Open Access

Approximating fixed points for continuous functions on an arbitrary interval

Journal of Inequalities and Applications20132013:214

https://doi.org/10.1186/1029-242X-2013-214

Received: 22 November 2012

Accepted: 15 April 2013

Published: 29 April 2013

Abstract

In this research article, we introduce a new iterative method for solving a fixed point problem of continuous functions on an arbitrary interval. We then prove the convergence theorem of the proposed algorithm. We finally give numerical examples to compare the result with Mann, Ishikawa and Noor iterations. Our main results extend the corresponding results in the literature.

MSC:47H09, 47H10.

Keywords

continuous functionconvergence theoremfixed pointiteration

1 Introduction

Let C be a closed interval on the real line and let f : C C be a continuous function. A point p C is called a fixed point of f if f ( p ) = p .

One classical way to approximate a fixed point of a nonlinear mapping was introduced in 1953 by Mann [1] as follows: a sequence { x n } defined by x 1 C and
x n + 1 = ( 1 α n ) x n + α n f ( x n )
(1.1)

for all n 1 , where { α n } is a sequence in [ 0 , 1 ] . Such an iteration process is known as Mann iteration. In 1991, Borwein and Borwein [2] proved the convergence theorem for a continuous function on the closed and bounded interval in the real line by using iteration (1.1).

Another classical iteration process was introduced by Ishikawa [3] as follows: a sequence { x n } defined by x 1 C and
y n = ( 1 β n ) x n + β n f ( x n ) , x n + 1 = ( 1 α n ) x n + α n f ( y n )
(1.2)

for all n 1 , where { α n } and { β n } are sequences in [ 0 , 1 ] . Such an iterative method is known as Ishikawa iteration. In 2006, Qing and Qihou [4] proved the convergence theorem of the sequence generated by iteration (1.2) for a continuous function on the closed interval in the real line (see also [5]).

In 2000, Noor [6] defined the following iterative scheme by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 β n ) x n + β n f ( z n ) , x n + 1 = ( 1 α n ) x n + α n f ( y n )
(1.3)

for all n 1 , where { α n } , { β n } and { μ n } are sequences in [ 0 , 1 ] . Such an iterative method is known as Noor iteration. Phuengrattana and Suantai [7] considered the convergence of Noor iteration for continuous functions on an arbitrary interval in the real line.

In this paper, motivated by the previous ones, we introduce a new modified iteration process for solving a fixed point problem for continuous functions on an arbitrary interval in the real line. Numerical examples are also presented to compare the result with Mann, Ishikawa and Noor iterations.

2 Main results

We begin this section by proving the following crucial lemmas.

Lemma 2.1 Let C be a closed interval on the real line (can be unbounded) and let f : C C be a continuous function. Let { α n } , { β n } , { μ n } , { γ n } and { τ n } be sequences in [ 0 , 1 ] with 0 τ n + β n 1 and 0 γ n + α n 1 . Let { x n } be a sequence generated iteratively by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 τ n β n ) x n + τ n z n + β n f ( z n ) , x n + 1 = ( 1 γ n α n ) z n + γ n y n + α n f ( y n ) , n 1 ,
(2.1)

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

If x n a , then a is a fixed point of f.

Proof Let x n a and suppose a f ( a ) . Then { x n } is bounded. So, { f ( x n ) } is bounded by the continuity of f. So are { y n } , { z n } , { f ( y n ) } and { f ( z n ) } . Moreover, z n a since x n a and μ n 0 . We also have y n a since z n a and β n 0 .

From (2.1) we obtain
x n + 1 = ( 1 γ n α n ) z n + γ n y n + α n f ( y n ) = z n + γ n ( y n z n ) + α n ( f ( y n ) z n ) = ( 1 μ n ) x n + μ n f ( x n ) + γ n ( y n z n ) + α n ( f ( y n ) z n ) = x n + μ n ( f ( x n ) x n ) + γ n ( y n z n ) + α n ( f ( y n ) z n ) = x n + μ n ( f ( x n ) x n ) + γ n ( ( 1 τ n β n ) ( x n z n ) + β n ( f ( z n ) z n ) ) + α n ( f ( y n ) z n ) = x n + μ n ( f ( x n ) x n ) + γ n ( ( 1 τ n β n ) μ n ( x n f ( x n ) ) + β n ( f ( z n ) z n ) ) + α n ( f ( y n ) z n ) = x n + μ n ( 1 γ n ( 1 τ n β n ) ) ( f ( x n ) x n ) + γ n β n ( f ( z n ) z n ) + α n ( f ( y n ) z n ) .
(2.2)
Let p k = f ( x k ) x k , q k = f ( z k ) z k and r k = f ( y k ) z k . Then we observe
lim k p k = lim n ( f ( x k ) x k ) = f ( a ) a 0 , lim k q k = lim n ( f ( z k ) z k ) = f ( a ) a 0 , lim k r k = lim n ( f ( y k ) z k ) = f ( a ) a 0 .
So, from (2.2) we obtain
x n = x 1 + k = 1 n μ k ( 1 γ k ( 1 τ k β k ) ) ( f ( x k ) x k ) + k = 1 n γ k β k ( f ( z k ) z k ) + k = 1 n α k ( f ( y k ) z k ) = x 1 + k = 1 n μ k ( 1 γ k ( 1 τ k β k ) ) p k + k = 1 n γ k β k q k + k = 1 n α k r k .

It is easy to see that k = 1 μ k ( 1 γ k ( 1 τ k β k ) ) p k < since lim k p k 0 and k = 1 μ k < . Similarly, we have k = 1 γ k β k q k < since lim k q k 0 and k = 1 β k < . This shows that { x n } is a divergent sequence since lim k r k 0 and k = 1 α k = . This contradicts the convergence of { x n } . Hence f ( a ) = a and a is a fixed point of f. □

Lemma 2.2 Let C be a closed interval on the real line (can be unbounded) and let f : C C be a continuous function. Let { α n } , { β n } , { μ n } , { γ n } and { τ n } be sequences in [ 0 , 1 ] with 0 τ n + β n 1 and 0 γ n + α n 1 . Let { x n } be a sequence generated iteratively by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 τ n β n ) x n + τ n z n + β n f ( z n ) , x n + 1 = ( 1 γ n α n ) z n + γ n y n + α n f ( y n ) , n 1 ,

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

If { x n } is bounded, then { x n } is convergent.

Proof Suppose { x n } is not convergent. Let a = lim inf n x n and b = lim sup n x n . Then a < b . We first show that if a < m < b , then f ( m ) = m . Suppose f ( m ) m . Without loss of generality, we suppose f ( m ) m > 0 . Since f is continuous, there exists δ with 0 < δ < b a such that for | x m | δ ,
f ( x ) x > 0 .
Since { x n } is bounded and f is continuous, { f ( x n ) } is bounded. Hence { z n } , { y n } , { f ( z n ) } and { f ( y n ) } are all bounded. Using
x n + 1 x n = ( 1 γ n α n ) ( z n x n ) + γ n ( y n x n ) + α n ( f ( y n ) x n ) , y n x n = τ n ( z n x n ) + β n ( f ( z n ) x n ) , z n x n = μ n ( f ( x n ) x n ) ,
we can easily show that | z n x n | 0 , | y n x n | 0 and | x n + 1 x n | 0 . Thus there exists a positive integer N such that for all n > N ,
| x n + 1 x n | < δ 2 , | y n x n | < δ 2 , | z n x n | < δ 2 .
(2.3)
Since b = lim sup n x n > m , there exists k 1 > N such that x n k 1 > m . Let n k 1 = k , then x k > m . For x k , there exist two cases as follows.
  1. (i)

    x k > m + δ 2 , then x k + 1 > x k δ 2 m using (2.3). So, we have x k + 1 > m .

     
  2. (ii)
    m < x k < m + δ 2 , then m δ 2 < y k < m + δ and m δ 2 < z k < m + δ by (2.3). So, we obtain | x k m | < δ 2 < δ , | y k m | < δ , | z k m | < δ . Hence
    f ( x k ) x k > 0 , f ( y k ) y k > 0 , f ( z k ) z k > 0 .
    (2.4)
     
We observe that
y k z k = ( 1 τ k β k ) ( x k z k ) + β k ( f ( z k ) z k ) = μ k ( 1 τ k β k ) ( x k f ( x k ) ) + β k ( f ( z k ) z k ) .
(2.5)
From (2.2), (2.4) and (2.5), we have
x k + 1 = x k + μ k ( 1 γ k ( 1 τ k β k ) ) ( f ( x k ) x k ) + γ k β k ( f ( z k ) z k ) + α k ( f ( y k ) z k ) = x k + μ k ( 1 γ k ( 1 τ k β k ) ) ( f ( x k ) x k ) + γ k β k ( f ( z k ) z k ) + α k ( f ( y k ) y k ) + α k ( y k z k ) = x k + μ k ( 1 γ k ( 1 τ k β k ) ) ( f ( x k ) x k ) + γ k β k ( f ( z k ) z k ) + α k ( f ( y k ) y k ) + α k ( μ k ( 1 τ k β k ) ( x k f ( x k ) ) + β k ( f ( z k ) z k ) ) = x k + μ k ( 1 ( γ k + α k ) ( 1 τ k β k ) ) ( f ( x k ) x k ) + α k ( f ( y k ) y k ) + β k ( γ k + α k ) ( f ( z k ) z k ) > x k .

Thus x k + 1 > x k > m . From (i) and (ii), we have x k + 1 > m . Similarly, we get that x k + 2 > m , x k + 3 > m , …. Thus we have x n > m for all n > k = n k 1 . So, a = lim k x n k m , which is a contradiction with a < m . Thus f ( m ) = m .

We next consider the following two cases.
  1. (i)
    There exists x M such that a < x M < b . Then f ( x M ) = x M . It follows that
    z M = ( 1 μ M ) x M + μ M f ( x M ) = x M
    and
    y M = ( 1 τ M β M ) x M + τ M z M + β M f ( z M ) = ( 1 τ M β M ) x M + τ M x M + β M f ( x M ) = x M .
    Hence, we obtain
    x M + 1 = ( 1 γ M α M ) z M + γ M y M + α M f ( y M ) = ( 1 γ M α M ) x M + γ M x M + α M f ( x M ) = x M .

    Similarly, we obtain x M = x M + 1 = x M + 2 = . So, we conclude that x n x M . Since there exists x n k a , x M = a . This shows that x n a , which is a contradiction.

     
  2. (ii)

    For all n , x n a or x n b . Since b a > 0 and lim n | x n + 1 x n | = 0 , there exists N ¯ such that | x n + 1 x n | < ( b a ) 2 for n > N ¯ . So, it is always that x n a for n > N ¯ ; or it is always that x n b for n > N ¯ . If x n a for n > N ¯ , then b = lim l x n l a , which is a contradiction with a < b . If x n b for n > N ¯ , then a = lim k x n k b , which is a contradiction with a < b . Thus we conclude that x n a . This completes the proof. □

     

We are now ready to prove the main results of this paper.

Theorem 2.3 Let C be a closed interval on the real line (can be unbounded) and let f : C C be a continuous function. Let { α n } , { β n } , { μ n } , { γ n } and { τ n } be sequences in [ 0 , 1 ] with 0 τ n + β n 1 and 0 γ n + α n 1 . Let { x n } be a sequence generated iteratively by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 τ n β n ) x n + τ n z n + β n f ( z n ) , x n + 1 = ( 1 γ n α n ) z n + γ n y n + α n f ( y n ) , n 1 ,

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

If { x n } is bounded, then { x n } converges to a fixed point of f.

Proof Let { x n } be a bounded sequence. Then, by Lemma (2.2), { x n } is a convergent sequence. Hence, by Lemma (2.1), it converges to a fixed point of f. □

As a direct consequence of Theorem 2.3, we obtain the following result.

Theorem 2.4 Let C be a closed interval on the real line (can be unbounded) and let f : C C be a continuous function. Let { α n } , { β n } , { μ n } , { γ n } and { τ n } be sequences in [ 0 , 1 ] with 0 τ n + β n 1 and 0 γ n + α n 1 . Let { x n } be a sequence generated iteratively by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 τ n β n ) x n + τ n z n + β n f ( z n ) , x n + 1 = ( 1 γ n α n ) z n + γ n y n + α n f ( y n ) , n 1 ,

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

Then { x n } converges to a fixed point of f if and only if { x n } is bounded.

Corollary 2.5 Let f : [ a , b ] [ a , b ] be a continuous function. Let { α n } , { β n } , { μ n } , { γ n } and { τ n } be sequences in [ 0 , 1 ] with 0 τ n + β n 1 and 0 γ n + α n 1 . Let { x n } be a sequence generated iteratively by x 1 [ a , b ] and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 τ n β n ) x n + τ n z n + β n f ( z n ) , x n + 1 = ( 1 γ n α n ) z n + γ n y n + α n f ( y n ) , n 1 ,

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

Then { x n } converges to a fixed point of f.

If we take τ n = γ n = 0 , then we obtain the following result.

Corollary 2.6 Let C be a closed interval on the real line (can be unbounded) and let f : C C be a continuous function. Let { α n } , { β n } and { μ n } be sequences in [ 0 , 1 ] . Let { x n } be a sequence generated iteratively by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 β n ) x n + β n f ( z n ) , x n + 1 = ( 1 α n ) z n + α n f ( y n ) , n 1 ,

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

Then { x n } converges to a fixed point of f if and only if { x n } is bounded.

If we take τ n + β n = 1 and γ n + α n = 1 , then we obtain the following result.

Corollary 2.7 Let C be a closed interval on the real line (can be unbounded) and let f : C C be a continuous function. Let { α n } , { β n } and { μ n } be sequences in [ 0 , 1 ] . Let { x n } be a sequence generated iteratively by x 1 C and
z n = ( 1 μ n ) x n + μ n f ( x n ) , y n = ( 1 β n ) z n + β n f ( z n ) , x n + 1 = ( 1 α n ) y n + α n f ( y n ) , n 1 ,

where n = 1 α n = , lim n α n = 0 , n = 1 β n < and n = 1 μ n < .

Then { x n } converges to a fixed point of f if and only if { x n } is bounded.

Remark 2.8 Corollary 2.7 extends the main result obtained in [8] from the modified Ishikawa iteration to the modified Noor iteration.

3 Numerical examples

In this section, we give numerical examples to demonstrate the convergence of the algorithm defined in this paper. For convenience, we call the iteration (2.1) the CP-iteration.

Example 3.1 Let f : [ 1 , ) [ 1 , ) be defined by f ( x ) = 0.9 ln x + 1 . Then f is a continuous function. Use the initial point x 1 = 3 and the control conditions α n = 1 ( n + 1 ) 0.5 , β n = 1 ( n + 1 ) 2.5 , μ n = 1 ( n + 1 ) 1.5 , τ n = 1 n + 1 and γ n = 1 7 .

Example 3.2 Let f : [ 1 , ) [ 1 , ) be defined by f ( x ) = 0.2 x 1 + x . Then f is a continuous function. Use the initial point x 1 = 5 and the control conditions α n = 1 ( n + 1 ) 0.5 , β n = 1 ( n + 1 ) 2 , μ n = 1 ( n + 1 ) 1.1 , τ n = 1 n + 1 and γ n = 1 5 .

Remark 3.3 From Table 1, Figure 1, Table 2 and Figure 2, we observe that the sequence generated by the CP-iteration converges to a fixed point faster than that of Mann, Ishikawa and Noor iterations.
Figure 1

Convergence behavior of Mann, Ishikawa, Noor and CP iterations for the function given in Example 3.1.

Figure 2

Convergence behavior of Mann, Ishikawa, Noor and CP iterations for the function given in Example 3.2.

Table 1

Comparison of the convergence rate of Mann, Ishikawa, Noor and CP iterations for the function given in Example 3.1

n

Mann

Ishikawa

Noor

CP-iteration

x n

x n

x n

x n

| f ( x n ) x n |

1

3.000000

3.000000

3.000000

3.000000

1.589769

10

1.074110

1.071538

1.071437

1.043308

0.024408

20

1.015478

1.014946

1.014925

1.008428

0.004658

30

1.004821

1.004656

1.004650

1.002531

0.001394

40

1.001822

1.001760

1.001757

1.000934

0.000514

50

1.000777

1.000750

1.000749

1.000392

0.000216

60

1.000360

1.000348

1.000348

1.000180

0.000099

70

1.000178

1.000172

1.000172

1.000088

0.000048

80

1.000093

1.000089

1.000089

1.000045

0.000025

90

1.000050

1.000048

1.000048

1.000024

0.000013

100

1.000028

1.000027

1.000027

1.000014

0.000007

Table 2

Comparison of the convergence rate of Mann, Ishikawa, Noor and CP iterations for the function given in Example 3.2

n

Mann

Ishikawa

Noor

CP-iteration

x n

x n

x n

x n

| f ( x n ) x n |

1

5.000000

5.000000

5.000000

5.000000

2.363932

10

1.529828

1.504824

1.501931

1.359649

0.073668

20

1.293319

1.284893

1.283943

1.224654

0.023218

30

1.223891

1.219794

1.219335

1.187049

0.011033

40

1.193599

1.191253

1.190991

1.171120

0.006204

50

1.177752

1.176279

1.176115

1.162994

0.003827

60

1.168542

1.167561

1.167452

1.158380

0.002505

70

1.162807

1.162126

1.162050

1.155568

0.001710

80

1.159055

1.158568

1.158513

1.153766

0.001205

90

1.156509

1.156152

1.156112

1.152565

0.000870

100

1.154730

1.154463

1.154433

1.151740

0.000641

110

1.153458

1.153255

1.153233

1.151160

0.000480

120

1.152530

1.152374

1.152357

1.150743

0.000365

130

1.151842

1.151721

1.151707

1.150438

0.000281

140

1.151325

1.151230

1.151219

1.150212

0.000218

150

1.150931

1.150856

1.150847

1.150042

0.000171

160

1.150628

1.150568

1.150561

1.149912

0.000136

170

1.150393

1.150345

1.150339

1.149813

0.000108

180

1.150209

1.150169

1.150165

1.149735

0.000087

190

1.150063

1.150031

1.150027

1.149675

0.000070

200

1.149947

1.149920

1.149918

1.149627

0.000057

Declarations

Acknowledgements

The authors wish to thank Professor Suthep Suantai for the valuable guidance and suggestion. The first author was supported by the Thailand Research Fund, the Commission on Higher Education, and University of Phayao under Grant No. MRG5580016.

Authors’ Affiliations

(1)
School of Science, University of Phayao

References

  1. Mann WR: Mean value methods in iteration. Proc. Am. Math. Soc. 1953, 4: 506–510. 10.1090/S0002-9939-1953-0054846-3View ArticleGoogle Scholar
  2. Borwein D, Borwein J: Fixed point iterations for real functions. J. Math. Anal. Appl. 1991, 157: 112–126. 10.1016/0022-247X(91)90139-QMathSciNetView ArticleGoogle Scholar
  3. Ishikawa S: Fixed points by a new iteration method. Proc. Am. Math. Soc. 1974, 44: 147–150. 10.1090/S0002-9939-1974-0336469-5MathSciNetView ArticleGoogle Scholar
  4. Qing Y, Qihou L: The necessary and sufficient condition for the convergence of Ishikawa iteration on an arbitrary interval. J. Math. Anal. Appl. 2006, 323: 1383–1386. 10.1016/j.jmaa.2005.11.058MathSciNetView ArticleGoogle Scholar
  5. Qing Y, Cho SY, Qin X: Convergence of Ishikawa iteration with error terms on an arbitrary interval. Commun. Korean Math. Soc. 2011, 26: 229–235. 10.4134/CKMS.2011.26.2.229MathSciNetView ArticleGoogle Scholar
  6. Noor MA: New approximation schemes for general variational inequalities. J. Math. Anal. Appl. 2000, 251: 217–229. 10.1006/jmaa.2000.7042MathSciNetView ArticleGoogle Scholar
  7. Phuengrattana W, Suantai S: On the rate of convergence of Mann Ishikawa, Noor and SP iterations for continuous functions on an arbitrary interval. J. Comput. Appl. Math. 2011, 235: 3006–3014. 10.1016/j.cam.2010.12.022MathSciNetView ArticleGoogle Scholar
  8. Suwana-adth W: The necessary and sufficient condition for the convergence of a new fixed point approximation method for continuous functions on an arbitrary interval. Thai J. Math. 2010, 8: 627–632.MathSciNetGoogle Scholar

Copyright

© Cholamjiak and Pholasa; licensee Springer 2013

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.