Data Dependence for Ishikawa Iteration When Dealing with Contractive-Like Operators

Abstract

We prove a convergence result and a data dependence for Ishikawa iteration when applied to contraction-like operators. An example is given, in which instead of computing the fixed point of an operator, we approximate the operator with a contractive-like one. For which it is possible to compute the fixed point, and therefore to approximate the fixed point of the initial operator.

    Authors

    S.M. Soltuz
    (Tiberiu Popoviciu Institute of Numerical Analysis, Romanian Academy)

    T. Grosan

    Keywords

    References

    See the expanding block below.

    Paper coordinates

    Ş. Şoltuz, T. Grosan, Data Dependence for Ishikawa Iteration When Dealing with Contractive-Like Operators. Fixed Point Theory Appl 2008, 242916 (2008).
    DOI: 10.1155/2008/242916

    PDF

    About this paper

    Publisher Name

    Springer

    Print ISSN

    0022-247X

    Online ISSN
    Google Scholar Profile

    google scholar

    soon

    Paper (preprint) in HTML form

    Data Dependence for Ishikawa Iteration When Dealing with Contractive-Like Operators

    Ş. M. Şoltuz 1,2 and Teodor Grosan 3
    1 Departamento de Matematicas, Universidad de los Andes, Carrera 1 No. 18A-10, Bogota, Colombia
    2 The Institute of Numerical Analysis, P.O. Box 68-1, Cluj-Napoca, Romania
    3 Department of Applied Mathematics, Babes-Bolyai University, Cluj-Napoca, Romania

    Correspondence should be addressed to Teodor Grosan, tgrosan@math.ubbcluj.ro
    Received 13 February 2008; Accepted 27 May 2008
    Recommended by Hichem Ben-El-Mechaiekh
    We prove a convergence result and a data dependence for Ishikawa iteration when applied to contraction-like operators. An example is given, in which instead of computing the fixed point of an operator, we approximate the operator with a contractive-like one. For which it is possible to compute the fixed point, and therefore to approximate the fixed point of the initial operator.

    Copyright (C) 2008 Ş. M. Şoltuz and T. Grosan. 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 XX be a real Banach space; let BXB\subset X be a nonempty convex closed and bounded set. Let T,S:BBT,S:B\rightarrow B be two maps. For a given x0,u0Bx_{0},u_{0}\in B, we consider the Ishikawa iteration (see [1]) for TT and SS :

    xn+1=(1αn)xn+αnTyn,yn=(1βn)xn+βnTxnun+1=(1αn)un+αnSvn,vn=(1βn)un+βnSun\begin{array}[]{ll}x_{n+1}=\left(1-\alpha_{n}\right)x_{n}+\alpha_{n}Ty_{n},&y_{n}=\left(1-\beta_{n}\right)x_{n}+\beta_{n}Tx_{n}\\ u_{n+1}=\left(1-\alpha_{n}\right)u_{n}+\alpha_{n}Sv_{n},&v_{n}=\left(1-\beta_{n}\right)u_{n}+\beta_{n}Su_{n}\end{array}

    where {αn}(0,1),{βn}0,1)\left.\left\{\alpha_{n}\right\}\subset(0,1),\left\{\beta_{n}\right\}\subset 0,1\right), and

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

    Set βn=0,nN\beta_{n}=0,\forall n\in N, to obtain the Mann iteration, see [2].
    The map TT is called Kannan mappings, see [3], if there exists b(0,1/2)b\in(0,1/2) such that for all x,yBx,y\in B,

    TxTyb(xTx+yTy)\|Tx-Ty\|\leq b(\|x-Tx\|+\|y-Ty\|) (1.4)

    Similar mappings are Chatterjea mappings, see [4], for which there exists c(0,1/2)c\in(0,1/2) such that for all x,yBx,y\in B,

    TxTyc(xTy+yTx)\|Tx-Ty\|\leq c(\|x-Ty\|+\|y-Tx\|) (1.5)

    Zamfirescu collected these classes. He introduced the following definition, see [5].
    Definition 1.1 (see [5,6]). The operator T:XXT:X\rightarrow X satisfies condition ZZ (Zamfirescu condition) if and only if there exist the real numbers a,b,ca,b,c satisfying 0<a<1,0<b,c<1/20<a<1,0<b,c<1/2 such that for each pair x,yx,y in XX, at least one condition is true:
    (i) (z1)TxTyaxy\left(z_{1}\right)\|Tx-Ty\|\leq a\|x-y\|,
    (ii) (z2)TxTyb(xTx+yTy)\left(z_{2}\right)\|Tx-Ty\|\leq b(\|x-Tx\|+\|y-Ty\|),
    (iii) (z3)TxTyc(xTy+yTx)\left(z_{3}\right)\|Tx-Ty\|\leq c(\|x-Ty\|+\|y-Tx\|).

    It is known, see Rhoades [7], that (z1),(z2)\left(z_{1}\right),\left(z_{2}\right), and (z3)\left(z_{3}\right) are independent conditions. Consider x,yBx,y\in B. Since TT satisfies condition ZZ, at least one of the conditions from (z1),(z2)\left(z_{1}\right),\left(z_{2}\right), and ( z3z_{3} ) is satisfied. If ( z2z_{2} ) holds, then

    TxTyb(xTx+yTy)b(xTx+(yx+xTx+TxTy)).\|Tx-Ty\|\leq b(\|x-Tx\|+\|y-Ty\|)\leq b(\|x-Tx\|+(\|y-x\|+\|x-Tx\|+\|Tx-Ty\|)). (1.6)

    Thus

    (1b)TxTybxy+2bxTx.(1-b)\|Tx-Ty\|\leq b\|x-y\|+2b\|x-Tx\|. (1.7)

    From 0b<10\leq b<1 one obtains,

    TxTyb1bxy+2b1bxTx\|Tx-Ty\|\leq\frac{b}{1-b}\|x-y\|+\frac{2b}{1-b}\|x-Tx\| (1.8)

    If (z3)\left(z_{3}\right) holds, then one gets

    TxTyc(xTy+yTx)c(xTx+TxTy+xy+xTx)\|Tx-Ty\|\leq c(\|x-Ty\|+\|y-Tx\|)\leq c(\|x-Tx\|+\|Tx-Ty\|+\|x-y\|+\|x-Tx\|) (1.9)

    Hence,

    (1c)TxTycxy+2cxTx,(1-c)\|Tx-Ty\|\leq c\|x-y\|+2c\|x-Tx\|, (1.10)

    that is,

    TxTyc1cxy+2c1cxTx\|Tx-Ty\|\leq\frac{c}{1-c}\|x-y\|+\frac{2c}{1-c}\|x-Tx\| (1.11)

    Denote

    δ:=max{a,b1b,c1c}\delta:=\max\left\{a,\frac{b}{1-b},\frac{c}{1-c}\right\} (1.12)

    to obtain

    0δ<10\leq\delta<1 (1.13)

    Finally, we get

    TxTyδxy+2δxTx,x,yB.\|Tx-Ty\|\leq\delta\|x-y\|+2\delta\|x-Tx\|,\quad\forall x,y\in B. (1.14)

    Formula (1.14) was obtained as in [8].
    Osilike and Udomene introduced in [9] a more general definition of a quasicontractive operator; they considered the operator for which there exists L0L\geq 0 and q(0,1)q\in(0,1) such that

    TxTyqxy+LxTx,x,yB.\|Tx-Ty\|\leq q\|x-y\|+L\|x-Tx\|,\quad\forall x,y\in B. (1.15)

    Imoru and Olatinwo considered in [10], the following general definition. Because they failed to name them, we will call them here contractive-like operators.

    Definition 1.2. One calls contractive-like the operator TT if there exist a constant q0,1q\in 0,1 ) and a strictly increasing and continuous function ϕ:[0,)[0,)\phi:[0,\infty)\rightarrow[0,\infty) with ϕ(0)=0\phi(0)=0 such that for each x,yXx,y\in X,

    TxTyqxy+ϕ(xTx).\|Tx-Ty\|\leq q\|x-y\|+\phi(\|x-Tx\|). (1.16)

    In both papers [9, 10], the TT-stability of Picard and Mann iterations was studied.

    2. Preliminaries

    The data dependence abounds in literature of fixed point theory when dealing with PicardBanach iteration, but is quasi-inexistent when dealing with Mann-Ishikawa iteration. As far as we know, the only data-dependence result concerning Mann-Ishikawa iteration is in [11]. There, the data dependence of Ishikawa iteration was proven when applied to contractions. In this note, we will prove data-dependence results for Ishikawa iteration when applied to the above contractive-like operators. Usually, Ishikawa iteration is more complicated but nevertheless more stable as Mann iteration. There is a classic example, see [12], in which Mann iteration does not converge while Ishikawa iteration does. This is the main reason for considering Ishikawa iteration in Theorem 3.2.

    The following remark is obvious by using the inequality (1x)exp(x),x0(1-x)\leq\exp(x),\forall x\geq 0.
    Remark 2.1. Let {θn}\left\{\theta_{n}\right\} be a nonnegative sequence such that θn(0,1],n\theta_{n}\in(0,1],\forall n\in\mathbb{N}. If n=1θn=\sum_{n=1}^{\infty}\theta_{n}=\infty, then n=1(1θn)=0\prod_{n=1}^{\infty}\left(1-\theta_{n}\right)=0.

    The following is similar to lemma from [13]. (Note that another proof for this lemma [13] can be found in [11].)

    Lemma 2.2. Let {an}\left\{a_{n}\right\} be a nonnegative sequence for which one supposes there exists n0n_{0}\in\mathbb{N}, such that for all nn0n\geq n_{0} one has satisfied the following inequality:

    an+1(1λn)an+λnσna_{n+1}\leq\left(1-\lambda_{n}\right)a_{n}+\lambda_{n}\sigma_{n} (2.1)

    where λn(0,1),n,n=1λn=\lambda_{n}\in(0,1),\forall n\in\mathbb{N},\sum_{n=1}^{\infty}\lambda_{n}=\infty, and σn0n\sigma_{n}\geq 0\forall n\in\mathbb{N}. Then,

    0limnsupanlimnsupσn0\leq\lim_{n\rightarrow\infty}\sup a_{n}\leq\lim_{n\rightarrow\infty}\sup\sigma_{n} (2.2)

    Proof. There exists n1n_{1}\in\mathbb{N} such that σnlim supσn,nn1\sigma_{n}\leq\limsup\sigma_{n},\forall n\geq n_{1}. Set n2=max{n0,n1}n_{2}=\max\left\{n_{0},n_{1}\right\} such that the following inequality holds, for all nn2n\geq n_{2} :

    an+1(1λn)(1λn1)(1λn1)an1+limnsupσna_{n+1}\leq\left(1-\lambda_{n}\right)\left(1-\lambda_{n-1}\right)\cdots\left(1-\lambda_{n_{1}}\right)a_{n_{1}}+\lim_{n\rightarrow\infty}\sup\sigma_{n} (2.3)

    Using the above Remark 2.1 with θn=λn\theta_{n}=\lambda_{n}, we get the conclusion. In order to prove (2.3), consider (2.1) and the induction step:

    an+2\displaystyle a_{n+2}\leq (1λn+1)an+1+λn+1σn+1(1λn+1)(1λn)(1λn1)(1λn1)an1\displaystyle\left(1-\lambda_{n+1}\right)a_{n+1}+\lambda_{n+1}\sigma_{n+1}\leq\left(1-\lambda_{n+1}\right)\left(1-\lambda_{n}\right)\left(1-\lambda_{n-1}\right)\cdots\left(1-\lambda_{n_{1}}\right)a_{n_{1}}
    +(1λn+1)limnsupσn+λn+1σn+1\displaystyle+\left(1-\lambda_{n+1}\right)\lim_{n\rightarrow\infty}\sup\sigma_{n}+\lambda_{n+1}\sigma_{n+1} (2.4)
    =\displaystyle= (1λn+1)(1λn)(1λn1)(1λn1)an1+limnsupσn.\displaystyle\left(1-\lambda_{n+1}\right)\left(1-\lambda_{n}\right)\left(1-\lambda_{n-1}\right)\cdots\left(1-\lambda_{n_{1}}\right)a_{n_{1}}+\lim_{n\rightarrow\infty}\sup\sigma_{n}.

    3. Main results

    Theorem 3.1. Let XX be a real Banach space, BXB\subset X a nonempty convex and closed set, and T:BBT:B\rightarrow B a contractive-like map with xx^{*} being the fixed point. Then for all x0Bx_{0}\in B, the iteration (1.1) converges to the unique fixed point of TT.

    Proof. The uniqueness comes from (1.16); supposing we have two fixed points xx^{*} and yy^{*}, we get

    xy=TxTyqxy+ϕ(xTx)=qxy,\left\|x^{*}-y^{*}\right\|=\left\|Tx^{*}-Ty^{*}\right\|\leq q\left\|x^{*}-y^{*}\right\|+\phi\left(\left\|x^{*}-Tx^{*}\right\|\right)=q\left\|x^{*}-y^{*}\right\|, (3.1)

    that is, (1q)xy=0(1-q)\left\|x^{*}-y^{*}\right\|=0. From (1.1) and (1.16) we obtain

    xn+1x\displaystyle\left\|x_{n+1}-x^{*}\right\| (1αn)xnx+αnTynTx\displaystyle\leq\left(1-\alpha_{n}\right)\left\|x_{n}-x^{*}\right\|+\alpha_{n}\left\|Ty_{n}-Tx^{*}\right\|
    (1αn)xnx+αnqynx\displaystyle\leq\left(1-\alpha_{n}\right)\left\|x_{n}-x^{*}\right\|+\alpha_{n}q\left\|y_{n}-x^{*}\right\|
    (1αn)xnx+αnq(1βn)xnx+qαnβnTxnTx\displaystyle\leq\left(1-\alpha_{n}\right)\left\|x_{n}-x^{*}\right\|+\alpha_{n}q\left(1-\beta_{n}\right)\left\|x_{n}-x^{*}\right\|+q\alpha_{n}\beta_{n}\left\|Tx_{n}-Tx^{*}\right\| (3.2)
    (1αn(1q))(1(1q)βn))xnx\displaystyle\left.\leq\left(1-\alpha_{n}(1-q)\right)\left(1-(1-q)\beta_{n}\right)\right)\left\|x_{n}-x^{*}\right\|
    (1αn(1q))xnx(k=1n(1αkq))x0x\displaystyle\leq\left(1-\alpha_{n}(1-q)\right)\left\|x_{n}-x^{*}\right\|\leq\cdots\leq\left(\prod_{k=1}^{n}\left(1-\alpha_{k}q\right)\right)\left\|x_{0}-x^{*}\right\|

    Use Remark 2.1 with θk=αkq\theta_{k}=\alpha_{k}q to obtain the conclusion.
    This result allows us to formulate the following data dependence theorem.
    Theorem 3.2. Let XX be a real Banach space, let BXB\subset X be a nonempty convex and closed set, and let ε>0\varepsilon>0 be a fixed number. If T:BBT:B\rightarrow B is a contractive-like operator with the fixed point xx^{*} and S:BBS:B\rightarrow B is an operator with a fixed point uu^{*}, (supposed nearest to xx^{*} ), and if the following relation is satisfied:

    TzSzε,zB\|Tz-Sz\|\leq\varepsilon,\quad\forall z\in B (3.3)

    then

    xuε1q\left\|x^{*}-u^{*}\right\|\leq\frac{\varepsilon}{1-q} (3.4)

    Proof. From (1.1) and (1.2), we have

    xn+1un+1=(1αn)(xnun)+αn(TynSvn).x_{n+1}-u_{n+1}=\left(1-\alpha_{n}\right)\left(x_{n}-u_{n}\right)+\alpha_{n}\left(Ty_{n}-Sv_{n}\right). (3.5)

    Thus

    xn+1un+1=\displaystyle\left\|x_{n+1}-u_{n+1}\right\|= (1αn)(xnun)+αn(SvnTyn)\displaystyle\left\|\left(1-\alpha_{n}\right)\left(x_{n}-u_{n}\right)+\alpha_{n}\left(Sv_{n}-Ty_{n}\right)\right\|
    \displaystyle\leq (1αn)xnun+αnSvnTvn+TvnTyn\displaystyle\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\left\|Sv_{n}-Tv_{n}+Tv_{n}-Ty_{n}\right\|
    \displaystyle\leq (1αn)xnun+αnTvnSvn+αnTvnTyn\displaystyle\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\left\|Tv_{n}-Sv_{n}\right\|+\alpha_{n}\left\|Tv_{n}-Ty_{n}\right\|
    \displaystyle\leq (1αn)xnun+αnε+qαnynvn+αnϕ(ynTyn)\displaystyle\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\varepsilon+q\alpha_{n}\left\|y_{n}-v_{n}\right\|+\alpha_{n}\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)
    \displaystyle\leq (1αn)xnun+αnε+qαn(1βn)xnun+qαnβnTxnSun+αnϕ(ynTyn)\displaystyle\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\varepsilon+q\alpha_{n}\left(1-\beta_{n}\right)\left\|x_{n}-u_{n}\right\|+q\alpha_{n}\beta_{n}\left\|Tx_{n}-Su_{n}\right\|+\alpha_{n}\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)
    \displaystyle\leq (1αn)xnun+αnε+qαn(1βn)xnun\displaystyle\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\varepsilon+q\alpha_{n}\left(1-\beta_{n}\right)\left\|x_{n}-u_{n}\right\|
    +αnβnq(TxnTun+TunSun)+αnϕ(ynTyn)\displaystyle+\alpha_{n}\beta_{n}q\left(\left\|Tx_{n}-Tu_{n}\right\|+\left\|Tu_{n}-Su_{n}\right\|\right)+\alpha_{n}\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)
    \displaystyle\leq (1αn)xnun+αnε+qαn(1βn)xnun\displaystyle\left(1-\alpha_{n}\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\varepsilon+q\alpha_{n}\left(1-\beta_{n}\right)\left\|x_{n}-u_{n}\right\|
    +q2αnβnxnun+qαnβnϕ(xnTxn)+qαnβnε+αnϕ(ynTyn)\displaystyle+q^{2}\alpha_{n}\beta_{n}\left\|x_{n}-u_{n}\right\|+q\alpha_{n}\beta_{n}\phi\left(\left\|x_{n}-Tx_{n}\right\|\right)+q\alpha_{n}\beta_{n}\varepsilon+\alpha_{n}\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)
    =\displaystyle= (1αn(1q(1βn)βnq2))xnun+αnε+qαnβnε\displaystyle\left(1-\alpha_{n}\left(1-q\left(1-\beta_{n}\right)-\beta_{n}q^{2}\right)\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\varepsilon+q\alpha_{n}\beta_{n}\varepsilon
    +qαnβnϕ(xnTxn)+αnϕ(ynTyn)\displaystyle+q\alpha_{n}\beta_{n}\phi\left(\left\|x_{n}-Tx_{n}\right\|\right)+\alpha_{n}\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)
    =\displaystyle= (1αn(1q)(1+qβn))xnun+αn(qβnϕ(xnTxn)+ϕ(ynTyn)+qβnε+ε)\displaystyle\left(1-\alpha_{n}(1-q)\left(1+q\beta_{n}\right)\right)\left\|x_{n}-u_{n}\right\|+\alpha_{n}\left(q\beta_{n}\phi\left(\left\|x_{n}-Tx_{n}\right\|\right)+\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)+q\beta_{n}\varepsilon+\varepsilon\right)
    \displaystyle\leq (1αn(1q))xnun+(αn(1q))qβnϕ(xnTxn)+ϕ(ynTyn)+qβnε+ε1q.\displaystyle\left(1-\alpha_{n}(1-q)\right)\left\|x_{n}-u_{n}\right\|+\left(\alpha_{n}(1-q)\right)\frac{q\beta_{n}\phi\left(\left\|x_{n}-Tx_{n}\right\|\right)+\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)+q\beta_{n}\varepsilon+\varepsilon}{1-q}. (3.6)

    Note that limnϕ(xnTxn)=limnϕ(ynTyn)=0\lim_{n\rightarrow\infty}\phi\left(\left\|x_{n}-Tx_{n}\right\|\right)=\lim_{n\rightarrow\infty}\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)=0 because ϕ\phi is a continuous map and both {xn},{yn}\left\{x_{n}\right\},\left\{y_{n}\right\} converge to the fixed point of TT. Set

    λn:=αn(1q)\displaystyle\lambda_{n}:=\alpha_{n}(1-q)
    σn:=qβnϕ(xnTxn)+ϕ(ynTyn)+qβnε+ε1q\displaystyle\sigma_{n}:=\frac{q\beta_{n}\phi\left(\left\|x_{n}-Tx_{n}\right\|\right)+\phi\left(\left\|y_{n}-Ty_{n}\right\|\right)+q\beta_{n}\varepsilon+\varepsilon}{1-q} (3.7)

    and use Lemma 2.2 to obtain the conclusion

    xuε1q\left\|x^{*}-u^{*}\right\|\leq\frac{\varepsilon}{1-q} (3.8)

    Remark 3.3. (i) Set βn=0,n\beta_{n}=0,\forall n\in\mathbb{N}, to obtain the data dependence for Mann iteration.
    (ii) The Zamfirescu operators and implicitly (Chatterjea and Kannan) are contractive-like operators, therefore our Theorem 3.2 remains true for these classes.

    4. Numerical example

    The following example follows the example from [8].
    Example 4.1. Let T:T:\mathbb{R}\rightarrow\mathbb{R} be given by

    Tx\displaystyle Tx =0, if x(,2]\displaystyle=0,\quad\text{ if }x\in(-\infty,2] (4.1)
    =0.5, if x(2,+).\displaystyle=-0.5,\quad\text{ if }x\in(2,+\infty).

    Then TT is contractive-like operator with q=0.2q=0.2 and ϕ=\phi= identity.
    Note the unique fixed point is 0 . Consider now the map S:RRS:R\rightarrow R,

    Sx\displaystyle Sx =1, if x(,2]\displaystyle=1,\quad\text{ if }x\in(-\infty,2] (4.2)
    =1.5, if x(2,+)\displaystyle=-1.5,\quad\text{ if }x\in(2,+\infty)

    with the unique fixed point 1 . Take ε\varepsilon to be the distance between the two maps as follows:

    SxTx1,x\|Sx-Tx\|\leq 1,\quad\forall x\in\mathbb{R} (4.3)

    Set u0=x0=0,αn=βn=1/(n+1)u_{0}=x_{0}=0,\alpha_{n}=\beta_{n}=1/(n+1). Independently of above theory, the Ishikawa iteration applied to SS, leads to

    Iteration step Ishikawa iteration
    1 0.5
    10 0.9
    100 0.99

    Note that for n=1n=1,

    0.5=1n+10+1n+1S(12)0.5=\frac{1}{n+1}0+\frac{1}{n+1}S\left(\frac{1}{2}\right) (4.5)

    since y1=(1/(n+1))0+(1/n+1)1=1/2y_{1}=(1/(n+1))0+(1/n+1)1=1/2. (The above computations can be obtained also by using a Matlab program.) This leads us to "conclude" that Ishikawa iteration applied to SS converges to fixed point, (x=1)\left(x^{*}=1\right). Eventually, one can see that the distance between the two fixed points is one. Actually, without knowing the fixed point of SS (and without computing it), via Theorem 3.2, we can do the following estimate for it:

    xu11q=110.2=108=1.2\left\|x^{*}-u^{*}\right\|\leq\frac{1}{1-q}=\frac{1}{1-0.2}=\frac{10}{8}=1.2 (4.6)

    As a conclusion, instead of computing fixed points of SS, choose TT more closely to SS and the distance between the fixed points will shrink too.

    Acknowledgments

    The authors are indebted to referee for carefully reading the paper and for making useful suggestions. This work was supported by CEEX ET 90/2006-2008.

    References

    [1] S. Ishikawa, "Fixed points by a new iteration method," Proceedings of the American Mathematical Society, vol. 44, no. 1, pp. 147-150, 1974.
    [2] W. R. Mann, "Mean value methods in iteration," Proceedings of the American Mathematical Society, vol. 4, no. 3, pp. 506-510, 1953.
    [3] R. Kannan, "Some results on fixed points," Bulletin of the Calcutta Mathematical Society, vol. 60, pp. 71-76, 1968.
    [4] S. K. Chatterjea, "Fixed-point theorems," Comptes Rendus de l’Académie Bulgare des Sciences, vol. 25, pp. 727-730, 1972.
    [5] T. Zamfirescu, "Fix point theorems in metric spaces," Archiv der Mathematik, vol. 23, no. 1, pp. 292-298, 1972.
    [6] B. E. Rhoades, "Fixed point iterations using infinite matrices," Transactions of the American Mathematical Society, vol. 196, pp. 161-176, 1974.
    [7] B. E. Rhoades, "A comparison of various definitions of contractive mappings," Transactions of the American Mathematical Society, vol. 226, pp. 257-290, 1977.
    [8] V. Berinde, "On the convergence of the Ishikawa iteration in the class of quasi contractive operators," Acta Mathematica Universitatis Comenianae, vol. 73, no. 1, pp. 119-126, 2004.
    [9] M. O. Osilike and A. Udomene, "Short proofs of stability results for fixed point iteration procedures for a class of contractive-type mappings," Indian Journal of Pure and Applied Mathematics, vol. 30, no. 12, pp. 1229-1234, 1999.
    [10] C. O. Imoru and M. O. Olatinwo, "On the stability of Picard and Mann iteration processes," Carpathian Journal of Mathematics, vol. 19, no. 2, pp. 155-160, 2003.
    [11] Ş. M. Şoltuz, "Data dependence for Ishikawa iteration," Lecturas Matemáticas, vol. 25, no. 2, pp. 149155, 2004.
    [12] C. E. Chidume and S. A. Mutangadura, "An example of the Mann iteration method for Lipschitz pseudocontractions," Proceedings of the American Mathematical Society, vol. 129, no. 8, pp. 2359-2363, 2001.
    [13] J. A. Park, "Mann-iteration process for the fixed point of strictly pseudocontractive mapping in some Banach spaces," Journal of the Korean Mathematical Society, vol. 31, no. 3, pp. 333-337, 1994.

    2008

    Related Posts