International Journal of Advances in Applied Sciences (IJAAS)

Vol. 10, No. 2, June 2021, pp. 141∼148

ISSN: 2252-8814, DOI: 10.11591/ijaas.v10.i2.pp141-148 r 141

A numerical algorithm to find a root of non-linear

equations using householder’s method
Srinivasarao Thota
Department of Applied Mathematics, School of Applied Natural Sciences, Adama Science and Technology University,
Adama, Ethiopia

Article Info ABSTRACT

Article history: In this paper, we propose a three step algorithm using Householder’s method for find-
ing an approximate root of the given non-linear equations in one variable. Several
Received Nov 24, 2020
numerical examples are presented to illustrate and validation of the proposed method.
Revised Feb 2, 2021 Implementation of the proposed algorithm in Maple is also discussed with sample
Accepted Apr 15, 2021 computations.

Keywords:
This is an open access article under the CC BY-SA license.
Non-linear equations
Iterative method
Householder’s method
Approximate root
Maple implementation

Corresponding Author:
Srinivasarao Thota
Department of Applied Mathematics
School of Applied Natural Sciences
Adama Science and Technology University
Post Box No. 1888, Adama, Ethiopia
Email: [email protected], [email protected]

1. INTRODUCTION
Finding an approximate solution of a given non-linear equation is one of the important topics in
numerical analysis. Many mathematical models of physics, economics, engineering, science and any other
disciplines come up with non-linear equations of the type . In resent time, several scientists and engineers have
been focused on to solve the non-linear equations both numerically and analytically. In the literature, there
are several iterative methods/algorithms available using different techniques such as homotopy, interpolation,
Taylor’s series, quadrature formulas, decomposition etc. with modifications and improvements, and different
hybrid iterative methods, see, for example [1]-[17]. There are several methods available to find an approximate
solution of non-linear systems also, see for example, [18]-[23]. In generally, the roots of non-linear or tran-
scendental equations cannot be expressed in closed form or cannot be computed analytically. The root-finding
algorithms provide us to compute approximations to the roots; these approximations are expressed either as
small isolating intervals or as isolating point numbers. In this paper, we use the Householder’s technique to
create a three-step iterative algorithm for solving the given non-linear equation of single variable with converg-
ing order more than or equal to four. Indeed, we focus on creating a tree-step algorithm to find a root of a given
non-linear equation using the well-known techniques namely, Householder’s technique and modified Newton’s
method. We also focus on its Maple implementation with sample computations. The proposed algorithm is
simple to understand and efficient than some existing methods.
Since, we focus on creating an efficient and simple three-step algorithm using Householder’s method
and modified Newton’s method, we recall basic concepts related to the existing methods. In this paper, we

Journal homepage: http://ijaas.iaescore.com

142 r ISSN: 2252-8814

consider the non-linear equations of the following type (1).

f (x) = 0, (1)

where f : I → R is a scalar function in single variable for an open interval I ⊆ R and continuously differ-
entiable function. In numerical analysis, the order of convergence of a convergent sequence are quantities that
represent how quickly the sequence approaches its limit. The order of converges of any iterative method is
defined [13] as
|en+1 | ≤ c|en |p ,

where p is the order of convergence and c is a positive finite constant. Householder’s method is one of the
famous methods in producing a sequence of approximation roots of (1) with initial point x0 . The Householder’s
method has cubic order of convergence [1]-[2] and is given as (2)

f (xn ) (f (xn ))2 f 00 (xn )

xn+1 = xn − − , n = 0, 1, 2, . . . . (2)
f 0 (xn ) 2(f 0 (xn ))3

In the literature, there are several modifications of Householder’s method have been developed and
analyzed by various techniques, for example, see [1]-[3], [24]. The modified Householder’s methods give
better performance than the Newton’s method.
Recall the predictor-corrector method developed by Traub [4] in 1864 as follows, for f 0 (xn ) 6= 0,

f (xn )
yn = xn − ,
f 0 (xn )
(3)
f (yn )
xn = yn − 0 .
f (yn )

A modified Newton’s method developed by McDougall

√ and Wotherspoon [25] is converged faster than
the Newton’s method with a convergence order 1 + 2, and it is given as follows: Let x0 = y0 , for n = 0
(initially), compute
f (x0 )
x1 = x0 − 0 ,
f (x0 )

and for n ≥ 1,

f (xn )
yn = xn −  ,
xn−1 +yn−1
f0 2
(4)
f (xn )
xn+1 = xn − xn +yn

.
f0 2

If xn = yn , for all n in equation (4), then the modified Newton’s method developed by McDougall
and Wotherspoon becomes Newton’s predictor-corrector method as given in equation (3). The rest of paper is
organized as follows: Section 2. presents the methodology and steps involving in the proposed algorithm and the
flow chart; Section 3. presents couple of numerical examples to illustrate and validate the proposed algorithm;
and Section 4. presents the Maple implementation of the proposed algorithm with sample computations.

2. PROPOSED ITERATIVE ALGORITHM

In this section, we present a three-step iterative algorithm using the well-known Householder’s method
given in (2) and a modified newton’s method given in equation (4). The proposed iterative formula under
consideration is given as: set x0 = y0 , for n = 0 (initially), compute

f (x0 ) (f (x0 ))2 f 00 (x0 )

x1 = x0 − − ,
f 0 (x0 ) 2(f 0 (x0 ))3

Int J Adv Appl Sci, Vol. 10, No. 2, June 2021 : 141 – 148
Int J Adv Appl Sci ISSN: 2252-8814 r 143

and for n ≥ 1,
f (xn ) (f (xn ))2 f 00 (xn )
yn = xn −  − 3 ,
xn−1 +yn−1
 
f0 2 2 f 0 xn−1 +y n−1
2
(5)
f (xn ) (f (xn ))2 f 00 (xn )
xn+1 = xn − xn +yn −



3 .
f0 2 2 f 0 xn +y n
2

If xn = yn , for all n and f 00 (xn ) 6= 0 in equation (4), then the proposed algorithm becomes Householder’s
predictor-corrector method and it has fourth-order convergence [3, 24]. If xn = yn for all n and f 00 (xn ) = 0
in (4), then the proposed algorithm becomes Newton’s predictor-corrector method as given in (3).
2.1. Steps for calculating root
The following steps are involved to compute an approximate root of a given using the proposed algo-
rithm.
1. Choose an initial approximation x0 ∈ R with x0 = y0 for n = 0.
2. Compute x1 using formula given in (5). i.e.,
f (x0 ) (f (x0 ))2 f 00 (x0 )
x1 = x0 − 0
− ,
f (x0 ) 2(f 0 (x0 ))3

3. For n ≥ 1, compute yn and xn+1 using the proposed algorithm. i.e.,

f (xn ) (f (xn ))2 f 00 (xn )
yn = xn −  − 3 ,
xn−1 +yn−1
 
f0 2 2 f 0 xn−1 +y n−1
2

f (xn ) (f (xn ))2 f 00 (xn )

xn+1 = xn − xn +yn

−

3 .
f0 2 2 f 0 xn +y n
2

4. Repeat Step 3 until we get desired approximate root, for n = 1, 2, 3 . . ..

Flow chat of the proposed algorithm is presented in Figure 1.

Figure 1. Flow chat of the proposed algorithm

A numerical algorithm to find a root of non-linear equations...(Srinivasarao Thota)

144 r ISSN: 2252-8814

3. NUMERICAL EXAMPLES
In this section, we present couple of numerical examples to illustrate the proposed algorithm and
comparisons are made to confirm that the algorithm gives solution faster than existing methods.
Consider the following nonlinear equation to compute an approximate root using the proposed method.

f (x) = xex − 1. (6)

The exact root of the equation is x = 0.5671432904. Suppose the initial approximation is x0 = 3. The
stopping condition is corrected to ten decimal places. Set x0 = y0 = 3, compute x1 as follows

f (x0 ) (f (x0 ))2 f 00 (x0 )

x1 = x0 − − = 1.922456285.
f 0 (x0 ) 2(f 0 (x0 ))3

Now, we compute yn and xn+1 , for n = 1, using the proposed algorithm, as follows

f (x1 ) (f (x1 ))2 f 00 (x1 )

y1 = x1 − x0 +y0

−

3 = 1.767472904,
f0 2 2 f 0 x0 +y 0
2
f (x1 ) (f (x1 ))2 f 00 (x1 )
x2 = x1 − x1 +y1

−

3 = 0.9087794052.
f0 2 2 f 0 x1 +y 1
2

Similarly, we can compute yn and xn+1 , for n > 1, using the proposed algorithm, as follows

y2 = 0.8380960197, x3 = 0.5661945014,
y3 = 0.5667781506, x4 = 0.5671428368,
y4 = 0.5671432908, x5 = 0.5671432903,
y5 = 0.5671432904, x6 = 0.5671432904.

Hence the required approximate root of the given (6) is 0.5671432904 up to ten decimal places and with
the tolerance error 1 × 10−10 . The graphical representations (using Maple) of the iterations values at different
iterations is shown in Figure 2, absolute errors at different iterations are shown in Figure 3 and absolute function
values at different iterations are shown in Figure 4.

Figure 2. Iterations at Figure 3. Absolute errors at Figure 4. Absolute function

different iterations different iterations values at different iterations

The following non-linear equations with initial approximations to compare the number of iterations
taken to obtain the approximate root using existing methods and proposed method. For simplicity, we take

Int J Adv Appl Sci, Vol. 10, No. 2, June 2021 : 141 – 148
Int J Adv Appl Sci ISSN: 2252-8814 r 145

tolerance error 1 × 10−8 .

f1 (x) = x2 − (1 − x)5 , x0 = 5.0,

3 −x
f2 (x) = x − e , x0 = 6.0,
x 1
f3 (x) = −20x5 − + , x0 = 1.5,
2 2
f4 (x) = cos(x) − x3 , x0 = 8.0,
x 3
f5 (x) = (e + x − 20) , x0 = 4.0,
f6 (x) = x − 3 ln(x), x0 = 0.5.

We compare the proposed method with various existing methods such as Newton’s Method (NM), Kang et al.
Method (KM) [26], McDougall-Wotherspoon Method (MWM) [25], Neamvonk Method (ANM) [27]-[29] and
Proposed Method (PM) to show the efficiency of the proposed method. Table 1, shows the number of iterations
taken by various methods to obtain the required approximate root.

Table 1. Comparison of iterations by different methods for example 3

Function Approx. Root NM KM MWM ANM PM
f1 (x) 0.345954815848 12 6 9 6 6
f2 (x) 0.772882959149 10 5 7 6 4
f3 (x) 0.427677296931 10 5 8 6 4
f4 (x) 0.865474033102 10 5 7 6 4
f5 (x) 0.865474033102 44 14 33 10 12
f6 (x) 0.865474033102 7 4 5 5 2

From Table 1, one can observe that the proposed method (PM) gives the required approximate root
faster than the other existing methods (i.e., in lowest iteration number).

4. MAPLE IMPLEMENTATION OF PROPOSED ALGORITHM

In this section, we present the implementation of the proposed algorithm in Maple.
Input: given f (x),
initial approximation x[0],
tolerance ,
correct to decimal places δ,
maximum number of iterations n.
Output: approximate solution.

4.1. Pseudo-code
− Set x[0] = y[0] and compute x[1] using formula in (5).

− for i from 1 to n do

− compute y[i] and x[i+1] using formulas in (5).

− if abs(x[i + 1] − x[i]) <  and abs(f (x[i + 1])) < δ then break;

− output x[i + 1]

4.2. Maple programming

We present the maple code of the proposed algorithm as follows.

> eps := TYPE; # epsilon value

> epsabs := TYPE; # delta value
> f(x):= TYPE; # given non-linear equation
> fd:=D(f);
> fdd:=D(D(f));

A numerical algorithm to find a root of non-linear equations...(Srinivasarao Thota)

146 r ISSN: 2252-8814

> x[0] := TYPE; # initial approximation

> y[0]:=x[0];
> n:= TYPE; # required number of iterations
> x[1] := x[0]-(f(x[0]))/(fd(x[0]))-
((f(x[0]))∧ (2)*fdd(x[0]))/ (2*(fd(x[0]))∧ (3)):
> for i from 1 to n do
> y[i]:=x[i]-(f(x[i]))/(fd((x[i-1]+y[i-1])/(2)))-
(f(x[i]))∧ (2)fdd(x[i]))/(2*(fd((x[i-1]+y[i-1])/(2)))∧ (3)):
> x[i+1] := x[i]-(f(x[i]))/(fd((x[i]+y[i])/(2)))-
((f(x[i]))∧ (2)*fdd(x[i]))/(2*(fd((x[i]+y[i])/(2)))∧ (3)):
> abserror[i]:=abs(x[i+1]-x[i]);
> absfun[i]:=abs(f(x[i+1]));
> if abs(x[i+1]-x[i]) < eps or abs(f(x[i+1])) < epsabs
> then break;
> end if:
> printf("Iteration No: %g = %10.10f\\n", i,x[i+1]);
> end do;
4.3. Sample computations
Recall the function f6 (x) from Example 3. to compute an approximate solution using the maple im-
plementation.
> eps := 0.00000001;
> epsabs := 0.00000001;
> f(x):= x-3 ln(x);
> fd:=D(f);
> fdd:=D(D(f));
> x[0] := 0.5;y[0]:=0.5;
> n:= 15;
> x[1] := x[0]-(f(x[0]))/(fd(x[0]))-
((f(x[0]))∧ (2)*fdd(x[0]))/ (2*(fd(x[0]))∧ (3)):
> for i from 1 to n do
> y[i]:=x[i]-(f(x[i]))/(fd((x[i-1]+y[i-1])/(2)))-
((f(x[i]))∧ (2)fdd(x[i]))/(2*(fd((x[i-1]+y[i-1])/(2)))∧ (3)):
> x[i+1] := x[i]-(f(x[i])) / (fd((x[i]+y[i])/(2)))
-((f(x[i]))∧ (2)*fdd(x[i])) /(2*(fd((x[i]+y[i])/(2)))∧ (3)):
> abserror[i]:=abs(x[i+1]-x[i]);
> absfun[i]:=abs(f(x[i+1]));
> if abs(x[i+1]-x[i]) < eps or abs(f(x[i+1])) < epsabs then
> break;
> end if:
> printf("Iteration No: %g = %10.10f\\n", i,x[i+1]);
> end do;

1.10−8
1.10−8
x → x − 3 ln(x)
3
x→1−
x
3
x→ 2
x
0.5
0.5
15

Int J Adv Appl Sci, Vol. 10, No. 2, June 2021 : 141 – 148
Int J Adv Appl Sci ISSN: 2252-8814 r 147

1.430307717
1.850449158
0.515191954
0.004163965
Iteration No:1 = 1.8504491580
1.854014254
1.857200752
0.006751594
0.000010394
Iteration No:2 = 1.857183861
1.857183978
1.857183861
0.000016891
1.10−9

5. CONCLUSION
In this paper, we proposed a three-step iterative algorithm for finding an approximate root of a given
non-linear equation in one variable with the help of Householder’s method and modified Newton’s method.
Maple implementation of the proposed algorithm is also discussed with sample computations. Couple of nu-
merical examples are presented to illustrate and validation of the proposed method. Comparisons of the pro-
posed method with existing methods are discussed and the results showed that the proposed algorithm required
less number of iterations (see, for example, Table 1). As future work, one can use this algorithm for finding
the approximate solution of the non-linear equations which have multiple solutions as well as for system of
non-linear equations.

REFERENCES
[1] A.S. Housholder, The Numerical Treatment of a Single Nonlinear Equation, McGraw-Hill, New York,
1970.
[2] C. Chun, ”Iterative methods improving Newton’s method by the decomposition method,” Comput. Math.
Appl. vol. 50, no. 10-12, pp. 1559–1568, 2005
[3] J.F. Traub, Iterative Methods for Solution of Equations, Prentice-Hall, Englewood Cliffs, NJ, 1964.
[4] J.F. Traub, Iterative Methods for the Solution of Equations (Englewood Cliffs, NJ: Prentice-Hall, 1964).
[5] J.H. He, ”A new iterative method for solving algebraic equations”, Applied Mathematics and Computa-
tion, vol. 135, no. 1, pp. 81–84, 2005.
[6] M.A. Noor, K.I. Noor, W.A. Khan, F. Ahmad, ”On iterative methods for nonlinear equations,” Applied
Mathematics and Computation, vol. 183, no. 1, pp. 128–133, 2006.
[7] M.A. Noor, F. Ahmad, ”Numerical comparison of iterative methods for solving nonlinear equations,”
Applied Mathematics and Computation, vol. 180, no. 1, pp. 167–172, 2006.
[8] M.A. Noor, K.I. Noor, ”Three-step iterative methods for nonlinear equations,” Applied Mathematics and
Computation, vol. 183, no. 1, pp. 322–327, 2006.
[9] M. Saqib, M. Iqbal, S. Ali, T. Ismaeel, ”New fourth and fifth-order iterative methods for solving nonlinear
equations,” Applied Mathematics, vol. 6, no. 8, pp. 1220–1227, 2015.
[10] Saied, Liao, ”A new modification of False-Position method based on homotopy analysis method,” Applied
Mathematics and Mechanics, vol. 29, no. 2, 223–228, 2008.
[11] Sagralo, Mehlhorn, Computing real roots of real polynomials, 2013.
[12] S. Hussain, V. K. Srivastav, S. Thota, ”Assessment of Interpolation Methods for Solving the Real Life
Problem,” International Journal of Mathematical Sciences and Applications, vol. 5, no. 1, pp. 91–95,
2015.
[13] S. Thota, V.K. Srivastav, ”Quadratically Convergent Algorithm for Computing Real Root of

A numerical algorithm to find a root of non-linear equations...(Srinivasarao Thota)

148 r ISSN: 2252-8814

Non-Linear Transcendental Equations,” BMC Research Notes, vol. 11, no. 909, pp. 1–6,2018,
https://doi.org/10.1186/s13104-018-4008-z.
[14] S. Thota, V.K. Srivastav, ”Interpolation based Hybrid Algorithm for Computing Real Root of Non-Linear
Transcendental Functions,” International Journal of Mathematics and Computer Research, vol. 2, no. 11,
pp. 729–735, 2014.
[15] S. Thota, ”A New Root-Finding Algorithm Using Exponential Series,” Ural Mathematical Journal, vol.
5, no. 1, pp. 83–90, 2019.
[16] S. Thota, T. Gemechu, ”A New Algorithm for Computing a root of Transcendental Equations Using Series
Expansion,” Southeast Asian Journal of Sciences, vol. 7, no. 2, pp. 106–114, 2019.
[17] T. Gemechu, S. Thota, ”On New Root Finding Algorithms for Solving Nonlinear Transcendental Equa-
tions,” International Journal of Chemistry, Mathematics and Physics, vol. 4, no. 2, pp. 18–24, 2020.
[18] Sharma, J.R., Kumar, D, ”On a reduced cost derivative-free higher-order numerical algorithm for nonlin-
ear systems,” Comp. Appl. Math. vol. 39, no. 202, 2020. https://doi.org/10.1007/s40314-020-01218-3.
[19] Argyros IK, Hilout S. Numerical Methods in Nonlinear Analysis. World Scientific Publishing Company,
New Jersey, 2013.
[20] Constantinides A, Mostoufi N. Numerical Methods for Chemical Engineers with MATLAB Applications.
Prentice Hall PTR, New Jersey, 1999.
[21] Moré JJ. A collection of nonlinear model problems. In, ”Allgower EL, Georg K (eds) Computational
Solution of Nonlinear Systems of Equations Lectures in Applied Mathematics,” American Mathematical
Society, Providence, RI, vol. 26, pp. 723–762, 1990.
[22] Ortega JM, Rheinboldt WC. Iterative Solution of Nonlinear Equations in Several Variables. Academic
Press, New York, 1970.
[23] Traub JF. Iterative Methods for the Solution of Equations. Chelsea Publishing Company, New York, 1982.
[24] W.A. Khan, M.A. Noor, A. Rauf, ”Higher-order iterative methods by using Householder’s method for
solving certain nonlinear equations,” Math. Sci. Lett. vol. 2, no. 2, pp. 107–120, 2013.
[25] T.J. McDougall, S.J. Wotherspoon, ”A simple modification of Newton’s method to achieve convergence
of order,” Journal of Applied and Mathematical letters, vol. 29, no. 1, pp. 20–25, 2014.
[26] S.M. Kang, A. Rafiq, F. Ali, W Mahmood, ”A New Newton’s Type Method for Nonlinear Equations,”
International Journal of Mathematical Analysis, vol. 10, no. 4, pp. 151–158, 2016.
[27] A. Neamvonk, ”Three-Step Method For Finding Root of Non-Linear Equations,” IOSR Journal of Math-
ematics, vol. 15, no. 1, pp. 01–06, 2019.
[28] T. Parveen, S. Singh, S. Thota, V. K. Srivastav, ”A New Hydride Root-Finding Algorithm for Transcen-
dental Equations using Bisection, Regula-Falsi and Newton-Raphson methods,” National Conference on
Sustainable & Recent Innovation in Science and Engineering (SUNRISE-19), (2019) (ISBN No. 978-93-
5391-715-9).
[29] V.K. Srivastav, S. Thota, M. Kumar, ”A New Trigonometrical Algorithm for Computing Real Root of
Non-linear Transcendental Equations,” International Journal of Applied and Computational Mathemat-
ics,, vol. 5, no. 44, pp. 1-8, 2019.

BIOGRAPHIES OF AUTHORS
Dr. Srinivasarao Thota (Corresponding Author) completed his M.Sc. in Mathematics from Indian
Institute of Technology (IIT) Madras and Ph.D. in Mathematics from Motilal Nehru National Institute
of Technology (NIT) Allahabad, India. Srinivasarao Thota’s areas of research interest are Computer
Algebra (symbolic methods for differential equations), Numerical Analysis (root finding algorithms)
and Mathematical Modelling (ecology). He has published his research work in several international
journals and presented his work in different national and international conferences as oral presenter
and expertise/invited presenter in different countries. Presently working at Department of Applied
Mathematics, Adama Science and Technology University, Ethiopia. Further info on his homepage:
https://srinivasaraothota.webs.com/

Int J Adv Appl Sci, Vol. 10, No. 2, June 2021 : 141 – 148