Improved TR
Improved TR
Improved TR
net/publication/356854078
CITATIONS READS
0 2,962
2 authors, including:
Qurban A Memon
United Arab Emirates University
121 PUBLICATIONS 464 CITATIONS
SEE PROFILE
All content following this page was uploaded by Qurban A Memon on 08 December 2021.
A. F Abdulhameed1, Q A Memon2
PhD Student1, Electrical Engineering, UAE University, Al Ain – UAE
Associate Professor2, Electrical Engineering, UAE University, Al Ain – UAE
[email protected], [email protected]
1. Introduction
Mathematical integration is a kind of summation, even the symbol used for integration is a capital ‘S’
stylized to show that integration and summation are close to each other [1]. For numerical data or
complex analytical solutions, the numerical methods, for example, for integration were developed to
solve problems for some practical engineering problems. Many numerical methods have been
developed to help make life easier in solving these complex problems. For integration, one of the most
common methods is the trapezoidal rule, which is a very common and simple method due to its certain
level of accuracy. Unfortunately, in most of the cases, the accuracy of this method is not achieved
except that the integration interval is divided into many small intervals, which may not be possible in
some cases, otherwise the result will be far from the actual solution. To achieve relatively better
accuracy for the smaller number of intervals, an improved numerical method is proposed in this paper
based on the trapezoidal rule concept, with slightly more complexity, but with much more accurate
results even with a single (whole integration) interval period. In the next section, literature review is
presented about this topic, followed by proposed method in section three, where error value is derived.
This section also contains experimental calculations conducted to show results compared with ordinary
trapezoidal rule. The conclusions are presented in section four.
2. Litrature review
Numerical methods topic is an essential subject, that all engineers study as part of their degree in
higher education, and the book [1] is one of the commonly used textbooks that are widely preferred in
teaching numerical methods to students in many universities. It covers several methods for integration
such as the Trapezoidal rule, Simpson rules, Romberg, etc. and, is considered a great tool for students
and engineers to learn, practice and solve engineering problems. In this paper, the equations and
examples related to the Trapezoidal rule are referenced to this book for comparative purposes.
In [2], the authors tried to solve the Runge phenomena for getting higher order accuracy compared to
the ones in the traditional quadrature methods which are limited to 4th order only. Their proposed
scheme is capable of utilizing the trapezoidal rule spectral accuracy over the interior of the intervals. In
another work [3], Trefethen and Weideman reviewed the trapezoidal rule both mathematically and
historically, and they have shown that it is linked with the computational methods all across scientific
computing. It was shown that it is linked to algorithms related to inverse Laplace transform, complex
analysis, special functions, integral equations, rational approximation, and the computation of
eigenvalues of matrices and operators. Similarly, the authors in [4] did a survey on the methods for
highly accurate numerical integration and reviewed many methods from 1980 till recent years, and
they highlighted some of the techniques that when combined with powerful integer detection methods
have resulted in the analytical evaluation of many “impossible” integrands before finding these
techniques.
As far as applications in engineering disciplines are concerned, the authors in [5], discuss a method to
improve the contrast of the images with low contrast. The method defines a novel contrast
enhancement algorithm, which is based on discrete integration method using Trapezoidal rule. The
work in [6] describes the design of half sample delay recursive digital integrators, in which a half
sample delay is applied on trapezoidal integration rule, after that, a modified finite impulse response
(FIR) fractional delay filter is used to design recursive digital integrators. That modified FIR fractional
delay filter is less complex but more efficient than original one, which leads to the design of lower
order recursive digital integrators that have been compared with the existing half sample delay and the
conventional recursive digital integrators. The results of the proposed method show effectiveness with
considerably lower percentage absolute magnitude relative error (PARE), and linear phase response
over almost 0% - 80% of the total Nyquist frequency range. For further reading related to numerical
integration applications, the reader is referred to [7-8].
The work in [9] discusses the classical composite trapezoidal rule for computing two dimensional
singular integrals, to obtain the convergence results O(h2) that is the same as the Riemann integral
convergence rate. The respective authors were able, at specific points of the classical composite
trapezoidal rule, to get the upper convergence phenomenon when the special function used in error
function equates to zero. The authors in [10] propose a new method derived from trapezoidal rule,
which is capable of computing fractional integrals and derivatives to within a given error tolerance.
In this paper, we present a new and simpler method that can, significantly, improve the ordinary
trapezoidal rule method. The proposed method is shown mathematically along with experimental results
on some engineering examples for comparative purposes.
3. Proposed method
In this section, we explain the method that we are proposing and solve some examples to show its
effectiveness, and later will compare the results with the conventional trapezoidal rule method results.
3.1 Concept of Proposed Method
Inspired by the trapezoidal rule, which is easy to understand, easy to implement method, but
unfortunately, the error value produced is substantial in some cases. In trapezoidal rule, the area under
the curve is replaced by the area under the line to simplify the solution. The line here is in between the
first point and the last point on the curve, known as “the integration limits”. This forms with the x-axis,
a half trapezoid and its area can be calculated using the rule:
Trapezoid Area = Base x Average Height (1)
In Figure 1, the f(a)-f(b) line form with the x axis for the area that we are interested in calculating. In
order to see the weakness of this method, we can observe that the line f(a)-f(b) falls under a big curve,
thus causes a significant error, as we will see in the examples later. On the other hand, the new method
attempts to reduce this error by dividing the quadrilateral located between the trapezoid line and the
parallel tangent line at the far edge of the curve (red line in Figure 1), into two and adding one half of
this rectangle (green line in Figure 1 below) to the calculated area “the integration value”.
Figure 1: Trapezoidal rule and new method
To do so mathematically, the method undergoes the following steps:
• Calculate the slope of the black line [f(a)-f(b)], using the formula:
𝑚 = [𝑓(𝑏) − 𝑓(𝑎)] / (𝑥𝑏 − 𝑥𝑎 ) (2)
• Find the derivative of the original function f(x), then make it equal to the slope ‘m’ obtained as above.
𝑓′ 𝑥 = 𝑚
This will give us the value of x at the tangency point.
• Now find the value of f(x) at the tangency point using the original equation. Once we get the slope
and a single point coordinates of the line, then we can find its equation, using the formula:
𝑓𝐼 𝑥 = 𝑚𝑥 + 𝑐
All are known except c. Once we get c value, we have the line equation that is parallel to the
trapezoid top line (red line in Figure 1).
• Now, we find f1(a) by putting the given value of “x = a”, in the new line equation we found above.
Then, we go back to the integration method. The trapezoidal method says:
𝐼 = 𝑏 − 𝑎 ∗ [𝑓 𝑎 + 𝑓 𝑏 ] /2 (3)
The new method will replace f(a) by the average of f(a) and f1(a), also average of f(b) and f1(b).
Thus, the new formula will be:
𝐼 = 𝑏 − 𝑎 ∗ [𝑓 𝑎 + 𝑓𝐼 𝑎 + 𝑓 𝑏 + 𝑓𝐼 (𝑏)] /4
(4)
And since
𝑓𝐼 𝑎 − 𝑓 𝑎 = 𝑓𝐼 𝑏 − 𝑓 𝑏 = 𝑘
because they are parrel lines, then we can also say:
𝑓 𝑎 +𝑓 𝑏 𝑓𝐼 𝑎 − 𝑓 𝑎
𝐼= 𝑏 − 𝑎 ∗ +
2 2
which also equals:
𝑓 𝑏 𝑓𝐼 𝑎
𝐼= 𝑏 − 𝑎 ∗ +
2 2 (5)
And this is the easiest formula to calculate the area under the green line as shown in Figure 2.
Figure 2: New method equivalent area under line
3.2 Error Estimation
To derive the error for the proposed method, same approach was used as performed in the calculation of
the trapezoidal rule error method. For trapezoidal rule, the error is:
Error = Area under the curve – Area under the trapezoid line.
Which is equal to:
𝑏 𝑏
(𝑓 𝑏 − 𝑓 𝑎 )
𝐸𝑒𝑟𝑟𝑜𝑟 = 𝑓(𝑥) 𝑑𝑥 − 𝑏 − 𝑎 𝑑𝑥
𝑎 𝑎
If we consider I as fixed at all times for the same function, then the new error will be:
1
𝐸𝑖𝑚𝑝𝑟𝑜𝑣𝑒𝑑 = − 𝑓 ′′ 𝜉 ℎ3 − ℎ ∗ 𝑘/2
12 (7)
4. Conclusions
We presented a new method for numerical integration that is based on trapezoidal rule. We have
shown with examples that the new method is much more accurate than the normal trapezoidal rule
when the function curve is above and far from the trapezoid line, but in some cases when the function
curve is equally oscillating around the trapezoid line, the normal trapezoidal rule gave better results,
though the new method was not far from the true solution. Despite that the new method involves little
more steps to implement as compared to the classical trapezoidal rule, we recommend that this method
to be practiced by engineering and mathematics students as it, in general, results in relatively higher
accuracy than traditional trapezoidal approach in solving problems. Likewise, the proposed approach
can also be applied in situations, where multiple segments are involved.
References
[1] Chapra, Steven C., and Raymond P. Canale. Numerical methods for engineers. 7th ed.,pp. 605-615, (2015).
[2] Fornberg, Bengt. "Improving the accuracy of the trapezoidal rule." SIAM Review 63.1 (2021): 167-180.
[3] Trefethen, Lloyd N., and J. A. C. Weideman. "The exponentially convergent trapezoidal rule." siam
REVIEW 56.3 (2014): 385-458.
[4] Bailey, David H., and Jonathan M. Borwein. "High-precision numerical integration: Progress and
challenges." Journal of Symbolic Computation 46.7 (2011): 741-754.
[5] Halder, Amiya, and Nikita Shah. "Contrast Enhancement Algorithm Using Definite Integration
Mathematical Method Trapezoidal Rule." Proceedings of International Conference on Frontiers in
Computing and Systems. Springer, Singapore, 2021.
[6] J. Madhu, M. Gupta, N. K. Jain. "Design of half sample delay recursive digital integrators using trapezoidal
integration rule."International Journal of Signal and Imaging Systems Engineering, 9.2 (2016): 126-134.
[7] Z. Ali, Q. Memon,” Time delay tracking for multiuser synchronization in CDMA networks,” Journal of
Networks, 8 (9), 1929-1935, 2013
[8] Q. Memon, Z. AlKassim, E. AlHassan, M. Omer, and M. Alsiddig, “Audio-Visual Biometric Authentication
for Secured Access into Personal Devices,” Proceedings of the 6th International Conference on
Bioinformatics and Biomedical Science, pp. 85–89, 2017
[9] Li, Jin, and Hongxing Rui. "Error expansion of trapezoidal rule for certain two-dimensional Cauchy
principal value integrals." Computers & Mathematics with Applications 74.10 (2017): 2608-2637.
[10] Harker, Matthew, and Paul O’Leary. "Trapezoidal rule and its error analysis for the Grünwald-Letnikov
operator." International Journal of Dynamics and Control 5.1 (2017): 18-29.