Life Science Journal 2014;11(7s) http://www.lifesciencesite.

com

Three-step Method for Finding Multiple Root of Non-linear Equation

Nazir Ahmad Mir1, Khudija Bibi², Naila Rafiq³

¹Riphah International University, Islamabad. ²International Islamic University, Islamabad,

³Centre for Advanced Studies in Pure and Applied Mathematics, Bahauddin Zakariya University, Multan, Pakistan
nazirahmad.mir@gmail.com

Abstract In this paper, we present a three-step method of tenth order for determining multiple roots of a nonlinear
equation. This method requires six evaluations per iteration, three evaluations of function and three evaluations of
first derivative. Numerical examples are tested to compare our method with some other methods in the literature. We
observe that our method is either comparable or has better performance in some cases than the other methods and
preserves the order of convergence.
[Nazir Ahmad Mir, Khudija Bibi, Naila Rafiq. Three-step Method for Finding Multiple Root of Non-linear
Equation. Life Sci J 2014; 11(7s): 387-389]. (ISSN: 1097-8135). http://www.lifesciencesite.com. 82

Keywords: Nonlinear equation, Multiple roots, Iterative methods, Modified Newton's Method, Order of
convergence.

1. Introduction Li, et. al. (2010) developed fourth order method

One of the most important and interesting given as:
problem in the field of numerical analysis,  yn  xn  ( m2m2 ) gg (( xxn )) ,
n
engineering, optimization theory and applied 
 m m g ( xn )
 nz  x n  2( m 2 ) g ( yn ) , (2.2)
mathematics is determining the multiple roots of a  g(x )
nonlinear equation. A function g(x) is said to have a  xn1  xn  ( m2m2 )( a1g ( xn )a2g ( ynn ) a3g ( zn ) ) ,
multiple root of a non-linear equation where
g ( x)  0 , (1.1) 6 5 4
m  m  14m  12m  48m  80 m  32
3 2

a1  ,
iff g(x) can be written as 3 2
8m ( m  2m  8m  4)
g ( x )  ( x  t ) m h ( x) , 4
 m 3m  6 m  20 m  40m  16
3 2

a2  ( )( 3 2
),
where Lim h(t )  0 and m is the known multiplicity. 16 a4 ( m  2m  8m  4)
xt
3 2
1 m ( m  4)
The modified Newton's method (E.Schroder, a3  ( )( 3 2
),
16 a4 (m  2m  8m  4)
1870) is the most popular method for finding
m
multiple roots of a nonlinear equation and is given as a4  (
m
) .
m2
:
g ( xn )
xn1  xn  m . (1.2) Thukral (2012) proposed a fifth order covergent
g ( xn )
method:
This method has second order of convergence. In
 yn  xn  m gg (( xxn )) ,
recent years, many variants of modified Newton's  n
(2.3)
 2
method (N. A. Mir and Naila Rafiq 2007 and Li, et al  xn1  yn  m(1  z m ) gg (( yyn )) ,
n
2010, ) have been developed and analyzed by using
various techniques, which have either equal or where z  g ( y n ) g ( xn ) 1.

improved order of convergence, efficiency and show

better performance than the Newton's methods Thukral (2010) proposed a three-step method:
 yn  xn  m gg (( xxn )) ,
n

2. The Iterative Method  s 1bp s g ( xn )
Thukral (2010) proposed a third order  zn  yn  mp ( 1( b2) ps ) g ( xn ) , (2.4)

convergent method for finding multiple roots of non- s ) g ( zn ) s g ( z n ) s g ( xn ) g ( z n ) s
 xn1  z n  m(u ( p  ( g ( y ) ag ( z ) )  4( g ( x ) ) ( g ( x ) )( g ( x ) ) ,
 n n n n n
linear equation: where
 yn  xn  g ( xn ) s s 2s 2 3s
 g ( xn ) u1 ( p )  1  2 p  (5  2b ) p  (12  12b  2b ) p ,
 ug ( xn )2 5  2b  (2  8b  2b ) p  (1  4b) p
2 s 2s
 xn1  xn  ( ug ( xn ) g ( xn )vg ( xn ) g ( yn ) ), (2.1) s
u2 ( p )  ,
2 s
 m1 m1
5  2b  (12  12b  2b ) p
 where u  (m  1) , v  m , n  N .
 g ( yn )
p .
g ( xn )

http://www.lifesciencesite.com 387 lifesciencej@gmail.com

 Life Science Journal 2014;11(7s) http://www.lifesciencesite.com

This method has eighth order of convergence. c 52 c 32 c 3 c 52

zn 1  3
g 2  e 5n 
oe 6n ,
We suggest the following three-step method for m4 m 3 m3
finding multiple roots which is the modification of c42 c22 c3 c42 c 32 c 3 c 5 c 2 c 23 c5 c2 c4
g
zn  3 2  e 5n  11 11 2 8 19 2 3 2
the method (2.3) for multiple roots and is given as: m4 m3 m 3 m3 m4 m3 m5 m3
 yn  xn  m gg (( xxn )) , c 52 c3 c3
n 3 28 2 4
 e 6n 
oe 7n 
.
 m3 m
 2
g ( yn )
 zn  yn  m(1  a m ) g ( yn ) , (2.5)
 2
Finally, putting the values of zn , g ( zn ) and g '( zn ) in
 x  z  m(1  b m ) g ( zn ) ,
 n1 n g ( zn ) (2.5) and after simplification, we have
where a  g ( y n ) g ( xn ) 1 and b  g ( zn ) g ( yn ) 1.  c29 5 2
c2 c3 c2
9 7
c2 c3
7
c2 c3 c2 
9
10 11
xn1  t   7
4 7
6 8
 12 8
4 7
9 9  en  o(en ) ,
m m m m m m 
5 2 7 7
3. Convergence Analysis  c29 c2 c3 c2
9
c2 c3 c 2 c3 9
c2  10 11
or en1  7
4 7
6 8
 12 8
4 7
9 9  en  o(en ) ,
Now, we prove a theorem for finding the order m m m m m m 

of convergence of method (2.5). which shows that the method defined in (2.5) has
tenth order of convergence.
Theorem: Let t be the multiple root of a nonlinear
equation (1.1) with multiplicity m , where the 4. Comparison of Method(2.5)
function g(x) is sufficiently differentiable in an open Here, we take some numerical examples from
Thukral (2012) and compare our method with some
interval I=(a,b). Let x 0 be an initial guess
other methods in the literature. The test functions,
sufficiently close to t . Then the method (2.5) has
their roots and their initial guesses are displayed in
tenth order of convergence and its error equation is
given by: Table 1, while Table 2 shows the absolute error en
 c29 5 2 9 7 7 9  xn  t for each method. Each method is compared on
c2 c3 c2 c2 c3 c2 c3 c2  10 11
en1   7
4 7
6 8
 12 8
4 7
9 9  en  o (en ). fourth iteration taking the following stopping
m m m m m m 
criterion:
Proof: xn  t   , where   10 300.
Let en  xn  t. Using the Taylor expansion of All the computations reported here are done
g ( xn ), g '( xn ) about t , we get: using 1000 significant digits. MNM represents
Modified Newton's Method, BNM represents Li's
 g ( m ) (t )  m
g ( xn )  
2
 en 1  c1 en  c2 en  ..... , method of fourth order given in (2.2) and NM
 m 
represents the new method defined in (2.5) .We can
 g ( m) (t )  m1   m  1   m  2  c e 2  ..... , observe that the present method is well behaved.
g '( xn )    en 1    c en    2 n
  m  1  m  
 m 1 
 mk  ( t )
where ck  mg
, k  1, 2,... . Table 1. Test functions, their roots, and their initial guesses
 m  k  g ( m ) (t ) Functions m Roots
x 0
Putting the values of g ( xn ) and g '( xn ) in (2.5) and after 10 4 4 0.01 0
g1 ( x )  (( x  1)  1)
simplification, we have:
g 2 ( x )  ( e x  x  20)3 3 3.0 2.842438
c2 2
yn  t  en  o(en3 ).
m 12 12 0.2 0.257530
 2
g 3 ( x )  x  e  3x  2
x

Also
g 4 ( x )  (Si n 2 x  x 2  1)8 8 1.5 1.404491
2 2
c2 en c c c3 c3  3 4
g ( yn )  1    2 2 3  22  2  en  o( en ) , 5 -1 -0.739085
m  m m m g 5 ( x )  (Cosx  x ) 5
2
1 2 c 2  c
3
cc cc  3 4 g 6 ( x )  (tan x  e x  1) 5 5 1.4 1.371045
g ( yn )  1  ( 2 c2  2 )en   2 22  2 2 23  2 2 3  en  o( en ).
m m  m m m 

Now, putting the values of yn , g ( yn ) and g '( y n ) in Table 2. Comparison with various methods

(2.5), we have:
 c24 2
c2 c3
4
c2  5 6
zn  t   3 4
2 3
 3  en  o(en ) .
 m m m 
Again, by Taylor expansion, we get:

http://www.lifesciencesite.com 388 lifesciencej@gmail.com

 Life Science Journal 2014;11(7s) http://www.lifesciencesite.com

Now we take few numerical examples from multiple as well as distinct zeros of non-linear
(Thukral,2012) and will compare new method (2.5) equations, Tamsui Oxford Journal of
with method Thukral 8 defined in (Thukral, 2012). Mathematical Sciences, 26(3), (2010) 321-333.
3. N.A.Mir, Naila Rafiq, Third order and Fourth
Table 3. Test functions, initial guesses and roots order iterative methods for finding multiple and
Functions x0 Roots distinct zeros of nonlinear equations, Applied
x
g  (e  x  20) 95
3.5 2.842438 Mathematics and Computation, 190(1),(2007)
1
432-440.
g 2  (sin x 2  x 2  1)200 1.7 1.404491 4. N.A.Mir, Naila Rafiq, Fourth order two-step
g 3  (tan x  e  1)
x 11
1.4 1.371045 iterative methods for determining multiple zeros
2
1.2 1.207647 of nonlinear equation, IJCM, 84(7), (2007) 971-
g4  ( xe x  sin x2  3 cos x  5)4 977.
5. R. Thukral, A New Fifth-Order Iterative Method
Table 4. Comparison of Method NM for Finding Multiple Roots of Nonlinear
Equations, American Journal of Comp. and Appl.
Functions en  xn  t
Math (2012) 260-264.
NM Thukral 6. R. Thukral, A new third-order iterative method
g1 0.100e - 998 0.273e - 291 for solving nonlinear equations with multiple
roots, J. Math. Comput. 6 (2010) 61-68.
g2 0.133e - 324 0.214e - 194 7. R.Thukral, A family of Three-point Methods of
g3 0.125e - 370 0.128e - 513 Eight-order for Finding Multiple Roots of
Nonlinear equations, Journal of Modern Methods
g4 0.770e - 930 0.946e - 201 in Numerical Mathematics. 2(2013) 1-9.
8. S.G.Li, L.Z. Cheng, B. Neta, Some fourth-order
Conclusions nonlinear solvers with closed formulae for
In this paper, we proposed a new method, namely multiple roots. Comp. Math. Appl. 58 (2010)
NM (2.5) for finding multiple roots of a nonlinear 126-135.
equation. We discussed the convergence analysis of
new method and proved that this method has tenth
order. At the end, we compared our method with some
other methods. We observed that our method is
comparable with other methods or has better
performance than the other existing methods given in
the Tables 2 and 4.

Corresponding Author:
Nazir Ahmad Mir
Professor, Deparment of Mathematics,
Riphah International University,
Islamabad, Pakistan.
E-mail: nazirahmad.mir@gmail.com

References
1. E.Schroder, Uber unend viele Algorithmen zur
der Gleichungen, Math. Ann.2(1870) 317-365.
2. N.A.Mir, Farooq Ahmad, Muhammad Raza,
Tahira Nawaz, Fourth order methods for finding

http://www.lifesciencesite.com 389 lifesciencej@gmail.com