 1 - \frac{r_{n}}{r_{n + 1}} | ∥ y_{n + 1} - x_{n + 1} ∥) .

It follows that

\begin{array}{rcl} ∥ y_{n + 1} - y_{n} ∥ & \leq & ∥ x_{n + 1} - x_{n} ∥ + | 1 - \frac{r_{n}}{r_{n + 1}} | ∥ y_{n + 1} - x_{n + 1} ∥ \\ \leq & ∥ x_{n + 1} - x_{n} ∥ + \frac{M_{1}}{m} | r_{n + 1} - r_{n} |, \end{array}

(2.3)

where

M_{1}

is some real constant such that

M_{1} \geq {sup}_{n \geq 1} {∥ y_{n} - x_{n} ∥}

On the other hand, we have

\begin{array}{rcl} ∥ ρ_{n + 1} - ρ_{n} ∥ & = & ∥ P_{C} (I - s_{n + 1} B) y_{n + 1} - P_{C} (I - s_{n} B) y_{n} ∥ \\ \leq & ∥ (I - s_{n + 1} B) y_{n + 1} - (I - s_{n} B) y_{n} ∥ \\ = & ∥ (I - s_{n + 1} B) y_{n + 1} - (I - s_{n + 1} B) y_{n} + (s_{n} - s_{n + 1}) B y_{n} ∥ \\ \leq & ∥ y_{n + 1} - y_{n} ∥ + | s_{n} - s_{n + 1} | M_{2}, \end{array}

(2.4)

where

M_{2} \geq {sup}_{n \geq 1} {∥ B y_{n} ∥}

. Substituting (2.3) into (2.4) yields

∥ ρ_{n + 1} - ρ_{n} ∥ \leq ∥ x_{n + 1} - x_{n} ∥ + M_{3} (| r_{n + 1} - r_{n} | + | s_{n} - s_{n + 1} |),

(2.5)

where

M_{3} = max {M_{1}, M_{2}}

. Notice that

\begin{array}{rcl} ∥ ζ_{n} - ζ_{n + 1} ∥ & = & ∥ (I - β_{n + 1} A) (W_{n + 1} ρ_{n + 1} - W_{n} ρ_{n}) - (β_{n + 1} - β_{n}) A W_{n} ρ_{n} \\ + γ [β_{n + 1} (f (y_{n + 1}) - f (z_{n})) + f (y_{n}) (β_{n + 1} - β_{n})] ∥ \\ \leq & (1 - β_{n + 1} \bar{γ}) (∥ ρ_{n + 1} - ρ_{n} ∥ + ∥ W_{n + 1} ρ_{n} - W_{n} ρ_{n} ∥) \\ + | β_{n + 1} - β_{n} | ∥ A W_{n} ρ_{n} ∥ + γ (β_{n + 1} κ ∥ y_{n + 1} - y_{n} ∥ + | β_{n + 1} - β_{n} | ∥ f (y_{n}) ∥) . \end{array}

(2.6)

Since

T_{i}

and

U_{n, i}

are nonexpansive, we see from (1.3) that

\begin{array}{rcl} ∥ W_{n + 1} ρ_{n} - W_{n} ρ_{n} ∥ & = & ∥ γ_{1} T_{1} U_{n + 1, 2} ρ_{n} - γ_{1} T_{1} U_{n, 2} ρ_{n} ∥ \\ \leq & γ_{1} ∥ U_{n + 1, 2} ρ_{n} - U_{n, 2} ρ_{n} ∥ \\ = & γ_{1} ∥ γ_{2} T_{2} U_{u + 1, 3} ρ_{n} - γ_{2} T_{2} U_{n, 3} ρ_{n} ∥ \\ \leq & γ_{1} γ_{2} ∥ U_{u + 1, 3} ρ_{n} - U_{n, 3} ρ_{n} ∥ \\ \leq & \dots \\ \leq & γ_{1} γ_{2} \dots γ_{n} ∥ U_{n + 1, n + 1} ρ_{n} - U_{n, n + 1} ρ_{n} ∥ \\ \leq & M_{4} \prod_{i = 1}^{n} γ_{i}, \end{array}

(2.7)

where

M_{4}

is a constant such that

M_{4} \geq {sup}_{n \geq 1} {∥ U_{n + 1, n + 1} ρ_{n} - U_{n, n + 1} ρ_{n} ∥}

. Substituting (2.3), (2.5) and (2.7) into (2.6) yields

∥ ζ_{n} - ζ_{n + 1} ∥ \leq ∥ x_{n + 1} - x_{n} ∥ + M_{5} (\prod_{i = 1}^{n} γ_{i} + | r_{n + 1} - r_{n} | + | s_{n} - s_{n + 1} | + | β_{n} - β_{n + 1} |),

where

M_{5}

is a constant such that

M_{5} = max {M_{3} + \bar{γ} \frac{M_{1}}{m}, M_{4}, sup_{n \geq 1} {∥ A W_{n} ρ_{n} ∥ + γ ∥ f (y_{n}) ∥}} .

It follows from the restrictions (b) and (c) that

\underset{n \to \infty}{lim sup} (∥ ζ_{n} - ζ_{n + 1} ∥ - ∥ x_{n + 1} - x_{n} ∥) \leq 0 .

By virtue of Lemma 1.8, we obtain that

lim_{n \to \infty} ∥ ζ_{n} - x_{n} ∥ = 0 .

(2.8)

On the other hand, we have

∥ x_{n + 1} - x_{n} ∥ = (1 - α_{n}) ∥ x_{n} - ζ_{n} ∥ .

This implies from (2.8) that

lim_{n \to \infty} ∥ x_{n + 1} - x_{n} ∥ = 0 .

(2.9)

This completes the proof of Step 2.

Step 3. Show that

{lim}_{n \to \infty} ∥ y_{n} - W y_{n} ∥ = 0

Notice that

ζ_{n} = β_{n} γ f (z_{n}) + (I - β_{n} A) W_{n} ρ_{n}

. It follows that

∥ ζ_{n} - W_{n} ρ_{n} ∥ = β_{n} ∥ γ f (y_{n}) - A W_{n} ρ_{n} ∥ .

This implies from the restriction (b) that

lim_{n \to \infty} ∥ ζ_{n} - W_{n} ρ_{n} ∥ = 0 .

(2.10)

For any

p \in F

, we find that

\begin{array}{rcl} {∥ y_{n} - p ∥}^{2} & = & {∥ T_{r_{n}} x_{n} - T_{r_{n}} p ∥}^{2} \\ \leq & 〈 T_{r_{n}} x_{n} - T_{r_{n}} p, x_{n} - p 〉 \\ = & 〈 y_{n} - p, x_{n} - p 〉 \\ = & 1 / 2 ({∥ y_{n} - p ∥}^{2} + {∥ x_{n} - p ∥}^{2} - {∥ x_{n} - y_{n} ∥}^{2}) . \end{array}

That is,

{∥ y_{n} - p ∥}^{2} \leq {∥ x_{n} - p ∥}^{2} - {∥ x_{n} - y_{n} ∥}^{2} .

This in turn implies that

\begin{matrix} {∥ x_{n + 1} - p ∥}^{2} \\ = {∥ α_{n} (x_{n} - p) + (1 - α_{n}) (ζ_{n} - p) ∥}^{2} \\ \leq α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) {∥ ζ_{n} - p ∥}^{2} \\ = α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) {∥ β_{n} (γ f (y_{n}) - A p) + (I - β_{n} A) (W_{n} ρ_{n} - p) ∥}^{2} \\ \leq α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) {(β_{n} ∥ γ f (y_{n}) - A p ∥ + (1 - β_{n} \bar{γ}) ∥ ρ_{n} - p ∥)}^{2} \\ \leq α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} + (1 - α_{n}) (1 - β_{n} \bar{γ}) {∥ ρ_{n} - p ∥}^{2} \\ + 2 (1 - α_{n}) β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ \\ \leq α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} + (1 - α_{n}) (1 - β_{n} \bar{γ}) {∥ y_{n} - p ∥}^{2} \\ + 2 (1 - α_{n}) β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ \\ \leq α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} + (1 - α_{n}) {∥ x_{n} - p ∥}^{2} \\ - (1 - α_{n}) (1 - β_{n} \bar{γ}) {∥ x_{n} - y_{n} ∥}^{2} + 2 (1 - α_{n}) β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥, \end{matrix}

from which it follows that

\begin{matrix} (1 - α_{n}) (1 - β_{n} \bar{γ}) {∥ x_{n} - y_{n} ∥}^{2} \\ \leq {∥ x_{n} - p ∥}^{2} - {∥ x_{n + 1} - p ∥}^{2} + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} \\ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ \\ \leq (∥ x_{n} - p ∥ - ∥ x_{n + 1} - p ∥) ∥ x_{n} - x_{n + 1} ∥ + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} \\ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ . \end{matrix}

It follows from the restriction (b) and (2.9) that

lim_{n \to \infty} ∥ y_{n} - x_{n} ∥ = 0 .

(2.11)

Notice that

\begin{array}{rcl} {∥ ρ_{n} - p ∥}^{2} & = & {∥ P_{C} (I - s_{n} B) y_{n} - P_{C} (I - s_{n} B) p ∥}^{2} \\ \leq & {∥ (y_{n} - p) - s_{n} (B y_{n} - B p) ∥}^{2} \\ = & {∥ y_{n} - p ∥}^{2} - 2 s_{n} 〈 y_{n} - p, B y_{n} - B p 〉 + s_{n}^{2} {∥ B y_{n} - B p ∥}^{2} \\ \leq & {∥ x_{n} - p ∥}^{2} - 2 s_{n} α {∥ B y_{n} - B p ∥}^{2} + s_{n}^{2} {∥ B y_{n} - B p ∥}^{2} \\ \leq & {∥ x_{n} - p ∥}^{2} - s_{n} (2 α - s_{n}) {∥ B y_{n} - B p ∥}^{2} . \end{array}

(2.12)

On the other hand, we have

https://static-content.springer.com/image/art%3A10.1186%2F1029-242X-2013-19/MediaObjects/13660_2012_Article_447_Equ16_HTML.gif

(2.13)

Substituting (2.12) into (2.13), we find that

\begin{array}{rcl} {∥ x_{n + 1} - p ∥}^{2} & \leq & {∥ x_{n} - p ∥}^{2} + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ \\ - (1 - α_{n}) s_{n} (2 α - s_{n}) {∥ B y_{n} - B p ∥}^{2} . \end{array}

This in turn implies that

\begin{matrix} (1 - α_{n}) s_{n} (2 α - s_{n}) {∥ B y_{n} - B p ∥}^{2} \\ \leq {∥ x_{n} - p ∥}^{2} + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} - {∥ x_{n + 1} - p ∥}^{2} \\ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ \\ \leq (∥ x_{n} - p ∥ + ∥ x_{n + 1} - p ∥) ∥ x_{n} - x_{n + 1} ∥ + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} \\ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ . \end{matrix}

It follows from the restrictions (a), (b) and (d) that

lim_{n \to \infty} ∥ B y_{n} - B p ∥ = 0 .

(2.14)

On the other hand, we have

\begin{array}{rcl} {∥ ρ_{n} - p ∥}^{2} & = & {∥ P_{C} (I - s_{n} B) y_{n} - P_{C} (I - s_{n} B) p ∥}^{2} \\ \leq & 〈 (I - s_{n} B) y_{n} - (I - s_{n} B) p, ρ_{n} - p 〉 \\ = & \frac{1}{2} ({∥ (I - s_{n} B) y_{n} - (I - s_{n} B) p ∥}^{2} + {∥ ρ_{n} - p ∥}^{2} \\ - {∥ (I - s_{n} B) y_{n} - (I - s_{n} B) p - (ρ_{n} - p) ∥}^{2}) \\ \leq & \frac{1}{2} ({∥ y_{n} - p ∥}^{2} + {∥ ρ_{n} - p ∥}^{2} - {∥ (y_{n} - ρ_{n}) - s_{n} (B y_{n} - B p) ∥}^{2}) \\ = & \frac{1}{2} ({∥ x_{n} - p ∥}^{2} + {∥ ρ_{n} - p ∥}^{2} - {∥ y_{n} - ρ_{n} ∥}^{2} - s_{n}^{2} {∥ B y_{n} - B p ∥}^{2} \\ + 2 s_{n} ∥ y_{n} - ρ_{n} ∥ B y_{n} - B p ∥), \end{array}

which yields

{∥ ρ_{n} - p ∥}^{2} \leq {∥ x_{n} - p ∥}^{2} - {∥ y_{n} - ρ_{n} ∥}^{2} + 2 s_{n} ∥ y_{n} - ρ_{n} ∥ ∥ B y_{n} - B p ∥ .

(2.15)

Substituting (2.15) into (2.13) yields

\begin{array}{rcl} {∥ x_{n + 1} - p ∥}^{2} & \leq & {∥ x_{n} - p ∥}^{2} + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} - (1 - α_{n}) {∥ y_{n} - ρ_{n} ∥}^{2} \\ + 2 s_{n} ∥ y_{n} - ρ_{n} ∥ ∥ B y_{n} - B p ∥ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ . \end{array}

It follows that

\begin{array}{rcl} (1 - α_{n}) {∥ y_{n} - ρ_{n} ∥}^{2} & \leq & {∥ x_{n} - p ∥}^{2} - {∥ x_{n + 1} - p ∥}^{2} + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} \\ + 2 s_{n} ∥ y_{n} - ρ_{n} ∥ ∥ B y_{n} - B p ∥ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ \\ \leq & (∥ x_{n} - p ∥ + ∥ x_{n + 1} - p ∥) ∥ x_{n} - x_{n + 1} ∥ + β_{n} {∥ γ f (y_{n}) - A p ∥}^{2} \\ + 2 s_{n} ∥ y_{n} - ρ_{n} ∥ ∥ B y_{n} - B p ∥ + 2 β_{n} ∥ γ f (y_{n}) - A p ∥ ∥ ρ_{n} - p ∥ . \end{array}

In view of the restrictions (a), (b) and (d), we find from (2.9) that

lim_{n \to \infty} ∥ y_{n} - ρ_{n} ∥ = 0 .

(2.16)

Notice that

\begin{array}{rcl} ∥ y_{n} - W_{n} y_{n} ∥ & \leq & ∥ W_{n} y_{n} - W_{n} ρ_{n} ∥ + ∥ W_{n} ρ_{n} - ζ_{n} ∥ + ∥ ζ_{n} - x_{n} ∥ + ∥ x_{n} - y_{n} ∥ \\ \leq & ∥ y_{n} - ρ_{n} ∥ + ∥ W_{n} ρ_{n} - ζ_{n} ∥ + ∥ ζ_{n} - x_{n} ∥ + ∥ x_{n} - y_{n} ∥ . \end{array}

In the light of (2.8), (2.10), (2.11) and (2.16), we find that

{lim}_{n \to \infty} ∥ y_{n} - W_{n} y_{n} ∥ = 0

. On the other hand, we have

∥ W y_{n} - y_{n} ∥ \leq ∥ W y_{n} - W_{n} y_{n} ∥ + ∥ W_{n} y_{n} - y_{n} ∥ .

It follows from Lemma 1.9 that

lim_{n \to \infty} ∥ y_{n} - W y_{n} ∥ = 0 .

(2.17)

This completes the proof of Step 3.

Step 4. Show that

{lim sup}_{n \to \infty} 〈 γ f (q) - A q, x_{n} - q 〉 \leq 0

, where

q = P_{F} (γ f + (I - A)) (q)

To see this, we choose a subsequence

{x_{n_{i}}}

{x_{n}}

such that

\underset{n \to \infty}{lim sup} 〈 γ f (q) - A q, x_{n} - q 〉 = lim_{i \to \infty} 〈 γ f (q) - A q, x_{n_{i}} - q 〉 .

(2.18)

Correspondingly, there exists a subsequence

{y_{n_{i}}}

{y_{n}}

. Since

{y_{n_{i}}}

is bounded, there exists a subsequence

{y_{n_{i_{j}}}}

{y_{n_{i}}}

which converges weakly to w. Without loss of generality, we can assume that

y_{n_{i}} ⇀ w

. Since

y_{n} = T_{r_{n}} x_{n}

, we have

F (y_{n}, z) + \frac{1}{r_{n}} 〈 z - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall z \in C .

It follows from (A2) that

〈 z - y_{n}, \frac{y_{n} - x_{n}}{r_{n}} 〉 \geq F (z, y_{n}) .

It follows that

〈 z - y_{n_{i}}, \frac{y_{n_{i}} - x_{n_{i}}}{r_{n_{i}}} 〉 \geq F (z, y_{n_{i}}) .

In view of the restriction (c), we obtain from (2.11) that

lim_{n \to \infty} \frac{y_{n} - x_{n}}{r_{n}} = 0 .

Since

y_{n_{i}} ⇀ w

, we have from (A4) that

F (z, w) \leq 0

for all

z \in C

. For t with

0 < t \leq 1

and

z \in C

, let

z_{t} = t z + (1 - t) w

. Since

z \in C

and

w \in C

, we have

z_{t} \in C

and hence

F (z_{t}, w) \leq 0

. So, from (A1) and (A4), we have

0 = F (z_{t}, z_{t}) \leq t F (z_{t}, z) + (1 - t) F (z_{t}, w) \leq t F (z_{t}, z) .

That is,

F (z_{t}, z) \geq 0

. It follows from (A3) that

F (w, z) \geq 0

for all

z \in C

and hence

w \in E P (F)

. On the other hand, we see that

w \in F (W) = ⋂_{i = 1}^{\infty} F (T_{i})

. If

w \neq W w

, then we have the following. Since Hilbert spaces are Opial’s spaces, we find from (2.17) that

\begin{array}{rcl} \underset{i \to \infty}{lim inf} ∥ y_{n_{i}} - w ∥ & < & \underset{i \to \infty}{lim inf} ∥ y_{n_{i}} - W w ∥ \\ = & \underset{i \to \infty}{lim inf} ∥ y_{n_{i}} - W y_{n_{i}} + W y_{n_{i}} - W w ∥ \\ \leq & \underset{i \to \infty}{lim inf} ∥ W y_{n_{i}} - W w ∥ \\ \leq & \underset{i \to \infty}{lim inf} ∥ y_{n_{i}} - w ∥, \end{array}

which derives a contradiction. Thus, we have

w \in F (W)

. Next, let us first show that

w \in V I (C, B)

. Put

S ξ = {\begin{matrix} B ξ + N_{C} ξ, & ξ \in C, \\ \emptyset, & ξ \bar{\in} C . \end{matrix}

Since B is monotone, we see that S is maximal monotone. Let

(ξ, ξ^{'}) \in Graph (S)

. Since

ξ^{'} - B ξ \in N_{C} ξ

and

ρ_{n} \in C

, we have

〈 ξ - ρ_{n}, ξ^{'} - B ξ 〉 \geq 0 .

On the other hand, we have from

ρ_{n} = P_{C} (I - s_{n} B) y_{n}

that

〈 ξ - ρ_{n}, ρ_{n} - (I - s_{n} B) y_{n} 〉 \geq 0 .

That is,

〈 ξ - ρ_{n}, \frac{ρ_{n} - y_{n}}{s_{n}} + B y_{n} 〉 \geq 0 .

It follows from the above that

\begin{array}{rcl} 〈 ξ - ρ_{n_{i}}, ξ^{'} 〉 & \geq & 〈 ξ - ρ_{n_{i}}, B ξ 〉 \\ \geq & 〈 ξ - ρ_{n_{i}}, B ξ - \frac{ρ_{n_{i}} - y_{n_{i}}}{s_{n_{i}}} - B y_{n_{i}} 〉 \\ = & 〈 ξ - ρ_{n_{i}}, B ξ - B ρ_{n_{i}} 〉 + 〈 ξ - ρ_{n_{i}}, B ρ_{n_{i}} - B y_{n_{i}} 〉 \\ - 〈 ξ - ρ_{n_{i}}, \frac{ρ_{n_{i}} - y_{n_{i}}}{s_{n_{i}}} 〉 \\ \geq & 〈 ξ - ρ_{n_{i}}, B ρ_{n_{i}} - B y_{n_{i}} 〉 - 〈 ξ - ρ_{n_{i}}, \frac{ρ_{n_{i}} - y_{n_{i}}}{s_{n_{i}}} 〉, \end{array}

which implies from (2.16) that

〈 ξ - w, ξ^{'} 〉 \geq 0

. We have

w \in S^{- 1} 0

and hence

w \in V I (C, B)

. This completes the proof

w \in F

. On the other hand, we find from (2.18) that

\begin{array}{rcl} \underset{n \to \infty}{lim sup} 〈 γ f (q) - A q, x_{n} - q 〉 & = & lim_{n \to \infty} 〈 γ f (q) - A q, x_{n_{i}} - q 〉 \\ = & 〈 γ f (q) - A q, w - q 〉 \leq 0 . \end{array}

(2.19)

This completes the proof of Step 4.

Step 5. Show

{lim}_{n \to \infty} ∥ x_{n} - q ∥ = 0

It follows from Lemma 1.3 that

\begin{array}{rcl} {∥ ζ_{n} - q ∥}^{2} & = & {∥ (I - β_{n} A) (W_{n} ρ_{n} - q) + β_{n} (γ f (y_{n}) - A q) ∥}^{2} \\ \leq & {∥ (I - β_{n} A) (W_{n} ρ_{n} - q) ∥}^{2} + 2 β_{n} 〈 γ f (z_{n}) - A q, ζ_{n} - q 〉 \\ \leq & {(1 - β_{n} \bar{γ})}^{2} {∥ ρ_{n} - q ∥}^{2} + 2 β_{n} 〈 γ f (y_{n}) - A q, ζ_{n} - q 〉 \\ \leq & {(1 - β_{n} \bar{γ})}^{2} {∥ y_{n} - q ∥}^{2} + 2 β_{n} γ 〈 f (y_{n}) - f (q), ζ_{n} - q 〉 \\ + 2 β_{n} 〈 γ f (q) - A q, ζ_{n} - q 〉 \\ \leq & {(1 - β_{n} \bar{γ})}^{2} {∥ x_{n} - q ∥}^{2} + 2 β_{n} γ κ ∥ y_{n} - q ∥ ∥ ζ_{n} - q ∥ \\ + 2 β_{n} 〈 γ f (q) - A q, ζ_{n} - q 〉 \\ \leq & {(1 - β_{n} \bar{γ})}^{2} {∥ x_{n} - q ∥}^{2} + β_{n} γ κ ({∥ y_{n} - q ∥}^{2} + {∥ ζ_{n} - q ∥}^{2}) \\ + 2 β_{n} 〈 γ f (q) - A q, ζ_{n} - q 〉 \\ \leq & {(1 - β_{n} \bar{γ})}^{2} {∥ x_{n} - q ∥}^{2} + β_{n} γ κ ({∥ x_{n} - q ∥}^{2} + {∥ ζ_{n} - q ∥}^{2}) \\ + 2 β_{n} 〈 γ f (q) - A q, ζ_{n} - q 〉, \end{array}

which implies that

\begin{array}{rcl} {∥ ζ_{n} - q ∥}^{2} & \leq & \frac{{(1 - β_{n} \bar{γ})}^{2} + β_{n} γ κ}{1 - β_{n} γ α} {∥ x_{n} - q ∥}^{2} + \frac{2 β_{n}}{1 - β_{n} γ κ} 〈 γ f (q) - A q, ζ_{n} - q 〉 \\ = & \frac{(1 - 2 β_{n} \bar{γ} + β_{n} κ γ)}{1 - β_{n} γ κ} {∥ x_{n} - q ∥}^{2} + \frac{β_{n}^{2} {\bar{γ}}^{2}}{1 - β_{n} γ κ} {∥ x_{n} - q ∥}^{2} \\ + \frac{2 β_{n}}{1 - β_{n} γ κ} 〈 γ f (q) - A q, ζ_{n} - q 〉 \\ \leq & (1 - \frac{2 β_{n} (\bar{γ} - κ γ)}{1 - β_{n} γ κ}) {∥ x_{n} - q ∥}^{2} \\ + \frac{2 β_{n} (\bar{γ} - κ γ)}{1 - β_{n} γ κ} (\frac{1}{\bar{γ} - κ γ} 〈 γ f (q) - A q, ζ_{n} - q 〉 + \frac{β_{n} {\bar{γ}}^{2}}{2 (\bar{γ} - κ γ)} M_{6}), \end{array}

(2.20)

where

M_{6}

is a constant such that

M_{6} \geq {sup}_{n \geq 1} {{∥ x_{n} - q ∥}^{2}}

. On the other hand, we have

{∥ x_{n + 1} - p ∥}^{2} \leq α_{n} {∥ x_{n} - p ∥}^{2} + (1 - α_{n}) {∥ ζ_{n} - p ∥}^{2} .

(2.21)

Substituting (2.20) into (2.21) yields

\begin{array}{rcl} {∥ x_{n + 1} - p ∥}^{2} & \leq & [1 - (1 - α_{n}) \frac{2 β_{n} (\bar{γ} - κ γ)}{1 - β_{n} γ κ}] {∥ x_{n} - q ∥}^{2} \\ + (1 - α_{n}) \frac{2 β_{n} (\bar{γ} - κ γ)}{1 - β_{n} γ α} \\ \times (\frac{1}{\bar{γ} - κ γ} 〈 γ f (q) - A q, ζ_{n} - q 〉 + \frac{β_{n} {\bar{γ}}^{2}}{2 (\bar{γ} - κ γ)} M_{6}) . \end{array}

(2.22)

Let

λ_{n} = (1 - α_{n}) \frac{2 β_{n} (\bar{γ} - κ γ)}{1 - β_{n} κ γ}

and

θ_{n} = \frac{1}{\bar{γ} - κ γ} 〈 γ f (q) - A q, ζ_{n} - q 〉 + \frac{β_{n} {\bar{γ}}^{2}}{2 (\bar{γ} - κ γ)} M_{6} .

This implies that

{∥ x_{n + 1} - q ∥}^{2} \leq (1 - λ_{n}) {∥ x_{n} - q ∥}^{2} + λ_{n} t_{n} .

(2.23)

In view of the restriction (b), we find from (2.8) and (2.11) that

lim_{n \to \infty} λ_{n} = 0, \sum_{n = 1}^{\infty} λ_{n} = \infty and \underset{n \to \infty}{lim sup} θ_{n} \leq 0 .

We can easily draw the desired conclusion with the aid of Lemma 1.4. This completes the proof of Step 5. The proof is completed. □

From Theorem 2.1, we have the following results.

Corollary 2.2 Let C be a nonempty closed convex subset of a Hilbert space H. Let

{T_{n}}_{n = 1}^{\infty}

be an infinite family of nonexpansive mappings of C into C. Let

B : C \to H

be an α-inverse-strongly monotone mapping. Let A be a strongly positive linear bounded self-adjoint operator on H with the coefficient

\bar{γ} > 0

. Assume that

0 < γ < \bar{γ} / κ

and

F : = ⋂_{i = 1}^{\infty} F (T_{i}) \cap V I (C, B) \neq \emptyset

. Let

f : C \to H

be a κ-contraction. Let

{x_{n}}

be a sequence generated in the following iterative process:

{\begin{matrix} x_{1} \in H, \\ x_{n + 1} = α_{n} x_{n} + (1 - α_{n}) β_{n} γ f (y_{n}) + (1 - α_{n}) (I - β_{n} A) W_{n} P_{C} (I - s_{n} B) P_{C} x_{n}, n \geq 1, \end{matrix}

where

W_{n}

is generated in (1.3),

{α_{n}}

{β_{n}}

are real number sequences in

(0, 1)

{r_{n}}

and

{s_{n}}

are positive real number sequences. Assume that the following restrictions are satisfied:

(a)

0 < {lim inf}_{n \to \infty} α_{n} \leq {lim sup}_{n \to \infty} α_{n} < 1

;

(b)

{lim}_{n \to \infty} β_{n} = 0

\sum_{n = 1}^{\infty} β_{n} = \infty

;

(c)

{lim}_{n \to \infty} | s_{n + 1} - s_{n} | = 0

;

(d)

{s_{n}} \subset [s, s^{'}]

for some s,

s^{'}

with

0 < s < s^{'} < 2 α

Then

{x_{n}}

converges strongly to

q \in F

, where

q = P_{F} (γ f + (I - A)) (q)

, which solves the following variational inequality:

〈 γ f (q) - A q, p - q 〉 \leq 0, \forall p \in F .

Proof Putting

F (x, y) = 0

and

r_{n} = 1

, we can immediately draw the desired conclusion from Theorem 2.1. □

Corollary 2.3 Let C be a nonempty closed convex subset of a Hilbert space H. Let F be a bifunction from

C \times C

to ℝ, which satisfies (A1)-(A4). Let

B : C \to H

be an α-inverse-strongly monotone mapping. Let A be a strongly positive linear bounded self-adjoint operator on H with the coefficient

\bar{γ} > 0

. Assume that

0 < γ < \bar{γ} / κ

and

F : = E P (F) \cap V I (C, B) \neq \emptyset

. Let

f : C \to H

be a κ-contraction. Let

{x_{n}}

be a sequence generated in the following iterative process:

{\begin{matrix} x_{1} \in H, \\ F (y_{n}, z) + \frac{1}{r_{n}} 〈 z - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall z \in C, \\ x_{n + 1} = α_{n} x_{n} + (1 - α_{n}) β_{n} γ u + (1 - α_{n}) (I - β_{n} A) P_{C} (I - s_{n} B) y_{n}, n \geq 1, \end{matrix}

where u is a fixed element in C,

{α_{n}}

{β_{n}}

are real number sequences in

(0, 1)

{r_{n}}

and

{s_{n}}

are positive real number sequences. Assume that the following restrictions are satisfied:

(a)

0 < {lim inf}_{n \to \infty} α_{n} \leq {lim sup}_{n \to \infty} α_{n} < 1

;

(b)

{lim}_{n \to \infty} β_{n} = 0

\sum_{n = 1}^{\infty} β_{n} = \infty

;

(c)

{lim}_{n \to \infty} | r_{n + 1} - r_{n} | = 0

{lim}_{n \to \infty} | s_{n + 1} - s_{n} | = 0

;

(d)

{lim inf}_{n \to \infty} r_{n} > 0

{s_{n}} \subset [s, s^{'}]

for some s,

s^{'}

with

0 < s < s^{'} < 2 α

Then,

{x_{n}}

converges strongly to

q \in F

, where

q = P_{F} (γ u + (q - A q))

, which solves the following variational inequality:

〈 γ u - A q, p - q 〉 \leq 0, \forall p \in F .

Proof Putting

T_{i} = I

, where I is the identity mapping and

f (x) = u

, for all

x \in C

, we can immediately draw the desired conclusion from Theorem 2.1. □

Corollary 2.4 Let C be a nonempty closed convex subset of a Hilbert space H. Let F be a bifunction from

C \times C

to ℝ which satisfies (A1)-(A4). Let

{T_{n}}_{n = 1}^{\infty}

be an infinite family of nonexpansive mappings of C into C. Let

B : C \to H

be an α-inverse-strongly monotone mapping. Assume that

F : = ⋂_{i = 1}^{\infty} F (T_{i}) \cap E P (F) \cap V I (C, B) \neq \emptyset

. Let

f : C \to H

be a κ-contraction. Let

{x_{n}}

be a sequence generated in the following iterative process:

{\begin{matrix} x_{1} \in H, \\ F (y_{n}, z) + \frac{1}{r_{n}} 〈 z - y_{n}, y_{n} - x_{n} 〉 \geq 0, \forall z \in C, \\ x_{n + 1} = α_{n} x_{n} + (1 - α_{n}) β_{n} f (y_{n}) + (1 - α_{n}) (1 - β_{n}) W_{n} P_{C} (I - s_{n} B) y_{n}, n \geq 1, \end{matrix}

where

W_{n}

is generated in (1.3),

{α_{n}}

{β_{n}}

are real number sequences in

(0, 1)

{r_{n}}

and

{s_{n}}

are positive real number sequences. Assume that the following restrictions are satisfied:

(a)

0 < {lim inf}_{n \to \infty} α_{n} \leq {lim sup}_{n \to \infty} α_{n} < 1

;

(b)

{lim}_{n \to \infty} β_{n} = 0

\sum_{n = 1}^{\infty} β_{n} = \infty

;

(c)

{lim}_{n \to \infty} | r_{n + 1} - r_{n} | = 0

{lim}_{n \to \infty} | s_{n + 1} - s_{n} | = 0

;

(d)

{lim inf}_{n \to \infty} r_{n} > 0

{s_{n}} \subset [s, s^{'}]

for some s,

s^{'}

with

0 < s < s^{'} < 2 α

Then

{x_{n}}

converges strongly to

q \in F

, where

q = P_{F} f (q)

, which solves the following variational inequality:

〈 f (q) - q, p - q 〉 \leq 0, \forall p \in F .

Proof Putting

A = I

, where I is the identity mapping and

γ = 1

, we can immediately draw the desired conclusion from Theorem 2.1. □

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Competing interests

The author declares that they have no competing interests.

Vorheriger Artikel Some inequalities for the Hadamard product of an M-matrix and an inverse M-matrix

Nächster Artikel A generalized Ostrowski-Grüss type inequality for bounded differentiable mappings and its applications

Rockafellar RT: On the maximality of sums of nonlinear monotone operators. Trans. Am. Math. Soc. 1970, 149: 75–88. 10.1090/S0002-9947-1970-0282272-5MATHMathSciNetCrossRef

Su Y, Shang M, Qin X: An iterative method of solution for equilibrium and optimization problems. Nonlinear Anal. 2008, 69: 2709–2719. 10.1016/j.na.2007.08.045MATHMathSciNetCrossRef

Zegeye H, Shahzad N, Alghamdi M: Strong convergence theorems for a common point of solution of variational inequality, solutions of equilibrium and fixed point problems. Fixed Point Theory Appl. 2012, 2012: 119. 10.1186/1687-1812-2012-119MathSciNetCrossRef

Kang SM, Cho SY, Liu Z: Convergence of iterative sequences for generalized equilibrium problems involving inverse-strongly monotone mappings. J. Inequal. Appl. 2010., 2010: Article ID 827082

Cho SY, Qin X, Kang SM: Hybrid projection algorithms for treating common fixed points of a family of demicontinuous pseudocontractions. Appl. Math. Lett. 2012, 25: 854–857. 10.1016/j.aml.2011.10.031MATHMathSciNetCrossRef

Yang S, Li W: Iterative solutions of a system of equilibrium problems in Hilbert spaces. Adv. Fixed Point Theory 2011, 1: 15–26.

Cho SY, Kang SM: Approximation of common solutions of variational inequalities via strict pseudocontractions. Acta Math. Sci. 2012, 32: 1607–1618.MATHCrossRef

Cho SY, Kang SM: Approximation of fixed points of pseudocontraction semigroups based on a viscosity iterative process. Appl. Math. Lett. 2011, 24: 224–228. 10.1016/j.aml.2010.09.008MATHMathSciNetCrossRef

Kim JK: Strong convergence theorems by hybrid projection methods for equilibrium problems and fixed point problems of the asymptotically quasi- ϕ -nonexpansive mappings. Fixed Point Theory Appl. 2011, 2011: 10. 10.1186/1687-1812-2011-10CrossRef

10.

Zegeye H, Shahzad N: Strong convergence theorem for a common point of solution of variational inequality and fixed point problem. Adv. Fixed Point Theory 2012, 2: 374–397.

11.

Ye J, Huang J: Strong convergence theorems for fixed point problems and generalized equilibrium problems of three relatively quasi-nonexpansive mappings in Banach spaces. J. Math. Comput. Sci. 2011, 1: 1–18.MathSciNetCrossRef

12.

Qin X, Cho YJ, Kang SM: Convergence theorems of common elements for equilibrium problems and fixed point problems in Banach spaces. J. Comput. Appl. Math. 2009, 225: 20–30. 10.1016/j.cam.2008.06.011MATHMathSciNetCrossRef

13.

Qin X, Cho SY, Kang SM: Strong convergence of shrinking projection methods for quasi- ϕ -nonexpansive mappings and equilibrium problems. J. Comput. Appl. Math. 2010, 234: 750–760. 10.1016/j.cam.2010.01.015MATHMathSciNetCrossRef

14.

Lv S, Wu C: Convergence of iterative algorithms for a generalized variational inequality and a nonexpansive mapping. Eng. Math. Lett. 2012, 1: 44–57.

15.

Blum E, Oettli W: From optimization and variational inequalities to equilibrium problems. Math. Stud. 1994, 63: 123–145.MATHMathSciNet

16.

Combettes PL, Hirstoaga SA: Equilibrium programming in Hilbert spaces. J. Nonlinear Convex Anal. 2005, 6: 117–136.MATHMathSciNet

17.

Takahashi S, Takahashi W: Viscosity approximation methods for equilibrium problems and fixed point problems in Hilbert spaces. J. Math. Anal. Appl. 2007, 331: 506–515. 10.1016/j.jmaa.2006.08.036MATHMathSciNetCrossRef

18.

Qin X, Chang SS, Cho YJ: Iterative methods for generalized equilibrium problems and fixed point problems with applications. Nonlinear Anal. 2010, 11: 2963–2972. 10.1016/j.nonrwa.2009.10.017MATHMathSciNetCrossRef

19.

Plubtieng S, Punpaeng R: A general iterative method for equilibrium problems and fixed point problems in Hilbert spaces. J. Math. Anal. Appl. 2007, 336: 455–469. 10.1016/j.jmaa.2007.02.044MATHMathSciNetCrossRef

20.

Shimoji K, Takahashi W: Strong convergence to common fixed points of infinite nonexpansive mappings and applications. Taiwan. J. Math. 2001, 5: 387–404.MATHMathSciNet

21.

Xu HK: An iterative approach to quadratic optimization. J. Optim. Theory Appl. 2003, 116: 659–678. 10.1023/A:1023073621589MATHMathSciNetCrossRef

22.

Marino G, Xu HK: A general iterative method for nonexpansive mappings in Hilbert spaces. J. Math. Anal. Appl. 2006, 318: 43–52. 10.1016/j.jmaa.2005.05.028MATHMathSciNetCrossRef

23.

Blum E, Oettli W: From optimization and variational inequalities to equilibrium problems. Math. Stud. 1994, 63: 123–145.MATHMathSciNet

24.

Combettes PL, Hirstoaga SA: Equilibrium programming in Hilbert spaces. J. Nonlinear Convex Anal. 2005, 6: 117–136.MATHMathSciNet

25.

Suzuki T: Strong convergence of Krasnoselskii and Mann’s type sequences for one-parameter nonexpansive semigroups without Bochner integrals. J. Math. Anal. Appl. 2005, 305: 227–239. 10.1016/j.jmaa.2004.11.017MATHMathSciNetCrossRef

26.

Ceng LC, Yao JC: Hybrid viscosity approximation schemes for equilibrium problems and fixed point problems of infinitely many nonexpansive mappings. Appl. Math. Comput. 2008, 198: 729–741. 10.1016/j.amc.2007.09.011MATHMathSciNetCrossRef

27.

Chang SS, Lee HWJ, Chan CK: A new method for solving equilibrium problem fixed point problem and variational inequality problem with application to optimization. Nonlinear Anal. 2009, 70: 3307–3319. 10.1016/j.na.2008.04.035MATHMathSciNetCrossRef

Titel: Strong convergence of a general iterative algorithm in Hilbert spaces
verfasst von: Songtao Lv
Publikationsdatum: 01.12.2013
Verlag: Springer International Publishing
Erschienen in: Journal of Inequalities and Applications / Ausgabe 1/2013
Elektronische ISSN: 1029-242X
DOI: https://doi.org/10.1186/1029-242X-2013-19

Springer Professional

Strong convergence of a general iterative algorithm in Hilbert spaces

Abstract

Competing interests

1 Introduction and preliminaries

2 Main results

Competing interests

Premium Partner

Springer Professional

Abstract

Competing interests

1 Introduction and preliminaries

2 Main results

Competing interests

Weitere Artikel der Ausgabe 1/2013

Functional equations and inequalities in paranormed spaces

Almost contractive coupled mapping in ordered complete metric spaces

Refinements of Hermite-Hadamard type inequalities for operator convex functions

Convergence theorems of a new iteration for asymptotically nonexpansive mappings in Banach spaces

Strong convergence of projection methods for a countable family of nonexpansive mappings and applications to constrained convex minimization problems

Sharp bounds for Hardy type operators on higher-dimensional product spaces

Premium Partner