The Equivalence of Mann Iteration and Ishikawa iteration for ψ-uniformly pseudocontractive or ψ-uniformly accretive maps

Abstract

Authors

B. E. Rhoades
Department of Mathematics, Indiana University, Bloomington, IN 47405-7106

Stefan M. ¸Soltuz
“Tiberiu Popoviciu” Institute of Numerical Analysis, Cluj-Napoca, Romania

Keywords

?

Paper coordinates

B.E. Rhoades and Ş.M. Şoltuz, The Equivalence of Mann Iteration and Ishikawa iteration for ψ-uniformly pseudocontractive or ψ-uniformly accretive maps, Internat. J. Math. Sci. 2004: 46, 2443-2451.

PDF

About this paper

Journal

International Journal of Mathematics and Mathematical Sciences

Publisher Name

Wiley Online Library

DOI
Print ISSN
Online ISSN

google scholar link

[1] S. Ishikawa, Fixed points by a new iteration method, Proc. Amer. Math. Soc. 44 (1974), 147–150.
[2] W. R. Mann, Mean value methods in iteration, Proc. Amer. Math. Soc. 4 (1953), 506–510.
[3] C. Moore and B. V. C. Nnoli, Iterative solution of nonlinear equations involving set-valueduniformly accretive operators, Comput. Math. Appl. 42 (2001), no. 1-2, 131–140.
[4] C. H. Morales and J. S. Jung, Convergence of paths for pseudocontractive mappings in Ba-nach spaces, Proc. Amer. Math. Soc. 128 (2000), no. 11, 3411–3419.
[5] B. E. Rhoades and ¸S. M. ¸Soltuz, The equivalence between T -stabilities of Mann and Ishikawaiterations, submitted to Math. Commun.
[6] , The equivalence of Mann and Ishikawa iteration for a Lipschitzian psi-uniformlypseudocontractive and psi-uniformly accretive map, to appear in Tamkang J. Math.[7] , The equivalence between the convergences of Ishikawa and Mann iterations for anasymptotically pseudocontractive map, J. Math. Anal. Appl. 283 (2003), no. 2, 681–688.
[8] , The equivalence of Mann iteration and Ishikawa iteration for non-Lipschitzian op-erators, Int. J. Math. Math. Sci. 2003 (2003), no. 42, 2645–2651.
[9] , On the equivalence of Mann and Ishikawa iteration methods, Int. J. Math. Math. Sci.2003 (2003), no. 7, 451–459.
[10] , The equivalence between the convergences of Ishikawa and Mann iterations foran asymptotically nonexpansive in the intermediate sense and strongly successivelypseudocontractive maps, J. Math. Anal. Appl. 289 (2004), no. 1, 266–278

Paper (preprint) in HTML form

THE EQUIVALENCE OF MANN ITERATION AND ISHIKAWA ITERATION FOR ψ\psi-UNIFORMLY PSEUDOCONTRACTIVE OR ψ\psi-UNIFORMLY ACCRETIVE MAPS

B. E. RHOADES and ŞTEFAN M. ŞOLTUZ

Received 2 December 2003

We show that the Ishikawa iteration and the corresponding Mann iteration are equivalent when applied to ψ\psi-uniformly pseudocontractive or ψ\psi-uniformly accretive maps.

2000 Mathematics Subject Classification: 47H10.

  1. 1.

    Introduction. Let XX be a real Banach space, BB a nonempty, convex subset of XX, and TT a self-map of BB, and let x0=u0Bx_{0}=u_{0}\in B. The Mann iteration (see [2]) is defined by

un+1=(1αn)un+αnTun,n=0,1,2,.u_{n+1}=\left(1-\alpha_{n}\right)u_{n}+\alpha_{n}Tu_{n},\quad n=0,1,2,\ldots. (1.1)

The Ishikawa iteration is defined (see [1]) by

xn+1\displaystyle x_{n+1} =(1αn)xn+αnTyn\displaystyle=\left(1-\alpha_{n}\right)x_{n}+\alpha_{n}Ty_{n}
yn\displaystyle y_{n} =(1βn)xn+βnTxn,n=0,1,2,.\displaystyle=\left(1-\beta_{n}\right)x_{n}+\beta_{n}Tx_{n},\quad n=0,1,2,\ldots. (1.2)

The sequences {αn}(0,1),{βn}[0,1)\left\{\alpha_{n}\right\}\subset(0,1),\left\{\beta_{n}\right\}\subset[0,1) satisfy

limnαn=limnβn=0,n=1αn=+.\lim_{n\rightarrow\infty}\alpha_{n}=\lim_{n\rightarrow\infty}\beta_{n}=0,\quad\sum_{n=1}^{\infty}\alpha_{n}=+\infty. (1.3)

The map J:X2XJ:X\rightarrow 2^{X^{*}} given by

Jx:={fX:x,f=x,f=x},xXJx:=\left\{f\in X^{*}:\langle x,f\rangle=\|x\|,\|f\|=\|x\|\right\},\quad\forall x\in X (1.4)

is called the normalized duality mapping.
Remark 1.1. The above JJ satisfies

x,j(y)xy,xX,j(y)J(y).\langle x,j(y)\rangle\leq\|x\|\|y\|,\quad\forall x\in X,\forall j(y)\in J(y). (1.5)

Proof. Denote j(y)j(y) by ff. Since fXf\in X^{*}, we have

x,ffx.\langle x,f\rangle\leq\|f\|\|x\|. (1.6)

From (1.4), we know that f=y\|f\|=\|y\|. Hence (1.5) holds.
Let

Ψ:={ψψ:[0,+)[0,+) is a nondecreasing map such that ψ(0)=0}.\Psi:=\{\psi\mid\psi:[0,+\infty)\rightarrow[0,+\infty)\text{ is a nondecreasing map such that }\psi(0)=0\}. (1.7)

The following definition is from [3].
DEFINITION 1.2. Let XX be a real Banach space. Let BB be a nonempty subset of XX. A map T:BBT:B\rightarrow B is called ψ\psi-uniformly pseudocontractive if there exist the map ψΨ\psi\in\Psi and j(xy)J(xy)j(x-y)\in J(x-y) such that

TxTy,j(xy)xy2ψ(xy),x,yB.\langle Tx-Ty,j(x-y)\rangle\leq\|x-y\|^{2}-\psi(\|x-y\|),\quad\forall x,y\in B. (1.8)

The map S:XXS:X\rightarrow X is called ψ\psi-uniformly accretive if there exist the map ψΨ\psi\in\Psi and j(xy)J(xy)j(x-y)\in J(x-y) such that

SxSy,j(xy)ψ(xy),x,yX.\langle Sx-Sy,j(x-y)\rangle\geq\psi(\|x-y\|),\quad\forall x,y\in X. (1.9)

Taking ψ(a):=ψ(a)a\psi(a):=\psi(a)\cdot a, for all a[0,+),ψΨa\in[0,+\infty),\psi\in\Psi, we get the usual definitions of ψ\psi-strongly pseudocontractivity and ψ\psi-strongly accretivity. Taking ψ(a):=γa2\psi(a):=\gamma\cdot a^{2}, γ(0,1)\gamma\in(0,1), for all a[0,+),ψΨa\in[0,+\infty),\psi\in\Psi, we get the usual definitions of strong pseudocontractivity and strong accretivity.

Denote by II the identity map.
REMARK 1.3. TT is ψ\psi-uniformly pseudocontractive if and only if S=(IT)S=(I-T) is ψ\psi uniformly accretive.

Let F(T)F(T) denote the fixed point set with respect to BB for the map TT.
In [9], the following conjecture was given: "if the Mann iteration converges, then so does the Ishikawa iteration." In a series of papers [5, 6, 7, 8, 9, 10], the authors have given a positive answer to this conjecture, showing the equivalence between Mann and Ishikawa iterations for several classes of maps. In this paper, we show that the convergence of Mann iteration is equivalent to the convergence of Ishikawa iteration, for the most general class of ψ\psi-uniformly pseudocontractive and ψ\psi-uniformly accretive maps.

Lemma 1.4 [4]. Let XX be a real Banach space and let J:X2XJ:X\rightarrow 2^{X^{*}} be the duality mapping. Then the following relation is true:

x+y2x2+2y,j(x+y),x,yX,j(x+y)J(x+y).\|x+y\|^{2}\leq\|x\|^{2}+2\langle y,j(x+y)\rangle,\quad\forall x,y\in X,\forall j(x+y)\in J(x+y). (1.10)

Lemma 1.5 [3]. Let {θn}\left\{\theta_{n}\right\} be a sequence of nonnegative real numbers, let {λn}\left\{\lambda_{n}\right\} be a real sequence satisfying

0λn1,n=0λn=+0\leq\lambda_{n}\leq 1,\quad\sum_{n=0}^{\infty}\lambda_{n}=+\infty (1.11)

and let ψΨ\psi\in\Psi. If there exists a positive integer n0n_{0} such that

θn+12θn2λnψ(θn+1)+σn,\theta_{n+1}^{2}\leq\theta_{n}^{2}-\lambda_{n}\psi\left(\theta_{n+1}\right)+\sigma_{n}, (1.12)

for all nn0n\geq n_{0}, with σn0\sigma_{n}\geq 0, for all nn\in\mathbb{N}, and σn=o(λn)\sigma_{n}=o\left(\lambda_{n}\right), then limnθn=0\lim_{n\rightarrow\infty}\theta_{n}=0.
2. Main result. We are now able to prove the following result.

Theorem 2.1. Let XX be a real Banach space, let BB be a nonempty, convex subset of XX, and let T:BBT:B\rightarrow B be a uniformly continuous and ψ\psi-uniformly pseudocontractive map with T(B)T(B) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3), and u0=x0Bu_{0}=x_{0}\in B, then the following are equivalent:
(i) the Mann iteration (1.1) converges (to xF(T)x^{*}\in F(T) ),
(ii) the Ishikawa iteration (1.2) converges (to the same xF(T)x^{*}\in F(T) ).

Proof. The implication (ii) \Rightarrow(i) is obvious by setting, in (1.2), βn=0\beta_{n}=0, for all nn\in\mathbb{N}. We will prove the implication (i) \Rightarrow (ii). Let xx^{*} be the fixed point of TT. Suppose that limnun=x\lim_{n\rightarrow\infty}u_{n}=x*. Using

limnxnun=0\displaystyle\lim_{n\rightarrow\infty}\left\|x_{n}-u_{n}\right\|=0 (2.1)
0xxnunx+xnun\displaystyle 0\leq\left\|x^{*}-x_{n}\right\|\leq\left\|u_{n}-x^{*}\right\|+\left\|x_{n}-u_{n}\right\| (2.2)

we get

limnxn=x\lim_{n\rightarrow\infty}x_{n}=x^{*} (2.3)

The proof is complete if we prove the relation (2.1).
Set

M:={x0u0+sup{TxTy,x,yB}}0.M:=\left\{\left\|x_{0}-u_{0}\right\|+\sup\{\|Tx-Ty\|,x,y\in B\}\right\}\geq 0. (2.4)

The condition that T(B)T(B) is bounded leads to

0M<+0\leq M<+\infty (2.5)

It is clear that x0u0M\left\|x_{0}-u_{0}\right\|\leq M. Supposing that xnunM\left\|x_{n}-u_{n}\right\|\leq M, we will prove that xn+1un+1M\left\|x_{n+1}-u_{n+1}\right\|\leq M. Indeed, from (1.1) and (1.2), we have

xn+1un+1\displaystyle\left\|x_{n+1}-u_{n+1}\right\| (1αn)xnun+αnTynTun\displaystyle\leq\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\left\|Ty_{n}-Tu_{n}\right\|
(1αn)M+αnM=M\displaystyle\leq\left(1-\alpha_{n}\right)M+\alpha_{n}M=M (2.6)

That is,

xnunM,n\left\|x_{n}-u_{n}\right\|\leq M,\quad\forall n\in\mathbb{N} (2.7)

The real function f:[0,+)[0,+),f(t)=t2f:[0,+\infty)\rightarrow[0,+\infty),f(t)=t^{2}, is increasing and convex. For all λ[0,1]\lambda\in[0,1] and t1,t2>0t_{1},t_{2}>0, we have

((1λ)t1+λt2)2(1λ)t12+λt22\left((1-\lambda)t_{1}+\lambda t_{2}\right)^{2}\leq(1-\lambda)t_{1}^{2}+\lambda t_{2}^{2} (2.8)

Set t1:=xnun,t2:=TynTun,λ:=αnt_{1}:=\left\|x_{n}-u_{n}\right\|,t_{2}:=Ty_{n}-Tu_{n},\lambda:=\alpha_{n} in (2.8), to obtain

xn+1un+12\displaystyle\left\|x_{n+1}-u_{n+1}\right\|^{2} =(1αn)(xnun)+αn(TynTun)2\displaystyle=\left\|\left(1-\alpha_{n}\right)\left(x_{n}-u_{n}\right)+\alpha_{n}\left(Ty_{n}-Tu_{n}\right)\right\|^{2}
((1αn)xnun+αnTynTun)2\displaystyle\leq\left(\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\left\|Ty_{n}-Tu_{n}\right\|\right)^{2}
(1αn)xnun2+αnM2\displaystyle\leq\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|^{2}+\alpha_{n}M^{2} (2.9)
xnun2+αnM2.\displaystyle\leq\left\|x_{n}-u_{n}\right\|^{2}+\alpha_{n}M^{2}.

From (1.1), (1.2), (1.5), and (1.10), with

x:=(1αn)(xnun)\displaystyle x:=\left(1-\alpha_{n}\right)\left(x_{n}-u_{n}\right)
y:=αn(TynTun)\displaystyle y:=\alpha_{n}\left(Ty_{n}-Tu_{n}\right) (2.10)
x+y=xn+1un+1\displaystyle x+y=x_{n+1}-u_{n+1}

we get

xn+1\displaystyle\|x_{n+1} un+12\displaystyle-u_{n+1}\|^{2}
=\displaystyle= (1αn)(xnun)+αn(TynTun)2\displaystyle\left\|\left(1-\alpha_{n}\right)\left(x_{n}-u_{n}\right)+\alpha_{n}\left(Ty_{n}-Tu_{n}\right)\right\|^{2}
\displaystyle\leq (1αn)2xnun2+2αnTynTun,j(xn+1un+1)\displaystyle\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\langle Ty_{n}-Tu_{n},j\left(x_{n+1}-u_{n+1}\right)\right\rangle
=\displaystyle= (1αn)2xnun2+2αnTxn+1Tun+1,j(xn+1un+1)\displaystyle\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\langle Tx_{n+1}-Tu_{n+1},j\left(x_{n+1}-u_{n+1}\right)\right\rangle
+2αnTynTxn+1,j(xn+1un+1)\displaystyle+2\alpha_{n}\left\langle Ty_{n}-Tx_{n+1},j\left(x_{n+1}-u_{n+1}\right)\right\rangle
+2αnTun+1Tun,j(xn+1un+1)\displaystyle+2\alpha_{n}\left\langle Tu_{n+1}-Tu_{n},j\left(x_{n+1}-u_{n+1}\right)\right\rangle
\displaystyle\leq (1αn)2xnun2+2αnxn+1un+122αnψ(xn+1un+1)\displaystyle\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\|x_{n+1}-u_{n+1}\right\|^{2}-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)
+2αnTynTxn+1,j(xn+1un+1)\displaystyle+2\alpha_{n}\left\langle Ty_{n}-Tx_{n+1},j\left(x_{n+1}-u_{n+1}\right)\right\rangle (2.11)
+2αnTun+1Tun,j(xn+1un+1)\displaystyle+2\alpha_{n}\left\langle Tu_{n+1}-Tu_{n},j\left(x_{n+1}-u_{n+1}\right)\right\rangle
\displaystyle\leq (1αn)2xnun2+2αnxn+1un+122αnψ(xn+1un+1)\displaystyle\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\|x_{n+1}-u_{n+1}\right\|^{2}-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)
+2αnTynTxn+1xn+1un+1\displaystyle+2\alpha_{n}\left\|Ty_{n}-Tx_{n+1}\right\|\left\|x_{n+1}-u_{n+1}\right\|
+2αnTun+1Tunxn+1un+1\displaystyle+2\alpha_{n}\left\|Tu_{n+1}-Tu_{n}\right\|\left\|x_{n+1}-u_{n+1}\right\|
\displaystyle\leq (1αn)2xnun2+2αnxn+1un+122αnψ(xn+1un+1)\displaystyle\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\|x_{n+1}-u_{n+1}\right\|^{2}-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)
+2αnTynTxn+1M+2αnTun+1TunM\displaystyle+2\alpha_{n}\left\|Ty_{n}-Tx_{n+1}\right\|M+2\alpha_{n}\left\|Tu_{n+1}-Tu_{n}\right\|M
=\displaystyle= (1αn)2xnun2+2αnxn+1un+122αnψ(xn+1un+1)\displaystyle\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\|x_{n+1}-u_{n+1}\right\|^{2}-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)
+2αn(bn+cn)\displaystyle+2\alpha_{n}\left(b_{n}+c_{n}\right)

where

bn\displaystyle b_{n} :=TynTxn+1M\displaystyle:=\left\|Ty_{n}-Tx_{n+1}\right\|M
cn\displaystyle c_{n} :=Tun+1TunM\displaystyle:=\left\|Tu_{n+1}-Tu_{n}\right\|M (2.12)

From (1.2), we have

xn+1yn\displaystyle\left\|x_{n+1}-y_{n}\right\| =(βnαn)xn+αnTynβnTxn\displaystyle=\left\|\left(\beta_{n}-\alpha_{n}\right)x_{n}+\alpha_{n}Ty_{n}-\beta_{n}Tx_{n}\right\|
(βnαn)xn+αnTyn+βnTxn\displaystyle\leq\left(\beta_{n}-\alpha_{n}\right)\left\|x_{n}\right\|+\alpha_{n}\left\|Ty_{n}\right\|+\beta_{n}\left\|Tx_{n}\right\| (2.13)

Analogously as for (2.6), we obtain the boundedness of {xn}\left\{x_{n}\right\}. Conditions (2.13) and (1.3) lead to

limnxn+1yn=0\lim_{n\rightarrow\infty}\left\|x_{n+1}-y_{n}\right\|=0 (2.14)

the uniform continuity of TT leads to

limnTynTxn+1=0\lim_{n\rightarrow\infty}\left\|Ty_{n}-Tx_{n+1}\right\|=0 (2.15)

thus, we have

limnbn=0\lim_{n\rightarrow\infty}b_{n}=0 (2.16)

The convergence of the Mann iteration {un}\left\{u_{n}\right\} implies limnun+1un=0\lim_{n\rightarrow\infty}\left\|u_{n+1}-u_{n}\right\|=0. The uniform continuity of TT implies limnTun+1Tun=0\lim_{n\rightarrow\infty}\left\|Tu_{n+1}-Tu_{n}\right\|=0, that is,

limncn=0\lim_{n\rightarrow\infty}c_{n}=0 (2.17)

Substituting (2.9) in (2.11) and using (2.7), we get

(1\displaystyle(1- αn)2xnun2+2αnxn+1un+12\displaystyle\left.\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\|x_{n+1}-u_{n+1}\right\|^{2}
(1αn)2xnun2+2αn(xnun2+αnM2)\displaystyle\leq\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left(\left\|x_{n}-u_{n}\right\|^{2}+\alpha_{n}M^{2}\right)
=[(1αn)2+2αn]xnun2+2αn2M2\displaystyle=\left[\left(1-\alpha_{n}\right)^{2}+2\alpha_{n}\right]\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}^{2}M^{2} (2.18)
=(1+αn2)xnun2+2αn2M2\displaystyle=\left(1+\alpha_{n}^{2}\right)\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}^{2}M^{2}
=xnun2+αn2xnun2+2αn2M2\displaystyle=\left\|x_{n}-u_{n}\right\|^{2}+\alpha_{n}^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}^{2}M^{2}
xnun2+3αn2M2.\displaystyle\leq\left\|x_{n}-u_{n}\right\|^{2}+3\alpha_{n}^{2}M^{2}.

Substituting (2.18) into (2.11), we obtain

xn+1un+12\displaystyle\left\|x_{n+1}-u_{n+1}\right\|^{2}
(1αn)2xnun2+2αnxn+1un+12\displaystyle\quad\leq\left(1-\alpha_{n}\right)^{2}\left\|x_{n}-u_{n}\right\|^{2}+2\alpha_{n}\left\|x_{n+1}-u_{n+1}\right\|^{2}
2αnψ(xn+1un+1)+2αn(bn+cn)\displaystyle\quad-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)+2\alpha_{n}\left(b_{n}+c_{n}\right) (2.19)
xnun2+3αn2M22αnψ(xn+1un+1)+2αn(bn+cn)\displaystyle\leq\left\|x_{n}-u_{n}\right\|^{2}+3\alpha_{n}^{2}M^{2}-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)+2\alpha_{n}\left(b_{n}+c_{n}\right)
=xnun22αnψ(xn+1un+1)+αn(3αnM2+2bn+2cn)\displaystyle=\left\|x_{n}-u_{n}\right\|^{2}-2\alpha_{n}\psi\left(\left\|x_{n+1}-u_{n+1}\right\|\right)+\alpha_{n}\left(3\alpha_{n}M^{2}+2b_{n}+2c_{n}\right)

Denote

θn\displaystyle\theta_{n} :=xnun2\displaystyle:=\left\|x_{n}-u_{n}\right\|^{2}
λn\displaystyle\lambda_{n} :=2αn\displaystyle:=2\alpha_{n} (2.20)
σn\displaystyle\sigma_{n} :=αn(3αnM2+2bn+2cn)\displaystyle:=\alpha_{n}\left(3\alpha_{n}M^{2}+2b_{n}+2c_{n}\right)

Condition (1.3) assures the existence of a positive integer n0n_{0} such that λn=2αn\lambda_{n}=2\alpha_{n}\leq 1, for all nn0n\geq n_{0}. Relations (1.3), (2.16), (2.17), (2.19), (2.20), and Lemma 1.5 lead to limnθn=0\lim_{n\rightarrow\infty}\theta_{n}=0; hence limnxnun=0\lim_{n\rightarrow\infty}\left\|x_{n}-u_{n}\right\|=0.

The above result does not completely generalize the main result, stated below, from [8], because the map TT in this result is not uniformly continuous.

Theorem 2.2 [8]. Let XX be a real Banach space with a uniformly convex dual and BB a nonempty, closed, convex, bounded subset of XX. Let T:BBT:B\rightarrow B be a continuous and strongly pseudocontractive operator. Then for u1=x1Bu_{1}=x_{1}\in B, the following assertions are equivalent:
(i) the Mann iteration (1.1) converges to the fixed point of TT;
(ii) the Ishikawa iteration (1.2) converges to the fixed point of TT.

Remark 2.3 [8]. (i) If TT has a fixed point, then Theorem 2.2 holds without the continuity of TT.
(ii) If BB is not bounded, then Theorem 2.2 holds if {xn}\left\{x_{n}\right\} is bounded.
3. The Lipschitzian case. The following result can be found in [6].

Corollary 3.1 [6]. Let XX be a real Banach space, BB a nonempty, convex subset of XX, and T:BBT:B\rightarrow B a Lipschitzian and ψ\psi-uniformly pseudocontractive map with T(B)T(B) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3), then the following are equivalent:
(i) the Mann iteration (1.1) converges (to xF(T)x^{*}\in F(T) ),
(ii) the Ishikawa iteration (1.2) converges (to the same xF(T)x^{*}\in F(T) ).

Proof. If the Lipschitzian constant L(0,1)L\in(0,1), then the conclusion holds on basis of [9, Theorem 3]. If L1L\geq 1, then all the assumptions in Theorem 2.1 are satisfied because a Lipschitzian map is uniformly continuous.

Corollary 3.1 does not completely generalize the main result, stated below, from [9], because neither boundedness of BB nor that of T(B)T(B) is required.

Theorem 3.2 [9]. Let BB be a closed, convex subset of an arbitrary Banach space XX and let TT be a Lipschitzian strongly pseudocontractive self-map of BB. Consider the Mann iteration and the Ishikawa iteration with the same initial point and {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfying (1.3). Then the following conditions are equivalent:
(i) the Mann iteration (1.1) converges to xF(T)x^{*}\in F(T),
(ii) the Ishikawa iteration (1.2) converges to xF(T)x^{*}\in F(T).
4. Application. Let SS be a ψ\psi-uniformly accretive map. Suppose the equation Sx=fSx=f has a solution for a given fXf\in X. Remark 1.3 assures that

Tx=x+fSx,xXTx=x+f-Sx,\quad\forall x\in X (4.1)

is a ψ\psi-uniformly pseudocontractive map. A fixed point for TT is a solution of Sx=fSx=f, and conversely. For the same {αn}(0,1),{βn}[0,1)\left\{\alpha_{n}\right\}\subset(0,1),\left\{\beta_{n}\right\}\subset[0,1) as in (1.3), the iterations (1.2) and (1.1) become

xn+1\displaystyle x_{n+1} =(1αn)xn+αn(f+(IS)yn)\displaystyle=\left(1-\alpha_{n}\right)x_{n}+\alpha_{n}\left(f+(I-S)y_{n}\right)
yn\displaystyle y_{n} =(1βn)xn+βn(f+(IS)xn),n=0,1,2,\displaystyle=\left(1-\beta_{n}\right)x_{n}+\beta_{n}\left(f+(I-S)x_{n}\right),\quad n=0,1,2,\ldots (4.2)
un+1\displaystyle u_{n+1} =(1αn)un+αn(f+(IS)un),n=0,1,2,\displaystyle=\left(1-\alpha_{n}\right)u_{n}+\alpha_{n}\left(f+(I-S)u_{n}\right),\quad n=0,1,2,\ldots (4.3)

We are now able to give the following result.
Corollary 4.1. Let XX be a real Banach space and S:XXS:X\rightarrow X a uniformly continuous and ψ\psi-uniformly accretive map with (IS)(X)(I-S)(X) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3) and u0=x0Bu_{0}=x_{0}\in B, then the following are equivalent:
(i) the Mann iteration (4.3) converges to a solution of Sx=fSx=f,
(ii) the Ishikawa iteration (4.2) converges to a solution of Sx=fSx=f.

Proof. Set Tx:=f+(IS)xTx:=f+(I-S)x. If SS is uniformly continuous, then TT is also uniformly continuous. The boundedness of (IS)(X)(I-S)(X) assures the boundedness of {yn+fSyn}\left\{\|y_{n}+f-\right.\left.Sy_{n}\|\right\} and {xn+fSxn}\left\{\left\|x_{n}+f-Sx_{n}\right\|\right\}. Hence Theorem 2.1 gives our conclusion.

From Corollary 3.1, we obtain, (see [6]) the following result.
Corollary 4.2 [6]. Let XX be a real Banach space and S:XXS:X\rightarrow X a Lipschitzian and ψ\psi-uniformly accretive map with (IS)(X)(I-S)(X) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3), then the following are equivalent:
(i) the Mann iteration (4.3) converges to a solution of Sx=fSx=f,
(ii) the Ishikawa iteration (4.2) converges to a solution of Sx=fSx=f.

Proof. Set, in Corollary 3.1, Tx:=(IS)xTx:=(I-S)x and use Remark 1.3.
5. The equivalence between TT-stabilities of Mann and Ishikawa iterations. All the arguments for the equivalence between TT-stabilities of Mann and Ishikawa iterations are similar to those from [5]. The following nonnegative sequences are well defined for all nn\in\mathbb{N} :

εn:=xn+1(1αn)xnαnTyn,\displaystyle\varepsilon_{n}:=\left\|x_{n+1}-\left(1-\alpha_{n}\right)x_{n}-\alpha_{n}Ty_{n}\right\|, (5.1)
δn:=un+1(1αn)unαnTun.\displaystyle\delta_{n}:=\left\|u_{n+1}-\left(1-\alpha_{n}\right)u_{n}-\alpha_{n}Tu_{n}\right\|. (5.2)

DEFINITION 5.1. If limnεn=0\lim_{n\rightarrow\infty}\varepsilon_{n}=0 (resp., limnδn=0\lim_{n\rightarrow\infty}\delta_{n}=0 ) implies that limnxn=x\lim_{n\rightarrow\infty}x_{n}=x^{*} (resp., limnun=x\lim_{n\rightarrow\infty}u_{n}=x^{*} ), then (1.2) (resp., (1.1)) is said to be TT-stable.

REMARK 5.2 [5]. Let XX be a normed space, BB a nonempty, convex, closed subset of XX, and T:BBT:B\rightarrow B a continuous map. If the Mann (resp., Ishikawa) iteration converges, then limnδn=0\lim_{n\rightarrow\infty}\delta_{n}=0 (resp., limnεn=0\lim_{n\rightarrow\infty}\varepsilon_{n}=0 ).

Theorem 5.3. Let XX be a real Banach space, BB a nonempty, convex subset of XX, and T:BBT:B\rightarrow B a uniformly continuous and ψ\psi-uniformly pseudocontractive map with T(B)T(B) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3) and u0=x0Bu_{0}=x_{0}\in B, then the following are equivalent:
(i) the Mann iteration (1.1) is TT-stable,
(ii) the Ishikawa iteration (1.2) is TT-stable.

Proof. The equivalence (i) \Leftrightarrow (ii) means that limnεn=0limnδn=0\lim_{n\rightarrow\infty}\varepsilon_{n}=0\Leftrightarrow\lim_{n\rightarrow\infty}\delta_{n}=0. The implication limnεn=0limnδn=0\lim_{n\rightarrow\infty}\varepsilon_{n}=0\Rightarrow\lim_{n\rightarrow\infty}\delta_{n}=0 is obvious by setting βn=0\beta_{n}=0, for all nn\in\mathbb{N}, in (1.2) and using (5.2). Conversely, we suppose that (1.1) is TT-stable. Using Definition 5.1, we get

limnδn=0limnun=x\lim_{n\rightarrow\infty}\delta_{n}=0\Longrightarrow\lim_{n\rightarrow\infty}u_{n}=x^{*} (5.3)

Theorem 2.1 assures that limnun=x\lim_{n\rightarrow\infty}u_{n}=x^{*} leads us to limnxn=x\lim_{n\rightarrow\infty}x_{n}=x*. Using Remark 5.2, we have limnεn=0\lim_{n\rightarrow\infty}\varepsilon_{n}=0. Thus, we get limnδn=0limnεn=0\lim_{n\rightarrow\infty}\delta_{n}=0\Rightarrow\lim_{n\rightarrow\infty}\varepsilon_{n}=0.

Set Tx=f+(IS)xTx=f+(I-S)x in Theorem 5.3. Corollary 3.1 leads to the following result.
Corollary 5.4. Let XX be a real Banach space and S:XXS:X\rightarrow X a uniformly continuous and ψ\psi-uniformly accretive map with (IS)(X)(I-S)(X) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3) and u0=x0Bu_{0}=x_{0}\in B, then the following are equivalent:
(i) the Mann iteration (4.3) is TT-stable,
(ii) the Ishikawa iteration (4.2) is TT-stable.

Analogously, we obtain the following corollary.
Corollary 5.5 [5]. Let XX be a real Banach space and S:XXS:X\rightarrow X a Lipschitzian and ψ\psi-uniformly accretive map with (IS)(X)(I-S)(X) bounded. If {αn},{βn}\left\{\alpha_{n}\right\},\left\{\beta_{n}\right\} satisfy (1.3), then the following are equivalent:
(i) the Mann iteration (4.3) is TT-stable,
(ii) the Ishikawa iteration (4.2) is TT-stable.

If the map TT is multivalued, then the definition of a ψ\psi-uniformly pseudocontractive map has the following form.

DEFINITION 5.6 Let XX be a real Banach space. Let BB be a nonempty subset. A map T:B2BT:B\rightarrow 2^{B} is called ψ\psi-uniformly pseudocontractive if there exist ψΨ\psi\in\Psi and j(xy)J(xy)j(x-y)\in J(x-y) such that

ξθ,j(xy)xy2ψ(xy),\langle\xi-\theta,j(x-y)\rangle\leq\|x-y\|^{2}-\psi(\|x-y\|), (5.4)

for all x,yB,ξTx,θTyx,y\in B,\xi\in Tx,\theta\in Ty.
Let S:X2XS:X\rightarrow 2^{X}. The map SS is called ψ\psi-uniformly accretive if there exist ψΨ\psi\in\Psi and j(xy)J(xy)j(x-y)\in J(x-y) such that

ξθ,j(xy)ψ(xy)\langle\xi-\theta,j(x-y)\rangle\geq\psi(\|x-y\|) (5.5)

for all x,yX,ξSx,θSyx,y\in X,\xi\in Sx,\theta\in Sy.

We remark that all the results from this paper hold in the multivalued case, provided that these multivalued maps admit an appropriate selection.

Acknowledgment. The authors are indebted to the referee for carefully reading the paper and for making useful suggestions.

References

[1] S. Ishikawa, Fixed points by a new iteration method, Proc. Amer. Math. Soc. 44 (1974), 147150.
[2] W. R. Mann, Mean value methods in iteration, Proc. Amer. Math. Soc. 4 (1953), 506-510.
[3] C. Moore and B. V. C. Nnoli, Iterative solution of nonlinear equations involving set-valued uniformly accretive operators, Comput. Math. Appl. 42 (2001), no. 1-2, 131-140.
[4] C. H. Morales and J. S. Jung, Convergence of paths for pseudocontractive mappings in Banach spaces, Proc. Amer. Math. Soc. 128 (2000), no. 11, 3411-3419.
[5] B. E. Rhoades and Ş. M. Şoltuz, The equivalence between TT-stabilities of Mann and Ishikawa iterations, submitted to Math. Commun.
[6] __, The equivalence of Mann and Ishikawa iteration for a Lipschitzian psi-uniformly pseudocontractive and psi-uniformly accretive map, to appear in Tamkang J. Math.
[7] _, The equivalence between the convergences of Ishikawa and Mann iterations for an asymptotically pseudocontractive map, J. Math. Anal. Appl. 283 (2003), no. 2, 681688.
[8] , The equivalence of Mann iteration and Ishikawa iteration for non-Lipschitzian operators, Int. J. Math. Math. Sci. 2003 (2003), no. 42, 2645-2651.
[9] __, On the equivalence of Mann and Ishikawa iteration methods, Int. J. Math. Math. Sci. 2003 (2003), no. 7, 451-459.
[10] ____\_\_\_\_ , The equivalence between the convergences of Ishikawa and Mann iterations for an asymptotically nonexpansive in the intermediate sense and strongly successively pseudocontractive maps, J. Math. Anal. Appl. 289 (2004), no. 1, 266-278.
B. E. Rhoades: Department of Mathematics, Indiana University, Bloomington, IN 47405-7106, USA

E-mail address: rhoades@indiana.edu
Ştefan M. Şoltuz: "Tiberiu Popoviciu" Institute of Numerical Analysis, P.O. Box 68-1, 400110 Cluj-Napoca, Romania

E-mail address: soltuzu1@yahoo.com

2004

Related Posts