We show that T-stability of Mann and Ishikawa iterations are equivalent.
Authors
B.E. Rhoades
S.M. Soltuz
(Tiberiu Popoviciu Institute of Numerical Analysis, Romanian Academy)
Keywords
Mann iteration; Ishikawa iteration; T-stability
References
See the expanding block below.
Paper coordinates
B.E. Rhoades, Ş.M . Şoltuz, The equivalence between T-stabilities of Mann and Ishikawa iterations, J. Math. Anal. Appl. 318 (2006), 472-475. doi: 10.1016/j.jmaa.2005.05.066
The equivalence between the TT-stabilities of Mann and Ishikawa iterations
B.E. Rhoades ^("a "){ }^{\text {a }}, Ştefan M. Şoltuz ^("b, "){ }^{\text {b, }} t^(a){ }^{\mathrm{a}} Department of Mathematics, Indiana University, Bloomington, IN 47405-7106, USA^(b){ }^{\mathrm{b}} "T. Popoviciu" Institute of Numerical Analysis, PO Box 68-1, 400110 Cluj-Napoca, Romania
Received 16 January 2005
Available online 11 July 2005
Submitted by G. Jungck
Keywords: Mann iteration; Ishikawa iteration; TT-stability
1. Introduction
Let XX be a normed space and TT a selfmap of XX. Let x_(0)x_{0} be a point of XX, and assume that x_(n+1)=f(T,x_(n))x_{n+1}=f\left(T, x_{n}\right) is an iteration procedure, involving TT, which yields a sequence {x_(n)}\left\{x_{n}\right\} of point from XX. Suppose {x_(n)}\left\{x_{n}\right\} converges to a fixed point x^(**)x^{*} of TT. Let {xi_(n)}\left\{\xi_{n}\right\} be an arbitrary sequence in XX, and set epsilon_(n)=||xi_(n+1)-f(T,xi_(n))||\epsilon_{n}=\left\|\xi_{n+1}-f\left(T, \xi_{n}\right)\right\| for all n inNn \in \mathbb{N}.
Definition 1.1. [2] If ((lim_(n rarr oo)epsilon_(n)=0)=>(lim_(n rarr oo)xi_(n)=p))\left(\left(\lim _{n \rightarrow \infty} \epsilon_{n}=0\right) \Rightarrow\left(\lim _{n \rightarrow \infty} \xi_{n}=p\right)\right), then the iteration procedure x_(n+1)=f(T,x_(n))x_{n+1}=f\left(T, x_{n}\right) is said to be TT-stable with respect to TT.
Remark 1.2. [2] In practice, such a sequence {xi_(n)}\left\{\xi_{n}\right\} could arise in the following way. Let x_(0)x_{0} be a point in XX. Set x_(n+1)=f(T,x_(n))x_{n+1}=f\left(T, x_{n}\right). Let xi_(0)=x_(0)\xi_{0}=x_{0}. Now x_(1)=f(T,x_(0))x_{1}=f\left(T, x_{0}\right). Because of rounding or discretization in the function TT, a new value xi_(1)\xi_{1} approximately equal to x_(1)x_{1} might be obtained instead of the true value of f(T,x_(0))f\left(T, x_{0}\right). Then to approximate xi_(2)\xi_{2}, the value f(T,xi_(1))f\left(T, \xi_{1}\right) is computed to yields xi_(2)\xi_{2}, an approximation of f(T,xi_(1))f\left(T, \xi_{1}\right). This computation is continued to obtain {xi_(n)}\left\{\xi_{n}\right\} an approximate sequence of {x_(n)}\left\{x_{n}\right\}.
The two most popular iteration procedures for obtaining fixed points of TT, when the Banach principle fails, are Mann iteration [3], defined by
{:(1.1)u_(n+1)=(1-alpha_(n))u_(n)+alpha_(n)Tu_(n)",":}\begin{equation*}
u_{n+1}=\left(1-\alpha_{n}\right) u_{n}+\alpha_{n} T u_{n}, \tag{1.1}
\end{equation*}
and Ishikawa iteration [1], defined by
{:[x_(n+1)=(1-alpha_(n))x_(n)+alpha_(n)Tz_(n)],[(1.2)z_(n)=(1-beta_(n))x_(n)+beta_(n)Tx_(n)]:}\begin{align*}
& x_{n+1}=\left(1-\alpha_{n}\right) x_{n}+\alpha_{n} T z_{n} \\
& z_{n}=\left(1-\beta_{n}\right) x_{n}+\beta_{n} T x_{n} \tag{1.2}
\end{align*}
The sequences {alpha_(n)}sub(0,1),{beta_(n)}sub[0,1)\left\{\alpha_{n}\right\} \subset(0,1),\left\{\beta_{n}\right\} \subset[0,1) satisfy
A reasonable conjecture is that the Ishikawa iteration and the corresponding Mann iteration are equivalent for all maps for which either method provides convergence to a fixed point. In an attempt to verify this conjecture the authors, in a series of papers [4-9] have shown the equivalence for several classes of maps. We shall prove the equivalence between TT-stabilities of (1.1) and (1.2). Throughout this paper, we shall assume that both Mann and Ishikawa iterations converge to a fixed point of TT.
2. The equivalence between T\boldsymbol{T}-stabilities
Let {x_(n)}\left\{x_{n}\right\} be the Ishikawa iteration and {u_(n)}\left\{u_{n}\right\} be the Mann iteration. Let {s_(n)},{p_(n)}sub X\left\{s_{n}\right\},\left\{p_{n}\right\} \subset X be such that s_(0)=p_(0)s_{0}=p_{0}, and let (alpha_(n))_(n)sub(0,1),(beta_(n))_(n)sub[0,1)\left(\alpha_{n}\right)_{n} \subset(0,1),\left(\beta_{n}\right)_{n} \subset[0,1) satisfy (1.3) and
{:(2.1)y_(n)=(1-beta_(n))s_(n)+beta_(n)Ts_(n):}\begin{equation*}
y_{n}=\left(1-\beta_{n}\right) s_{n}+\beta_{n} T s_{n} \tag{2.1}
\end{equation*}
We consider the following nonnegative sequences, for all n inNn \in \mathbb{N} :
{:[(2.2)epsi_(n):=||s_(n+1)-(1-alpha_(n))s_(n)-alpha_(n)Ty_(n)||],[(2.3)delta_(n):=||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Tp_(n)||.]:}\begin{align*}
& \varepsilon_{n}:=\left\|s_{n+1}-\left(1-\alpha_{n}\right) s_{n}-\alpha_{n} T y_{n}\right\| \tag{2.2}\\
& \delta_{n}:=\left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T p_{n}\right\| . \tag{2.3}
\end{align*}
Definition 2.1. Definition 1.1 for (2.2) and (2.3) gives:
(i) If lim_(n rarr oo)epsi_(n)=0\lim _{n \rightarrow \infty} \varepsilon_{n}=0 implies that lim_(n rarr oo)s_(n)=x^(**)\lim _{n \rightarrow \infty} s_{n}=x^{*}, then the Ishikawa iteration (1.2), is said to be TT-stable.
(ii) If lim_(n rarr oo)delta_(n)=0\lim _{n \rightarrow \infty} \delta_{n}=0 implies that lim_(n rarr oo)p_(n)=x^(**)\lim _{n \rightarrow \infty} p_{n}=x^{*}, then the Mann iteration (1.1) is said to be TT-stable.
Remark 2.2. Let XX be a normed space and T:X rarr XT: X \rightarrow X a map. The following are equivalent:
(i) for all {alpha_(n)}sub(0,1),{beta_(n)}sub[0,1)\left\{\alpha_{n}\right\} \subset(0,1),\left\{\beta_{n}\right\} \subset[0,1) satisfying (1.3), the Ishikawa iteration is TT-stable,
(I) for all {alpha_(n)}sub(0,1),{beta_(n)}sub[0,1)\left\{\alpha_{n}\right\} \subset(0,1),\left\{\beta_{n}\right\} \subset[0,1) satisfying (1.3), AA{s_(n)}sub X\forall\left\{s_{n}\right\} \subset X :
Remark 2.3. Let XX be a normed space and T:X rarr XT: X \rightarrow X a map. The following are equivalent:
(ii) for all {alpha_(n)}sub(0,1)\left\{\alpha_{n}\right\} \subset(0,1) satisfying (1.3), the Mann iteration is TT-stable,
(II) for all {alpha_(n)}sub(0,1)\left\{\alpha_{n}\right\} \subset(0,1) satisfying (1.3), AA{p_(n)}sub X\forall\left\{p_{n}\right\} \subset X :
Theorem 2.4. Let XX be a normed space and T:X rarr XT: X \rightarrow X a map. Then the following are equivalent:
(i) for all {alpha_(n)}sub(0,1),{beta_(n)}sub[0,1)\left\{\alpha_{n}\right\} \subset(0,1),\left\{\beta_{n}\right\} \subset[0,1) satisfying (1.3), the Ishikawa iteration (1.2) is TT-stable,
(ii) for all {alpha_(n)}sub(0,1)\left\{\alpha_{n}\right\} \subset(0,1), satisfying (1.3), the Mann iteration (1.1) is TT-stable.
Proof. Let
M:=max{s u p_(n inN){||T(y_(n))||},s u p_(n inN){||T(s_(n))||},s u p_(n inN){||T(p_(n))||}}M:=\max \left\{\sup _{n \in \mathbb{N}}\left\{\left\|T\left(y_{n}\right)\right\|\right\}, \sup _{n \in \mathbb{N}}\left\{\left\|T\left(s_{n}\right)\right\|\right\}, \sup _{n \in \mathbb{N}}\left\{\left\|T\left(p_{n}\right)\right\|\right\}\right\}
Since the Mann and Ishikawa iterations converge, M < ooM<\infty. Remarks 2.2 and 2.3 assure that (i) ⇔ (ii) is equivalent to (I) ⇔ (II). We shall prove that (I) ⇒ (II). In (I) and (2.4) set s_(n):=p_(n)s_{n}:=p_{n} to obtain
{:[||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Tp_(n)||],[quad <= ||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Ty_(n)||+||alpha_(n)Ty_(n)-alpha_(n)Tp_(n)||],[quad <= ||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Ty_(n)||+alpha_(n)(||Ty_(n)||+||Tp_(n)||)],[(2.6)quad <= ||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Ty_(n)||+2alpha_(n)M rarr0quad" as "n rarr oo.]:}\begin{align*}
& \left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T p_{n}\right\| \\
& \quad \leqslant\left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T y_{n}\right\|+\left\|\alpha_{n} T y_{n}-\alpha_{n} T p_{n}\right\| \\
& \quad \leqslant\left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T y_{n}\right\|+\alpha_{n}\left(\left\|T y_{n}\right\|+\left\|T p_{n}\right\|\right) \\
& \quad \leqslant\left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T y_{n}\right\|+2 \alpha_{n} M \rightarrow 0 \quad \text { as } n \rightarrow \infty . \tag{2.6}
\end{align*}
Condition (I) assures that lim_(n rarr oo)||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Ty_(n)||=0=>lim_(n rarr oo)p_(n)=x^(**)\lim _{n \rightarrow \infty}\left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T y_{n}\right\|=0 \Rightarrow \lim _{n \rightarrow \infty} p_{n}=x^{*}. Thus, for a {p_(n)}\left\{p_{n}\right\} satisfying lim_(n rarr oo)||p_(n+1)-(1-alpha_(n))p_(n)-alpha_(n)Tp_(n)||=0\lim _{n \rightarrow \infty}\left\|p_{n+1}-\left(1-\alpha_{n}\right) p_{n}-\alpha_{n} T p_{n}\right\|=0, we have shown that lim_(n rarr oo)p_(n)=x^(**)\lim _{n \rightarrow \infty} p_{n}=x^{*}.
Conversely, we prove (II) =>\Rightarrow (I). In (II) and (2.5) set p_(n):=s_(n)p_{n}:=s_{n} to obtain
{:[||s_(n+1)-(1-alpha_(n))s_(n)-alpha_(n)Ty_(n)||],[quad <= ||s_(n+1)-(1-alpha_(n))s_(n)-alpha_(n)Ts_(n)||+||alpha_(n)Ty_(n)-alpha_(n)Ts_(n)||],[(2.7)quad <= ||s_(n+1)-(1-alpha_(n))s_(n)-alpha_(n)Ts_(n)||+2alpha_(n)M rarr0quad" as "n rarr oo]:}\begin{align*}
& \left\|s_{n+1}-\left(1-\alpha_{n}\right) s_{n}-\alpha_{n} T y_{n}\right\| \\
& \quad \leqslant\left\|s_{n+1}-\left(1-\alpha_{n}\right) s_{n}-\alpha_{n} T s_{n}\right\|+\left\|\alpha_{n} T y_{n}-\alpha_{n} T s_{n}\right\| \\
& \quad \leqslant\left\|s_{n+1}-\left(1-\alpha_{n}\right) s_{n}-\alpha_{n} T s_{n}\right\|+2 \alpha_{n} M \rightarrow 0 \quad \text { as } n \rightarrow \infty \tag{2.7}
\end{align*}
Condition (II) assures that lim_(n rarr oo)||s_(n+1)-(1-alpha_(n))s_(n)-alpha_(n)Ts_(n)||=0=>lim_(n rarr oo)s_(n)=x^(**)\lim _{n \rightarrow \infty}\left\|s_{n+1}-\left(1-\alpha_{n}\right) s_{n}-\alpha_{n} T s_{n}\right\|=0 \Rightarrow \lim _{n \rightarrow \infty} s_{n}=x^{*}. Thus, for a {s_(n)}\left\{s_{n}\right\} satisfying lim_(n rarr oo)||s_(n+1)-(1-alpha_(n))s_(n)-alpha_(n)Ty_(n)||=0\lim _{n \rightarrow \infty}\left\|s_{n+1}-\left(1-\alpha_{n}\right) s_{n}-\alpha_{n} T y_{n}\right\|=0, we have shown that lim_(n rarr oo)s_(n)=x^(**)\lim _{n \rightarrow \infty} s_{n}=x^{*}.
Set in (1.1) and (1.2), T:=T^(n)T:=T^{n}, to obtain the modified Mann and modified Ishikawa iterations. We suppose that both modified Mann and modified Ishikawa iterations converge to a fixed point of TT. Note that Definition 2.1, Remarks 2.2 and 2.3, and Theorem 2.4 hold in this case too.
Corollary 2.5. Let XX be a normed space and T:X rarr XT: X \rightarrow X a map. Then the following are equivalent:
(i) for all {alpha_(n)}sub(0,1),{beta_(n)}sub[0,1)\left\{\alpha_{n}\right\} \subset(0,1),\left\{\beta_{n}\right\} \subset[0,1) satisfying (1.3), the modified Ishikawa iteration is TT-stable,
(ii) for all {alpha_(n)}sub(0,1)\left\{\alpha_{n}\right\} \subset(0,1), satisfying (1.3), the modified Mann iteration is TT-stable.
References
[1] S. Ishikawa, Fixed points by a new iteration method, Proc. Amer. Math. Soc. 44 (1974) 147-150.
[2] A.M. Harder, T. Hicks, Stability results for fixed point iteration procedures, Math. Japonica 33 (1988) 693706.
[3] W.R. Mann, Mean value in iteration, Proc. Amer. Math. Soc. 4 (1953) 506-510.
[4] B.E. Rhoades, Ş.M. Şoltuz, On the equivalence of Mann and Ishikawa iteration methods, Internat. J. Math. Math. Sci. 2003 (2003) 451-459.
[5] B.E. Rhoades, Ş.M. Şoltuz, The equivalence of Mann iteration and Ishikawa iteration for non-Lipschitzian operators, Internat. J. Math. Math. Sci. 2003 (2003) 2645-2652.
[6] B.E. Rhoades, Ş.M. Şoltuz, The equivalence of Mann and Ishikawa iteration for psi\psi-uniformly pseudocontractive or psi\psi-uniformly accretive maps, Internat. J. Math. Math. Sci. 46 (2004) 2443-2452.
[7] B.E. Rhoades, Ş.M. Şoltuz, The equivalence of Mann and Ishikawa iteration for a Lipschitzian psi-uniformly pseudocontractive and psi-uniformly accretive maps, Tamkang J. Math. 35 (2004) 235-245.
[8] B.E. Rhoades, Ş.M. Şoltuz, The equivalence between the convergences of Ishikawa and Mann iterations for asymptotically nonexpansive in the intermediate sense and strongly successively pseudocontractive maps, J. Math. Anal. Appl. 289 (2004) 266-278.
[9] B.E. Rhoades, Ş.M. Şoltuz, The equivalence between Mann-Ishikawa iteration and the multistep iteration, Nonlinear Anal. 58 (2004) 219-228.