See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/299657382

Ninth Order Method for Nonlinear Equations and its Dynamic Behaviour

Article · April 2016

DOI: 10.17114/j.aua.2016.45.06

CITATIONS READS
5 303

3 authors:

Muhammad Rafiullah Diyashvir Kreetee Rajiv Babajee

COMSATS University Islamabad Independent Scholar
8 PUBLICATIONS 55 CITATIONS 51 PUBLICATIONS 653 CITATIONS

SEE PROFILE SEE PROFILE

D. Jabeen
Sir Syed University of Engineering and Technology
6 PUBLICATIONS 20 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Digital Signal Processing with applications to Digital Communications View project

Development and Analysis of Higher Order Multipoint variants of Newton's method for solving nonlinear equations View project

All content following this page was uploaded by Muhammad Rafiullah on 05 April 2016.

The user has requested enhancement of the downloaded file.

Acta Universitatis Apulensis No. 45/2016
ISSN: 1582-5329 pp. 73-86
http://www.uab.ro/auajournal/ doi: 10.17114/j.aua.2016.45.06

NINTH ORDER METHOD FOR NONLINEAR EQUATIONS AND

ITS DYNAMIC BEHAVIOUR

M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen

Abstract. The aim of this paper is to construct a new efficient iterative method
to solve nonlinear equations and discuss the dynamic behaviour of it. This method is
based on ”A fifth-order iterative method for solving nonlinear equations, Numerical
Analysis and Applications, 4 (3) (2011), pp. 239–243”. The finite difference and
Hermite interpolation are used to improve the convergence order and efficiency index
of this method. The new method is of the ninth order of convergence and it is
compared with other ninth order methods. Some numerical test problems are given
to show the accuracy and fast convergence of the method proposed. The dynamic
behaviour of the methods for finding the roots of unity are also studied.

2010 Mathematics Subject Classification: 65H05, 65D05, 41A25.

Keywords: Nonlinear equations, Iterative methods, Convergence analysis, Her-

mite interpolation, Dynamic behaviour, Polynomiograph

1. Introduction
Solving nonlinear equations f (x) = 0 is one of the most important and challenging
problem in scientific and engineering applications. These equations largely occur
in daily life with useful applications such as, to measure the speed of rocket, to find
the eigen values of a system, to measure compressibility of gasses, to discuss aging
model of a cell’s energy producing organelle (mitochondria), to calculate the simple
harmonic oscillation, to measure the variation of the local heat transfer, to measure
the interior temperature of a material, to the assessment of drug concentration
in plasma, to produce the methanol from CO and H2 by using the equilibrium
equation, to measure the velocity of a falling parachutist, etc. Finding the solution
of nonlinear equations is not an easy task. Mostly, the analytical methods fail to
find their solutions. Ultimately, for this purpose we move towards the numerical
methods. There are numerous and well known methods which help us to deal

73
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

with nonlinear equations such as Bisection, Regula False, Newton-Raphson, Secant,

Steffensen, Halley, Jarratt, Ostrowski, King’s methods, etc.
Now, consider a nonlinear equation

f (x) = 0. (1.1)

The Newton-Raphson method is largely used to solve such nonlinear equations

(1.1) and it is written as

f (xn )
xn+1 = xn − , f 0 (xn ) 6= 0. (1.2)
f 0 (xn )

This is an important and basic method [8], which converges quadratically.

Recently, a large variety of methods which based on the Newton’s method are
proposed to solve nonlinear equations. All these modified methods are in the direc-
tion of improving the efficiency index and order of convergence by using lower-order
derivatives as possible.

2. Preliminaries
Definition 1. Let x0 , x1 , x2 , ... be a sequence which converges to α. Let en = xn −α.
If there exist a real number p and a positive constant C such that |e|en+1n|
|
p → C, as n

becomes large, i.e.

|en+1 |
lim p =C
n→∞ |en |

then p is called the order of convergence and C is called asymptotic constant [6].
∗
Definition 2. According to Ostrowski [6, 8] the efficiency index ef f is defined as
∗ 1
ef f = p d ,

where d is called the number of function evaluations used in method.

−
Definition 3. The computational order of convergence (COC) p of a sequence
{xn }n≥0 is defined by
en+1
−
ln en
pn = ,
en
ln en−1

where xn−1 , xn and xn+1 are three consecutive iterations near the root α and en =
xn − α [10].

74
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Definition 4. Suppose that a set of points x0 , x1 , x2 , ....xn and respective function

values on these points are y0 , y1 , y2 , ....yn then the first order divided difference of
−y0
first two points can be define as f [x0 , x1 ] = xy11 −x 0
and second order divided difference
f [x1 ,x2 ]−f [x0 ,x1 ]
can be defined as define as f [x0 , x1 , x2 ] = x2 −x0 . the general formula is

f [x1 , x2 , · · · , xk ] − f [x0 , x1 , · · · , xk−1 ]

f [x0 , x1 , · · · , xk−1 , xk ] =
xk − x0

where k = 2, 3, · · · , n [5].

Theorem 1. The order of points does not matter such as f [x0 , x1 ] = f [x1 , x0 ] or
f [x0 , x1 , x2 ] = f [x2 , x1 , x0 ].

Theorem 2. For any two points of x0 , x1 , x2 , . · · · , xn happen to be equal, then the

divided difference can be define as follows [10]:

f (x1 ) − f (x0 )
f [x0 , x0 ] = lim = f 0 (x0 ).
x1 →x0 x1 − x0
Definition 5. The Newton’s polynomial for the data points (x0 , y0 ), (x1 , y1 ),· · · ,
(xn , yn ) using the divided differences definition can be written as

Pn (x) = f [x0 ] + f [x0 , x1 ](x − x0 ) + f [x0 , x1 , x2 ](x − x0 )(x − x1 ) +

· · · + f [x0 , x1 , · · · , xn ](x − x0 )(x − x1 ) · · · (x − xn−1 )
n
P i−1
Q
Pn (x) = f [x0 ] + f [x0 , · · · , xi ] (x − xj ).
i=1 j=0

Definition 6. Hermite interpolation is a method to find out a polynomial of degree

2n + 1 which satisfy the function values at n known data points and first m deriva-
tives. Hermite Interpolation is likely Newton’s divided difference formula but these
both differs by the polynomial degree. Lets consider n data points with m derivatives

(x0 , f (x0 )) (x1 , f (x1 )) ··· (xn−1 , f (xn−1 ))

(x0 , f 0 (x0 )) (x1 , f 0 (x1 )) ··· (xn−1 , f 0 (xn−1 ))
.. .. .. ..
. . . .
(x0 , f (m) (x0 )) (x1 , f (m) (x1 )) · · · (xn−1 , f (m) (xn−1 ))

The Hermite interpolation polynomial can be determine by using a modification of

Newton’s divided difference method.

75
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Modification of Newton’s divided differences for Hermite Interpola-

tion
Suppose that the distinct numbers x0 , · · · , xn are given together with the val-
ues of f and f 0 at these numbers. Define a new sequence z0 , z1 , · · · , z2n+1 by
z2i = z2i+1 = xi for each i = 0, 1, · · · , n, and contract divided difference table.
We can not define f [z2i , z2i+1 ] by the divided difference formula, so the reasonable
substitution in this situation is f [z2i , z2i+1 ] = f 0 (x2i ) = f 0 (xi ), we can use the en-
tries f 0 (x0 ), f 0 (x1 ), · · · , f 0 (xn ) in the place of the undefined first divided differences
f [z0 , z1 ], f [z2 , z3 ], · · · , f [z2n , z2n+1 ]. The remaining divided differences are pro-
duced as usual, and the appropriate divided differences are employed in Newton’s
interpolatory divided-difference formula [2, p. 137]. The Hermite polynomial is
given by
2n+1
X
H2n+1 (x) = f [z0 ] + f [z0 , z1 , · · · , zk ](x − z0 )(x − z1 ) · · · (x − zk−1 ).
k=0

The divided difference table constructed as for Hermite polynomial

z f (z) 1st DD 2nd DD

z0 = x0 f [z0 ] = f (x0 )
f [z0 , z1 ] = f 0 (x0 )
f [z1 ,z2 ]−f [z0 ,z1 ]
z1 = x0 f [z1 ] = f (x0 ) f [z0 , z1 , z2 ] = z2 −z0
f [z2 ]−f [z1 ]
f [z1 , z2 ] = z2 −z1
f [z2 ,z3 ]−f [z1 ,z2 ]
z2 = x1 f [z2 ] = f (x1 ) f [z1 , z2 , z3 ] = z3 −z1
f [z2 , z3 ] = f 0 (x1 )
f [z3 ,z4 ]−f [z2 ,z3 ]
z3 = x1 f [z3 ] = f (x1 ) [z2 , z3 , z4 ] = z4 −z2
f [z4 ]−f [z3 ]
f [z3 , z4 ] = z4 −z3
f [z4 ,z5 ]−f [z3 ,z4 ]
z4 = x2 f [z4 ] = f (x2 ) [z3 , z4 , z5 ] = z5 −z3
f [z4 , z5 ] = f 0 (x2 )
z5 = x2 f [z5 ] = f (x2 )

This Table shows the entries that are used for the first three divided-difference
columns when determining the Hermite polynomial H5 (x) for x0 , x1 and x2 ”. The
Hermite polynomial H2n+1 (x) can also be written as

H2n+1 (x) = f [z0 ] + f [z0 , z1 ](x − x0 ) + f [z0 , z1 , z2 ](x − x0 )2

+f [z0 , z1 , z2 , z3 ](x − x0 )2 (x − x1 ) +
... + f [z0 , z1 , ..., z2n+1 ](x − x0 )2 (x − x1 )2 ...(x − xn−1 ).

76
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

3. Derivation of new methods

Consider the following iterative method [7]

f (xn )
yn = xn − ,
f 0 (xn )
1 f (yn ) 2 f 00 (yn )
 
f (yn )
xn+1 = yn − 0 − . (3.1)
f (yn ) 2 f 0 (xn ) f 0 (yn )

This method is fifth-order and it involves five functions f (xn ), f 0 (xn ), f (yn ), f 0 (yn ),
1 1
f 00 (yn ) at each step, so the efficiency index of this method is p d = 5 5 = 1. 379 7.
00 (y )
First of all we eliminate the term ff 0 (y n
n)
from (3.1) to improve the efficiency index of
the method (3.1) by approximating f (yn ) and f 0 (yn ) using divided differences with
00
0 0 (x )
known functions and their derivatives as f (yynn)−f −xn
n
and f (yynn)−f
−xn
(xn )
respectively,
so we have
f (xn )
yn = x n −
f 0 (xn )
f (yn ) f (yn )2 (f 0 (xn ) − f 0 (yn ))
xn+1 = yn − 0 − . (3.2)
f (yn ) 2(f (xn ) − f (yn ))f 0 (xn )2

This method involves only four functions evaluation such as f (xn ), f 0 (xn ), f (yn ),
f 0 (yn ). The order of this method remains same as fifth with error term 21 (10c42 −
1 1
3c22 c3 )e5n . But efficiency index of this method is improved as p d = 5 4 = 1. 495 3.
Now at the third step, introducing Newton-Raphson method gives

f (xn )
yn = x n −
f 0 (xn )
f (yn ) f (yn )2 (f 0 (xn ) − f 0 (yn ))
z n = yn − 0 −
f (yn ) 2(f (xn ) − f (yn ))f 0 (xn )2
f (zn )
xn+1 = zn − 0 . (3.3)
f (zn )

By suggesting this new step the function evaluations also increases as six. We reduce
the number of functions by approximating the f 0 (zn ) using the Hermite interpolation
polynomial with known data points (xn , f (xn )), (yn , f (yn )), (zn , f (zn )), (xn , f 0 (xn )),

77
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

(yn , f 0 (yn )). So we have,

H4 (x) = f [z0 ] + f [z0 , z1 ](x − xn ) + f [z0 , z1 , z2 ](x − xn )2
+f [z0 , z1 , z2 , z3 ](x − xn )2 (x − yn ) + f [z0 , z1 , z2 , z3 , z4 ](x − xn )2 (x − yn )2
H40 (x) = f [z0 , z1 ] + 2f [z0 , z1 , z2 ](x − xn )
+f [z0 , z1 , z2 , z3 ] (x − xn )2 + (x − xn )(x − yn )

+f [z0 , z1 , z2 , z3 , z4 ] (x − xn )2 (x − yn ) + (x − xn )(x − yn )2 .

we approximate df z = f 0 (z n ) = H40 (zn ) as

f (zn )(xn − yn )3 + f 0 (yn )(xn − yn )(xn − zn
)

   
  xn + 2yn + xn yn − (3 + xn + yn )zn + zn2  
   + (yn − zn ) 
 
 −f (yn )(xn − zn )  

1 
 (−6y
 n + 6zn  + (xn − zn )(xn − 3yn + 2zn )) 
.
yn2 − 3xn (2 + yn − zn ) + yn zn
 
(xn − yn )3  
 f (xn ) 
−2(−3 + zn )zn
   
    
 2xn + yn + xn yn 
+f 0 (xn )(xn − yn )
 
−(3 + xn + yn )zn + zn2

So the new method is

f (xn )
yn = xn −
f 0 (xn )
f (yn ) f (yn )2 (f 0 (xn ) − f 0 (yn ))
z n = yn − 0 −
f (yn ) 2(f (xn ) − f (yn ))f 0 (xn )2
f (zn )
xn+1 = zn − . (3.4)
df z
This method involves five function evaluations and the order of convergence is im-
proved up to the ninth. The error term is (10c52 c4 − 3c32 c3 c4 )e9n . The efficiency
1 1
index is p d = 9 5 = 1. 551 8 which is better than (3.1) and (3.2).

4. Convergence Analysis
Theorem 3. Let α be a simple zero of sufficiently differentiable function f: I ⊆
R → R for an open interval I. If x0 closed to α, then the method defined by (3.2)
is of fifth order and moreover satisfies the following error equation
1
en+1 = (10c42 − 3c22 c3 )e5n .
2
f (k) (α)
where en = xn − α, en+1 = xn+1 − α and ck = k!f 0 (α) .

78
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Proof. Using Taylor series we have

en + c2 en 2 + c3 en 3 + c4 en 4
 
0
f (xn ) = f (α) (4.1)
+ c5 en 5 + c6 en 6 + c7 en 7 + c5 en 8 + · · ·
1 + 2c2 en + 3c3 e2n + 4c4 e3n
 
0 0
f (xn ) = f (α) (4.2)
+5c5 e4n + 6c6 en 5 + 7c7 en 6 + 8c8 en 7 + · · ·
f (xn )
= α + c2 e2n + −2c22 + 2c3 e3n + 4c32 − 7c2 c3 + 3c4 e4n



yn = xn − 0
f (xn )
− 2 4c2 − 10c22 c3 + 3c23 + 5c2 c4 − 2c5 e5n + · · ·
4


(4.3)

Expanding f (yn ) around root and using (4.3), we have

f (yn ) = f 0 (α)[c2 e2n + −2c22 + 2c3 e3n + 5c32 − 7c2 c3 + 3c4 e4n

− 2 6c42 − 12c22 c3 + 3c23 + 5c2 c4 − 2c5 e5n + ...]

(4.4)
0 0 2 2 3

3 3

4
f (yn ) = f (α)[1 + 2c2 en + −4c2 + 4c2 c3 en + c2 8c2 − 11c2 c3 + 6c4 en
− 4 c2 4c42 − 7c22 c3 + 5c2 c4 − 2c5 e5n + · · · ]



(4.5)

Using (4.1) to (4.5) in (3.2), we have

1
10c42 − 3c22 c3 e5n + −40c52 + 41c32 c3 − 6c2 c23 − 2c22 c4 e6n + · · · (4.6)



xn+1 = α +
2
so
1
10c42 − 3c22 c3 e5n + O(e6n ).


en+1 =
2

Theorem 4. Let α be a simple zero of sufficiently differentiable function f: I ⊆

R → R for an open interval I. If x0 closed to α, then the method defined by (3.4)
is of ninth order and moreover satisfies the following error equation

en+1 = (10c52 c4 − 3c32 c3 c4 )e9n .

f (k) (α)
where en = xn − α, en+1 = xn+1 − α and ck = k!f 0 (α) .

Proof. from (3.4) and (4.6)

1
10c42 − 3c22 c3 e5n + −40c52 + 41c32 c3 − 6c2 c23 − 2c22 c4 e6n + · · ·



zn = α + (4.7)
2

79
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Expanding f (zn ) around root and using (4.7), we have

5c42 − 32 c22 c3 e5n



4 − 41c2 c + 6c2 + 2c c e6



 −c2 40c 2 2 3 3 2 4 n


6 − 683c4 c − 12c3 + 110c3 c

 398c 3 4

 + 12 2 2 3
2 249c2 − 5c

2 e7n 

 −34c 2 c3 c4 + c 2 3 5


−1584c72 + 3777c52 c3 − 926c42 c4
− 52c23 c4
   
 
 +c32 −2319c23 + 142c5

+ 12 
  8
  en 
 3 2
+4c2 83c3 − 6c4 − 11c3 c5

0
 
f (z n ) = f (α)   (4.8)
 
2
+c2 (670c3 c4 − 6c6 )
 
5530c82 − 16789c62 c3 + 5184c 5c
   
 2 4 
+c42 14759c23 − 1202c5



   
3 2
 
+c3 164c3 − 75c4 − 68c3 c5
 
 1  9
 en + · · · 
 +2 

3

  +c 2 (−6242c 3 c4 + 174c 6 ) 
 
2


  +2c2 671c3 c4 − 31c4 c5 − 27c3 c6
 
+c22 −3919c33 + 446c24 + 866c3 c5 − 7c7

Using (4.1) to (4.5) and (4.7) to (4.8) in (3.4), we have

 
5 3

9 9 61125 7
xn+1 = α + 10c2 c4 − 3c2 c3 c4 en + 8382c2 − c2 c3 + ... e10
n (4.9)
2
so
en+1 = 10c52 c4 − 3c32 c3 c4 e9n + O(e10


n ).

5. Applications
We have compared our new method (3.4) with the following methods having order
nine respectively.
Gradimir et al. in 2007 [12] suggested a ninth order method (GM) which is
given below,

f (xn )
yn = x n −
f 0 (xn )
(xn − yn )f (yn )
zn = yn −
f (xn ) − 2f (yn )
f (zn )f 0 (zn )
xn+1 = zn − 0 0 (x )
[f 0 (zn )]2 − 12 f (zn ) f (zznn)−f
−xn
n

80
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Hu et al. in 2011 [4] proposed a ninth-order method (HU)

f (xn )
yn = xn −
f 0 (xn )
"  #
f (yn ) 2 f (yn )

zn = yn − 1 +
f (xn ) f 0 (yn )
" #
f (yn ) 2
 
f (zn ) f (zn )
xn+1 = zn − 1 + 2 +2
f (xn ) f (yn ) f 0 (yn )

In 2012, Hafiz and Salwa [11] proposed ninth order method (HS) using Halley it-
erative method and the weight combination of mid-point with Simpson’s quadrature
formulas and using predictor–corrector technique, as
f (xn )
yn = xn −
f 0 (xn )
xn + yn
wn =
2
12f (xn )
zn = x n −
f 0 (x 0 0
n ) + 10f (wn ) + f (yn )
2f (zn )f 0 (zn )
xn+1 = zn − .
2 [f 0 (zn )]2 − f (zn )f 00 (zn )
Now, consider some test problems to illustrate the efficiency of the proposed
method, namely R-1 (3.4) which is of ninth order. First we show first three iterations
of given examples using method R-1 (3.4), accuracy |x3 − x2 | and computational
order of convergence (COC) of R-1 in table-1. Secondly in table-2, we have compared
the results (|x3 − x2 |) of new method R-1 (3.4) with the results of existing methods
namely GM, HU and HS which are also ninth order.
Examples:

Function x0 α (exact root)

f1 (x) = 4x5 − 3x4 + 2x3 − 3 2 1
x 2
f2 (x) = e sin x + log(1 + x ) 1.5 0
f3 (x) = 3 tan x − x 1 0
Table-1:

f x1 x2 x3 |x3 − x2 | COC
f 1 (x) 1.1717e + 000 1.0000e + 000 1.0000e + 000 2.1236e − 005 8.92375
f 2 (x) 5.0159e − 003 −1.8309e − 019 3.6059e − 167 1.8309e − 019 8.98578
f 3 (x) 3.9926e + 000 4.0781e + 000 4.0781e + 000 3.0268e − 008 8.95792

81
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Table-2:
f GM HU HS R−1
f 1 (x) 1.8235e − 003 4.1047e − 005 6.0517e − 009 2.1236e − 005
f 2 (x) 1.5768e − 013 6.8481e − 009 2.4763e − 012 1.8309e − 019
f 3 (x) 1.6744e − 006 4.5605e − 004 1.6837e − 004 3.0268e − 008

Bahman Kalantari [14] coined the term ”polynomiography” to be the art and sci-
ence of visualization in the approximation of roots of polynomial using Iteration
Functions. We describe the method to produce the polynomiographs for finding the
roots of unity using MATLAB in the following section.

6. Polynomiographs of the ninth order methods for finding the roots

of unity
Let x0 ∈ C be the initial point. A square grid of 65536 points, composed of 256
columns and 256 rows corresponding to the pixels of a computer display would
represent a region of the complex plane [15]. We consider the square R × R =
[−2, 2] × [−2, 2]. Each grid point is used as a starting value x0 of the sequence
generated by the ninth order methods and the number of iterations until convergence
is counted for each gridpoint. We consider the polynomial f (x) = xr − 1, x ∈ C for
finding the rth roots of unity. The rth roots of unity are given by
   
2π(j − 1) 2π(j − 1)
αj = cos + i sin , j = 1, 2...r.
r r

The basin of attraction corresponding to a zero αj of the polynomial f (x) is the set of
all starting points x0 which are attracted to αj . If the sequence generated by iterative
method attempt a zero αj of the polynomial with a tolerance |xk − αj | < 1e − 4
and a maximum of 100 iterations, we decide that x0 is in the basin of attraction
of these zero and assign a color to that zero. In this way, the basin of attraction
for each root would be assigned a characteristic colour. The common boundaries of
these basins of attraction constitute the Julia set of the Iteration Function. If the
iterates do not satisfy the above criterion for convergence we assign the dark blue
colour (The iterates either diverge or converge to additional fixed points ). Let us
denote Nd as the number of diverging points and µ as the mean number of iterations
for the starting points. We choose r = 2, 3, 4, 5 for our numerical experiments.

82
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Table-3:
M ethod r=2 r=3 r=4 r=5
Nd µ Nd µ Nd µ Nd µ
GM 10893 2.68 2976 3.15 2954 4.21 18806 9.39
HU 28360 3.70 8778 10.1 25702 17.5 12360 16.5
HS 0 2.86 0 3.22 430 4.32 0 3.77
R−1 0 3.78 0 6.14 358 11.5 4000 13.0

Table 3 show that the new method R-1 performs better than the HU and GM methods
because it is globally convergent for the cases r = 2 and r = 3 and our method is
the most efficient method for the case r = 4 with the smallest number of diverging
points. Our method has many diverging points for the case r = 5 compared to the
HS with no diverging points. However, the HS method requires the calculation of
second derivative.
Figure 1: Polynomiographs for f (x) = x2 − 1

(a) GM (b) HU (c) HS (d) R-1

Fig. 1 shows the polynomiographs of the 4 ninth order methods for the quadratic
polynomial with roots 1 (green) and −1 (blue). The Julia set for the HS is the
imaginary axis. The behaviour of the R-1 method is less dynamic than the HU and
GM methods.
Figure 2: Polynomiographs for f (x) = x3 − 1

(a) GM (b) HU (c) HS (d) R-1

Fig. 2 shows the polynomiographs of the 4 ninth order methods for the cubic
polynomial with roots 1 (orange), −0.5000 − 0.8660i (blue) and −0.5000 + 0.8660i
(green). The behaviour of the R-1 method is less dynamic than the HU and GM

83
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

methods.

Figure 3: Polynomiographs for f (x) = x4 − 1

(a) GM (b) HU (c) HS (d) R-1

Fig. 3 shows the polynomiographs of the 4 ninth order methods for the quartic
polynomial with roots 1 (reddish-brown), −1 (green), i (orange) and −i (blue). The
behaviour of the R-1 method is less dynamic than the HU and GM methods.

Figure 4: Polynomiographs for f (x) = x5 − 1

(a) GM (b) HU (c) HS (d) R-1

Fig. 4 shows the polynomiographs of the 4 ninth order methods for the quintic
polynomial with roots 0.3090 + 0.9511i (reddish-brown), 0.3090 − 0.9511i (blue) ,
−0.8090 + 0.5878i (orange), −0.8090 − 0.5878i (green) and 1 (dark brown). The
behaviour of the R-1 method is less dynamic than the HU and GM methods.

7. Conclusions
In this paper we have proposed a new ninth order iterative method. The conver-
gence order of the suggested method is proved, the efficiency is measured and the
computational order of convergence (COC) is also calculate. With the help of
some numerical test problems, comparison of the obtained results with the existing
methods such as the GM, HU and HS is also given and it is observed that the new
method is efficient in many cases as compared to the existing methods. The dynamic
of the methods for finding the roots of unity are also studied.

84
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

References
[1] J. Biazar, A. Amirteimoori, An improvement to the fixed point iterative method,
Applied mathematics and computation, 182 (2006) pp. 567-571.
[2] R. L. Burden, J. D. Faires , Numerical analysis 8th ed, Brooks-Cole, 2004.
[3] L. Hou and X. Li, Twelfth-order method for nonlinear equation, International
Journal of Research and Reviews in Applied Sciences, 3 (1) (2010), pp. 30-36.
[4] Z. Hu, L. Guocai, L. Tian , An iterative method with ninth-order convergence
for solving nonlinear equations, Int. J. contemp. Math. Science, 6(1) (2011) 17-23.
[5] M. K. Jain,S. R. K. Iyengar, R. K Jain, Numerical Methods For Scientic and
Engineering Computation, 4th edition, New age international publishers, 2003.
[6] A.M.Ostrowski Solutions of equations and system of equations. New York: Aca-
demic Press; 1960.
[7] M. Rafiullah, A fifth-order iterative method for solving nonlinear equations,
Numerical Analysis and Applications, 4(3) (2011), pp. 239–243.
[8] J.F. Traub, Iterative Methods for the Solution of Equations, Prentice Hall, 1964.
[9] F. Ahmed, Elements of Numerical Analysis, National book foundation Pak-
istan, 2001.
[10] S. Weerakoon, T.G.I. Fernando, A variant of Newton’s method with accelerated
third-order convergence, Appl. Math. Lett. 13(2000) 87–93.
[11] M.A. Hafiz, Salwa M. H. Al-Goria, New ninth and seventh order methods for
solving nonlinear eqautions, European Scientific Journal November edition, 8(27)
(2012), PP. 83-95.
[12] G.V. Milovanov’c and A.S. Cvetkovi’c, A note on three-step iterative methods
for nonlinear equations, Studia Univ. ”Babes-Bolyai”, Mathematica, Vol. LII, No. 3,
September 2007.
[13] Y. I. Kim and C. Chun, New Twelfth-order modification of Jarratt’s method for
solving nonlinear equations, Studies in Nonlinear Sciences, 1(1) 2010, 14-18.
[14] B. Kalantari, Polynomial root-finding and polynomiography, World Scientific
Publishing Co. Pte. Ltd. Singapore, 2009.
[15] E. R. Vrscay, Julia Sets and Mandelbrot-Like Sets Associated With Higher Order
Schröder Rational Iteration Functions: A Computer Assisted Study, Math. Comp.
46(1986), 151–169.

Muhammad Rafiullah
Dept. of Mathematics,
COMSATS Institute of Information Technology,

85
 M. Rafiullah, D. K. R Babajee and Dur-e-Jabeen – Ninth Order Method . . .

Lahore, Pakistan
email: [email protected], [email protected]

D. K. R. Babajee
African Network for Policy Research & Actions for Sustainability (ANPRAS)
Mauritius
email: [email protected]

Dur-e-Jabeen
Faculty of Engineering Science and Technology,
Hamdard Univeristy,
Karachi, Pakistan
email: [email protected]

86

View publication stats