See discussions, stats, and author profiles for this publication at: [Link]

net/publication/332632633

New Two-Step Method with Fifth-Order Convergence for Solving Nonlinear

Equations

Article · November 2014

CITATIONS READS

19 111

2 authors, including:

Oghovese Ogbereyivwe
Delta State University of Science and Technology, Ozoro
25 PUBLICATIONS 66 CITATIONS

SEE PROFILE

All content following this page was uploaded by Oghovese Ogbereyivwe on 31 December 2020.

The user has requested enhancement of the downloaded file.

International Journal Of Mathematics and Statistics Invention (IJMSI)
E-ISSN: 2321 – 4767 P-ISSN: 2321 - 4759
[Link] || Volume 2 Issue 10 || November. 2014 || PP-24-27

New Two-Step Method with Fifth-Order Convergence for Solving

Nonlinear Equations
1,
Ogbereyivwe Oghovese, 2,Atoma O. Johnson
Department of Mathematics and Statistics, Delta State Polytechnic, Ozoro, Nigeria

ABSTRACT : In this paper, a new iterative method for solving nonlinear equations is presented. The method
is a modified Newton’s method. Per iteration the presented method require two evaluations of the function and
two evaluations of the first-order derivatives. The convergence order of the method is established to five and the
efficiency index is 1.4953. Numerical comparisons are made with several other existing methods to show the
performance of the presented methods.

KEYWORDS: Convergence order, Taylor’s series expansion, Nonlinear equations

I. INTRODUCTION
Solving nonlinear equations is an important issue in pure and applied mathematics. Researchers have
developed various effective methods to find a single root x* of the nonlinear equation f (x) = 0 , where 𝑓: 𝐷 ⊂
𝑅 → 𝑅 is a scalar function on an open interval D. Newton’s method is one of the best iterative methods to find
x* by using
𝑓 𝑥𝑛
𝑥𝑛+1 = 𝑥𝑛 − ′ (1)
𝑓 𝑥𝑛
that converges quadratically in some neighborhood of x * [1].
There are plenty of modified iterative methods to improve the order of convergence or to simplify the
computation in open literatures. For more details, see [1-17] and the references therein. Chun [2] developed two
one-parameter fourth-order methods, which are given by:
𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − ′ ,
𝑓 𝑥𝑛
(2)
𝑓(𝑥𝑛 )2 𝑓 𝑦𝑛
𝑥𝑛+1 = 𝑦𝑛 −
𝑓(𝑥𝑛 )2 − 2𝑓 𝑥𝑛 𝑓 𝑦𝑛 + 2𝛽𝑓(𝑦𝑛 )2 𝑓 ′ 𝑥𝑛
𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − ′ ,
𝑓 𝑥𝑛
(3)
𝑓(𝑥𝑛 )3 𝑓 𝑦𝑛
𝑥𝑛+1 = 𝑦𝑛 −
𝑓(𝑥𝑛 )2 𝑆2 𝑥𝑛 , 𝑦𝑛 + 2𝛽𝑓(𝑦𝑛 )2 𝑆𝛽 𝑥𝑛 , 𝑦𝑛 𝑓 ′ 𝑥𝑛
where 𝑆𝛽 = 𝑓 𝑥𝑛 − 𝛽𝑓 𝑦𝑛 and 𝛽𝜖𝑅 is a constant. We note that the methods defined by (2) and (3) reduce to
the Traub-Ostrowski method [3] when 𝛽 = 0 . These two-point methods are special cases of the more general
family of two-point methods
𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − ′ ,
𝑓 𝑥𝑛
(4)
𝑓 𝑦𝑛
𝑥𝑛+1 = 𝑦𝑛 − 𝐺 𝑡 ′
𝑓 𝑥𝑛
with particular cases
1
𝐺 𝑡 = (5)
1 − 2𝑡 + 2𝛽𝑡 2
and
1
𝐺 𝑡 = (6)
1 − 2𝑡 + 2𝛽𝑡 2 (1 − 𝛽𝑡)
where
𝑓 𝑦𝑛
𝑡= ′ (7)
𝑓 𝑥𝑛

Ostrowski’s method [4], given by

[Link] 24 | P a g e
 New Two-Step Method With Fifth-Order Convergence…

𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − ,
𝑓 ′ 𝑥𝑛
(8)
𝑓 𝑥𝑛 𝑓 𝑦𝑛
𝑥𝑛+1 = 𝑦𝑛 −
𝑓 𝑥𝑛 − 2𝑓 𝑦𝑛 𝑓 ′ 𝑥𝑛
is an improvement of (1). The order increases by at least two at the expense of additional function evaluation at
another point iterated by the Newton’s method. This is also a special case of (4) with particular case
1
𝐺 𝑡 = (9)
1 − 2𝑡
To improve the local order of convergence and efficiency index, many more modified methods of (8) have been
developed. These include Chun and Ham [5], Kou et al [6], Bi et al [7] and reference therein.
In this paper, we develop new Newton-type iterative method to find a single root of nonlinear equations. The
method is a special case of (4) with better computational order and efficiency.

II. THE METHOD AND THE ANALYSIS OF CONVERGENCE

In this section we present the new fifth order iterative methods. We begin with the following theorem and
definitions.

Definition 1: Let 𝛼 ∈ 𝑅, 𝑥𝑛 ∈ 𝑅, 𝑛 = 0,1,2, … Then the sequence 𝑥𝑛 is said to converge to 𝛼 if

𝑙𝑖𝑚 𝑥𝑛 − 𝛼 = 0 (10)
𝑛 →∞
In addition, there exists a constant 𝑐 ≥ 0, an integer 𝑛0 ≥ 0 and 𝑝 ≥ 0 such that for all 𝑛 ≥ 𝑛0
𝑥𝑛 +1 − 𝛼 ≤ 𝑥𝑛 − 𝛼 𝑝 (11)
then 𝑥𝑛 is said to converge to 𝛼 of order 𝑝.

Definition 2: The computational efficiency of an iterative method of order 𝑝, requiring 𝑘 function

evaluations per iteration [8], is calculated by
𝑘
𝑝. (12)
Theorem 1. Let 𝜑1 𝑥 , 𝜑2 𝑥 , … , 𝜑𝑠 (𝑥) be iterative functions with the orders 𝑟1 , 𝑟2 , … , 𝑟𝑠 , respectively. Then the
composition of iterative functions
𝜑 𝑥 = 𝜑1 𝜑2 … … (𝜑𝑠 (𝑥) … ) (13)
defines the iterative method of the order 𝑟1 , 𝑟2 , … , 𝑟𝑠 [9].

We now present the method. Consider the following iteration scheme

𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − ′ ,
𝑓 𝑥𝑛
(14)
𝑓 𝑦𝑛
𝑥𝑛+1 = 𝑦𝑛 − ′
𝑓 𝑦𝑛
Equation (14) is a composite Newton method and by Theorem 1, has fourth-order convergence. To achieve
iteration with (14), it requires four evaluations. Hence by definition 2, the computational efficiency of (14) is
4 2
4 which does not increases the computational efficiency of the Newton’s method that is 2. Our aim is to
improve the order and computational efficiency of (14) by presenting a fifth-order method of special case (4).
We achieve this by introducing a weight function expressed as

𝐻 𝑡 = (1 − 𝑡 2 )−1 , (15)
where 𝑡 is as defined in (7).
Then the improved iteration scheme (14) becomes
𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − ′ ,
𝑓 𝑥𝑛
(16)
𝑓 𝑦𝑛
𝑥𝑛+1 = 𝑦𝑛 − 𝐻(𝑡) ′
𝑓 𝑦𝑛
Equation (16) is based on composition of two steps, the Newton’s method step which is a predictor-type and
weighted Newton step which is a corrector-type. As a consequence, the order of convergence is improved from
four for double Newton method to five for the new method. In order to establish the fifth-order convergence of
the proposed method (16), we state the following theorem.

Theorem 2: Let 𝛼 be a simple zero of sufficiently differentiable function 𝑓: 𝑅 → 𝑅 for an open interval 𝐼. If 𝑥0
is sufficiently close to , then the two-step method defined by (16) has convergence at least of order five.

[Link] 25 | P a g e
 New Two-Step Method With Fifth-Order Convergence…

Proof
Let 𝑒𝑛 = 𝑥𝑛 − 𝛼 be the error in the iterate 𝑥𝑛 . Using Taylor’s series expansion, we get
𝑓 𝑥𝑛 = 𝑓 ′ 𝛼 𝑒𝑛 + 𝑐2 𝑒𝑛2 + 𝑐3 𝑒𝑛3 + 𝑐4 𝑒𝑛4 + 𝑐5 𝑒𝑛5 + 𝑐6 𝑒𝑛6 + 𝑂 𝑒𝑛 7 (17)
and

𝑓 ′ 𝑥𝑛 = 𝑓 ′ 𝛼 1 + 2𝑐2 𝑒𝑛 + 3𝑐3 𝑒𝑛2 + 4𝑐4 𝑒𝑛3 + 5𝑐5 𝑒𝑛4 + 6𝑐6 𝑒𝑛5 + 7𝑐7 𝑒𝑛6 + 𝑂 𝑒𝑛 8 (18)

1 𝑓𝑘 𝛼
where 𝑐𝑘 = for 𝑘 ∈ 𝑁..
𝑘 ! 𝑓′ 𝛼

Now
𝑓 𝑥𝑛
𝑦𝑛 = 𝑥𝑛 − = 𝛼 + 𝑐2 𝑒𝑛2 + 2 𝑐22 − 𝑐3 𝑒𝑛3 + 4𝑐23 − 7𝑐2 𝑐3 + 3𝑐4 𝑒𝑛4
𝑓 ′ 𝑥𝑛
+ −8842 + 20𝑐22 𝑐3 − 6𝑐32 − 10𝑐2 𝑐4 + 4𝑐5 𝑒𝑛5 + 16𝑐25 − 52𝑐23 𝑐3 +
28𝑐22 𝑐4 − 17𝑐3 𝑐4 + 𝑐2 33𝑐32 − 13𝑐5 + 5𝑐6 𝑒𝑛6 + 𝑂 𝑒𝑛7 (19)

𝑓 𝑦𝑛 = 𝑓 ′ 𝛼 𝑐2 𝑒𝑛2 + 2𝑐3 − 2𝑐22 𝑒𝑛3 + 5𝑐23 − 7𝑐2 𝑐3 + 3𝑐4 𝑒𝑛4

−2 6𝑐24 − 12𝑐22 𝑐3 + 3𝑐32 + 5𝑐2 𝑐4 − 2𝑐5 𝑒𝑛5 + 28𝑐25 − 73𝑐23 𝑐3 + 34𝑐22 𝑐4
−17𝑐3 𝑐4 + 𝑐2 37𝑐22 − 13𝑐5 + 5𝑐6 𝑒𝑛6 + 𝑂 𝑒𝑛7 (20)

𝑓 ′ 𝑦𝑛 = 𝑓 ′ 𝛼 1 + 2𝑐22 𝑒𝑛2 + 4𝑐2 𝑐3 − 4𝑐23 𝑒𝑛3 + 𝑐2 8𝑐23 − 11𝑐2 𝑐3 + 6𝑐4 𝑒𝑛4

−4 𝑐2 4𝑐24 − 7𝑐22 𝑐3 + 5𝑐2 𝑐4 − 2𝑐5 𝑒𝑛4 + 2 16𝑐26 − 34𝑐24 𝑐3 + 6𝑐33
+30𝑐23 𝑐4 − 13𝑐22 𝑐5 + 𝑐2 − 8𝑐3 𝑐4 + 5𝑐6 𝑐𝑛6 + 𝑂 𝑒𝑛7 (21)

𝑓 𝑦𝑛
= 𝑐2 𝑒𝑛 + 2 2𝑐3 − 3𝑐22 𝑒𝑛2 + 8𝑐23 − 10𝑐2 𝑐3 + 3𝑐4 𝑒𝑛3 + 𝑂 𝑒𝑛4 (22)
𝑓 ′ 𝑥𝑛

𝐻 𝑡 = 1 + 𝑐22 𝑒𝑛2 + 4𝑐2 𝑐3 − 𝑐22 𝑒𝑛3 + 2𝑐2 7𝑐23 − 10𝑐2 𝑐3 + 3𝑐4 + 𝑐23 𝑒𝑛4 + 𝑂 𝑐𝑛5 (23)

𝑓 𝑦𝑛
= 𝑐2 𝑒𝑛2 + 2 𝑐22 − 𝑐3 𝑒𝑛3 + 3𝑐23 − 7𝑐2 𝑐3 + 3𝑐4 𝑒𝑛4
𝑓 ′ 𝑦𝑛
− 18𝑐24 − 16𝑐22 𝑐3 + 6𝑐32 + 10𝑐2 𝑐4 − 4𝑐25 𝑒25 + 𝑂 𝑐𝑛6 (24)

𝑓 𝑦𝑛
𝐻 𝑡 = 𝑐2 𝑒𝑛2 + 2 𝑐3 − 𝑐22 𝑒𝑛3 + 4𝑐23 − 7𝑐2 𝑐3 + 3𝑐4 𝑒𝑛4
𝑓 ′ 𝑦𝑛
−2 4𝑐24 − 11𝑐22 𝑐3 + 3𝑐32 + 5𝑐2 𝑐4 − 𝑐5 𝑒25 + 𝑂 𝑒𝑛6 (25)

using (17) to (25) in (16), we get

𝑒𝑛+1 = 2𝑐22 2𝑐22 − 𝑐3 𝑒𝑛5 + 𝑂 𝑒𝑛6 (26)

which shows that the method is at least fifth order convergent method.
We now discuss the efficiency index of the method by using definition 2. The number of function evaluations
per iteration of the method is 4. Therefore from definition 2 the efficiency index of the method is 1.4953, which
1 1 1
is better than Newton’s method 22 = 1.4142, Homeier’s method 33 = 1.4422 [10], Wang’s method 65 =
1 1
1.4310 eq. (8) in [11], Siyyam’s method 55 = 1.3792 [12] and equivalent to Kou’s method 54 = 1.4953 [13]
1
and Sharma’s method 54 = 1.4953 eq. (22) in [14].

III. APPLICATIONS
Now, consider some test problems to illustrate the efficiency of the developed method namely OM and
compare it with classical Newton Method (NM), the method of Abbasbandy [17] (AM), the method of Homeier
[10] (HM), and the mehods of Chun [15] (CM2 is referred to method 10 in Chun [15] with fourth-order
convergence) and (CM3 is referred to method 11 in Chun [15] with fifth-order convergence), the method of
Noor and Noor [16] (NNM), the Siyyam’s method [12] (SM), and the method of Sharma [14] (M2 is referred to
method 22 in Sharma [14]). The comparisons are given in Table [Link] test examples used are chosen from Chun
[15]. They are given below:

[Link] 26 | P a g e
 New Two-Step Method With Fifth-Order Convergence…

𝑓1 𝑥 = 𝑠𝑖𝑛2 𝑥 − 𝑥 2 + 1
𝑓2 𝑥 = 𝑥 2 − 𝑒 𝑥 − 3𝑥 + 2
𝑓3 𝑥 = cos 𝑥 − 𝑥
𝑓4 𝑥 = 𝑥−1 3 −1 (27)
𝑓5 𝑥 = 𝑥 3 − 10
2
𝑓6 𝑥 = 𝑥𝑒 𝑥 − 𝑠𝑖𝑛2 𝑥 + 3 cos 𝑥 + 5
2 +7𝑥−30
𝑓7 𝑥 = 𝑒 𝑥 −1

All calculation was done using 15-digit floating-point arithmetic. The following stopping criteria are used for
computations.

𝑖. 𝑥𝑛+1 − 𝑥𝑛 < 𝜖 𝑖𝑖. 𝑓 𝑥𝑛+1 < 𝜖 (28)

where 𝜖 = 10−15 .
Table 1 shows the calculated root and the number of iterations necessary to reach the root up to the desired
accuracy by each method.

Table 1: Comparison of number of iterations of various iterative method

Functions 𝒙𝟎 NM AM HM CM2 CM3 NNM SM M2 OM

𝑓1 1 7 5 4 5 4 4 5 4 4
𝑓2 2 6 5 5 4 4 4 3 4 3
𝑓3 1.7 5 4 4 4 3 3 3 4 2
𝑓4 3.5 8 5 5 5 5 5 4 Failed 3
𝑓5 1.5 7 5 4 5 5 4 4 4 4
𝑓6 -2 9 6 6 6 5 5 5 Failed 4
𝑓7 3.5 13 7 8 8 7 7 7 Failed 6

IV. CONCLUSION
A fifth order two-step method is proposed to solve nonlinear equations without evaluation of second
derivative of the function and it requires three functions and one first derivative per iteration. Its efficiency
index is 1.4953 which is better than some existing methods. With the help of some test problems, comparison of
obtained results with some existing methods is also given.

Competing Interests: The authors declare that no competing interests exist.

REFERENCES
[1] J.M. Ortega, W.C. Rheinbolt. Iterative solution of nonlinear equations in several variables. Academic Press, New York, 1970.
[2] C. Chun: Some variants of King's fourth-order family of methods for nonlinear equations. Appl. Math. Comput., 190 (2007), 57-
62.
[3] J. F. Traub: Iterative Methods for the Solution of Equations. Chelsea publishing, New York, 1977.
[4] A. M. Ostrowski, Solution of Equations in Euclidean and Banach Space, third ed, Academic Press, New York, 1973.
[5] C. Chun, YoonMee Ham, Some sixth-order variants of Ostrowski root-finding methods, Applied Mathematics and Computation,
193(2007) 389-394.
[6] J. Kou, Y. Li, Xiuhua Wang, Some variants of Ostrowskis method with seventh-order convergence. Journal of Computational
and Applied Mathematics, 209 (2007) 153-159.
[7] W. Bi, H. Ren, Q. Wu, Three-step iterative methods with eight-order convergence for solving nonlinear equations, Journal of
Computational and Applied Mathematics, 225 (2009) 105-112.
[8] Gautschi W. Numerical Analysis: an Introduction, Birkhauser; 1997.
[9] M. Heydari, S. M. Hosseini, G. B. Loghmoni, On Two New Families of Iterative Methods for Solving Nonlinear Equations with
Optimal Order, Appl. Anal. Discrete Math. 5 (2011),93-109
[10] H. H.H. Homeier, On Newton-type methods with cubic convergence. J. Comput. Appl. Math. 176,425-435,(2005).
[11] X. Wang et al, Modified Jarratt method with sixth-order convergence, App. Math. Lett., (2009), doi:10.10.16/jaml.2009.06.022.
[12] H. |I. Siyyam, An Iterative Method with Fifth-Order Convergence for Nonlinear Equations, Applied Mathematical Sciences,
3(41)2041-2053, 2009.
[13] J. Kou et al, Modified Halley’s method free from second derivative, Appl. Math. Comput. 183 (2006) 704-708.
[14] R. Sharma, Some Fifth and Sixth Order Iterative Methods for Solving Nonlinear Equations, J. of Engineering Research and
Applications 4(2) 268-273, 2014.
[15] C. Chun, Iterative methods improving Newton’s method by the decomposition method, Comput. Math. Appl., 50, (2005), 1559 -
1568.
[16] K. I. Noor and K. I. Noor, Iterative methods with fourth-order convergence for nonlinear equations, Appl. Math. Comput., 189,
(2007), 221-227.
[17] S. Abbasbandy, Improving Newton-Raphson method for nonlinear equations by modified Adomian decomposition method,
Appl. Math. Comput., 145, (2003), 887-893.

[Link] 27 | P a g e

View publication stats