International Journal of Pure and Applied Mathematics

————————————————————————–
Volume 48 No. 1 2008, 11-19

SOME IMPROVED ITERATIVE METHODS FOR

SOLVING NON-LINEAR EQUATIONS
Nazir Ahmad Mir1 § , Nusrat Yasmin2 , Naila Rafiq3
1 Departmentof Mathematics
COMSATS – Institute of Information Technology
Plot No. 30, Sector H-8/1, Islamabad, 44000, PAKISTAN
e-mail: namir@comsats.edu.pk
2,3 Centre for Advanced Studies in Pure and Applied Mathematics

Bahauddin Zakariya University

Multan, PAKISTAN
2 e-mail: nusyasmin@yahoo.com
3 e-mail: rafiqnaila@yahoo.com

Abstract: In this paper, we propose a class of one-step, two-step and three-

step iterative methods for solving non-linear single variable equations using
decomposition technique. The convergence analysis of the methods is discussed.
Numerical tests show that the new methods are comparable with the well known
existing methods and in many cases give better results. Our results can be
considered as improvement and refinement of the previously known results in
the literature.

AMS Subject Classification: 47H17, 58C15

Key Words: iterative methods, three-step methods, quadrature rule, predictor-
corrector methods, nonlinear equations

1. Introduction

Let us consider a single variable non-linear equation

f (x) = 0. (1.1)
Finding zeros of a single variable nonlinear equation (1.1) efficiently, is an inter-

Received: November 26, 2007 c 2008, Academic Publications Ltd.

§ Correspondence author
12 N.A. Mir, N. Yasmin, N. Rafiq

esting and very old problem in numerical analysis and has many applications
in applied sciences. We assume that α is a simple root of (1.1) and z is a an
initial guess close to α.
In recent years, researchers have developed many iterative methods for solv-
ing equation (1.1). These methods can be classified as one-step, two-step and
three-step methods, see [1]-[13]. These methods have been proposed using Tay-
lor series, decomposition techniques, error analysis and quadrature rules, etc.
Abbasbandy [1] and Chun [3] have proposed many two-step and three-step
methods using decomposition technique of Adomian [2].
We, here however use decomposition technique of Noor, see [4], [11] to
propose and analyse one-step, two-step and three-step methods. Newton and
Householder methods [1], [3] and the methods suggested by Noor [11] are special
cases of our proposed methods. The three-step method proposed by Noor in
[11] has convergence order twoand not four as pointed out by Chun in [4].
We emphasize here on three-step method which has efficiency index equal to
(3)(1/3) = 1.44. Several numerical examples are given to test the efficiency and
performance of the new iterative method. Our results can be considered as an
improvement and refinement of the previously known results in the literature.

2. The Iterative Methods

Weerakoon and Fernando [13] and M. Frontini and E. Sormani [5], [6] have
proposed various methods by the approximation of the indefinite integral
Zx
′
f (x) = f (xn ) + f (t)dt, (2.1)
xn
using Newton Cotes formulae of order zero and one. We approximate, here
however the integral (1.2) by rectangular rule at λx + (1 − λ)zn with the end-
points x and zn . We thus have:
Zx
′ ′
f (t)dt = (x − zn )f (λx + (1 − λ)zn ) .
zn
This gives
′ ′′
0 = f (zn ) + (x − zn )(f (zn ) + λ(x − zn )2 f (zn ). (2.2)
Equation (2.2) can be written as:
x = c + N (x), (2.3)
SOME IMPROVED ITERATIVE METHODS FOR... 13

where
′
c = zn − f (zn )/f (zn ) (2.4)
and
′′
λ(x − zn )2 f (zn )
N (x) = . (2.5)
f ′ (zn )
Using the decomposition technique of Noor, see [4], [12], equation (2.3) has the
series solution:
X∞
x= xi . (2.6)
i=0
The nonlinear operator N can be decomposed as
  
∞ ∞ ∞
!
X X X
N xi = N (x0 ) + N  xj  . (2.7)
i=0 i=1 j=0

Combining (2.3), (2.6) and (2.7), we have

  
∞
X ∞
X i
X
xi = c + N (x0 ) + N  xj   . (2.8)
i=0 i=1 j=0
Thus, we have the following iterative schemes:
x0 = c,
x1 = N (x0 ) ,
x2 = N (x0 + x1 ) ,
..
.
xn+1 = N (x0 + x1 + · · · + xn ) , n = 0, 2, . . . . (2.9)
Then
x1 +x2 +· · ·+xn+1 = N (x0 )+N (x0 + x1 )+· · ·+N (x0 + x1 + · · · + xn ) , (2.10)
and
∞
X
x=c+ xi . (2.11)
i=1
Thus from (2.3), (2.4), (2.5), (2.10), (2.11),
′
x0 = c = zn − f (zn )/f (zn ), (2.12)
′′
(x0 − zn )2 f (zn )
x1 = N (x0 ) = −λ , (2.13)
f ′ (zn )
14 N.A. Mir, N. Yasmin, N. Rafiq

′′
(x0 + x1 − zn )2 f (zn )
x2 = N (x0 + x1 ) = −λ , (2.14)
f ′ (zn )
and so on.
This formulation allows to suggest many one-step, two-step and three-step
methods. Thus by a first order approximation, namely
′
x ≃ x0 = zn − f (zn )/f (zn ),
we can define the following one-step iterative scheme:
Algorithm 2.1. For a given initial guess xo , find the first order approxi-
mate solution by the iterative scheme:
′ ′
xn+1 = xn − f (xn )/f (xn ), f (xn ) 6= 0, n = 0, 1, 2, . . . (2.15)
which is a well known Newton’s method.
Now using (2.10), (2.12) and (2.13), we have the second order approximate
solution
′′
′ (x0 − zn )2 f (zn )
x ≃ c + x1 = x0 + N (x0 ) = zn − f (zn )/f (zn ) − λ . (2.16)
f ′ (zn )
Using (2.15) and (2.16), we can suggest the following two-step iterative method.
Algorithm 2.2. For a given initial guess xo , find the approximate solution
by the iterative scheme:
Predictor:
′ ′
yn = xn − f (xn )/f (xn ), f (xn ) 6= 0 .
Corrector:
′′
′ (yn − xn )2 f (xn )
xn+1 = xn − f (xn )/f (xn ) − λ , n = 0, 1, 2, . . . .
f ′ (xn )
Algorithm 2.2 can be written in the following form.
Algorithm 2.3. For a given initial guess xo , find the approximate solution
by the iterative scheme:
′ f (xn )2 ′′
xn+1 = xn − f (xn )/f (xn ) − λ ′ f (xn ), n = 0, 1, 2, . . . . (2.17)
f (xn )3
We observe that for λ = 0, we have Newton’s method and for λ = 1/2, we
have Householder’s method [1], [3]. Using Maple, we can very easily establish
that Algorithm 2.3 has third order convergence when λ = 21 .
From (2.12), (2.13) and (2.14), we have third order approximation:
x = c + x1 + x2 = x0 + N (x0 ) + N (x0 + x1 )
SOME IMPROVED ITERATIVE METHODS FOR... 15

′′ ′′
′ (x0 − zn )2 f (zn ) (x0 + x1 − zn )2 f (zn )
= zn − f (zn )/f (zn ) − λ ′ −λ . (2.18)
f (zn ) f ′ (zn )
Using the relations (2.15) , (2.16) and (2.17), we can suggest the following three-
step method.
Algorithm 2.4. For a given initial guess xo , find the approximate solution
by the iterative scheme:
′ ′
yn = xn − f (xn )/f (xn ), f (xn ) 6= 0 , (2.19)
′′
(yn − xn )2 f (xn )
zn = −λ , (2.20)
f ′ (xn )
′′ ′′
′ (yn − xn )2 f (xn ) (yn + zn − xn )2 f (xn )
xn+1 = xn − f (xn )/f (xn ) − λ − λ ,
f ′ (xn ) f ′ (xn )
n = 0, 1, 2, . . . (2.21)
(2.19)-(2.21) constitute three-step method. Now using Maple 7.0, we prove that
Algorithm 2.4 has third order of convergence for λ = 1/4.

3. Convergence Analysis

Theorem. Let α ∈ I be a simple zero of sufficiently differentiable function

f : I ⊆ R → R for an open interval I. If x0 is sufficiently close to α, then
Algorithm 2.4 has cubic convergence for λ = 1/4.
Proof. Let α be a simple zero of f and xn = α + en . By Taylor’s expansion,
we have:
′
f (xn ) = f (α) en + c2 e2n + c3 e3n + c4 e4n + c5 e5n + c6 e6n + O e7n



(3.1)
′ ′
f (xn ) = f (α) 1 + 2c2 en + 3c3 e2n + 4c4 e3n + 5c5 e4n + 6c65 e5n + O(e6n ),


′′ ′
f (xn ) = f (α) 2c2 + 6c3 en + 12c4 e2n + 20c5 e3n + 30c6 e4n + O(e5n ),


(3.2)
where
1 f (k) (α)


ck = , k = 2, 3, . . . and en = xn − α . (3.3)
k! f ′ (α)
Using (3.1) and (3.2), we have
f (xn )
= en − c2 e2n + 2 c22 − c3 e3n + 7c2 c3 − 3c4 − 4c32 e4n + O(e5n ) . (3.4)



′
f (xn )
Using (3.4) in (2.3), we thus have
yn = α + c2 e2n + −2c22 + 2c3 e3n − 7c2 c3 − 4c32 − 3c4 e4n + O e5n .




16 N.A. Mir, N. Yasmin, N. Rafiq

Using (3.1), (3.2), (3.4) and (3.5) in (2.19), we have

zn = α + (−2λc2 + c2 )e2n + (8λc22 − 2c22 + 2c3 − 6λc3)e3n + O e4n .


(3.6)
Using (3.1), (3.2), (3.5) and (3.6), we have
′′
(yn + zn − xn )2 f (xn )
= (−4λ2 c2 + 2λc2 )e3n − λe2n + O e4n .


λ ′ (3.7)
f (xn )
Using (3.5), (3.6) and (3.7) in (2.20), we have
xn+1 := α+(−4λc2 +c2 )e2n +(−2c22 +2c3 +16λc22 −12λc3 −8λ2 c22 )e3n +O(e4n ) .
Thus, for λ = 1/4, we have:
 
3 2 3
xn+1 = α + −c3 + c2 en + O(e4n ),
2
 
3
or en+1 = −c3 + c22 e3n + O(e4n ). (3.12)
2
Thus, we observe that Algorithm 2.4 has third order convergence.

4. Numerical Examples

We consider here some numerical examples to demonstrate the performance of

the new developed three-step iterative method, namely Algorithm 2.4. We com-
pare the classical Newton’s method (CN ), the method of Abbasbandy (AM [1]),
the method of Homier (HM )[7], the method of Chun (CM 2) [3] and the new
developed three-step Algorithm (M N ), in this paper. All the computations are
performed using Maple 7.0. We take ε = 10−15 as tolerance. The following
criteria is used for estimating the zero:
(i) δ = |xn+1 − xn | < ε ,
(ii) |f (xn )| < ε .
The following examples as in Chun [3] are used for numerical testing in
Table 4.1.
Examples.
f1 = sin2 (x) − x2 + 1, f2 = x2 − ex − 3x + 2,
f3 = cos(x) − x, f4 = (x − 1)3 − 1,
2 2 +7x−30
f5 = x3 − 10, f6 = xex − sin2 (x) + 3 cos(x) + 5, f7 = ex − 1.
For convergence criteria, it was required that δ, the distance between two
consecutive iterates was less than 10−15 , n represents the number of iterations
and f (xn ), the value of the function. The numerical comparison is given in
SOME IMPROVED ITERATIVE METHODS FOR... 17

Table 4.1.
Ex n f (xn ) δ
f1 , x0 = −1, w =-1.4044916482153412260350868178
NM 7 -1.04e50 7.33e-26
AM 5 -5.81e-55 1.39e-18
HM 4 -5.4e-62 7.92e-21
CM 2 5 -2.0e-63 1.31e-17
MN 5 5.19e-73 1.3e-23
f2 , x0 = 2, w = .25753028543986076045536730494
NM 6 2.93e-55 9.1e-28
AM 5 1.0e-63 1.45e-26
HM 5 0 9.33e-43
CM 2 4 1.0e-63 9.46e-29
MN 5 0 3.4e-41
f3 , x0 = 1.7 w = .73908513321516064165531208767
NM 5 -2.03e-32 2.34-16
AM 4 -7.14e-47 8.6e-16
HM 4 -5.02e-59 9.64e-20
CM 2 4 0 1.86e-53
MN 5 -3.1e-127 2.4e-41
f4 , x0 = 3.5, w = 2.0000000000000000000000000000
NM 8 2.06e-42 8.28e-22
AM 5 0 4.3e-22
HM 5 0 1.46e-24
CM 2 5 0 2.74e-24
MN 6 3.4e-109 9.9e-36
f5 , x0 = 1.5, w = 2.1544346900318837217592935665
NM 7 2.06e-54 5.64e-28
AM 5 -5.0e-63 1.18e-25
HM 4 -5.0e-63 9.8e-23
CM 2 5 -5.0e-63 1.57e-22
MN 5 -3.7e-89 4.7e-29
Ex n f (xn ) δ
f6 , x0 = −2, w = −1.2076478271309189270094167584
NM 9 -2.27e-40 2.73e-21
AM 6 -4.0e-63 4.35-45
HM 6 -4.0e-63 2.57e-32
CM 2 6 -4.0e-63 2.15e-36
MN 6 -1.1e-69 7.4e-23
f7 , x0 = 3.5 α = 3.0000000000000000000000000000
NM 13 1.52e-47 4.2e-25
AM 7 -4.33e-48 2.25e-17
HM 8 2.0e-62 2.4e-33
CM 2 8 2.0e-62 2.21e-23
MN 8 -3.7e-89 4.7e-29

Table 4.1:
18 N.A. Mir, N. Yasmin, N. Rafiq

5. Conclusion

We have suggested here a class of one-step, two-step and three-step iterative

methods for nonlinear equations using quadrature rule, Taylor series and de-
composition technique. These methods involve terms only upto first derivative
and do not require the computation of higher derivatives as compared to other
methods of the same order, such as Halley’s method. Further, we note from
Table 4.1 that our three-step iterative method, namely (N M N ) is comparable
with all the methods cited in the table and in many cases gives better results
in terms of δ and the function evaluation f (xn ). Using these techniques, many
higher order methods can be proposed.

References

[1] S. Abbasbandy, Improving Newton-Raphson method for nonlinear equa-

tions by modified Adomian decomposition method, Appl. Math. Comput.,
145 (2003), 887-893.

[2] G. Adomian, Nonlinear Stochastic Systems and Applications to Physics,

Kluwer Academic Publishers, Dordrecht (1989).

[3] C. Chun, Iterative methods improving Newton’s method by the decompo-

sition method, Comput. and Math. with Appl., 50 (2005), 1559-1568.

[4] C. Chun, A note on some iterative schemes for nonlinear equations, Appl.
Math. Comput., In Press.

[5] J.E. Dennis, R.B. Schnable, Numerical Methods of Unconstraind Optimiza-

tion and Non-Linear Equations, Prentice Hall (1983).

[6] M. Frontini, E. Sormani, Some variants of Newton’s method with third

order convergence and multiple roots, J. Comput. Appl. Math., 156 (2003),
345-354.

[7] M. Frontini, E. Sormani, Third order methods for quadrature formulae for
solving system of nonlinear equations, Appl. Math. Comput., 149 (2004),
771-782.

[8] H.H. Homier, On Newton type methods with cubic convergence, J. Com-
put. Appl. Math., 176 (2005), 1559-1568.
SOME IMPROVED ITERATIVE METHODS FOR... 19

[9] Jisheng Kou, Yitian Li, Xiuhua Wang, Third order modification of New-
ton’s method, Appl. Math. and Comput., In Press.

[10] M.V. Kanwar, V.K. Kukreja, S. Singh, On a class of quadratically conver-

gent iteration formulae, Appl. Math. Comput., 166, No. 3 (2005), 633-637.

[11] Mamta, V. Kanwar, V.K. Kukreja, S. Singh, On some third order itera-
tive methods for solving non-linear equations, Appl. Math. Comput., 171
(2005), 272-280.

[12] M. Aslam Noor, K. Inayat Noor. Three-step iterative methods for nonlinear
equations, Appl. Math. Comput., In Press.

[13] S. Weerakoon, T.G.I. Fernando, A variant of Newton’s method with accel-

erated third-order convergence, Appl. Math. Lett., 13 (2000), 87-93.
20