Mohd Ali et al. / Malaysian Journal of Fundamental and Applied Sciences Vol. 13, No.
4 (2017) 728-731
RESEARCH ARTICLE
The n-th section method: A modification of Bisection
Mohd Rivaie Mohd Ali a,*, Muhammad Imza Fakhri b, Nujma Hayati a, Nurul Atikah Ramli b,
Ibrahim Jusoh a
a
Deparment of Mathematics,Faculty of Computer and Mathematical Sciences,Universiti Teknologi Mara Terengganu,21300 Kuala Terengganu,
Terengganu,Malaysia
b
Deparment of Mathematics,Faculty of Computer and Mathematical Sciences,Universiti Teknologi Mara Shah Alam,40450 Shah Alam, Selangor,
Malaysia
Article history Abstract
Received 18 February 2017
Accepted 4 October 2017 Bisection method is the easiest method to find the root of a function. This method is based on the
existence of a root on a specified interval. This interval is then halved or divided into two parts. The
root is known to be laying in either one of these interval. The iterative sequence is continued until a
desired stopping criterion is reached. In this research, a new modification of bisection method
namely fourth section and sixth section methods are introduced. These methods are tested for
several selected functions by using Maple software. The results are then analyzed based on the
number of iterations and the CPU times. Based on the results, it is shown that when the interval
increases, the CPU times will also increase. However, the number of iterations is reduced
significantly.
Keywords: Bisection method; Roots finding; Number of iterations; CPU times
© 2017 Penerbit UTM Press. All rights reserved
INTRODUCTION secant line L joining the points (𝑎, 𝑓 (𝑎)) and (𝑏, 𝑓 (𝑏)) crosses the
𝑥 −axis (Mathews et. al, 2004). The main advantage of Bisection
Numerical analysis is one of the areas of mathematics and method is that the behaviour itself is always convergent since the
computer science that create, analyses, and implements algorithms for method brackets the root much more quickly than the incremental
solving numerically problems of continuous mathematics. There are search method does. Hence, Bisection method is easy to use, apply and
many equations whose roots cannot be evaluated analytically by any has wide range of applications in other developments.
methods. The approximate values of the roots of such equations can be However, the division of the interval into two section leads to
found either by a graphical approach, or number of iterative methods or slow convergence of Bisection method. Hence, a new method is
by a combination of both processes. In numerical methods of solving proposed named as n-th section method. This methods are compared
non-linear equations or root finding, the most popular methods are with the original Bisection method based on number iterations, CPU
Bisection method, Newton’s method and Secant method. Several times and accuracy. The bisection algorithm is the most primitive but
studies shown that many researchers are interested in the development robust for finding a real root because it is always converges to the root
and the application of Bisection method such as Muller method by Park (Burden et. al, 1993).
and Hltotumatu (1987) and bisection-exclusion method by Yakoubsohn
(2005).
THE N-TH SECTION METHOD
BISECTION METHOD The n-th section method is a modification of classical Bisection
method which is fourth and sixth section method. Bisection method
Bisection method is the easiest method using an interval which is which divides the interval into two section leads to slow convergence.
divided into two or half. The root is known to be lying in either one of This new scheme divided the interval into four and six section. The root
these interval. The division of this interval into two continue until the is then identify either in the first, second, third, fourth, fifth or sixth
root is found, up to certain degree of accuracy. Doron (2010) assume interval. This will lead to faster convergence of the root and provide
that this method has opposite signs at both edges of the intervals where faster calculation of the roots. For an even number of n, the algorithm
𝑓(𝑎)𝑓(𝑏) < 0. Then, it is known that 𝑓(𝑥) has at least one root in the will be similar to Bisection method. However, for an odd number of n
interval, [𝑎, 𝑏]. The existing method proceeds and continue its iteration the algorithm needs to be modified. Therefore, in this research only the
until it converges to a point within the tolerance range and finds the even number of n is considered and compared with the Bisection
value of 𝑥 such that 𝑓(𝑥) = 0 or approximately 0 (Solanki, 2014). A method. A comparison between bisection method and n-th section
better approximation is obtained if a point (𝑐, 0) is found where the method will be analysed based on its efficiency in terms of number of
728
Mohd Ali et al. / Malaysian Journal of Fundamental and Applied Sciences Vol. 13, No. 4 (2017) 728-731
iterations and CPU times. Fig. 1, 2 and 3 show the schematic Table 1: Algorithm of Bisection method
representation of Bisection and n-th section line search method.
No. Step
1 Identify two numbers a and b at which f has different signs.
2 𝑎+𝑏
Define midpoint, c = and 𝑓(𝑐).
2
3 Determine if root exists
i. 𝑓(𝑎)𝑓(𝑐) < 0 then 𝑟 𝜖 (𝑎, 𝑐) or
ii. 𝑓(𝑏)𝑓(𝑐) < 0 then 𝑟 𝜖 (𝑐, 𝑏)
4 Choose the desire interval either (i) or (ii)
5 Repeated until desired iteration/ accuracy
Next, Table 2 and Table 3 are the calculation of n-th section method
of fourth section method and sixth section method, respectively.
Table 2: Algorithm of Fourth section method
No. Step
1 Identify two numbers a and b at which f has different signs.
2 𝑎+𝑏
Define midpoint, c = where 𝑛 = 4 and 𝑓(𝑐).
𝑛
Fig. 1 The schematic representation of bisection method 3 Determine if
i. 𝑓(𝑎)𝑓(𝑏) < 0 then 𝑟 𝜖 (𝑎, 𝑏)
ii. 𝑓(𝑏)𝑓(𝑐) < 0 then 𝑟 𝜖 (𝑏, 𝑐)
iii. 𝑓(𝑐)𝑓(𝑑) < 0 then 𝑟 𝜖 (𝑐, 𝑑)
4 Choose the desire interval either (i) or (ii) or (iii)
5 Repeated until desired iteration/ accuracy
Table 3: Algorithm of Sixth section method
No. Step
1 Identify two numbers a and b at which f has different signs.
2 𝑎+𝑏
Define midpoint, c = where 𝑛 = 6 and 𝑓(𝑐).
𝑛
3 Determine if
i. 𝑓(𝑎)𝑓(𝑏) < 0 then 𝑟 𝜖 (𝑎, 𝑏)
ii. 𝑓(𝑏)𝑓(𝑐) < 0 then 𝑟 𝜖 (𝑏, 𝑐)
iii. 𝑓(𝑐)𝑓(𝑑) < 0 then 𝑟 𝜖 (𝑐, 𝑑)
.
Fig. 2 The schematic representation of fourth section method .
.
𝑓(𝑛)𝑓(𝑛 + 1) < 0 then 𝑟 𝜖 (𝑛, 𝑛 + 1)
4 Choose the desire interval either (i) or (ii) or (iii) or infinity
5 Repeated until desired iteration/ accuracy
RESULTS AND DISCUSSION
The results are presented for several common functions such as
cubic polynomial, logarithm, exponential and trigonometric. This
function has been tested through Maple to check for the number of
iterations and CPU times based on Bisection method, Fourth Section
and Sixth Section. Numerical result based on the test run of both
methods using the standard test functions is listed below.
Fig. 3 The schematic representation of sixth section method
METHODOLOGY
A computer code programming is constructed by using Maple 12
software based on the Bisection method. The Maple 12 software is
chosen because the programming code is relatively simple and easy to
apply. This method is initialized with the limits of function on the
interval [a, b] where the function is defined. The function 𝑓(𝑥) is
assumed to be continuous on the interval [𝑎, 𝑏] is choosen such that
𝑓(𝑎)𝑓(𝑏) < 0. Thus, it is divided into two and more intervals. At each
𝑎+𝑏
interval, the midpoint of the interval 𝑐 = is computed and 𝑓(𝑐) is
2
evaluated. Lastly, the process is repeated until a desired accuracy is
achieved. Table 1 is the algorithm for Bisection method. Fig. 4 Graph function of 𝑓(𝑥) = 𝑠𝑖𝑛(𝑥) − 0.75
729
Mohd Ali et al. / Malaysian Journal of Fundamental and Applied Sciences Vol. 13, No. 4 (2017) 728-731
Table 5: Number of iterations for 𝑓(𝑥) = 𝑥 + 𝑙𝑛(𝑥) – 3
3 decimal 4 decimal 5 decimal
Bisection 10 12 17
Fourth Section 5 7 9
Sixth Section 3 5 6
Table 6: Number of iterations for 𝑓(𝑥) = 𝑥 3 − 𝑥 − 1
3 decimal 4 decimal 5 decimal
Bisection 10 13 17
Fourth Section 5 8 8
Sixth Section 4 5 5
Fig. 5 Graph function of 𝑓(𝑥) = 𝑥 + 𝑙𝑛(𝑥) – 3
Table 7: Number of iterations for 𝑓(𝑥) = 𝑒 𝑥 + 𝑥 − 1
3 decimal 4 decimal 5 decimal
Bisection 7 10 13
Fourth Section 5 7 7
Sixth Section 4 5 7
Based on the table 4, 5, 6 and 7, it shows that the number of iteration is
reduced when the number of interval increased. Therefore, the sixth and
fourth section possess lower number of iteration compared to Bisection.
Table 8: CPU times for 𝑓(𝑥) = 𝑠𝑖𝑛(𝑥) – 0.75
CPU times
Bisection 0.34375 0.390625 0.4375
Fig. 6 Graph function of 𝑓(𝑥) = 𝑥 3 − 𝑥 − 1 Fourth Section 0.640625 0.65625 0.65625
Sixth Section 1.328125 1.28125 1.359375
Table 9: CPU times for 𝑓(𝑥) = 𝑥 + 𝑙𝑛(𝑥) – 3
CPU times
Bisection 0.421875 0.437500 0.468750
Fourth Section 0.687500 0.640625 0.703125
Sixth Section 1.328125 1.281250 1.625000
Table 10: CPU times for 𝑓(𝑥) = 𝑥 3 − 𝑥 − 1
CPU times
Bisection 0.468750 0.390625 0.406250
Fourth Section 0.640625 0.656250 0.656250
Sixth Section 1.328125 1.421875 1.421875
𝑥
Fig. 7 Graph function of 𝑓(𝑥) = 𝑒 − 𝑥 − 1
Table 11: CPU times for f(x) = 𝑒 𝑥 + 𝑥 − 1
Table 4, Table 5, Table 5, and Table 7 are the numerical results
for number of iterations and Table 8, Table 9, Table 10 and Table 11 CPU times
are the CPU times. The decimal places indicate the tolerance of the Bisection 0.328125 0.359375 0.296875
roots for the current iteration and previous iterations.
Fourth Section 0.625000 0.578125 0.578125
Table 4: Number of iterations for 𝑓(𝑥) = 𝑠𝑖𝑛(𝑥) – 0.75 Sixth Section 1.265625 1.234375 1.234375
3 decimal 4 decimal 5 decimal
However, in term of CPU time, the CPU time increased
Bisection 5 9 12 significantly when the interval increased. Therefore, sixth and fourth
Fourth Section 2 5 6 section methods possessed higher CPU time compared to Bisection.
Sixth Section 2 6 5 The significant increased of CPU time is due to the division of interval
and determination of 𝑓(𝑎)𝑓(𝑏) < 0 in each interval.
730
Mohd Ali et al. / Malaysian Journal of Fundamental and Applied Sciences Vol. 13, No. 4 (2017) 728-731
CONCLUSION
In this research, a new modification of Bisection has been
proposed. This modification is based on even number of interval. The
new method is named as the Fourth section and Sixth section method.
If the number of iterations reduced, then the CPU times for Fourth
section and Sixth section are increased. Based on this research, if a
researcher is interested only with the method which possesses lower
number of iteration without concerning the CPU time, it is suggested
that the researcher use the Sixth section method as it converge faster
than Bisection and Fourth section method. On the other hand, if the
researcher is only interested on the CPU time without concerning the
number of iterations, it is suggested to use the Bisection method as it
possessed lower CPU time.
ACKNOWLEDGEMENT
This research is conducted under the fundamental research grant
scheme (FRGS) grant number : 600-RMI/FRGS 5/3 (64/2015).
REFERENCES
Burden, R. L. and J. D. Faires (1993). Numerical analysis (5th edition). Boston:
PWS Kent Pub. Co. Print.
Doron L. (2010). Introduction to numerical analysis. Departments of
Mathematics and Center of Scientific Computation and Mathematical
Modeling (CSCAMM) University of Maryland. Retrieved from
http://www2.math.umd.edu/~dlevy/books/na.pdf
Mathews, J. H. and K. D. Fink (2004). Numerical methods using MATLAB (4th
edition). Upper Saddle River, N.J.: Pearson. Print.
Park, B. and S. Hitotumatu (1987). A study on New Muller’s method.
Publications of the Research Institute for Mathematical Sciences. 23(4),
667–672. Retrieved from: https://www.ems-ph.org/journals/show_pdf.
php?issn=0034-5318&vol=23&iss=4&rank=5
Solanki, C., Thapliyal, P. and Tomar, K. (2014). Role of bisection method,
International Journal of Computer Applications Technology and
Research, 3(8), 533–535.
Yakoubsohn, J. (2005). Numerical analysis of a bisection-exclusion method to
find zeros of univariate analytic functions, Journal of Complexity, 21(5),
652–690.
731
