Journal of Kirkuk University Scientific Studies , vol.5, No.

2,2010

A New Technique to Compute Complex Roots

Sasan A. Muheadden
Department Of Mathematics
College of Science - University of Kirkuk
Received:1/2/2010, Accepted:1/6/2010

Abstract

In this paper we modified Newton Raphson and Secant methods to Complex

form, then from that we get new iterative formula to them, named Newton (Secant) real
iteration formula and Newton (Secant) imaginary iteration formula, real iteration used to
find real roots while imaginary iteration one will be used to find complex roots of non-
linear equations such as ( x a 0 ). To find real roots, we start with a real initial
2

point in real iteration getting a sequence of real numbers, to find complex roots start
with a complex initial point in complex formula getting a sequence of complex
numbers, also we write a new algorithm for this technique and write the program by
using Matlab application system version 7.8 for this new method such that it will
determine the real roots when we enter the initial real point but when we enter the initial
complex point it will determine the complex roots.

Introduction
Complex analysis can roughly be though of as a subject that applies the
theory of calculus to imaginary numbers. But what is exactly are imaginary
numbers ? we can take the square root of a negative number, but let pretend
we can and being by using the symbol i 1 . Also numerical analysis
is concerned with the mathematical derivation description and analysis of
methods of obtaining numerical solutions of mathematical problems; such
that find complex roots of linear and non linear equations by using
iteration methods. (Abbas & Sasan, 2009)
If R is the real field and C is the Complex field then consider the following
definitions:
Newton Raphson method can be used to approximate the root of any non-
linear equation of any degree by the following iterative formula (Atkinson ,
1998)
f (x )
xk 1 xk \ k
f ( xk ) k 0,1,2..
If the function f (x) has no derivative then we have difference
approximation in this case, Newton Raphson method is called Secant
method (Steven, 2008).

401
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

know we will show comparison between Newton and Secant method

(Abbas&Sasan, 2008):

Newton Raphson method Secant method

1.Iterative formula 1.Iterative formula

f ( xk )
xk 1 xk x k x k 1
f \ ( xk ) x k 1 x k f ( x k ) . , k 1,2,3..
f ( x k ) f ( x k 1 )
k 0,1,2..

2.Need for one initial point . 2. Need for two initial points.

3.Unstable . 3. Stable .

4. Need to derivative . 4. No need to derivative.

5.Faster . 5. Slow .

6. Quadratic speed . 6. Linear speed .

Complex iterative formula

1-Complex Newton Raphson formula (Complex Iteration ) :
In Newton formula Replace the variable x with the variable z in both
side then we get the following form
f (z )
z k 1 z k \ k , k 0,1,2,... (1)
f ( zk )
then we substitute z ( x i y) in the above form we get the formula
f ( x iy) k
( x i y ) k 1 ( x iy) k \ k 0,1,2... (2)
f ( x iy) k
f ( x iy k )
xk 1 iy k 1 ( xk iy k ) \ k k 0,1,2,... (3)
f ( xk iy k )
then we get from that

401
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

f (x ) f (y )
xk 1 iyk 1 xk \ k i y k \ k k 0,1,2,... (4)
f ( x k
) f ( y k
)
know equate the real and the imaginary parts in (4) we get two iterative
formula named the real iterative formula(Newton Raphson ) (R.I.F)(5)and
the imaginary iterative formula (I.I.F)(6):
f (x )
xk 1 xk \ k , k 0,1,2,.... (5)
f ( xk )
f (y )
y k 1 yk \ k , k 0,1,2,..... (6)
f ( yk )
R.I.F and I.I.F were used to evaluate real and imaginary roots of non
linear equations respectively.
2- Complex Secant formula:
In secant method if we replace the variable x by the variable z then we get
the complex secant formula which can separated into two formulas, the
first called the real iterative formula starts with real initial point ( x0 ) the
second called is the imaginary iterative formula starts with complex initial
point ( y0 ) , and where used to find real and imaginary roots of non linear
equations respectively.
Now
z k z k 1
z k 1 z k f ( z k ) . ) k 1,2,3,... (7)
f ( z k ) f ( z k 1 )
putting z ( x i y) in the above form we get the formula

( x iy) k ( x iy) k 1
( x iy) k 1 ( x iy) k f ( xk iyk ). , k 1,2,3,.. (8)
f ( xk iyk ) f ( xk 1 iyk 1 )

( xk iyk ) ( xk 1 iyk 1 )
xk 1 iyk 1 xk iyk f ( xk ) if ( y k ) . , k 1,2,3,.. (9)
f ( x k iy k ) f ( x k 1 iy k 1
)

x k x k 1 y k y k 1
x k 1 iy k 1 x k f ( x k ). i y k f ( y k ). ,
(10)
f ( x k ) f ( x k 1 ) f ( y k ) f ( y k 1 )
k 1,2,3.....
Now equating the real and the imaginary parts of both sides we get that

x k x k 1
xk 1 xk f ( xk ). , k=1,2,3, (11)
f ( xk ) f ( xk 1 )

401
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

Equation (11) called the real secant iterative starting with two real initial
points .
y k y k 1 (12)
y k 1 y k f ( y k ). , k 1,2,3,...
f ( y k ) f ( y k 1 )

Equation (12) called the imaginary secant iterative starting with two
imaginary initial points .
3-A new modified Newton Raphson algorithm :
For computing complex and real root of non linear equations f(x)=0
depended on given initial point, we write the real iterative formula and the
imaginary iterative formula of Newton Raphson in the following
procedure such that it is computing all roots of equations step by step and
the number of iterations depended on the value of tolerance .
Now
Input informatios : The initial point z and tolerance
Output informations: The complex root w1 or real root w2
Procedure :
Step1 : If z is real , Go to step9 .
Step2 : Replace y0 z .
Step3 : Evaluate new value by
f (y )
y1 y0 \ 0
f ( y0 )
Step4 : If | y1 y0 | , go to step7 .
Step5 : Replace y0 y1 .
Step6 : Back to step3 .
Step7 : w1 y1 .
Step8 : Print the complex root w1 .
Step9 : Replace x0 z .
Step10: Evaluate new value by
f (x )
x1 x0 \ 0
f ( x0 )
Step11: If | x1 x0 | , go to step14 .
Step12: Replace x0 x1 .
Step13: Back to step10 .
Step14: w2 x1
Step15: Print the real root w2 .
Step16: Stop.

401
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

4-A new modified Secant algorithm:

For compute the complex and the real roots of non linear equations
f(x)=0 where f(x) has no derivative we write the following procedure step
by step depended on two initial points and tolerance.
Input informations: the initial points z1 , z0 and tolerance .
Output informations: complex root w1 or real root w2 .
procedures :
Step1: If z1 , z0 are real , Go to step9 .
Step2: Replace y0 z0 & y1 z1 .
Step3: Evaluate new value by y y f ( y ) y y 1 0
2 1 1 f (y ) f (y )
1 0

Step4: If | y2 y1 | , go to step7 .
Step5: Replace y0 y1 & y1 y2 .
Step6: Back to step3 .
Step7: w1 y2 .
Step8: Print the complex root w1 .
Step9: Replace x0 Z 0 & x1 Z1 .
Step10: Evaluate new value by
x1 x0
x2 x1 f ( x1 )
f ( x1 ) f ( x0 )
Step11: If | x2 x1 | , go to step14 .
Step12: Replace x0 x1 & x1 x2 .
Step13: Back to step10 .
Step14: w2 x2
Step15: Print the real root w2 .
Step16: Stop.
5-The program by using Matlab application system version 7.8 for
new modified Newton algorithm :
Before writing the program script , we must define a function and its
derivative for the given non linear equation and save them as a function
m file .Then we define these functions as funcnew01 and funcnew02 and
save them as funcnew01.m and funcnew02 .m
Now :
Z=input('the complex value of Z=');
tol=input('the value tolerance tol=');
k=0;
if real(Z)==0
y=Z;

401
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

disp('y funcnew01(y) funcnew02(y) k')

disp('........................................')
while abs(feval('funcnew01',y))>tol
y=y-(feval('funcnew01',y)/feval('funcnew02',y))
k=k+1;
end;
else
x=Z;
disp('x funcnew01(x) funcnew(x) k')
disp('......................................')
while abs(feval('funcnew01',x))>tol
x=x-
(feval('funcnew01',x)/feval('funcnew02',x))
k=k+1;
end;
end;

6-The program by using Matlab application system version 7.8 for

new modified Secant algorithm :
Z1=input('the complex value Z1=');
Z2=input('the complex value Z2=');
tol=input('the value tolerance tol=');
k=0;
if real(Z1)==0 & real(Z2)==0
y1=Z1;
y2=Z2;
disp('y f1 f2 k')
disp('...........')
while abs(y2-y1)>tol
f1=feval('funsct01',y1)
f2=feval('funsct01',y2)
y=y2-((f2)*((y2-y1)/(f2-f1)))
y1=y2;
y2=y;
k=k+1;
end;
else
x1=Z1;
x2=Z2;
disp('x f1 f2 k')
disp('..........')

401
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

while abs(x2-x1)>tol
f1=feval('funsct01',x1)
f2=feval('funsct01',x2)
x=x2-((f2)*((x2-x1)/(f2-f1)))
x1=x2;
x2=x;
k=k+1;
end;
end;

Numerical Computation to evaluate complex roots

Example 1 :
Find the root of the following non linear equation by using the new
modified Newton and Secant algorithm :
f ( x) x 2 1 where ( tol = 0.00001 ) .
Solution :
1 By the new modified Newton algorithm:
We know that the equation f(x)=0 has no real roots
f ( xk )
if we use real Newton iterative formula xk 1 xk with real initial
f \ ( xk )
point ( x0 ) we get no solution .
If we use the new modified algorithm (imaginary iterative formula)
f (y )
y k 1 y k \ k starting by complex initial point ( y o ) then after some
f ( yk )
iteration we get a sequence of complex number y k k 0 converges to the

complex root on the y axis

Now start with the complex initial point y0 0.5i ,we get the following table
table No.1
K yk f ( yk ) f \ ( yk )
0 0+0.5i 0.7500 0+10000i
1 0+1.2500i -0.5625 0+2.5000i
2 0+1.0250i -0.0506 0+2.0500i
3 0+1.0003i -6.0985e-004 0+2.0006i
4 0+1.0000i -9.2922e-008 0+2.0012i

after 5 iteration we will be end with complex root y = 0 + 1.0000i. If we

start with the initial point y0 0.5i we get another root y = 0 1.0000i
also when the distance between the initial point and the exact root is small

440
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

again if we start with y0 1.1i then by 2 iteration we get the complex

root .
2By the new modified secant algorithm starting with initial point y0 0.5i
( y 2 y1 )
in imaginary iterative y3 y2 ( f ( y2 ) ) we get
f ( y2 ) f ( y1 )
table No.2
k yk y k 1 f ( yk ) f ( yk 1 )
1 0.0000+0.5000i 0.0000+0.8000i 0.7500 0.3600
2 0.0000+0.8000i 0.0000+1.0769i 0.3600 -0.1598
3 0.0000+1.0769i 0.0000+0.9918i -0.1598 0.0163
4 0.0000+0.9918i 0.0000+0.9997i 0.0163 6.0948e-004
5 0.0000+0.9997i 0.0000+1.0000i 6.0948e-004 -2.5089-006
then after 5 iterations we have the complex root y = 0.0000+1.0000i
Example 2 :
Find all roots of the following non linear equation by using the new
modified Newton algorithm .
f ( x) x 5 x 4 7 x 3 5x 2 4 x 4 when (tol = 0.00001)
Solution:
we know that the above function has 5 roots x i , x 2i and x 1
in this algorithm if we start by the real initial point x0 0.5 , we get a
sequence of real numbers after some iterations using real Newton Iteration
formula we get the real root x 1 .Now to find Complex roots let y 0 be any
complex initial point we get a sequence of complex numbers tends to the
complex root . As shown in table No.3 :
table No.3
K yk f ( yk ) f \ ( yk )
0 0.0000+0.5000i -2.8125+1.4063i 0.5625-4.5000i
1 0.3846+1.0769i -2.2112-2.2944i -8.2107-3.6410i
2 0.0560+0.9432i -0.6447+0.0009i -5.3867-5.4629i
3 -0.0029+1.0031i 0.0362-0.0013i -6.0372-6.0225i

4 -0.0000+1.0000i 0.0000 -6.0000-6.0000i

In table No.2 we used the new modified Newton algorithm starting with
initial point y0 0.5i to get the complex root y = 1.0000i after 4
iterations .If take y0 1.5i we will get another root y = 2.0000i again if we
take y0 1.5i we will get the negative roots.

444
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

Conclusion
In this paper we modified Newton Raphson and Secant algorithms to
Compute the complex roots of non linear equations, then we wrote down
their programs by Matlab applications version 7.8 by taking complex initial
point to get the complex roots if it exists if not we real initial point to get
real root .In this paper we study initial point have only real or imaginary
part , as future work we modified these algorithm to compute the complex
roots if the initial point is complex and real and imaginary part not equal to
zero .

References
Abbas Y.Al-bayati and Runak M. Abdullah, (2008): A New Family Of
Spectral Cg Algorithm, alrafiden Journal, published by college of
Computer Science and Mthematics Mosul university, Vol.5, No.1.
Abbas Y.Al-bayati and Sasan A. Al-Shwani, (2008): A New Numerical
Procedure To Compute the residue of a Complex Functions, al
rafiden Journal ,published by college of Computer Science and
Mthematics Mosul university, Vol.5,No.1.
Abbas Y. Al Bayati and Sasan A. Al Shwani, (2009): Residue of
Complex Functions With Definite and Infinite Poles On X axis
,Journal of Mathematics and Statistics, Science publication new
York USA, Vol.5,No.3, pp.152 158. Web Site : www.scipub.org
Atkinson, K. E. (1998): An Introduction To Numerical Analysis, Second
Edition, John and wiley and Sons, Inc.
Steven C Chapra, (2008): Applied Numerical Methods with Matlab For
Engineering and science, 2nd edition tata McGrawHill Publishing
Company.

441
Journal of Kirkuk University Scientific Studies , vol.5, No.2,2010

2212/2/1: 2212/6/1:

()
()
) )(2.3) 2.4


. 7.8 ) .(3.1 & 3.2

441