Hindawi Publishing Corporation Fixed Point Theory and Applications Volume 2007, Article ID 73246, 9 pages doi:10.1155/2007/73246
Research Article Nonlinear Mean Ergodic Theorems for Semigroups in Hilbert Spaces
Seyit Temir and Ozlem Gul
Received 26 December 2006; Accepted 4 April 2007
Recommended by Nan-Jing Huang
Let K be a nonempty subset (not necessarily closed and convex) of a Hilbert space and let Γ = {T(t); t ≥ 0} be a semigroup on K and let α(·) : [0, ∞) → K be an almost orbit of Γ. In this paper, we prove that every almost orbit of Γ is almost weakly and strongly convergent to its asymptotic center.
Copyright © 2007 S. Temir and O. Gul. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
1. Introduction
Let K be a nonempty subset of a Hilbert space (cid:2), where K is not necessarily closed and convex. A family Γ = {T(t); t ≥ 0} of mappings T(t) is called a semigroup on K if
(S1) T(t) is a mapping from K into itself for t ≥ 0, (S2) T(0)x = x and T(t + s)x = T(t)T(s)x for x ∈ K and t,s ≥ 0, (S3) for each x ∈ K, T(·)x is strongly measurable and bounded on every bounded
subinterval of [0, ∞).
Let Γ be a semigroup on K. Then F = {x ∈ K : T(t)x = x, t ≥ 0} is said to be fixed- points set of Γ. We state a condition introduced by Miyadera [1]. If, for every bounded set B ⊂ K, v ∈ K, and s ≥ 0, there exists a δs(B,v) ≥ 0 with lims→∞ δs(B,v) = 0 such that
(cid:2) (cid:2)T(s)u − T(s)v (cid:2) (cid:2) ≤ (cid:8)u − v(cid:8) + δs(B,v)
(1.1)
for u ∈ B, then Γ is said to be an asymptotically nonexpansive semigroup.
2
Fixed Point Theory and Applications
Definition 1.1. A function a(·) : [0, ∞) → K is called almost-orbit of Γ if a(·) : [0, ∞) → K is strongly measurable and bounded on every bounded subinterval of [0, ∞) and if
(cid:2) (cid:2)p (cid:2) (cid:2)a(s + t) − T k(s)a(t) = 0.
(1.2)
lim t → ∞
sup s≥0
Using these conditions, we prove that every almost-orbit of Γ is weakly and strongly convergent to its asymptotic center (see [1]). Xu [2] studied strong asymptotic behavior of almost-orbits of both of nonexpansive and asymptotically nonexpansive semigroups. Takahashi [3] generalized the nonlinear ergodic theorems for general semigroups of non- expansive mappings. Kada and Takahashi [4] proved a strong ergodic theorem for general semigroups of nonexpansive mappings. Oka [5] proved nonlinear ergodic theorems for commutative semigroups of asymptotically nonexpansive mappings. All of the above- mentioned authors studied, except Miyadera’s works, K as a closed and convex subset of a Hilbert space. Miyadera [1] studied almost convergence of almost-orbits of semigroup of non-Lipschitzian mappings in Hilbert spaces. Miyadera [1] proved the following the- orem. If Γ is asymptotically nonexpansive in the weak sense and F is nonempty set, then the following conditions holds:
(a1) a(·) is weakly almost convergent to its asymptotic center y, (a2) if y is an element of K and if T(t0) : K → K is continuous for some t0 > 0, then y
is a fixed point of Γ, that is, y belongs to F.
There are some conditions in the discrete case in [6–8]. Miyadera [7, 8] showed that the condition in [6] could be replaced by a weaker condition introduced in [7, 8]. Miyadera [7, 8] and Wittmann [6] proved nonlinear ergodic theorems where the closed- ness and convexity of K and the asymptotically nonexpansivity of T were not assumed. In this paper, in the light of these papers we establish weak ergodic theorem for semigroups of mappings on K satisfying condition (I) given in the statement of Theorem 3.1. We also establish strong ergodic theorem for semigroups of mappings on K satisfying condition (II) given before statement of Theorem 4.1. This paper is organized as follows.
In Section 2, we prove the covering lemmas we need for establishing weakly conver- gence result. In Section 3, we deal with a(·) almost-orbit weakly almost-convergent to its asymptotic center with respect to condition (I). In the last section, we investigate strong convergence using condition (II). We establish that every almost-orbit of Γ is strongly almost-convergent to its asymptotic center.
2. Lemmas Let a(·) : [0, ∞) → (cid:2) be a function strongly measurable and bounded on every bounded subinterval of [0, ∞) and let (cid:8)a(t)(cid:8) be convergent as t → ∞. Lemma 2.1 [1]. For r,s,t ≥ 0, the following statements are mutually equivalent:
(i) lims→∞ limt→∞ limr→∞[(a(t + r),a(t)) − (a(s + r),a(s))] ≤ 0; (ii) lims→∞ limt→∞ limr→∞[(cid:8)a(t + r) + a(t)(cid:8)2 − (cid:8)a(s + r) + a(s)(cid:8)2] ≤ 0; (iii) lims→∞ limt→∞ limr→∞[(cid:8)a(s + r) − a(s)(cid:8)2 − (cid:8)a(t + r) − a(t)(cid:8)2] ≤ 0. If a(·) satisfies the equivalent conditions (i), (ii), and (iii), then a(·) is weakly almost-
convergent to its asymptotic center y.
S. Temir and O. Gul
3
Lemma 2.2 [1]. Let a(·) : [0, ∞) → (cid:2) be a function strongly measurable and bounded on every bounded subinterval of [0, ∞) and let (cid:8)a(t)(cid:8) be convergent as t → ∞. Then, one has that the following statements are mutually equivalent:
(i) lims→∞ limt→∞ supr≥0[(a(t + r),a(t)) − (a(s + r),a(s))] ≤ 0; (ii) lims→∞ limt→∞ supr≥0[(cid:8)a(t + r) + a(t)(cid:8)2 − (cid:8)a(s + r) + a(s)(cid:8)2] ≤ 0; (iii) lims→∞ limt→∞ supr≥0[(cid:8)a(s + r) − a(s)(cid:8)2 − (cid:8)a(t + r) − a(t)(cid:8)2] ≤ 0. (cid:8)a(t)(cid:8) is convergent as t → ∞. Moreover, if a(·) satisfies the equivalent conditions (i),
(ii), and (iii), then a(·) is strongly almost-convergent to its asymptotic center y.
Remark 2.3. We can take the following conditions instead of (ii) and (iii) in Lemma 2.2, for A,C > 0,
(ii(cid:9)) lims→∞ limt→∞ supr≥0[(cid:8)a(t + r) + a(t)(cid:8)2 − A(cid:8)a(s + r) + a(s)(cid:8)2] ≤ 0; (iii(cid:9)) lims→∞ limt→∞ supr≥0[(cid:8)a(s + r) − a(s)(cid:8)2 − A(cid:8)a(t + r) − a(t)(cid:8)2] ≤ 0.
We can obtain
lim s → ∞
lim t → ∞
sup r≥0
(2.1)
(cid:2) (cid:2) (cid:2) (cid:3)(cid:2) (cid:2)2(cid:4) (cid:2)a(s + r) + a(s) (cid:2)2 − A (cid:2)a(t + r) + a(t) (cid:2) (cid:2) (cid:2) (cid:3)(cid:2) (cid:2)2(cid:4) (cid:2)2 − (cid:2)a(s + r) + a(s) (cid:2)a(t + r) + a(t) ≤ 0, ≤ lim s → ∞
lim t → ∞
sup r≥0
and
(cid:2) (cid:2)2(cid:4)
lim s → ∞
lim t → ∞
sup r≥0
(2.2)
(cid:2) (cid:2)2(cid:4) (cid:3)(cid:2) (cid:2)a(s + r) − a(s) (cid:3)(cid:2) (cid:2)a(s + r) − a(s) (cid:2) (cid:2) (cid:2)a(t + r) − a(t) (cid:2)2 − A (cid:2) (cid:2) (cid:2)a(t + r) − a(t) (cid:2)2 − ≤ 0. ≤ lim s → ∞
lim t → ∞
sup r≥0
Moreover, we can write
(cid:4) (cid:2) (cid:2)2 − C (cid:3)(cid:2) (cid:2)a(s + r) − a(s) (cid:2) (cid:2) (cid:2)a(t + r) − a(t) (cid:2)2 − A ≤ 0.
(2.3)
lim s → ∞
lim t → ∞
sup r≥0
Note that Lemma 2.2 holds for this condition.
3. Weak ergodic theorems Let (cid:2) be a Hilbert space with inner product (·, ·) and (cid:8) · (cid:8) norm, and let K be a nonempty subset of (cid:2), where K is not necessarily closed and convex. Let Γ = {T(t); t ≥ 0} be a semigroup acting on K. Theorem 3.1. Suppose that for every bounded set B ⊂ K, v ∈ K, u ∈ B and r ≥ 0, there exists δr(B,v) ≥ 0 with limr→∞ δr(B,v) = 0 such that
(cid:2) (cid:2)p (cid:2) (cid:2)T k(r)u − T k(r)v (cid:3)
(I)
(cid:2) (cid:2)p (cid:2) (cid:2)p(cid:4)
λr (cid:8)u(cid:8)p −
≤ λr (cid:8)u − v(cid:8)p + c (cid:2) (cid:2)T k(r)u
+ λr (cid:8)v(cid:8)p −
(cid:2) (cid:2)T k(r)v
+ δr(B,v),
4
Fixed Point Theory and Applications
where λr, c are nonnegative constants such that limr→∞ λr = 1, and p ≥ 1. If F (cid:10)= ∅ or c > 0, then a(·) is almost weakly convergent to its asymptotic center, which is y. Proof. Suppose F (cid:10)= ∅ and c = 0 for the semigroup Γ = {T(t); t ∈ R+}. Then for u = x and f ∈ F, we can take B = {x}. If we write u = x and v = f in (I), then we have
(cid:2) (cid:2)p (cid:2) (cid:2)p = (cid:2) (cid:2)T k(r)x − T k(r) f (cid:2) (cid:2)T k(r)x − f ≤ λr (cid:8)x − f (cid:8)p + δr(B, f ).
(3.1)
Thus, for every x ∈ K, the sequence {T k(r)x − f + f } = {T k(r)x} is bounded. Let
a(·) : [0, ∞) → K be almost-orbit of Γ.
From Definition 1.1, we have limt→∞ sups≥0[(cid:8)a(t + s) − T k(s)a(t)(cid:8)p] = 0. There is t0 =
t0(ε) > 0 for ε > 0, t ≥ t0, and s ≥ 0 such that (cid:8)a(t + s) − T k(s)a(t)(cid:8)p < ε.
In particular, for s ≥ 0, we have (cid:8)a(s + t0) − T t0(s)a(t0)(cid:8)p < ε. If we consider both this
inequality and boundness of sequence {T k(s)x}, we have
(cid:5) (cid:6) (cid:6) (cid:2) (cid:2)a
s + t0
(cid:6)(cid:2) (cid:2)p(cid:4) (cid:2) (cid:5) (cid:2)T t0(s)a
+
(3.2)
t0
(cid:5) − T t0(s)a t0 (cid:3)(cid:2) (cid:6) (cid:5) (cid:2)a s + t0 (cid:2) (cid:2)T t0(s)
< 2p−1 < 2p−1ε + 2p−1
,
(cid:6)(cid:2) (cid:5) (cid:2)p + T t0(s)a t0 (cid:6)(cid:2) (cid:5) (cid:2)p − T t0(s)a t0 (cid:2)p(cid:2) (cid:2) (cid:6)(cid:2) (cid:5) (cid:2)p (cid:2)a t0
then {a(s); s ∈ R+} is bounded.
If we take in (I), B = {a(t); t ∈ R+}, and v = f , then we obtain
(cid:2) (cid:2)p (cid:2) (cid:2)p. ≤ λr (cid:2) (cid:2)T k(r)a(t) − T k(r) f (cid:2) (cid:2)a(t) − f
(3.3)
Thus
(cid:2) (cid:2)p (cid:2) (cid:2)p (cid:2) (cid:2)a(r + t) − f (cid:2) (cid:2)a(r + t) − T k(r)a(t) + T k(r)a(t) − f (cid:2) (cid:2)p(cid:4)
+
(cid:2) (cid:2)T k(r)a(t) − f
(3.4)
(cid:2) (cid:2)p(cid:6) ≤ (cid:2) (cid:3)(cid:2) (cid:2)p (cid:2)a(r + t) − T k(r)a(t) ≤ 2p−1 (cid:2) (cid:5) (cid:2)a(t) − f < 2p−1 ε + λr
(since (cid:8)T k(r)a(t) − T k(r) f (cid:8)p ≤ λr (cid:8)a(t) − f (cid:8)p).
Taking limit as r → ∞, because of limr→∞ λr = 1, for arbitrary ε,
(cid:2) (cid:2)p. (cid:2) (cid:2)a(t) − f
(3.5)
(cid:2) (cid:2)a(r + t) − f limr→∞ (cid:2) (cid:2)p < 2p−1 limr→∞
Therefore {(cid:8)a(t) − f (cid:8)} is convergent. Let t > s > 0. We know that sequence {T k(r); r ∈ R+} is bounded. Moreover, since sequence {a(s);s ∈ R+} is bounded, {T k(h)a(s); h ∈ R+} is also bounded. Then we can take B = {T k(h)a(s); h ∈ R+} and a(s) ∈ K. Taking u = T k(h)a(s),v = a(s), and r = t − s,
S. Temir and O. Gul
5
for h ≥ 0, we have
(cid:2) (cid:2)p (cid:2) (cid:2)p (cid:6) . ≤ λt−s (cid:2) (cid:2)T k(t − s)T k(h)a(s) − T k(t − s)a(s) (cid:2) (cid:2)T k(h)a(s) − a(s)
+ δt−s
(cid:5) B,a(s)
(3.6)
Consequently,
(cid:2) (cid:2)p (cid:2) (cid:2)p (cid:2) (cid:2)a(t + h) − a(t) (cid:2) (cid:2)a(t + h) − T k(t + h − s)a(s) + T k(t + h − s)a(s) − a(t) (cid:2) (cid:2)p(cid:4) (cid:2) (cid:2)T k(t + h − s)a(s) − a(t)
+
(cid:2) (cid:2)p (cid:2) (cid:2)p ≤ ≤ 2p−1 = 2p−1 (cid:3)(cid:2) (cid:2)a(t + h) − T k(t + h − s)a(s) (cid:3)(cid:2) (cid:2)a(t + h) − T k(t + h − s)a(s) (cid:2) (cid:2)p(cid:4)
(cid:2) (cid:2)p ≤ 2p−1 − T k(t + h − s)a(s) (cid:2) (cid:2)p
(cid:2) (cid:2)p
+ (cid:2) (cid:2)a + 22(p−1) + 22(p−1)
(cid:6) (cid:2) (cid:2)p (cid:2) (cid:2)p (cid:6) (cid:2) (cid:2)T k(t + h − s)a(s) + T k(t − s)a(s) − T k(t − s)a(s) − a(t) (cid:6) (cid:5) (t + h − s) + s (cid:2) (cid:2)T k(t − s)T k(h)a(s) − T k(t − s)a(s) (cid:2) (cid:2)T k(t − s)a(s) − a(t) (cid:2) (cid:2)T k(h)a(s) − a(s) < 2p−1ε + 22(p−1)λt−s (cid:2) (cid:2)T k(t − s)a(s) − a(t − s + s) + 22(p−1)λt−s
+ 22(p−1) (cid:5) 1 + 2p−1
< 2p−1ε
(cid:6) (cid:5) B,a(s) + δt−s (cid:5)(cid:2) (cid:2)T k(h)a(s) − a(s + h) (cid:2) (cid:2)p(cid:6) + a(s + h) − a(s)
+ δt−s
(cid:5) B,a(s) (cid:6)
< 2p−1ε
(cid:6) . (cid:5) 1 + 2p−1 + 22(p−1)ελt−s (cid:2) (cid:2) (cid:2)p (cid:2)a(s + h) − a(s)
+ 22(p−1)λt−s
+ δt−s
(cid:5) B,a(s)
(3.7)
Then
(cid:6) (cid:6) (cid:2) (cid:2)p (cid:2) (cid:2)a(t+h)−a(t) −22(p−1)λt−s (cid:2) (cid:2)a(s + h)−a(s) (cid:2) (cid:2)p <2p−1ε (cid:5) 1+2p−1+2p−1λt−s
+ δt−s
(cid:5) B,a(s)
. (3.8)
Taking limit as t,s → ∞, for h ≥ 0 and arbitrary ε, from the last inequality, we obtain
(cid:2) (cid:2)p (cid:2) (cid:2)p(cid:4) (cid:3)(cid:2) (cid:2)a(t + h) − a(t) − 22(p−1) (cid:2) (cid:2)a(s + h) − a(s) ≤ 0.
(3.9)
lim t → ∞
lim s → ∞
sup h≥0
From Remark 2.3, a(·) is weakly almost convergent to its asymptotic center. Now, we investigate the case F (cid:10)= ∅ and c > 0. For x ∈ K, if we write B = {x} and v = x in (I), then we obtain
(cid:3) (cid:2) (cid:2)p(cid:4)
2λr (cid:8)x(cid:8)p − 2
(cid:2) (cid:2)T k(r)x
+ δr(B,x),
(3.10)
0 ≤ λr0 + c
6
Fixed Point Theory and Applications
and from this we can write
(cid:2) (cid:2)p
.
(cid:2) (cid:2)T k(r)x ≤ λr (cid:8)x(cid:8)p +
(3.11)
δr(B,x) 2c
Then for every x ∈ K, {T k(t)x; t ∈ R+} is bounded. By Definition 1.1, for ε > 0 taking t ≥ t0, s ≥ 0, there exists t0 = t0(ε) such that (cid:8)a(t0 + s) − T k(s)a(t0)(cid:8)p < ε. Since {T k(t)x; t ∈ R+} is bounded,
(cid:6) (cid:2) (cid:2)a (cid:6) (cid:5) t0 + s
(3.12)
(cid:6)(cid:2) (cid:2)p (cid:6)(cid:2) (cid:2)p (cid:6)(cid:2) (cid:2)p(cid:4) (cid:2)p(cid:2) (cid:2) (cid:2)a ≤ 2(p−1) (cid:2) (cid:2)T k(s)
+
(cid:5) (cid:5) + T k(s)a − T k(s)a t0 t0 (cid:3)(cid:2) (cid:5) (cid:6) (cid:5) (cid:2)a − T k(s)a t0 t0 + s (cid:5) t0
{a(t); t ∈ R+} is bounded. We can take B = {T k(h)a(s) : h ≥ 0} , if we write v = a(s) and r = t − s in (I), then we obtain
(cid:2) (cid:2)p
(3.13)
(cid:2) (cid:2)p (cid:2) (cid:2)p − (cid:2) (cid:2)p(cid:4) − (cid:2) (cid:2) (cid:2)p (cid:2)T k(t − s)T k(h)a(s) − T k(t − s)a(s) (cid:2) (cid:3) (cid:2)T k(h)a(s) + c λt−s (cid:2) (cid:2) (cid:2)p (cid:2)a(s) + λt−s (cid:6) (cid:5) . B,a(s) (cid:2) (cid:2)T k(h)a(s) − a(s) ≤ λt−s (cid:2) (cid:2)T k(t − s)T k(h)a(s) (cid:2) (cid:2)T k(t − s)a(s)
+ δt−s
Consequently,
(cid:2) (cid:2)p (cid:2) (cid:2)p (cid:2) (cid:2)a(t + h) − a(t) (cid:2) (cid:2)a(t + h) − T k(t + h − s)a(s) + T k(t + h − s)a(s) − a(t) (cid:2) (cid:2)p(cid:6) (cid:2) (cid:2)T k(t + h − s)a(s) − a(t)
+
= ≤ 2p−1 ≤ 2p−1 (cid:2) (cid:5)(cid:2) (cid:2)p (cid:2)a(t + h) − T k(t + h − s)a(s) (cid:2) (cid:2) (cid:2)p (cid:2)a(t + h) − T k(t + h − s)a(s) (cid:2) (cid:2)p
(cid:2) (cid:2)T k(t + h − s)a(s) − T k(t − s)a(s) (cid:2) (cid:2)p (cid:2) (cid:2)p
+ 2p−1 (cid:2) (cid:2)T k(t − s)a(s) − a(t)
(cid:2) (cid:2)a(t + h) − T k(t + h − s)a(s) (cid:2) (cid:2)p ≤ 2p−1 (cid:3)(cid:2) (cid:2)T k(t − s)T k(h)a(s) − T k(t − s)a(s)
+ + 22(p−1)
(cid:2) (cid:2)p(cid:4)
+
(cid:2) (cid:2)T k(t − s)a(s) − a(t − s + s) (cid:2) (cid:2)p
< 2p−1ε + 22(p−1)
(cid:2) (cid:2)p
(cid:2) (cid:2)T k(t − s)T k(h)a(s) (cid:2) (cid:2)p(cid:4)(cid:4) (cid:6) (cid:2) (cid:3) (cid:2)T k(h)a(s) − a(s) λt−s (cid:2) (cid:2) (cid:3) (cid:2)p (cid:2)T k(h)a(s) λt−s + c − (cid:2) (cid:2) (cid:2) (cid:2)p (cid:2)T k(t − s)a(s) (cid:2)a(s) + λt−s − (cid:6) (cid:5) 22(p−1) .
+ ε
+ δt−s
(cid:5) B,a(s)
(3.14)
S. Temir and O. Gul
7
(cid:6) (cid:2) (cid:2)p (cid:6) (cid:6)p (cid:5) B,a(s) (cid:6)
+ 22(p−1)λt−s (cid:5) M2N + cλt−sM p
(cid:6)
+ δt−s (cid:5) M p − 1 (cid:2) (cid:2)p
(cid:5) B,a(s) (cid:6) (cid:6)
+ δt−s (cid:6)
Taking (cid:8)a(s)(cid:8) ≤ M and (cid:8)T k(h)(cid:8) ≤ N, (cid:2) (cid:5) (cid:2)T k(h)a(s) − a(s) 1 + 2p−1 < 2p−1ε (cid:5) (cid:6) λt−sM pN p − + λt−sM p − M pN p + c (cid:6) (cid:5) (cid:6) (cid:5) 1 + 2p−1 < 2p−1ε − cM pN p N p + 1 (cid:2) (cid:2)T k(h)a(s) − a(s + h) + a(s + h) − a(s) (cid:5) N p + 1
+ 22(p−1)λt−s (cid:5) 1 + 2p−1
< 2p−1ε
(cid:5) M p − 1
(3.15)
− cM pN p (cid:2) (cid:2)p (cid:5) (cid:6)
+ 22(p−1)λt−s2p−1 + 22(p−1)λt−s2p−1 (cid:6)
(cid:2) (cid:2)p (cid:6) (cid:6)
< 2p−1ε
B,a(s) (cid:5) M p − 1
+ 23(p−1)λt−sε
.
+ 23(p−1)λt−s
+ cλt−sM p (cid:2) (cid:2)T k(h)a(s) − a(s + h) (cid:2) (cid:2)a(s + h) − a(s) (cid:5) (cid:5) 1 + 2p−1 N p + 1 + cλt−sM p (cid:2) (cid:2) (cid:2)p (cid:2)a(s + h) − a(s)
+ δt−s
+ δt−s − cM pN p (cid:6) (cid:5) B,a(s)
Taking limit as t, s → ∞, for h ≥ 0,
(cid:2) (cid:2)p (cid:2) (cid:2)p(cid:4) ≤ ´A. (cid:3)(cid:2) (cid:2)a(t + h) − a(t) − 23(p−1) (cid:2) (cid:2)a(s + h) − a(s)
(3.16)
lim t → ∞
lim s → ∞
sup h≥0
That is,
(cid:2) (cid:2)p (cid:2) (cid:2)p(cid:4) (cid:3)(cid:2) (cid:2)a(t + h) − a(t) − 23(p−1) (cid:2) (cid:2)a(s + h) − a(s) − ´A ≤ 0.
(3.17)
lim t → ∞
lim s → ∞
sup h≥0
Then from Remark 2.3, a(·) is weakly almost convergent to its asymptotic center. Thus, (cid:2) the proof is completed.
4. Strong ergodic theorems Let Γ = {T(t); t ≥ 0} be semigroup on K. Suppose that for every bounded set B ⊂ K and integer k ≥ 0, there exists a δr(B,v) ≥ 0 with limr→∞ δr(B,v) = 0 such that
(cid:2) (cid:2)p (cid:2) (cid:2)T k(r)u + T k(r)v (cid:3)
(II)
(cid:2) (cid:2)p (cid:2) (cid:2)p(cid:4)
λr (cid:8)u(cid:8)p −
≤ λr (cid:8)u + v(cid:8)p + c (cid:2) (cid:2)T k(r)u
+ λr (cid:8)v(cid:8)p −
(cid:2) (cid:2)T k(r)v
+ δr(B)
for u,v ∈ B, where λr, c, and p are nonnegative constants such that limr→∞ λr = 1 and p ≥ 1. Theorem 4.1. If Γ = {T(t); t ≥ 0} is a semigroup on K satisfying condition (II), then every almost-orbit of Γ is strongly almost convergent to its asymptotic center. Proof. Let a(·) : [0, ∞) → K be almost-orbit of Γ. For t ≥ 0, we set
(cid:2) (cid:2). (cid:2) (cid:2)a(t + s) − T k(t)a(s)
(4.1)
ϕ(s) = sup t≥0
When s → ∞, ϕ(s) → 0 and from condition (II) by taking B = {x} and v = x, we have
(cid:6)
δr
(cid:2) (cid:2)p (cid:2) (cid:2)T k(r)x ≤ λr (cid:8)x(cid:8)p +
+ 2c.
(4.2)
(cid:5) {x} 2p
8
Fixed Point Theory and Applications
Thus for x ∈ K, {T k(r)x : r ≥ 0} is bounded. By Definition 1.1, since {T k(r)x : r ≥ 0} is bounded, we have
(cid:2) (cid:2)p ≤ ε. (cid:2) (cid:2)a(s + t) − T k(s)a(t)
(4.3)
(cid:2) (cid:2)p
Therefore {a(s) : s ≥ 0} is bounded. Let r > h ≥ 0. Since {T k(h)x : h ≥ 0} and {a(s) : s ≥ 0} are bounded then {T k(h)a(s) : h ≥ 0} is bounded, by using (II) with B = {T k(h)a(s) : h ≥ 0}, v = a(s) and r = t − s we have (cid:2) (cid:2)T k(t − s)T k(h)a(s) + T k(t − s)a(s)
(cid:2) (cid:2)p
(cid:2) (cid:2)p(cid:4) (cid:2) (cid:2) (cid:2)p (cid:2)T k(h)a(s) + a(s) ≤ λt−s (cid:2) (cid:2) (cid:3) (cid:2)p (cid:2)T k(h)a(s) λt−s + c − (cid:2) (cid:2) (cid:2) (cid:2)p (cid:2)a(s) (cid:2)T k(t − s)a(s) + λt−s − (cid:2) (cid:2)T k(t − s)T k(h)a(s) + δt−s(B).
(4.4)
(cid:6) (cid:2) (cid:2)p
.
≤ λt−s
For c = 0, we have (cid:2) (cid:2)T k(t − s)T k(h)a(s) + T k(t − s)a(s)
(cid:2) (cid:2) (cid:2)p (cid:2)T k(h)a(s) + a(s)
+ δt−s
(cid:5) B,a(s)
(4.5)
Consequently, (cid:2) (cid:2) (cid:2)p (cid:2)a(t + h) + a(t)
≤ 2p−1 (cid:2) (cid:2)p(cid:6) (cid:2) (cid:2)p (cid:5)(cid:2) (cid:2)a + 22(p−1)
(cid:6) (cid:5) − T k(t + h − s)a(s) (t + r) + s − s (cid:5)(cid:2) (cid:2)T k(t + h − s)a(s) + T k(t − s)a(s) (cid:2) (cid:2) (cid:2)p(cid:6) (cid:2)a(t − s + s) − T k(t − s)a(s)
+
(cid:4) (cid:2) (cid:2)p (cid:3) (cid:4) (cid:6) (cid:3)(cid:2) (cid:2)T k(t − s)T k(h)a(s) + T k(t − s)a(s) λt−s (cid:2) (cid:2) (cid:2)p (cid:2)T k(h)a(s) + a(s)
+ ϕp(s)
+ ϕp(s) (cid:5) B,a(s)
+ δt−s
(cid:6) (cid:2) (cid:2)p
+ δt−s
(cid:5) B,a(s) (cid:6) ≤ 2p−1ϕp(s) (cid:2) (cid:2)p (cid:2) (cid:2)p(cid:4) ≤ 2p−1ϕp(s) + 22(p−1) ≤ 2p−1ϕp(s) + 22(p−1) ≤ 2p−1ϕp(s) + 22(p−1)λt−sϕp(s) (cid:2) (cid:2)T k(h)a(s) + a(s) + a(h + s) − a(h + s) + 22(p−1)λt−s (cid:5) 1 + 2p−1λt−s (cid:3)(cid:2) (cid:2)a(h + s) + a(s)
+
(cid:2) (cid:2)T k(h)a(s) − a(h + s) (cid:6)
.
+ 22(p−1)2p−1λt−s (cid:5) B,a(s) + δt−s
(4.6)
Taking limit as s, t → ∞, for h ≥ 0,
(cid:2) (cid:2)p(cid:4) (cid:2) (cid:3)(cid:2) (cid:2)p (cid:2)a(t + h) + a(t) − 23(p−1)λt−s (cid:2) (cid:2)a(h + s) + a(s)
lim t → ∞
sup h≥0
(4.7)
(cid:6) . ≤ 2p−1ϕp(s) (cid:5) 1 + 2p−1λt−s + λt−s2p−1
Since ϕ(s) → 0, we have
(cid:2) (cid:2)p(cid:4) (cid:2) (cid:3)(cid:2) (cid:2)p (cid:2)a(t + h) + a(t) − 23(p−1) (cid:2) (cid:2)a(h + s) + a(s) ≤ 0,
(4.8)
lim s → ∞
lim t → ∞
sup h≥0
S. Temir and O. Gul
9
that is, condition (ii) in Lemma 2.2 is satisfied. Thus, every almost-orbit of Γ is strongly (cid:2) almost convergent to its asymptotic center.
Remark 4.2. Our results presented in this paper generalize the results of Miyadera [7, 8] to the case of F (cid:10)= ∅ and c > 0 for semigroups of asymptotically nonexpansive mappings in Hilbert spaces.
References
[1] I. Miyadera, “Nonlinear ergodic theorems for semigroups of non-Lipschitzian mappings in
Hilbert spaces,” Taiwanese Journal of Mathematics, vol. 4, no. 2, pp. 261–274, 2000.
[2] H.-K. Xu, “Strong asymptotic behavior of almost-orbits of nonlinear semigroups,” Nonlinear
Analysis, vol. 46, no. 1, pp. 135–151, 2001.
[3] W. Takahashi, “A nonlinear ergodic theorem for an amenable semigroup of nonexpansive map- pings in a Hilbert space,” Proceedings of the American Mathematical Society, vol. 81, no. 2, pp. 253–256, 1981.
[4] O. Kada and W. Takahashi, “Strong convergence and nonlinear ergodic theorems for commu- tative semigroups of nonexpansive mappings,” Nonlinear Analysis, vol. 28, no. 3, pp. 495–511, 1997.
[5] H. Oka, “Nonlinear ergodic theorems for commutative semigroups of asymptotically nonex-
pansive mappings,” Nonlinear Analysis, vol. 18, no. 7, pp. 619–635, 1992.
[6] R. Wittmann, “Mean ergodic theorems for nonlinear operators,” Proceedings of the American
Mathematical Society, vol. 108, no. 3, pp. 781–788, 1990.
[7] I. Miyadera, “Nonlinear mean ergodic theorems,” Taiwanese Journal of Mathematics, vol. 1, no. 4,
pp. 433–449, 1997.
[8] I. Miyadera, “Nonlinear mean ergodic theorems—II,” Taiwanese Journal of Mathematics, vol. 3,
no. 1, pp. 107–114, 1999.
Seyit Temir: Department of Mathematics, Arts and Science Faculty, Harran University, 63200 Sanliurfa, Turkey Email address: temirseyit@harran.edu.tr
Ozlem Gul: Department of Mathematics, Arts and Science Faculty, Harran University, 63200 Sanliurfa, Turkey Email address: ozlemgul@harran.edu.tr
