Nonlinear Analysis 70 (2009) 719–734

www.elsevier.com/locate/na

Viscosity approximation methods for countable families of

nonexpansive mappings in Banach spaces
Wataru Takahashi ∗
Department of Mathematical and Computing Sciences, Tokyo Institute of Technology, Oh-okayama, Meguro-ku, Tokyo, 152-8552, Japan

Received 24 January 2007; accepted 8 January 2008

Abstract

Let C be a nonempty closed convex subset of a Banach space E and let {Sn } be a family of nonexpansive mappings of C into
itself such that the set of common fixed points of {Sn } is nonempty. We first introduce a sequence {xn } of C defined by x1 = x ∈ C
and

xn+1 = αn f (xn ) + (1 − αn )Sn xn for all n ∈ N,

where {αn } ⊂ (0, 1) and f is a contraction of C into itself. Further, we give the conditions of {αn } and {Sn } under which {xn }
converges strongly to a common fixed point of {Sn }. This result generalizes the strong convergence theorem for nonexpansive
mappings by Suzuki [T. Suzuki, A sufficient and necessary condition for Halpern-type strong convergence to fixed points of
nonexpansive mappings, Proc. Amer. Math. Soc. 135 (2007) 99–106] and the strong convergence theorem for accretive operators by
Kamimura and Takahashi [S. Kamimura, W. Takahashi, Weak and strong convergence of solutions to accretive operator inclusions
and applications, Set-Valued Anal. 8 (2000) 361–374], simultaneously. Using this result, we improve and extend the two above-
mentioned results.
c 2008 Elsevier Ltd. All rights reserved.

MSC: 47H06; 47H09; 47H10

Keywords: Banach space; Nonexpansive mapping; Strong convergence theorem; Resolvent; Iteration; Fixed point; Accretive operator

1. Introduction

Throughout this paper, let E be a real Banach space with norm k · k and let N be the set of all positive integers. Let
C be a nonempty closed convex subset of E. Then, a mapping T : C → C is called nonexpansive if
kT x − T yk ≤ kx − yk for all x, y ∈ C.
We denote by F(T ) the set of fixed points of T . On the other hand, an operator A ⊂ E × E is called accretive if for
(x1 , y1 ), (x2 , y2 ) ∈ A, there exists j ∈ J (x1 −x2 ) such that (y1 − y2 , j) ≥ 0, where J is the duality mapping on E. For
an accretive operator A ⊂ E × E and r > 0, we can define a mapping Jr : R(I + r A) → D(A) by Jr = (I + r A)−1 ,

∗ Fax: +81 03 5734 3208.

E-mail address: [email protected].

0362-546X/$ - see front matter c 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.na.2008.01.005
720 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

where R(I + r A) and D(A) are the range of I + r A and the domain of A, respectively. An accretive operator A is
said to be m-accretive if R(I + r A) = E for all r > 0. Recently, Suzuki [32] proved the following strong convergence
theorem for nonexpansive mappings in a Banach space; see also [38,29,35,40].

Theorem 1.1. Let E be a reflexive Banach space with a uniformly Gatêaux differentiable norm. Let C be a nonempty
closed convex subset of E which has the fixed-point property for nonexpansive mappings and let T : C → C be a
nonexpansive mapping such that F(T ) is nonempty. Define a sequence {xn } of C as follows: x1 , u ∈ C and
xn+1 = αn u + (1 − αn )((1 − λ)xn + λT xn ) for all n ∈ N,
where λ ∈ (0, 1) and {αn } ⊂ (0, 1) satisfies the following conditions:
∞
X
αn → 0 and αn = ∞.
n=1

Then, the sequence {xn } converges strongly to a fixed point of T .

Kamimura and Takahashi [9] also proved the following strong convergence theorem for accretive operators in a
Banach space; see also [2,3,8,14,19,22,27,30].

Theorem 1.2. Let E be a reflexive Banach space with a uniformly Gatêaux differentiable norm which has the fixed-
point property for nonexpansive mappings. Let A ⊂ E × E be an m-accretive operator with A−1 0 6= ∅. Define a
sequence {xn } of E as follows: x1 , u ∈ E and
xn+1 = αn u + (1 − αn )Jtn xn for all n ∈ N,
where {αn } ⊂ (0, 1) and {tn } ⊂ (0, ∞) satisfy the following conditions:
∞
X
αn → 0, αn = ∞ and tn → ∞.
n=1

Then, the sequence {xn } converges strongly to u ∈ A−1 0.

In this paper, motivated by Suzuki [32], Kamimura and Takahashi [9], Moudafi [15] and Xu [39], we prove a strong
convergence theorem for countable families of nonexpansive mappings in a Banach space which unifies the results
of [32,9]. Using this result, we improve and extend the results of [32,9]. The proof is closely related to Takahashi [36],
Nakajo, Shimoji and Takahashi [18], and Kikkawa and Takahashi [10,11].

2. Preliminaries

Let E be a real Banach space with norm k · k and let E ∗ denote the dual of E. We denote the value of y ∗ ∈ E ∗ at
∗
x ∈ E by hx, y ∗ i. The duality mapping J from E into 2 E is defined by
J x = {x ∗ ∈ E ∗ : x, x ∗ = kxk2 = kx ∗ k2 }
for every x ∈ E. Let U = {x ∈ E : kxk = 1}. The norm of E is said to be Gâteaux differentiable if for each x, y ∈ U ,
the limit
kx + t yk − kxk
lim (2.1)
t→0 t
exists. In the case, E is called smooth. The norm of E is said to be uniformly Gâteaux differentiable if for each y ∈ U ,
the limit (2.1) is attained uniformly for x ∈ U . We know that if E is smooth, then the duality mapping J is single
valued. Further, if the norm of E is uniformly Gâteaux differentiable, then J is uniformly norm to weak* continuous
on each bounded subset of E; see [25,33]. Let C be a closed convex subset of E. A mapping T : C → C is said
to be nonexpansive if kT x − T yk ≤ kx − yk for all x, y ∈ C. We denote by F(T ) the set of all fixed points of T .
Let I denote the identity operator on E. An operator A ⊂ E × E with domain D(A) = {x ∈ E : Az 6= ∅} and
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 721

range R(A) = {Az : z ∈ D(A)} is said to be accretive if for each xi ∈ D(A) and yi ∈ Axi , i = 1, 2, there exists
S
j ∈ J (x1 − x2 ) such that hy1 − y2 , ji ≥ 0. If A is accretive, then we have
kx1 − x2 k ≤ kx1 − x2 + r (y1 − y2 )k
for all r > 0 and yi ∈ Axi , i = 1, 2. If A is accretive, then we can define, for each r > 0, a nonexpansive
single-valued mapping Jr : R(I + r A) → D(A) by Jr = (I + r A)−1 . It is called the resolvent of A. We also
define the Yosida approximation Ar by Ar = (I − Jr )/r . We know that Ar x ∈ A Jr x for all x ∈ R(I + r A) and
kAr xk ≤ inf{kyk : y ∈ Ax} for all x ∈ D(A) ∩ R(I + r A). We also know that for an accretive operator A,
A−1 0 = F(Jr ) for all r > 0, where A−1 0 = {u ∈ E : 0 ∈ Au}. An accretive operator A is said to be m-accretive
if R(I + r A) = E for all r > 0. A closed convex subset C of a Banach space E is said to have normal structure if
for each bounded closed convex subset of K of C which contains at least two points, there exists an element x of K
which is not a diametral point of K , i.e.,
sup{kx − yk : y ∈ K } < δ(K ),
where δ(K ) is the diameter of K . It is well known that a closed convex subset of a uniformly convex Banach space has
normal structure and a compact convex subset of a Banach space has normal structure; see [33,34] for more details.
The following result was proved by Kirk [13].

Theorem 2.1 (Kirk [13]). Let E be a reflexive Banach space and let C be a nonempty bounded closed convex subset
of E which has normal structure. Let T be a nonexpansive mapping of C into itself. Then F(T ) is nonempty.

A closed convex subset C of a Banach space E is said to have the fixed-point property for nonexpansive mappings if
every nonexpansive mapping of a nonempty bounded closed convex subset of K of C into itself has a fixed point in
K . If C is a closed convex subset of a reflexive Banach space which has normal structure, from Theorem 2.1, C has
the fixed-point property for nonexpansive mappings. See Goebel and Reich [6] and Reich [21,23] for more details.
We denote by N the set of all natural numbers and let µ be a mean on N, i.e., a continuous linear functional on `∞
satisfying kµk = 1 = µ(1). We know that µ is a mean on N if and only if
inf an ≤ µ( f ) ≤ sup an
n∈N n∈N

for each f = (a1 , a2 , . . .) ∈ `∞ . Occasionally, we use µn (an ) instead of µ( f ). Let f = (a1 , a2 , . . .) ∈ `∞ with
an → a and let µ be a Banach limit on N. Then, µ( f ) = µn (an ) = a; see [33,34] for more details. Further, we know
the following result [37]. See also [6].

Theorem 2.2 (Takahashi and Ueda [37]). Let C be a nonempty closed convex subset of a Banach space E with a
uniformly Gâteaux differentiable norm, let {xn } be a bounded sequence of E and let µ be a mean on N. Let z ∈ C.
Then
µn kxn − zk2 = min µn kxn − yk2
y∈C

if and only if µn hy − z, J (xn − z)i ≤ 0 for all y ∈ C, where J is the duality mapping of E.

Let C be a nonempty subset of a Banach space E. Let D be a subset of C and let P be a retraction of C onto D,
i.e., P x = x for each x ∈ D. Then P is said to be sunny [24] if for each x ∈ C and t ≥ 0 with P x + t (x − P x) ∈ C,
P(P x + t (x − P x)) = P x.
A subset D of C is said to be a sunny nonexpansive retract of C if there exists a sunny nonexpansive retraction P of
C onto D. We know that if E is smooth and P is a retraction of C onto D, then P is sunny and nonexpansive if and
only if for each x ∈ C and z ∈ D,
hx − P x, J (z − P x)i ≤ 0. (2.2)
For more details, see [12,20,33,34].
722 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

3. NST-conditions

Let E be a Banach space and let C be a nonempty closed convex subset of E. Motivated by Nakajo, Shimoji and
Takahashi [18], we give the following definitions: Let {Tn } and T be families of nonexpansive mappings of C into
itself such that ∅ 6= F(T ) ⊂ ∩∞n=1 F(Tn ), where F(Tn ) is the set of all fixed points of Tn and F(T ) is the set of
all common fixed points of T . Then, {Tn } is said to satisfy NST-condition (I) with T if for each bounded sequence
{z n } ⊂ C,
lim kz n − Tn z n k = 0
n→∞

implies that limn→∞ kz n − T z n k = 0 for all T ∈ T . In particular, if T = {T }, i.e., T consists of one mapping T ,
then {Tn } is said to satisfy NST-condition (I) with T . {Tn } is also said to satisfy NST-condition (II) if for each bounded
sequence {z n } ⊂ C,
lim kz n+1 − Tn z n k = 0
n→∞

implies that limn→∞ kz n − Tm z n k = 0 for all m ∈ N; see also [16,17]. The following lemma is related to Edelstein
and O’Brien [4, Theorem 1]. We follow the ideas of [5,31] for the proof.

Lemma 3.1. Let C be a nonempty closed convex subset of a Banach space E and let T be a nonexpansive mapping
of C into itself with F(T ) 6= ∅. Let {βn } be a sequence of real numbers with 0 < a ≤ βn ≤ b < 1 and let B be a
nonempty bounded subset of C. Define a mapping Sn of C into itself by
Sn x = S(βn )x = (1 − βn )x + βn T x for all x ∈ C
and put an = supx∈B kT S n x − S n xk for all n ∈ N, where S n = Sn Sn−1 · · · S1 . Then, an → 0. In particular, for any
m ∈ N,
lim sup kSm S n x − S n xk = 0.
n→∞ x∈B

Proof. For any x, y ∈ C and n ∈ N, we have

kSn x − Sn yk = k(1 − βn )x + βn T x − (1 − βn )y − βn T yk
≤ (1 − βn )kx − yk + βn kT x − T yk
≤ (1 − βn )kx − yk + βn kx − yk
= kx − yk.
We also have, for any n, i ∈ N and x ∈ C,
kT S n+i x − S n+i xk ≤ kT S n+i x − T S n+i−1 xk + kT S n+i−1 x − S n+i xk
≤ kS n+i x − S n+i−1 xk + kT S n+i−1 x − (1 − βn+i )S n+i−1 x − βn+i T S n+i−1 xk
= βn+i kT S n+i−1 x − S n+i−1 xk + (1 − βn+i )kT S n+i−1 x − S n+i−1 xk
= kT S n+i−1 x − S n+i−1 xk.
Similarly, we have, for any n, i ∈ N and x ∈ C,

kT S n+i x − S n+i xk ≤ kT S n x − S n xk. (3.1)

In particular, we have, for any n ∈ N,

an = sup kT S n x − S n xk ≤ sup kT S n−1 x − S n−1 xk = an−1 . (3.2)

x∈B x∈B
Let k ∈ N and take j ∈ N ∪ {0} with 0 ≤ j ≤ k. For any n ∈ N, define An ( j, k) = 0 if j = k and
An ( j, k) = βn+ j+1 + · · · + βn+k if j < k. We have, for any n ∈ N,

sup kT S n+k x − T S n+ j xk ≤ An ( j, k) sup kT S n x − S n xk. (3.3)

x∈B x∈B
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 723

In fact, (3.3) holds for j = k. Suppose that j < k. Then, by (3.1) we have, for any x ∈ C,
k−1
X
kT S n+k x − T S n+ j xk ≤ kT S n+i+1 x − T S n+i xk
i= j
k−1
X
≤ kS n+i+1 x − S n+i xk
i= j
k−1
X
= k(1 − βn+i+1 )S n+i x + βn+i+1 T S n+i x − S n+i xk
i= j
k−1
X
= βn+i+1 kT S n+i x − S n+i xk
i= j
k−1
X
≤ βn+i+1 kT S n x − S n xk
i= j
= An ( j, k)kT S n x − S n xk.
So, (3.3) holds. Put c = limn→∞ supx∈B kT S n x − S n xk and fix k ∈ N. Then, for any ε > 0, by the definition of c
and (3.3), we can take m ∈ N such that for all n ∈ N with n ≥ m, the following (3.4)–(3.6) hold:
sup kT S n x − S n xk ≤ c + ε, (3.4)
x∈B

sup kT S n+k x − S n+k xk ≥ c − ε (3.5)

x∈B

and
sup kT S n+k x − T S n+ j xk ≤ An ( j, k) sup kT S n x − S n xk ( j = 0, 1, 2, . . . , k). (3.6)
x∈B x∈B

For such k, n, j ∈ N with n ≥ m and 0 ≤ j ≤ k, we have

(k − j + 1)(k − j + 2)
sup kT S n+k x − S n+ j xk ≥ (1 + An ( j, k))c − ε. (3.7)
x∈B 2(1 − b)k− j
In fact, for j ∈ N with j = k, from (3.5) we have
1·2
sup kT S n+k x − S n+k xk ≥ c − ε = c − ε.
x∈B 2
So, (3.7) holds. Suppose that 0 ≤ j ≤ k − 1. By the mathematical induction, let us prove (3.7). Assume that
(k − j)(k − j + 1)
sup kT S n+k x − S n+ j+1 xk ≥ (1 + An ( j + 1, k))c − ε.
x∈B 2(1 − b)k− j−1
Then, from (3.4) and (3.6) and 0 < b < 1, we have
(k − j)(k − j + 1)
(1 + An ( j + 1, k))c − ε ≤ sup kT S n+k x − S n+ j+1 xk
2(1 − b)k− j−1 x∈B
 
≤ sup (1 − βn+ j+1 )kT S n+k
x−S n+ j
xk + βn+ j+1 kT S n+k x − T S n+ j xk
x∈B
≤ (1 − βn+ j+1 ) sup kT S n+k x − S n+ j xk + βn+ j+1 An ( j, k) sup kT S n x − S n xk
x∈B x∈B
≤ (1 − βn+ j+1 ) sup kT S n+k
x−S n+ j
xk + βn+ j+1 An ( j, k)(c + ε)
x∈B
≤ (1 − βn+ j+1 ) sup kT S n+k x − S n+ j xk + βn+ j+1 An ( j, k)c + (k − j)ε
x∈B
724 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

(k − j + 1)
≤ (1 − βn+ j+1 ) sup kT S n+k x − S n+ j xk + βn+ j+1 An ( j, k)c + ε
x∈B (1 − b)k− j−1
and hence
(1 − βn+ j+1 ) sup kT S n+k x − S n+ j xk
x∈B
(k − j + 2)(k − j + 1)
≥ (1 + An ( j + 1, k))c − βn+ j+1 An ( j, k)c − ε
2(1 − b)k− j−1
(k − j + 2)(k − j + 1)
= (1 − βn+ j+1 )(1 + An ( j, k))c − ε.
2(1 − b)k− j−1
So, we get
(k − j + 2)(k − j + 1)
sup kT S n+k x − S n+ j xk ≥ (1 + An ( j, k))c − ε.
x∈B 2(1 − b)k− j
This implies that (3.7) holds for any j ∈ N ∪ {0} with 0 ≤ j ≤ k. In the case of j = 0, (3.7) implies that
(k + 2)(k + 1)
sup kT S n+k x − S n xk ≥ (1 + An (0, k))c − ε. (3.8)
x∈B 2(1 − b)k
On the other hand, from (3.4) and (3.6) and 0 < b < 1, we have
 
sup kT S n+k x − S n xk ≤ sup kT S n+k x − T S n xk + kT S n x − S n xk
x∈B x∈B
≤ An (0, k) sup kT S n x − S n xk + sup kT S n x − S n xk
x∈B x∈B
≤ An (0, k)(c + ε) + (c + ε)
= (1 + An (0, k))(c + ε)
≤ (1 + An (0, k))c + (1 + k)ε
(k + 2)(k + 1)
≤ (1 + An (0, k))c + ε. (3.9)
2(1 − b)k
By (3.8) and (3.9) we have
(k + 1)(k + 2)
sup kT S n+k x − S n xk − (1 + An (0, k))c ≤ ε. (3.10)
x∈B 2(1 − b)k
From (3.10), we have

lim sup kT S n+k x − S n xk − (1 + An (0, k))c = 0. (3.11)

n→∞ x∈B

For any k ∈ N, we have

(1 + ka)c ≤ (1 + An (0, k))c ≤ (1 + An (0, k))c − sup kT S n+k x − S n xk + sup kT S n+k x − S n xk

x∈B x∈B

≤ (1 + An (0, k))c − sup kT S n+k x − S n xk + M,

x∈B

where supx∈B kT S n+k x − S n xk ≤ M. From (3.11) we have

(1 + ka)c ≤ M. (3.12)
Since k ∈ N is arbitrary and 0 < a < 1, from (3.12) we have
c = lim sup kT S n x − S n xk = 0.
n→∞ x∈B
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 725

This implies that limn→∞ an = 0. Since kSm S n x − S n xk = βm kT S n x − S n xk for every m ∈ N, we have

sup kSm S n x − S n xk = sup βm kT S n x − S n xk.
x∈B x∈B

So, we have limn→∞ supx∈B kSm S n x − S n xk = 0. 

Lemma 3.2. Let C be a nonempty closed convex subset of a Banach space E and let T be a nonexpansive mapping
of C into itself with F(T ) 6= ∅. For a nonempty bounded subset B of C and n ∈ N, define a mapping f n of [0, 1]n
into (−∞, ∞) by
f n (βn , βn−1 , . . . , β1 ) = sup kT U n x − U n xk
x∈B

for all (βn , βn−1 , . . . , β1 ) ∈ [0, 1]n , where U n = S(βn )S(βn−1 ) · · · S(β1 ) and S(βk )x = (1 − βk )x + βk T x for all
x ∈ C and k ∈ {1, 2, . . . , n}. Then, f n is continuous.

Proof. Take z ∈ F(T ). For k ∈ N, put U k = S(βk )S(βk−1 ) . . . S(β1 ) and V k = S(αk )S(αk−1 ) . . . S(α1 ), where
(βk , βk−1 , . . . , β1 ), (αk , αk−1 , . . . , α1 ) ∈ [0, 1]k . Then, we can prove that for any x ∈ B,
n
X
kU n x − V n xk ≤ 2M |βi − αi |,
i=1

where M = supx∈B kx − zk. In fact, we prove this inequality by the mathematical induction. For x ∈ B, we have

kU 1 x − V 1 xk = kS(β1 )x − S(α1 )xk

= |β1 − α1 |kx − T xk
= |β1 − α1 |kx − z + z − T xk
≤ 2M|β1 − α1 |.
Suppose that for any x ∈ B,
k
X
kU k x − V k xk ≤ 2M |βi − αi |
i=1

for some k ∈ N. Then, we have

kU k+1 x − V k+1 xk = kU k+1 x − S(αk+1 )U k x + S(αk+1 )U k x − V k+1 xk
≤ k(1 − βk+1 )U k x + βk+1 T U k x − (1 − αk+1 )U k x − αk+1 T U k xk
+ kS(αk+1 )U k x − S(αk+1 )V k xk
≤ |βk+1 − αk+1 |kU k x − T U k xk + kU k x − V k xk
Xk
≤ 2M|βk+1 − αk+1 | + 2M |βi − αi |
i=1
k+1
X
= 2M |βi − αi |.
i=1

So, we have that for any x ∈ B,

n
X
|kT U n x − U n xk − kT V n x − V n xk| ≤ 2kU n x − V n xk ≤ 4M |βi − αi |.
i=1

This implies that

n
X
sup kT U n x − U n xk ≤ sup kT V n x − V n xk + 4M |βi − αi |
x∈B x∈B i=1
726 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

and
n
X
sup kT V n x − V n xk ≤ sup kT U n x − U n xk + 4M |βi − αi |.
x∈B x∈B i=1
Therefore, we have
n
X
| f n (βn , βn−1 , . . . , β1 ) − f n (αn , αn−1 , . . . , α1 )| ≤ 4M |βi − αi |
i=1
and hence f n is continuous. 
Using Lemmas 3.1 and 3.2 we obtain the following:

Theorem 3.3. Let C be a nonempty closed convex subset of a Banach space E and let T be a nonexpansive mapping
of C into itself such that F(T ) is nonempty. For any n ∈ N and βn ∈ R with 0 < a ≤ βn ≤ b < 1, define Sn : C → C
as follows:
Sn x = (1 − βn )x + βn T x for all x ∈ C.
Then, {Sn } satisfies NST-condition (I) with T and NST-condition (II).
Proof. Since F(T ) = F(Sn ) for every n ∈ N, we have ∅ 6= F(T ) = ∩∞
n=1 F(Sn ). Assume that {z n } is a bounded
sequence of C and
lim kz n − Sn z n k = 0. (3.13)
n→∞
Since kz n − Sn z n k = βn kz n − T z n k and 0 < a ≤ βn ≤ b < 1, from (3.13) we have
lim kz n − T z n k = 0.
n→∞
This implies that {Sn } satisfies NST-condition (I) with T . Next, assume that {z n } is a bounded sequence of C and
lim kz n+1 − Sn z n k = 0. (3.14)
n→∞
Fix k ∈ N and set S = Sk . For any ε > 0, from Lemmas 3.1 and 3.2 there exists m ∈ N such that for any n ∈ N
sup kSSn+m−1 Sn+m−2 · · · Sn z n − Sn+m−1 Sn+m−2 · · · Sn z n k ≤ ε.
n∈N

Put Wnn+m−1 = Sn+m−1 Sn+m−2 · · · Sn . Then, for any n ∈ N, we have

kz n+m − Sz n+m k
≤ kz n+m − Wnn+m−1 z n k + kWnn+m−1 z n − SWnn+m−1 z n k + kSWnn+m−1 z n − Sz n+m k
≤ 2kz n+m − Wnn+m−1 z n k + sup kWnn+m−1 z n − SWnn+m−1 z n k
n∈N
≤ 2(kz n+m − Sn+m−1 z n+m−1 k + kSn+m−1 z n+m−1 − Sn+m−1 Sn+m−2 z n+m−2 k
n+m−1
+ · · · + kWn+1 z n+1 − Wnn+m−1 z n k) + sup kWnn+m−1 z n − SWnn+m−1 z n k
n∈N
m
X
≤2 kz n+m−(i−1) − Sn+m−i z n+m−i k + sup kWnn+m−1 z n − SWnn+m−1 z n k
i=1 n∈N
Xm
≤2 kz n+m−(i−1) − Sn+m−i z n+m−i k + ε.
i=1
So, we have
limn→∞ kz n − Sz n k = limn→∞ kz n+m − Sz n+m k ≤ ε.
Since ε > 0 is arbitrary, we have limn→∞ kz n − Sz n k ≤ 0. This implies that limn→∞ kz n − Sz n k = 0. Therefore,
{Sn } satisfies NST-condition (II). 
We know the following lemma for resolvents of accretive operators; see [33].
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 727

Lemma 3.4. Let E be a Banach space and let A ⊂ E × E be an accretive operator. Let r, λ > 0 and D(A) ⊂
R(I + λA). Then,
1 1
kJr x − Jλ Jr xk ≤ kx − Jr xk
λ r
for every x ∈ R(I + r A).
Proof. For any x ∈ R(I + r A), we have
1 1
kJr x − Jλ Jr xk = kAλ Jr xk ≤ |A Jr x| ≤ kAr xk = kx − Jr xk,
λ r
where |Ax| = inf{kyk : y ∈ Ax}. 
Using Lemma 3.4, we also have the following theorem.

Theorem 3.5. Let C be a nonempty closed convex subset of a Banach space E and let A ⊂ E × E be an accretive
operator such that
\
D(A) ⊂ C ⊂ R(I + λA)
λ>0

and A−1 0 6= ∅. Let {λn } be a sequence of real numbers such that λn ∈ (0, ∞) and limn→∞ λn = ∞. Define Sn = Jλn
for any n ∈ N. Then, {Sn } satisfies NST-condition (I) with J1 and NST-condition (II), where J1 = (I + A)−1 .
Proof. From [33], we know that Jr is a nonexpansive mapping of C into itself for each r > 0 and ∞ n=1 F(Jλn ) =
T
F(J1 ) = A−1 0. Assume that {z n } is a bounded sequence of C and limn→∞ kz n − Sn z n k = 0. Since
kz n − J1 z n k ≤ kz n − Jλn z n k + kJλn z n − J1 Jλn z n k + kJ1 Jλn z n − J1 z n k
1
≤ 2kz n − Jλn z n k + kz n − Jλn z n k
λn
for all n ∈ N, we have
lim kz n − J1 z n k = 0
n→∞
and hence {Sn } satisfies NST-condition (I) with J1 . Assume that {z n } is a bounded sequence of C and limn→∞ kz n+1 −
Jλn z n k = 0. Fix m ∈ N. Then, by Lemma 3.4 we have
kz n+1 − Jλm z n+1 k
≤ kz n+1 − Jλn z n k + kJλn z n − Jλm Jλn z n k + kJλm Jλn z n − Jλm z n+1 k
λm
≤ 2kz n+1 − Jλn z n k + kz n − Jλn z n k
λn
for all n ∈ N. So, we have limn→∞ kz n − Jλm z n k = 0 and hence {Sn } satisfies NST-condition (II). 

4. Strong convergence theorem of Browder’s type

We prove a strong convergence theorem of Browder’s type for a countable family of nonexpansive mappings in a
Banach space. The proof is closely related to [24,26,37].

Theorem 4.1. Let E be a reflexive Banach space with a uniformly Gâteaux differentiable norm and let C be a
nonempty closed convex subset of E which has the fixed-point property for nonexpansive mappings. Let T be a
nonexpansive mappingTof C into itself and let {Tn } be a family of nonexpansive mappings of C into itself which
satisfies ∅ 6= F(T ) ⊂ ∞
n=1 F(Tn ). Further, suppose that {Tn } satisfies NST-condition (I) with T . Define a sequence
{xn } ⊂ C as follows:
xn = αn f (xn ) + (1 − αn )Tn xn , n = 1, 2, 3, . . . ,
where {αn } ⊂ (0, 1) satisfies limn→∞ αn = 0 and f : C → C is a-contractive with 0 ≤ a < 1. Then, {xn } converges
strongly to u ∈ F(T ), where u = PF(T ) f (u) and PF(T ) is a sunny nonexpansive retraction of C onto F(T ).
728 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

Proof. Let Un x = αn f (x) + (1 − αn )Tn x for all x ∈ C. Then, we have

kUn x − Un yk = kαn f (x) − (1 − αn )Tn x − (αn f (y) + (1 − αn )Tn y)k
≤ αn k f (x) − f (y)k + (1 − αn )kTn x − Tn yk
≤ αn akx − yk + (1 − αn )kx − yk
= (1 − αn (1 − a))kx − yk
and hence Un is a contraction of C into itself. So, for each n ∈ N, there exists a unique element xn ∈ C such that
xn = αn f (xn ) + (1 − αn )Tn xn .
By (I), we get F(T ) = ∞
T∞
n=1 F(Tn ). Let z ∈ n=1 F(Tn ). We obtain, for any n ∈ N,
T

kxn − zk = kαn ( f (xn ) − z) + (1 − αn )(Tn xn − z)k

≤ αn k f (xn ) − zk + (1 − αn )kxn − zk
≤ αn (k f (xn ) − f (z)k + k f (z) − zk) + (1 − αn )kxn − zk
≤ αn (akxn − zk + k f (z) − zk) + (1 − αn )kxn − zk
and hence
αn (1 − a)kxn − zk ≤ αn k f (z) − zk.

1−a k f (z) −
1
So, we have kxn − zk ≤ zk for all n ∈ N and hence {xn } is bounded. Further, we have
kxn − Tn xn k = kαn f (xn ) + (1 − αn )Tn xn − Tn xn k
= αn k f (xn ) − Tn xn k
≤ αn (k f (xn ) − f (z)k + k f (z) − zk + kz − Tn xn k)
≤ αn (akxn − zk + k f (z) − zk + kxn − zk).
So, from αn → 0, we get limn→∞ kxn − Tn xn k = 0. From (I), we have
lim kxn − T xn k = 0.
n→∞
We get
αn h f (xn ) − z, J (xn − z)i = hαn f (xn ) − αn z, J (xn − z)i
= hxn − (1 − αn )Tn xn − αn z, J (xn − z)i
= αn hxn − z, J (xn − z)i + (1 − αn ) hxn − Tn xn − (z − Tn z), J (xn − z)i
= αn kxn − zk2 + (1 − αn ) {hxn − z, J (xn − z)i − hTn xn − Tn z, J (xn − z)i}
n o
≥ αn kxn − zk2 + (1 − αn ) kxn − zk2 − kTn xn − Tn zkkxn − zk
≥ αn kxn − zk2
T∞
for every n ∈ N and z ∈ n=1 F(Tn ). So, we obtain
kxn − zk2 ≤ h f (xn ) − z, J (xn − z)i .
We also get
1 − αn
hxn − f (xn ), J (xn − z)i = hTn xn − xn , J (xn − z)i
αn
1 − αn
= {hTn xn − z, J (xn − z)i − hxn − z, J (xn − z)i}
αn
1 − αn n o
= hTn xn − z, J (xn − z)i − kxn − zk2
αn
1 − αn n o
≤ kxn − zkkxn − zk − kxn − zk2 = 0 (4.1)
αn
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 729

for every n ∈ N and z ∈ ∞ n=1 F(Tn ). Let {x n i } be a subsequence of {x n } and let µ be a Banach limit. Without loss of
T
generality, put u i = xn i for all i ∈ N and let g be a real-valued function on C defined by
g(y) = µi ku i − yk2
for every y ∈ C. By [33], we know that g is continuous and convex. Further, g satisfies that g(yn ) → ∞ as
kyn k → ∞. So, there exists x0 ∈ C such that g(x0 ) = inf y∈C g(y). Define the set
K = {v ∈ C : g(v) = min{g(x) : x ∈ C}}.
Then, K is nonempty, bounded, closed and convex; see [33] for more details. Further, we have that K is invariant
under T , that is, T K ⊂ K . In fact, since
kT u i − u i k → 0 (4.2)
as i → ∞, we have, for any v ∈ K ,
µi ku i − T vk2 ≤ µi (ku i − T u i k + kT u i − T vk)2
= µi kT u i − T vk2 ≤ µi ku i − vk2
and hence T v is an element of K . Since C has the fixed-point property for nonexpansive mappings, T has a fixed
point v0 in K . Since v0 is a minimizer of the function g on C, by Theorem 2.2 we have
µi hx − v0 , J (u i − v0 )i ≤ 0
for all x ∈ C. Putting x = f (v0 ), we have
µi h f (v0 ) − v0 , J (u i − v0 )i ≤ 0. (4.3)
Since
ku i − v0 k2 = hu i − v0 , J (u i − v0 )i
= hαni f (u i ) + (1 − αn i )Tn i u i − v0 , J (u i − v0 )i
= (1 − αn i )hTn i u i − v0 , J (u i − v0 )i + αn i h f (u i ) − v0 , J (u i − v0 )i
≤ (1 − αn i )ku i − v0 k2 + αn i h f (u i ) − v0 , J (u i − v0 )i ,
we have
ku i − v0 k2 ≤ h f (u i ) − v0 , J (u i − v0 )i (4.4)
and hence, from (4.3) and (4.4),
µi ku i − v0 k2 ≤ µi h f (u i ) − v0 , J (u i − v0 )i
= µi h f (u i ) − f (v0 ) + f (v0 ) − v0 , J (u i − v0 )i
≤ µi h f (u i ) − f (v0 ), J (u i − v0 )i
≤ µi (k f (u i ) − f (v0 )k ku i − v0 k)
≤ aµi ku i − v0 k2 .
So, we have (1 − a)µi ku i − v0 k2 ≤ 0. This implies that µi ku i − v0 k2 = 0. From
0 ≤ limn→∞ ku i − v0 k2 ≤ µi ku i − v0 k2 = 0,
there exists a subsequence {u i j } of {u i } such that u i j → v0 as j → ∞. On the other hand, let {xn i } and {xn j } be
subsequences of {xn } such that xni → z 1 and xn j → z 2 . Then, we have
kz 1 − T z 1 k ≤ kz 1 − xni k + kxn i − T xn i k + kT xn i − T z 1 k → 0
and hence z 1 = T z 1 . So, we have z 1 ∈ ∞
T∞
n=1 F(Tn ). Similarly, z 2 ∈ n=1 F(Tn ). By (4.1) we have
T

xni − f (xn i ), J (xn i − z 2 ) ≤ 0

730 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

and
xn j − f (xn j ), J (xn j − z 1 ) ≤ 0.
Since
xn i − f (xn i ), J (xni − z 2 ) − hz 1 − f (z 1 ), J (z 1 − z 2 )i
≤ xni − f (xni ), J (xn i − z 2 ) − z 1 − f (z 1 ), J (xn i − z 2 )
+ z 1 − f (z 1 ), J (xn i − z 2 ) − hz 1 − f (z 1 ), J (z 1 − z 2 )i → 0,
we have
hz 1 − f (z 1 ), J (z 1 − z 2 )i ≤ 0.
Similarly, we have
hz 2 − f (z 2 ), J (z 2 − z 1 )i ≤ 0.
So, we get
hz 1 − f (z 1 ) − (z 2 − f (z 2 )), J (z 1 − z 2 )i ≤ 0
and hence
kz 1 − z 2 k2 ≤ h f (z 1 ) − f (z 2 ), J (z 1 − z 2 )i
≤ akz 1 − z 2 k2 .
T∞
This implies that z 1 = z 2 . Therefore, {xn } converges strongly to some element of n=1 F(Tn ) = F(T ). Let xn → u.
From (4.1), we have that
hxn − f (xn ), J (xn − z)i ≤ 0
for all z ∈ F(T ) and hence
hu − f (u), J (u − z)i ≤ 0
for all z ∈ F(T ). This implies that P f (u) = u. 

5. Strong convergence theorem of Halpern’s type

In this section, we first prove a strong convergence theorem of Halpern’s type for a countable family of
nonexpansive mappings in a Banach space.

Theorem 5.1. Let E be a reflexive Banach space with a uniformly Gâteaux differentiable norm and let C be a
nonempty closed convex subset of E which has the fixed-point property for nonexpansive mappings. Let T be a
nonexpansive mapping of T C into itself and let {Tn } be a family of nonexpansive mappings of C into itself which
∞
satisfies ∅ 6= F(T ) ⊂ n=1 F(Tn ). Further, suppose that {Tn } satisfies NST-condition (I) with T and NST-
condition (II). Let {xn } be a sequence in C as follows: x1 = x ∈ C and
xn+1 = αn f (xn ) + (1 − αn )Tn xn , n = 1, 2, 3, . . . ,
where {αn } ⊂ [0, 1) satisfies limn→∞ αn = 0, andT ∞ n=1 αn = ∞, and f : C → C is a-contractive with 0 ≤ a < 1.
P
Then, {xn } converges strongly to u ∈ F(T ) = ∞ n=1 F(Tn ), where u = P f (u) and P is a sunny nonexpansive
retraction of C onto F(T ).
Proof. We have F(T ) = ∞
T∞
n=1 F(Tn ) by (I). Let z ∈ n=1 F(Tn ). We have kx n − zk ≤ max{kx − zk, 1−a k f (z) −
T 1

zk} = A for all n ∈ N. In fact, kx1 − zk = kx − zk ≤ A is obvious. Suppose that kxk − zk ≤ A for some k ∈ N.
Then, we have
kxk+1 − zk = kαk f (xk ) + (1 − αk )Tk xk − zk
≤ αk k f (xk ) − zk + (1 − αk )kTk xk − zk
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 731

≤ αk k f (xk ) − zk + (1 − αk )kxk − zk
≤ αk akxk − zk + αk k f (z) − zk + (1 − αk )kxk − zk
1
= (1 − αk (1 − a))kxk − zk + αk (1 − a) k f (z) − zk
1−a
≤ (1 − αk (1 − a))A + αk (1 − a)A = A.
So, {xn } is bounded. We also have that { f (xn )} and {Tn xn } are bounded. Since αn → 0 and
kxn+1 − Tn xn k = kαn f (xn ) + (1 − αn )Tn xn − Tn xn k
= αn k f (xn ) − Tn xn k,
we have limn→∞ kxn+1 − Tn xn k = 0. From (II), we have limn→∞ kxn − Tm xn k = 0 for all m ∈ N. Let {rm } be a
sequence in (0, 1) such that rm → 0 and let {ym } be a sequence in C such that
ym = rm f (ym ) + (1 − rm )Tm ym
for every m ∈ N. We know from Theorem 4.1 that ym → z ∈ F(T ). Fix m ∈ N. Then, we have
kxn − Tm ym k2 ≤ kxn − Tm xn k2 + kxn − ym k2 + 2kxn − Tm xn kkxn − ym k. (5.1)
From (1 − rm )(xn − Tm ym ) = (xn − ym ) − rm (xn − f (xm )), we also have that
(1 − rm )2 kxn − Tm ym k2 ≥ kxn − ym k2 − 2rm hxn − f (xm ), J (xn − ym )i
= (1 − 2rm )kxn − ym k2 + 2rm h f (xm ) − ym , J (xn − ym )i (5.2)
for every n, m ∈ N. From (5.1) and (5.2), we get
(1 − rm )2 kxn − Tm xn k2 + (1 − rm )2 kxn − ym k2 + 2(1 − rm )2 kxn − Tm xn kkxn − ym k
≥ (1 − 2rm )kxn − ym k2 + 2rm h f (xm ) − ym , J (xn − ym )i
and hence
rm (1 − rm )2
kxn − ym k2 + (kxn − Tm xn k2 + 2kxn − Tm xn kkxn − ym k) ≥ h f (xm ) − ym , J (xn − ym )i
2 2rm
for every m, n ∈ N. Since limn→∞ kxn − Tm xn k = 0 for every m ∈ N, we have
rm
lim sup h f (xm ) − ym , J (xn − ym )i ≤ lim sup kxn − ym k2
n→∞ 2 n→∞
for all m ∈ N. Let ε > 0. Since limm→∞ rm = 0, there exists m 1 ∈ N such that for every m ≥ m 1 ,
ε
lim sup h f (xm ) − ym , J (xn − ym )i < .
n→∞ 3
Since E is norm to weak* uniformly continuous on bounded subsets of E and ym → z, we have that for every
m ≥ m2,
ε
| h f (z) − z, J (xn − z)i − h f (z) − z, J (xn − ym )i | <
3
and
ε
| h f (z) − z, J (xn − ym )i − h f (ym ) − ym , J (xn − ym )i | < .
3
So, there exists m 3 ∈ N such that for every m ≥ m 3
lim sup h f (z) − z, J (xn − z)i ≤ lim sup | h f (z) − z, J (xn − z)i − h f (z) − z, J (xn − ym )i |
n→∞ n→∞
+ lim sup | h f (z) − z, J (xn − ym )i − h f (ym ) − ym , J (xn − ym )i |
n→∞
+ lim sup | h f (ym ) − ym , J (xn − ym )i |
n→∞
ε ε ε
< + + = ε.
3 3 3
732 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

So, we obtain that

lim sup h f (z) − z, J (xn − z)i ≤ 0. (5.3)
n→∞
From xn+1 − z = αn ( f (xn ) − z) + (1 − αn )(Tn xn − z), we have
(1 − αn )2 kTn xn − zk2 ≥ kxn+1 − zk2 − 2αn h f (xn ) − z, J (xn+1 − z)i
and hence
kxn+1 − zk2 ≤ (1 − αn )2 kTn xn − zk2 + 2αn h f (xn ) − z, J (xn+1 − z)i
≤ (1 − αn )2 kxn − zk2 + 2αn h f (xn ) − f (z), J (xn+1 − z)i + 2αn h f (z) − z, J (xn+1 − z)i
≤ (1 − αn )2 kxn − zk2 + 2αn a kxn − zk kxn+1 − zk + 2αn h f (z) − z, J (xn+1 − z)i
≤ (1 − αn )2 kxn − zk2 + αn a{kxn − zk2 + kxn+1 − zk2 } + 2αn h f (z) − z, J (xn+1 − z)i .
This implies that
(1 − αn )2 + aαn 2αn
kxn+1 − zk2 ≤ kxn − zk2 + h f (z) − z, J (xn+1 − z)i
1 − aαn 1 − aαn
1 − 2αn + aαn αn2 2αn
= kxn − zk2 + kxn − zk2 + h f (z) − z, J (xn+1 − z)i
1 − aαn 1 − aαn 1 − aαn
αn M
   
2(1 − a)αn 2(1 − a)αn 1
≤ 1− kxn − zk2 + + h f (z) − z, J (xn+1 − z)i ,
1 − aαn 1 − aαn 2(1 − a) 1 − a
where M = supn kxn − zk2 . Put
2(1 − a)αn
βn = .
1 − aαn
We obtain ∞n=1 βn = ∞ and
P

lim βn = 0.
n→∞
αn M ε
Let ε > 0. From (5.3), there exists m ∈ N such that 2(1−a) ≤ 2 and
1 ε
h f (z) − z, J (xn − z)i ≤
1−a 2
for all n ≥ m. Then we have
ε ε
kxn+1 − zk2 ≤ (1 − βn ) kxn − zk2 + βn +
2 2
= (1 − βn ) kxn − zk2 + (1 − (1 − βn ))ε.
Similarly, we have
!
m+n−1
Y m+n−1
Y
kxm+n − zk ≤ 2
(1 − βk ) kxm − zk + 1 −
2
(1 − βk ) ε.
k=m k=m
P∞ Q∞
We know that k=m βk = ∞ implies k=m (1 − βk ) = 0; see [34]. Therefore, we have
lim sup kxn − zk = lim sup kxm+n − zk ≤ ε.
2 2
n→∞ n→∞
Since ε > 0 is arbitrary, we obtain
lim sup kxn − zk2 ≤ 0.
n→∞
So, we conclude that {xn } converges strongly to z ∈ F(T ) = ∩∞
n=1 F(Tn ). We also know from Theorem 4.1 that
z = P f (z). 
Using Theorems 3.3 and 5.1, we obtain the following result:
 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734 733

Theorem 5.2. Let E be a reflexive Banach space with a uniformly Gatêaux differentiable norm. Let C be a nonempty
closed convex subset of E which has the fixed-point property for nonexpansive mappings and let T : C → C be a
nonexpansive mapping such that F(T ) is nonempty and let f be a contraction of C into itself. Define a sequence {xn }
of C as follows: x1 = x ∈ C and
xn+1 = αn f (xn ) + (1 − αn )((1 − βn )xn + βn T xn ) for all n ∈ N,
where {αn } ⊂ (0, 1) and {βn } ⊂ (0, 1) satisfy the following conditions:
∞
X
αn → 0, αn = ∞ and 0 < a ≤ βn ≤ b < 1.
n=1

Then, the sequence {xn } converges strongly to a fixed point of T .

Proof. Put Tn x = (1 − βn )x + βn T x for all n ∈ N and x ∈ C. Then, from Theorem 3.3, {Tn } satisfies NST-condition
(I) with T and NST-condition (II). So, from Theorem 5.1 we obtain the desired result. 
Using Theorems 3.5 and 5.1, we also obtain the following strong convergence theorem which was proved by
Takahashi [36]. This result is of particular interest because in general the proximal point algorithm [28] provides only
weak convergence; see [7,1] for the counterexamples.

Theorem 5.3. Let E be a reflexive Banach space with a uniformly Gatêaux differentiable norm and let C be a
nonempty closed convex subset of E which has the fixed-point property for nonexpansive mappings. Let A ⊂ E × E
be an accretive operator with A−1 0 6= ∅ satisfying
\
D(A) ⊂ C ⊂ R(I + t A),
t>0

where D(A) is the closure of D(A) and let f be a contraction of C into itself. Let {xn } be a sequence of C defined
by x1 = x ∈ C and
xn+1 = αn f (xn ) + (1 − αn )Jtn xn for all n ∈ N,
where {αn } ⊂ (0, 1) and {tn } ⊂ (0, ∞) satisfy the following conditions:
∞
X
αn → 0, αn = ∞ and tn → ∞.
n=1

Then, the sequence {xn } converges strongly to u ∈ A−1 0, where u = P f (u) and P is a sunny nonexpansive retraction
of C onto A−1 0.
Proof. Put Tn x = Jtn x for all n ∈ N and x ∈ C. Then, from Theorem 3.5, {Tn } satisfies NST-condition (I) with T and
NST-condition (II). So, from Theorem 5.1 we obtain the desired result. 

References

[1] H.H. Bauschke, E. Matoušková, S. Reich, Projection and proximal point methods: convergence results and counterexamples, Nonlinear Anal.
56 (2004) 715–738.
[2] H. Brézis, P.-L. Lions, Produits infinis de résolvantes, Israel J. Math. 29 (1978) 329–345.
[3] R.E. Bruck, S. Reich, Nonexpansive projections and resolvents of accretive operators in Banach spaces, Houston J. Math. 3 (1977) 459–470.
[4] M. Edelstein, R.C. O’Brien, Nonexpansive mappings, asymptotic regularity and successive approximations, J. London Math. Soc. 17 (1978)
547–554.
[5] K. Eshita, W. Takahashi, Suzuki’s lemma in convex metric spaces (in press).
[6] K. Goebel, S. Reich, Uniform Convexity, Hyperbolic Geometry, and Nonexpansive Mappings, Marcel Dekker, New York and Basel, 1984.
[7] O. Güller, On the convergence of the proximal point algorithm for convex minimization, SIAM J. Control Optim. 29 (1991) 403–419.
[8] S. Kamimura, W. Takahashi, Approximating solutions of maximal monotone operators in Hilbert space, J. Approx. Theory 106 (2000)
226–240.
[9] S. Kamimura, W. Takahashi, Weak and strong convergence of solutions to accretive operator inclusions and applications, Set-Valued Anal. 8
(2000) 361–374.
734 W. Takahashi / Nonlinear Analysis 70 (2009) 719–734

[10] M. Kikkawa, W. Takahashi, Strong convergence theorems by the viscosity approximation method for a countable family of nonexpansive
mappings, Taiwanese J. Math. (in press).
[11] M. Kikkawa, W. Takahashi, in: W. Takahashi, T. Tanaka (Eds.), Strong convergence theorems by the viscosity approximation methods for
nonexpansive mappings in Banach spaces, in: Convex Analysis and Nonlinear Analysis, Yokohama Publishers, Yokohama, 2007, pp. 227–238.
[12] E. Kopecká, S. Reich, Nonexpansive retracts in Banach spaces, vol. 77, Banach Center Publications, 2007, pp. 161–174.
[13] W.A. Kirk, A fixed point theorem for mappings which do not increase distances, Amer. Math. Monthly 72 (1965) 1004–1006.
[14] B. Martinet, Régularisation d’inéquations variationnelles par approximations successives, Rev. Francaise Informat. Recherche Opérationnelle
4 (1970) 154–158.
[15] A. Moudafi, Viscosity approximation methods for fixed-point problems, J. Math. Anal. Appl. 241 (2000) 46–55.
[16] K. Nakajo, K. Shimoji, W. Takahashi, in: W. Takahashi, T. Tanaka (Eds.), A weak convergence theorem by products of mappings in Hilbert
spaces, in: Nonlinear Analysis and Convex Analysis, Yokohama Publishers, Yokohama, 2004, pp. 381–390.
[17] K. Nakajo, K. Shimoji, W. Takahashi, Strong convergence theorems of Halpern’s type for families of nonexpansive mappings in Hilbert spaces
(in press).
[18] K. Nakajo, K. Shimoji, W. Takahashi, Strong convergence to a common fixed point of families of nonexpansive mappings in Banach spaces,
J. Nonlinear Convex Anal. 8 (2007) 11–34.
[19] O. Nevanlinna, S. Reich, Strong convergence of contraction semigroups and of iterative methods for accretive operators in Banach spaces,
Israel J. Math. 32 (1979) 44–58.
[20] S. Reich, Asymptotic behavior of contractions in Banach spaces, J. Math. Anal. Appl. 44 (1973) 57–70.
[21] S. Reich, The fixed point property for nonexpansive mappings I, Amer. Math. Monthly 83 (1976) 266–268.
[22] S. Reich, On infinite products of resolvents, Atti Accad. Naz. Lincei 63 (1977) 338–340.
[23] S. Reich, The fixed point property for nonexpansive mappings II, Amer. Math. Monthly 87 (1980) 292–294.
[24] S. Reich, Strong convergence theorems for resolvents of accretive operators in Banach spaces, J. Math. Anal. Appl. 75 (1980) 287–292.
[25] S. Reich, On the asymptotic behavior of nonlinear semigroups and the range of accretive operators, J. Math. Anal. Appl. 79 (1981) 113–126.
[26] S. Reich, Convergence, resolvent consistency, and the fixed point property for nonexpansive mappings, Contemp. Math. 18 (1983) 167–174.
[27] S. Reich, A.J. Zaslavski, Infinite products of resolvents of accretive operators, Topol. Methods Nonlinear Anal. 15 (2000) 153–168.
[28] R.T. Rockafellar, Monotone operators and the proximal point algorithm, SIAM J. Control Optim. 14 (1976) 877–898.
[29] N. Shioji, W. Takahashi, Strong convergence of approximated sequences for nonexpansive mappings in Banach spaces, Proc. Amer. Math.
Soc. 125 (1997) 3641–3645.
[30] M.V. Solodov, B.F. Svaiter, Forcing strong convergence of proximal point iterations in a Hirbert space, Math. Program. 87 (2000) 189–202.
[31] T. Suzuki, Strong convergence theorem to common fixed points of two nonexpansive mappings in general Banach spaces, J. Nonlinear Convex
Anal. 3 (2002) 381–391.
[32] T. Suzuki, A sufficient and necessary condition for Halpern-type strong convergence to fixed points of nonexpansive mappings, Proc. Amer.
Math. Soc. 135 (2007) 99–106.
[33] W. Takahashi, Nonlinear Functional Analysis, Yokohama Publishers, Yokohama, 2000.
[34] W. Takahashi, Convex Analysis and Approximation of Fixed Points, Yokohama Publishers, Yokohama, 2000 (in Japanese).
[35] W. Takahashi, Iterative methods for approximation of fixed points and their applications, J. Oper. Res. Soc. Japan 43 (2000) 87–108.
[36] W. Takahashi, Viscosity approximation methods for resolvents of accretive operators in Banach spaces, J. Fixed Point Theory Appl. 1 (2007)
135–147.
[37] W. Takahashi, Y. Ueda, On Reich’s strong convergence theorems for resolvents of accretive operators, J. Math. Anal. Appl. 104 (1984)
546–553.
[38] R. Wittmann, Approximation of fixed points of nonexpansive mappings, Arch. Math. 58 (1992) 486–491.
[39] H.K. Xu, Viscosity approximation methods for nonexpansive mappings, J. Math. Anal. Appl. 298 (2004) 279–291.
[40] Y. Yao, R. Chen, J.C. Yao, Strong convergence and certain control conditions of modified Mann iteration, Nonlinear Anal. 68 (6) (2008)
1687–1693.