The International Journal of Engineering and Science (IJES)

|| Volume || 13 || Issue || 4 || Pages || PP 58-63 || 2024 ||

ISSN (e): 2319 – 1813 ISSN (p): 2319 – 1805

Numerical Methods for Solving First Order Ordinary

Differential Equations
M.U. UWAEZUOKE
Department of Mathematics , Faculty of physical sciences Imo State University. P.m.b 2000, Owerri, Nigeria.

--------------------------------------------------------ABSTRACT----------------------------------------------------------------
Abstract. In this paper, we use both Newton’s Interpolation and Lagrange Polynomial to create cubic
polynomials for solving the initial value problem. By this new method, it is simple to solve linear and nonlinear
first order ordinary differential equations and to yield and implement actual precise results.Some numerical
examples are provided to test the performance and illustrate the efficiency of the method.
Key Words and Phrases: Numerical method, Initial value problems, Newton’s interpolation, Lagrange
polynomial.
---------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: 02-04-2024 Date of acceptance: 13-04-2024
---------------------------------------------------------------------------------------------------------------------------------------

I. Introduction
Many mathematical models in science and engineering fields ([3],[4],[5],[6],[7],[10],[13]) can be
formulated in the form linear and nonlinear ordinary differential equations ([9],[12]) which need an analytical
method ([2],[15].[17]) to solve the exact equations. However in some problems, we can obtain the exact
solution by analytical method in [7] for example y′ = x2 + y2. Therefore the numerical method is an important
tool to solve this kind of problems such as Euler’s method, Runge-Kutta method ([11],[14]) and Runge-Kutta-
Fehlberg method ([8],[18]). Many methods have been widely developed by a lot of researchers ([1],[11],[14]
and [16]) to solve these problems. Some problems in form of partial differential equations can be converted to
ordinary differential equations form ([10],[13]). In this article, we consider only the first order ordinary
differential equations with an initial condition (initial value problems) in form:

(1)

Where is a known function and the value of initial conditions are known numbers. In 2018, to
find solutions of (1), [14] used Newton’s interpolation and three points to build a quadratic equation by
Lagrange method. Also [11] and [16] used Newton’s interpolation and constructed a quadratic equation by
Aitken’s method to solve the same problems. Moreover [20] gave an idea to solve these problems for combining
Pi-card’s method and Taylor’s Series. Furthermore [21] proposed a new solving technique by improved Euler’s
method.
Ultimately [1] applied all techniques from [11], [14], [16], [19], [20] and [21] to approximate the solution of (1)
and other systems of first order ordinary differential equations. The goal of this study is to estimate
approximated solutions and relative errors by comparing the results of our new method with other methods such
as Euler’s method and methods in [11] and [14].

II. Our proposed method

To improve the accuracy of in [11] and [14] which are the same quadratic solutions, we combine
Newton’s interpolation and Lagrange's method for solving (1) to create higher degree polynomial. Newton’s
interpolation is applied to find four values for = 0,1,2,3 to form a cubic polynomial,
. This polynomial approximates solutions of (1) by using Lagrange’s method as following.

2.1 Newton’s interpolation method

This method uses the initial point from (1) to estimate the values of a function from any
intermediate values of the independent variables. The general form of the degree polynomial that goes
through n+1 points (( is written as

DOI:10.9790/1813-13045863 www.theijes.com Page 58

 Numerical Methods For Solving First Order Ordinary Differential Equations

(2)

Where , , , , … are given by (3)

=
y1-y0 (4)
a1 =
x1-x0 ,
y2-y1 y1-y0 (5)
x2a-x1= x1-x0 ,
2
x2 - x0

y3-y2 - y2-y1 y2-y1 - y1-y0 (6)

x3-x2 - x2-x1 x2-x1 - x1-x0
x3-x1 x2-x0
a3 = ,….
x3-x0

Since y1, y2 and y3 in (4) - (6) are unknown values, we use differential values for approximation i.e
yi-yi - 1 𝑑𝑦 . .
xi-xi -1 ≈ 𝑑𝑥 | 𝑥𝑖 𝑦𝑖
For our method, we require three more values y1, y2 and y3 which are given by

y1 = a0 + a1(x1 - x0), (7)

y2 = a0 + a1(x2 - x0) + a2(x2 - x0)(x2 - x1),
y3 = a0 + a1(x3 - x0) + a2(x3 - x0)(x3 - x1) + a3(x3 - x0)(x3-x3), (8)
(9)
where xi+1 = xi+h and the step size h is very small constant.

2.2. Lagrange’s method

Lagrange’s method is a method to find a nth degree polynomial that takes on certain values at an
arbitrary point. We have only (x0,y0), (x1,y1),(x2,y2) and (x3,y3) to build the 3rd degree polynomial equation (cubic
function), P3(xi) =c0 + c1x +c2x2 + c3x3

(10)

Then we apply (10) to approximate yi = P3(xi) where i = 4,5, …8 which are solutions of (1)as shown in Figure 1

2.3. Algorithm of our method

# Consider IVP dy/dx = f(x,y) with initial conditions y(x0) = y0

1. Define function f(x,y)

DOI:10.9790/1813-13045863 www.theijes.com Page 59

 Numerical Methods For Solving First Order Ordinary Differential Equations

2. Set values of initial condition (x0, y0), number of steps (n) and steps size (h)
3. Use Newton’s interpolation method to find (x1, y1), (x2, y2) and (x3, y3)
4. Use Lagrange’s method to find P3(x)
5. Set i = 4
6. Loop while i <= n
yi = P3(xi)
xi = xi + h
i=i+1
7. Display yi as results

III. Numerical results

We will use our method to find the numerical solutions and relative errors and compare results with
Euler’s method, methods of [11] and [14] in the following examples.

Example 1. Consider the initial value problem

.
Then take the step size h = 0.1 and use Newton’s interpolation
a0 = 0.0, a1 = 1.0, a2 = -0.499999, a3 = 0.166667
and
y1 = 0.1, y2 = 0.19, y3 = 0.271.
Apply (0,0), (0.1, 0.1), (0.2, 0.19) and (0.3, 0.271) to find cubic polynomial by (10). Then we obtain

P3(x) = 0.166667x3 - 0.549997x2 + 1.053333x

(11)

In order to approximate the solutions, we substitute xi in P3(x) to get yi = P3(xi) for i = 4,5, …20 and compute
relative error, y -y
xi i
yxi

where yxi, is exact solutions at xi, as shown in Table 1 and Figure 2.

The approximate solutions are close to exact solutions where x and relative errors when x ≥ 0.9
highly increase as shown in Figure 2.

DOI:10.9790/1813-13045863 www.theijes.com Page 60

 Numerical Methods For Solving First Order Ordinary Differential Equations

Example 2. Consider the initial value problem

1.

Then take the step size h = 0.1 and use Newton’s interpolation
a0 =1.0, a1 = -1.0, a2 = 0.549999, a3 = 0.149999
and
y1 = 0.9, y2 = 0.811, y3 = 0.7339.

Apply (0, 1.0), (0.1,0.9), (0.2, 0.811) and (0.3, 0.7339) to find cubic polynomial by (10).
Then we obtain
P3(x) = 0.149999x3 + 0.505x2 - 1.052x +1 (12)

In order to approximate the solutions, we substitute xi in P3(x) to get yi = P3(xi) for i =4,5,…20 in (12) and
relative error as shown in Table 2 and Figure 3. The approximate solutions are close to exact solutions where
and relative errors when x ≥ 0.7 highly increase as shown in Figure 3.

DOI:10.9790/1813-13045863 www.theijes.com Page 61

 Numerical Methods For Solving First Order Ordinary Differential Equations

Example 3. Consider the initial value problem

.
Then take the step h = 0.1 and use Newton’s interpolation
a0 = 1.0, a1 = -0.367879, a2 = 0.348868, a3 = -0.129608
and
y1 = 0.963212, y2 = 0.933401, y3 = 0.909791.

Apply (0,0.1), (0.1, 0.963212), (0.2, 0.933401) and (0.3, 0.909791) to find cubic polynomial by (10).Then we
obtain
P3(x) = -0.129608x3 + 0.387751x2 - 0.405358x + 1 (13)

In order to approximate the solutions, we substitute xi in P3(x) to get yi = P3(xi) for i = 4,5, …20 in(13) and
relative error as shown in Table 3 and Figure 4. The approximate solutions are close to exact solutions where x
and relative error when x ≥ 0.9 highly increase as shown in Figure 4.

IV. Conclusions
In this article, we studied many numerical techniques for solving initial value problems. Then we
decided to combine the Newton’s interpolation and Lagrange’s method to construct cubic polynomials as the
solutions of linear and non-linear of ordinary differential equations. We compared our numerical results and
relative errors with the results of Euler’s method, methods from [11] and [14] and exact solutions. From
example 1-3, where we compare the result in other methods, our method gives numerical approximated
solutions which is much closer to the exact solution as shown in Table 1-3. Also our relative errors are nearly
close to zero and much smaller that errors from [11] and [14] methods are shown in Figure 2-4. Notice that to
compute each yi in each step of Euler’s method is so much time-consuming. Therefore, our method is much
more accurate and simpler to find approximated solutions directly from yi = P3(xi) at any value of xi. For the
future work, we aim to apply our method to solve other problems of first differential equations or higher order
differential equations.

DOI:10.9790/1813-13045863 www.theijes.com Page 62

 Numerical Methods For Solving First Order Ordinary Differential Equations

References
[1]. J. D. Abraha. Comparison of numerical methods for system of first order ordinary differential equations. Pure Appl. Math. J., 9:32-
36, 2020.
[2]. F. M. Al-Askar, C. Cesarano, and W. W. Mohammed. The analytical solutions of stochastic-fractional drinfel'd-sokolov-wilson
equations via (g'/g)-expansion method. Symmetry, 14(10), 2022.
[3]. M. Alkasassbeh, Z. Omar, F. Mebarek-Oudina, J. Raza, and A. Chamkha. Heat transfer study of convective fin with temperature-
dependent internal heat generation by hybrid block method. Heat Transfer, 48(4):1225-1244, 2019.
[4]. O. A. Arqub. The reproducing kernel algorithm for handling differential algebraic systems of ordinary differential equations. Math.
Methods in the App. Sci., 39(15):4549-4562, 2016.
[5]. O. A. Arqub, H. O. Alsulami, and M. Alhodaly. Numerical hilbert space solution of fractional sobolev equation in (1+1)-
dimensional space. Math. Sct., 16, 2022.
[6]. O. A. Arqub and B.'Maayah. Adaptive the dirichlet model of mobile/immobile advection/dispersion in a time-fractional sense with
the reproducing kernel computational approach: Formulations and approximations. Int. J. Modern Physics B, 37, 2022.
[7]. H. Badawi, N. Shawagfeh, and O. A. Arqub. Fractional conformable stochastic inte-grodifferential equations: existence,
uniqueness, and numerical simulations utilizing the shifted legendre spectral collocation algorithm. Math. Problems in Engineering,
2022,
[8]. R. Djebali, F. Mebarek-Oudina, and C. Rajashekhar. Similarity solution analysis of dynamic and thermal boundary layers: further
formulation along a vertical flat plate. Physica Scripta, 96(8), 2021.
[9]. M. Farhan, Z. Omar, F. Mebarek-Oudina, Z. Shah J. Raza, R. V Choudhari, and O. D. Makinde. Implementation of the one-step
one-hybrid block method on the equation of a circular sector oscillator. Comput. Math. Model, 31: 116-132, 2020.
[10]. M. Hassan, F. Mebarek-Oudina, A. Faisal, A. Ghafar, and A. I. Ismail. Thermal energy and mass transport of shear thinning fluid
under effects of low to high shear rate viscosity. Int. J. Thermofluids, 15, 2022.
[11]. N. D. Ide. Using newton's interpolation and aitken's method for solving first order differential equation. World Appl. Sci. J., 38:191-
194, 2022.
[12]. A. Ishkhanyan and C. Cesarano. Generalized-hypergeometric solutions of the general fuchsian linear ode having five regular
singularities. Axioms, 8(3): 102, 2019.
[13]. U. Khan, F. Mebarek-Oudina, A. Zaib, A. Ishak, S. A. Bakar, E. M. Sherif, and D. Baleanu. An exact solution of a casson fluid
flow induced by dust particles with hybrid nanofluid over a stretching sheet subject to lorentz forces. Waves in Random and
Complex Media, 32, 2022.
[14]. F. C. Kosgei. Solution of first order differential equation using numerical newton's interpolation and Lagrange method. Int. J.
Dev. Res., 8: 18973-18976, 2018.
[15]. A. Lupica, C. Cesarano, F. Crisanti, and A. Ishkhanyan. Analytical solution of the three-dimensional laplace equation in terms of
linear combinations of hypergeometric functions. Mathematics, 9(24), 2021.
[16]. J. C. Mbagwu and N. D. Ide. Comparison of newton's interpolation and aitken's methods with some numerical methods for solving
system of first and second order differential equation. World Sci. News, 164: 108-121, 2022.
[17]. W. W. Mohammed, F. M. Al-Askar, C. Cesarano, T. Botmart, and M. El-Morshedy. Wiener process effects on the solutions of the
fractional (2 + 1)-dimensional heisenberg ferromagnetic spin chain equation. Mathematics, 10(12), 2022.
[18]. J. Raza, F. Mebarek-Oudina, and L. Ali Lund. The flow of magnetized convective casson liquid via a porous channel with shrinking
and stationary walls. Pramana, 96, 2022.
[19]. A. Setia, Y.Liu, and A. S. Vatsala. Numerical solutions of fractional integro-differential equations with nonlocal boundary
conditions. J. Frac. Cal. Appl., 8:155-165, 2014.
[20]. I. Singh and G. Singh. Numerical solutions of first order ordinary differential equations. Int. J. Sci. Res. Rev., 8:593-602, 2019.
[21]. S. Tapaswini and S. Chakraverty. A new Approach to fuzzy initial value problem by improved euler method. Fuzzy Information
and Engineering, 4:293-312, 2012

DOI:10.9790/1813-13045863 www.theijes.com Page 63