Research Article A New Sixth-Order Steffensen-Type …downloads.hindawi.com/journals/ijanal/2014/685796.pdfA New Sixth-Order Steffensen-Type Iterative Method for Solving Nonlinear
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Research ArticleA New Sixth-Order Steffensen-Type Iterative Method forSolving Nonlinear Equations
Tahereh Eftekhari
Faculty of Mathematics University of Sistan and Baluchestan Zahedan 987-98155 Iran
Correspondence should be addressed to Tahereh Eftekhari teftekhari2009gmailcom
Received 11 November 2013 Revised 6 January 2014 Accepted 9 January 2014 Published 12 February 2014
Academic Editor Ahmed Zayed
Copyright copy 2014 Tahereh Eftekhari This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
Based on iterative method proposed by Basto et al (2006) we present a new derivative-free iterative method for solving nonlinearequations The aim of this paper is to develop a new method to find the approximation of the root 120572 of the nonlinear equation119891(119909) = 0 This method has the efficiency index which equals 614 = 15651 The benefit of this method is that this method doesnot need to calculate any derivative Several examples illustrate that the efficiency of the newmethod is better than that of previousmethods
1 Introduction
Solving nonlinear equations is one of themost important andchallenging problems in scientific and engineering applica-tions In this paper we consider an iterative method to findthe root of a nonlinear equation 119891(119909) = 0
Newtonrsquos method is the best known iterative method forsolving nonlinear equations [1] given by
119909119899+1= 119909119899minus
119891 (119909119899)
1198911015840(119909119899)
119899 = 0 1 2 (1)
which converges quadratically But it has a major weaknessone has to calculate the derivative of 119891(119909) at each approxima-tion Frequently 1198911015840(119909
119899) is far more difficult to evaluate and
needs more arithmetic operations to calculate than 119891(119909)It is well known that the forward-difference approxima-
tion for 1198911015840(119909119899) at 119909 is
1198911015840(119909) asymp
119891 (119909 + ℎ) minus 119891 (119909)
ℎ
(2)
If the derivative 1198911015840(119909119899) is replaced by the forward-difference
approximation with ℎ = 119891(119909119899) that is
1198911015840(119909119899) asymp
119891 (119909119899+ 119891 (119909
119899)) minus 119891 (119909
119899)
119891 (119909119899)
(3)
the Newtonrsquos method becomes
119909119899+1= 119909119899minus
119891(119909119899)2
119891 (119909119899+ 119891 (119909
119899)) minus 119891 (119909
119899)
119899 = 0 1 2 (4)
which is the famous Steffensenrsquos method [2] The Steffensenrsquosmethod is based on forward-difference approximation toderivative This method is a tough competitor of Newtonrsquosmethod Both methods are of quadratic convergence andboth require two functions evaluation per iteration butin contrast to Newtonrsquos method Steffensenrsquos method isderivative-free Based on this method many derivative-freeiterative methods have been proposed
In [3] a sixth-order derivative-free iterative method hasbeen proposed by Cordero et al as follows
119910119899= 119909119899minus
2119891(119909119899)2
119891 (119909119899+ 119891 (119909
119899)) minus 119891 (119909
119899minus 119891 (119909
119899))
119911119899= 119910119899minus
119910119899minus 119909119899
2119891 (119910119899) minus 119891 (119909
119899)
119891 (119910119899)
119909119899+1= 119911119899minus
119910119899minus 119909119899
2119891 (119910119899) minus 119891 (119909
119899)
119891 (119911119899)
(5)
which has efficiency index 1430
Hindawi Publishing CorporationInternational Journal of AnalysisVolume 2014 Article ID 685796 5 pageshttpdxdoiorg1011552014685796
2 International Journal of Analysis
Soleymani andHosseinabadi [4] suggested another sixth-order derivative-free scheme in the form below
119910119899= 119909119899minus
119891 (119909119899)
119891 [119909119899 119908119899]
119911119899= 119909119899minus
119891 (119909119899)
119891 [119909119899 119908119899]
(1 +
119891 (119910119899)
119891 (119909119899)
(1 + 2
119891 (119910119899)
119891 (119909119899)
))
119909119899+1= 119911119899minus
119891 (119911119899)
119891 [119910119899 119911119899]
(1 minus
1 + 119891 [119909119899 119908119899]
119891 [119909119899 119908119899]
119891 (119911119899)
119891 (119908119899)
)
(6)
wherein the efficiency index is 15651In [5] a sixth-order derivative-free algorithm has been
derived by Yasmin and Iftikhar which is written as
119891 [119910119899 119911119899] minus 119891 [119909
119899 119910119899] + 119891 [119909
119899 119911119899]
(7)
which has efficiency index 15651Theprocesses of removing the derivatives usually increase
the number of function evaluations per iteration In ourmethod we used the technique of composition of Newtonrsquosmethod with the known methods which not only increasethe order of themethod as high as possible but also reduce thenumber of function evaluations and improve the efficiencyindex of the composed method In view of this fact thenew steffensen-like methods are significantly better whencompared with the established methods described above
This paper is organized as follows some basic definitionsrelevant to the present work are presented in Section 2 InSection 3 we present a new three-step sixth-order iterativemethod for solving nonlinear equations The new method isfree from derivative We prove that the order of convergenceof the new method is six Numerical examples show betterperformance of our method in Section 4 Section 5 is a shortconclusion
2 Basic Definitions
Definition 1 Let 119891(119909) be a real function with a simple root 120572and let 119909
119899119899isinN be a sequence of real numbers that converges
towards 120572 Then we say that the order of convergence of thesequence is 119901 if there exists a number 119901 isin R+ such that
lim119899rarrinfin
119909119899+1minus 120572
(119909119899minus 120572)119901= 119862 (8)
where for some 119862 = 0 119862 is known as the asymptotic errorconstant
If 119901 = 1 2 or 3 the sequence is said to have linearconvergence quadratic convergence or cubic convergencerespectively
Definition 2 Let 119890119899= 119909119899minus 120572 be the error in the 119899th iteration
One calls the relation
119890119899+1= 119862119890119901
119899+ 119874 (119890
119901+1
119899) (9)
the error equation
Definition 3 Let 119902 be the number of function evaluationsof the new method The efficiency of the new method ismeasured by the concept of efficiency index [6] and definedas
119864 = 1199011119902 (10)
where 119901 is the order of the method
Definition 4 Suppose that119909119899+1
119909119899 and 119909
119899minus1are three succes-
sive iterations closer to the root 120572 Then the computationalorder of convergence [7] is approximated by using (9) asfollows
119891 [119910119899 119911119899] minus 119891 [119909
119899 119910119899] + 119891 [119909
119899 119911119899]
(20)
where 119908119899= 119909119899+ 119891(119909
119899)
Theorem 5 demonstrates its convergence analysis
Theorem 5 Let 120572 isin 119868 be a simple root of a sufficientlydifferentiable function 119891 119868 sube R rarr R in an open interval 119868If the initial approximation 119909
0is sufficiently close to 120572 then the
derivative-freemethod defined by (20) has order of convergencesix
Proof Let 120572 be the simple root of 119891(119909) that is 119891(120572) = 01198911015840(120572) = 0 and the error equation is 119890
119899= 119909119899minus 120572
By Taylorrsquos expansion of 119891(119909119899) about 119909 = 120572 and putting
Now dividing (33) by (29) and using the last step of (20) wehave
119890119899+1=
(1 + 1198881)2
1198883
2(1198882
2minus 11988811198883)
1198885
1
1198906
119899+ 119874 (119890
7
119899) (34)
This proves that our first proposed method defined by (20) isa sixth-order derivative-free algorithm and satisfies the aboveerror equation This completes the proof
Nowwe discuss the efficiency index of the newmethod byusing Definition 3 as 1199011119902 where 119901 is the order of the methodand 119902 is the number of function evaluations per iterationrequired of the method It is easy to know that the numberof function evaluations per iteration required by the methoddefined in algorithm 1 is four So the efficiency index is 614 =15651
4 Numerical Results
In this section we test the effectiveness of our new methodWe have used second-order method of Steffensen (SM) [2]sixth-order method of Cordero et al (CM) [3] six-ordermethod of Soleymani and Hosseinabadi (SHM) [4] andsix-order method of Yasmin and Iftikhar (YIM) [5] forcomparison with our method to find the simple root ofnonlinear equations The test functions of 119891(119909) are listed inTable 1
Numerical computations reported here have been carriedout in a 119872119886119905ℎ119890119898119886119905119894119888119886 80 environment Table 2 shows thedifference of the root 120572 and the approximation 119909
119899to 120572
where 120572 is the exact root computed with 800 significant digits(Digits = 800) The absolute values of the function (|119891(119909
119899)|)
the difference between the approximated root 119909119899and the
exact root 120572 the number of function evaluations (TNFE)and the computational order of convergence (COC) are alsoshown in Table 2 Here COC is defined in Definition 4
5 Conclusions
We have obtained a new Steffensen-type iterative method forsolving nonlinear equations The convergence order of thismethod is six and consists of four evaluations of the functionper iteration so it has an efficiency index equal to 614 =15651 Numerical examples also show that the numericalresults of our new method in equal iterations improve theresults of other existing methods
Conflict of Interests
The author declares that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
The author is grateful to the referees for their comments andsuggestions that helped to improve the paper
References
[1] W Bi H Ren and Q Wu ldquoNew family of seventh-ordermethods for nonlinear equationsrdquo Applied Mathematics andComputation vol 203 no 1 pp 408ndash412 2008
[2] D Kincaid and W Cheney Numerical Analysis BrooksColePacific Grove Calif USA 2nd edition 1996
[3] A Cordero J L Hueso E Martınez and J R TorregrosaldquoSteffensen type methods for solving nonlinear equationsrdquoJournal of Computational andAppliedMathematics vol 236 no12 pp 3058ndash3064 2012
[4] F Soleymani and V Hosseinabadi ldquoNew third- and sixth-orderderivativefree techniques for nonlinear equationsrdquo Journal ofMathematics Research vol 3 no 2 pp 107ndash112 2011
[5] N Yasmin and S Iftikhar ldquoSome new Steffensen like three-stepmethods for solving nonlinear equationsrdquo International Journalof Pure and Applied Mathematics vol 82 pp 557ndash572 2013
[6] W Gautschi Numerical Analysis Birkhauser Boston MassUSA 1997
[7] S Weerakoon and T G I Fernando ldquoA variant of Newtonrsquosmethod with accelerated third-order convergencerdquo AppliedMathematics Letters vol 13 no 8 pp 87ndash93 2000
[8] M Basto V Semiao and F L Calheiros ldquoA new iterativemethod to compute nonlinear equationsrdquo Applied Mathematicsand Computation vol 173 no 1 pp 468ndash483 2006
[9] J Kou Y Li and XWang ldquoModified Halleyrsquos method free fromsecond derivativerdquo Applied Mathematics and Computation vol183 no 1 pp 704ndash708 2006
[10] M Dehghan and M Hajarian ldquoSome derivative free quadraticand cubic convergence iterative formulas for solving nonlinearequationsrdquoComputational andAppliedMathematics vol 29 no1 pp 19ndash30 2010
119891 [119910119899 119911119899] minus 119891 [119909
119899 119910119899] + 119891 [119909
119899 119911119899]
(7)
which has efficiency index 15651Theprocesses of removing the derivatives usually increase
the number of function evaluations per iteration In ourmethod we used the technique of composition of Newtonrsquosmethod with the known methods which not only increasethe order of themethod as high as possible but also reduce thenumber of function evaluations and improve the efficiencyindex of the composed method In view of this fact thenew steffensen-like methods are significantly better whencompared with the established methods described above
This paper is organized as follows some basic definitionsrelevant to the present work are presented in Section 2 InSection 3 we present a new three-step sixth-order iterativemethod for solving nonlinear equations The new method isfree from derivative We prove that the order of convergenceof the new method is six Numerical examples show betterperformance of our method in Section 4 Section 5 is a shortconclusion
2 Basic Definitions
Definition 1 Let 119891(119909) be a real function with a simple root 120572and let 119909
119899119899isinN be a sequence of real numbers that converges
towards 120572 Then we say that the order of convergence of thesequence is 119901 if there exists a number 119901 isin R+ such that
lim119899rarrinfin
119909119899+1minus 120572
(119909119899minus 120572)119901= 119862 (8)
where for some 119862 = 0 119862 is known as the asymptotic errorconstant
If 119901 = 1 2 or 3 the sequence is said to have linearconvergence quadratic convergence or cubic convergencerespectively
Definition 2 Let 119890119899= 119909119899minus 120572 be the error in the 119899th iteration
One calls the relation
119890119899+1= 119862119890119901
119899+ 119874 (119890
119901+1
119899) (9)
the error equation
Definition 3 Let 119902 be the number of function evaluationsof the new method The efficiency of the new method ismeasured by the concept of efficiency index [6] and definedas
119864 = 1199011119902 (10)
where 119901 is the order of the method
Definition 4 Suppose that119909119899+1
119909119899 and 119909
119899minus1are three succes-
sive iterations closer to the root 120572 Then the computationalorder of convergence [7] is approximated by using (9) asfollows
119891 [119910119899 119911119899] minus 119891 [119909
119899 119910119899] + 119891 [119909
119899 119911119899]
(20)
where 119908119899= 119909119899+ 119891(119909
119899)
Theorem 5 demonstrates its convergence analysis
Theorem 5 Let 120572 isin 119868 be a simple root of a sufficientlydifferentiable function 119891 119868 sube R rarr R in an open interval 119868If the initial approximation 119909
0is sufficiently close to 120572 then the
derivative-freemethod defined by (20) has order of convergencesix
Proof Let 120572 be the simple root of 119891(119909) that is 119891(120572) = 01198911015840(120572) = 0 and the error equation is 119890
119899= 119909119899minus 120572
By Taylorrsquos expansion of 119891(119909119899) about 119909 = 120572 and putting
Now dividing (33) by (29) and using the last step of (20) wehave
119890119899+1=
(1 + 1198881)2
1198883
2(1198882
2minus 11988811198883)
1198885
1
1198906
119899+ 119874 (119890
7
119899) (34)
This proves that our first proposed method defined by (20) isa sixth-order derivative-free algorithm and satisfies the aboveerror equation This completes the proof
Nowwe discuss the efficiency index of the newmethod byusing Definition 3 as 1199011119902 where 119901 is the order of the methodand 119902 is the number of function evaluations per iterationrequired of the method It is easy to know that the numberof function evaluations per iteration required by the methoddefined in algorithm 1 is four So the efficiency index is 614 =15651
4 Numerical Results
In this section we test the effectiveness of our new methodWe have used second-order method of Steffensen (SM) [2]sixth-order method of Cordero et al (CM) [3] six-ordermethod of Soleymani and Hosseinabadi (SHM) [4] andsix-order method of Yasmin and Iftikhar (YIM) [5] forcomparison with our method to find the simple root ofnonlinear equations The test functions of 119891(119909) are listed inTable 1
Numerical computations reported here have been carriedout in a 119872119886119905ℎ119890119898119886119905119894119888119886 80 environment Table 2 shows thedifference of the root 120572 and the approximation 119909
119899to 120572
where 120572 is the exact root computed with 800 significant digits(Digits = 800) The absolute values of the function (|119891(119909
119899)|)
the difference between the approximated root 119909119899and the
exact root 120572 the number of function evaluations (TNFE)and the computational order of convergence (COC) are alsoshown in Table 2 Here COC is defined in Definition 4
5 Conclusions
We have obtained a new Steffensen-type iterative method forsolving nonlinear equations The convergence order of thismethod is six and consists of four evaluations of the functionper iteration so it has an efficiency index equal to 614 =15651 Numerical examples also show that the numericalresults of our new method in equal iterations improve theresults of other existing methods
Conflict of Interests
The author declares that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
The author is grateful to the referees for their comments andsuggestions that helped to improve the paper
References
[1] W Bi H Ren and Q Wu ldquoNew family of seventh-ordermethods for nonlinear equationsrdquo Applied Mathematics andComputation vol 203 no 1 pp 408ndash412 2008
[2] D Kincaid and W Cheney Numerical Analysis BrooksColePacific Grove Calif USA 2nd edition 1996
[3] A Cordero J L Hueso E Martınez and J R TorregrosaldquoSteffensen type methods for solving nonlinear equationsrdquoJournal of Computational andAppliedMathematics vol 236 no12 pp 3058ndash3064 2012
[4] F Soleymani and V Hosseinabadi ldquoNew third- and sixth-orderderivativefree techniques for nonlinear equationsrdquo Journal ofMathematics Research vol 3 no 2 pp 107ndash112 2011
[5] N Yasmin and S Iftikhar ldquoSome new Steffensen like three-stepmethods for solving nonlinear equationsrdquo International Journalof Pure and Applied Mathematics vol 82 pp 557ndash572 2013
[6] W Gautschi Numerical Analysis Birkhauser Boston MassUSA 1997
[7] S Weerakoon and T G I Fernando ldquoA variant of Newtonrsquosmethod with accelerated third-order convergencerdquo AppliedMathematics Letters vol 13 no 8 pp 87ndash93 2000
[8] M Basto V Semiao and F L Calheiros ldquoA new iterativemethod to compute nonlinear equationsrdquo Applied Mathematicsand Computation vol 173 no 1 pp 468ndash483 2006
[9] J Kou Y Li and XWang ldquoModified Halleyrsquos method free fromsecond derivativerdquo Applied Mathematics and Computation vol183 no 1 pp 704ndash708 2006
[10] M Dehghan and M Hajarian ldquoSome derivative free quadraticand cubic convergence iterative formulas for solving nonlinearequationsrdquoComputational andAppliedMathematics vol 29 no1 pp 19ndash30 2010
119891 [119910119899 119911119899] minus 119891 [119909
119899 119910119899] + 119891 [119909
119899 119911119899]
(20)
where 119908119899= 119909119899+ 119891(119909
119899)
Theorem 5 demonstrates its convergence analysis
Theorem 5 Let 120572 isin 119868 be a simple root of a sufficientlydifferentiable function 119891 119868 sube R rarr R in an open interval 119868If the initial approximation 119909
0is sufficiently close to 120572 then the
derivative-freemethod defined by (20) has order of convergencesix
Proof Let 120572 be the simple root of 119891(119909) that is 119891(120572) = 01198911015840(120572) = 0 and the error equation is 119890
119899= 119909119899minus 120572
By Taylorrsquos expansion of 119891(119909119899) about 119909 = 120572 and putting
Now dividing (33) by (29) and using the last step of (20) wehave
119890119899+1=
(1 + 1198881)2
1198883
2(1198882
2minus 11988811198883)
1198885
1
1198906
119899+ 119874 (119890
7
119899) (34)
This proves that our first proposed method defined by (20) isa sixth-order derivative-free algorithm and satisfies the aboveerror equation This completes the proof
Nowwe discuss the efficiency index of the newmethod byusing Definition 3 as 1199011119902 where 119901 is the order of the methodand 119902 is the number of function evaluations per iterationrequired of the method It is easy to know that the numberof function evaluations per iteration required by the methoddefined in algorithm 1 is four So the efficiency index is 614 =15651
4 Numerical Results
In this section we test the effectiveness of our new methodWe have used second-order method of Steffensen (SM) [2]sixth-order method of Cordero et al (CM) [3] six-ordermethod of Soleymani and Hosseinabadi (SHM) [4] andsix-order method of Yasmin and Iftikhar (YIM) [5] forcomparison with our method to find the simple root ofnonlinear equations The test functions of 119891(119909) are listed inTable 1
Numerical computations reported here have been carriedout in a 119872119886119905ℎ119890119898119886119905119894119888119886 80 environment Table 2 shows thedifference of the root 120572 and the approximation 119909
119899to 120572
where 120572 is the exact root computed with 800 significant digits(Digits = 800) The absolute values of the function (|119891(119909
119899)|)
the difference between the approximated root 119909119899and the
exact root 120572 the number of function evaluations (TNFE)and the computational order of convergence (COC) are alsoshown in Table 2 Here COC is defined in Definition 4
5 Conclusions
We have obtained a new Steffensen-type iterative method forsolving nonlinear equations The convergence order of thismethod is six and consists of four evaluations of the functionper iteration so it has an efficiency index equal to 614 =15651 Numerical examples also show that the numericalresults of our new method in equal iterations improve theresults of other existing methods
Conflict of Interests
The author declares that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
The author is grateful to the referees for their comments andsuggestions that helped to improve the paper
References
[1] W Bi H Ren and Q Wu ldquoNew family of seventh-ordermethods for nonlinear equationsrdquo Applied Mathematics andComputation vol 203 no 1 pp 408ndash412 2008
[2] D Kincaid and W Cheney Numerical Analysis BrooksColePacific Grove Calif USA 2nd edition 1996
[3] A Cordero J L Hueso E Martınez and J R TorregrosaldquoSteffensen type methods for solving nonlinear equationsrdquoJournal of Computational andAppliedMathematics vol 236 no12 pp 3058ndash3064 2012
[4] F Soleymani and V Hosseinabadi ldquoNew third- and sixth-orderderivativefree techniques for nonlinear equationsrdquo Journal ofMathematics Research vol 3 no 2 pp 107ndash112 2011
[5] N Yasmin and S Iftikhar ldquoSome new Steffensen like three-stepmethods for solving nonlinear equationsrdquo International Journalof Pure and Applied Mathematics vol 82 pp 557ndash572 2013
[6] W Gautschi Numerical Analysis Birkhauser Boston MassUSA 1997
[7] S Weerakoon and T G I Fernando ldquoA variant of Newtonrsquosmethod with accelerated third-order convergencerdquo AppliedMathematics Letters vol 13 no 8 pp 87ndash93 2000
[8] M Basto V Semiao and F L Calheiros ldquoA new iterativemethod to compute nonlinear equationsrdquo Applied Mathematicsand Computation vol 173 no 1 pp 468ndash483 2006
[9] J Kou Y Li and XWang ldquoModified Halleyrsquos method free fromsecond derivativerdquo Applied Mathematics and Computation vol183 no 1 pp 704ndash708 2006
[10] M Dehghan and M Hajarian ldquoSome derivative free quadraticand cubic convergence iterative formulas for solving nonlinearequationsrdquoComputational andAppliedMathematics vol 29 no1 pp 19ndash30 2010
Now dividing (33) by (29) and using the last step of (20) wehave
119890119899+1=
(1 + 1198881)2
1198883
2(1198882
2minus 11988811198883)
1198885
1
1198906
119899+ 119874 (119890
7
119899) (34)
This proves that our first proposed method defined by (20) isa sixth-order derivative-free algorithm and satisfies the aboveerror equation This completes the proof
Nowwe discuss the efficiency index of the newmethod byusing Definition 3 as 1199011119902 where 119901 is the order of the methodand 119902 is the number of function evaluations per iterationrequired of the method It is easy to know that the numberof function evaluations per iteration required by the methoddefined in algorithm 1 is four So the efficiency index is 614 =15651
4 Numerical Results
In this section we test the effectiveness of our new methodWe have used second-order method of Steffensen (SM) [2]sixth-order method of Cordero et al (CM) [3] six-ordermethod of Soleymani and Hosseinabadi (SHM) [4] andsix-order method of Yasmin and Iftikhar (YIM) [5] forcomparison with our method to find the simple root ofnonlinear equations The test functions of 119891(119909) are listed inTable 1
Numerical computations reported here have been carriedout in a 119872119886119905ℎ119890119898119886119905119894119888119886 80 environment Table 2 shows thedifference of the root 120572 and the approximation 119909
119899to 120572
where 120572 is the exact root computed with 800 significant digits(Digits = 800) The absolute values of the function (|119891(119909
119899)|)
the difference between the approximated root 119909119899and the
exact root 120572 the number of function evaluations (TNFE)and the computational order of convergence (COC) are alsoshown in Table 2 Here COC is defined in Definition 4
5 Conclusions
We have obtained a new Steffensen-type iterative method forsolving nonlinear equations The convergence order of thismethod is six and consists of four evaluations of the functionper iteration so it has an efficiency index equal to 614 =15651 Numerical examples also show that the numericalresults of our new method in equal iterations improve theresults of other existing methods
Conflict of Interests
The author declares that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
The author is grateful to the referees for their comments andsuggestions that helped to improve the paper
References
[1] W Bi H Ren and Q Wu ldquoNew family of seventh-ordermethods for nonlinear equationsrdquo Applied Mathematics andComputation vol 203 no 1 pp 408ndash412 2008
[2] D Kincaid and W Cheney Numerical Analysis BrooksColePacific Grove Calif USA 2nd edition 1996
[3] A Cordero J L Hueso E Martınez and J R TorregrosaldquoSteffensen type methods for solving nonlinear equationsrdquoJournal of Computational andAppliedMathematics vol 236 no12 pp 3058ndash3064 2012
[4] F Soleymani and V Hosseinabadi ldquoNew third- and sixth-orderderivativefree techniques for nonlinear equationsrdquo Journal ofMathematics Research vol 3 no 2 pp 107ndash112 2011
[5] N Yasmin and S Iftikhar ldquoSome new Steffensen like three-stepmethods for solving nonlinear equationsrdquo International Journalof Pure and Applied Mathematics vol 82 pp 557ndash572 2013
[6] W Gautschi Numerical Analysis Birkhauser Boston MassUSA 1997
[7] S Weerakoon and T G I Fernando ldquoA variant of Newtonrsquosmethod with accelerated third-order convergencerdquo AppliedMathematics Letters vol 13 no 8 pp 87ndash93 2000
[8] M Basto V Semiao and F L Calheiros ldquoA new iterativemethod to compute nonlinear equationsrdquo Applied Mathematicsand Computation vol 173 no 1 pp 468ndash483 2006
[9] J Kou Y Li and XWang ldquoModified Halleyrsquos method free fromsecond derivativerdquo Applied Mathematics and Computation vol183 no 1 pp 704ndash708 2006
[10] M Dehghan and M Hajarian ldquoSome derivative free quadraticand cubic convergence iterative formulas for solving nonlinearequationsrdquoComputational andAppliedMathematics vol 29 no1 pp 19ndash30 2010
Now dividing (33) by (29) and using the last step of (20) wehave
119890119899+1=
(1 + 1198881)2
1198883
2(1198882
2minus 11988811198883)
1198885
1
1198906
119899+ 119874 (119890
7
119899) (34)
This proves that our first proposed method defined by (20) isa sixth-order derivative-free algorithm and satisfies the aboveerror equation This completes the proof
Nowwe discuss the efficiency index of the newmethod byusing Definition 3 as 1199011119902 where 119901 is the order of the methodand 119902 is the number of function evaluations per iterationrequired of the method It is easy to know that the numberof function evaluations per iteration required by the methoddefined in algorithm 1 is four So the efficiency index is 614 =15651
4 Numerical Results
In this section we test the effectiveness of our new methodWe have used second-order method of Steffensen (SM) [2]sixth-order method of Cordero et al (CM) [3] six-ordermethod of Soleymani and Hosseinabadi (SHM) [4] andsix-order method of Yasmin and Iftikhar (YIM) [5] forcomparison with our method to find the simple root ofnonlinear equations The test functions of 119891(119909) are listed inTable 1
Numerical computations reported here have been carriedout in a 119872119886119905ℎ119890119898119886119905119894119888119886 80 environment Table 2 shows thedifference of the root 120572 and the approximation 119909
119899to 120572
where 120572 is the exact root computed with 800 significant digits(Digits = 800) The absolute values of the function (|119891(119909
119899)|)
the difference between the approximated root 119909119899and the
exact root 120572 the number of function evaluations (TNFE)and the computational order of convergence (COC) are alsoshown in Table 2 Here COC is defined in Definition 4
5 Conclusions
We have obtained a new Steffensen-type iterative method forsolving nonlinear equations The convergence order of thismethod is six and consists of four evaluations of the functionper iteration so it has an efficiency index equal to 614 =15651 Numerical examples also show that the numericalresults of our new method in equal iterations improve theresults of other existing methods
Conflict of Interests
The author declares that there is no conflict of interestsregarding the publication of this paper
Acknowledgment
The author is grateful to the referees for their comments andsuggestions that helped to improve the paper
References
[1] W Bi H Ren and Q Wu ldquoNew family of seventh-ordermethods for nonlinear equationsrdquo Applied Mathematics andComputation vol 203 no 1 pp 408ndash412 2008
[2] D Kincaid and W Cheney Numerical Analysis BrooksColePacific Grove Calif USA 2nd edition 1996
[3] A Cordero J L Hueso E Martınez and J R TorregrosaldquoSteffensen type methods for solving nonlinear equationsrdquoJournal of Computational andAppliedMathematics vol 236 no12 pp 3058ndash3064 2012
[4] F Soleymani and V Hosseinabadi ldquoNew third- and sixth-orderderivativefree techniques for nonlinear equationsrdquo Journal ofMathematics Research vol 3 no 2 pp 107ndash112 2011
[5] N Yasmin and S Iftikhar ldquoSome new Steffensen like three-stepmethods for solving nonlinear equationsrdquo International Journalof Pure and Applied Mathematics vol 82 pp 557ndash572 2013
[6] W Gautschi Numerical Analysis Birkhauser Boston MassUSA 1997
[7] S Weerakoon and T G I Fernando ldquoA variant of Newtonrsquosmethod with accelerated third-order convergencerdquo AppliedMathematics Letters vol 13 no 8 pp 87ndash93 2000
[8] M Basto V Semiao and F L Calheiros ldquoA new iterativemethod to compute nonlinear equationsrdquo Applied Mathematicsand Computation vol 173 no 1 pp 468ndash483 2006
[9] J Kou Y Li and XWang ldquoModified Halleyrsquos method free fromsecond derivativerdquo Applied Mathematics and Computation vol183 no 1 pp 704ndash708 2006
[10] M Dehghan and M Hajarian ldquoSome derivative free quadraticand cubic convergence iterative formulas for solving nonlinearequationsrdquoComputational andAppliedMathematics vol 29 no1 pp 19ndash30 2010