Research Article Journal of Applied & Computational Mathematics

Volume 11:8, 2022

ISSN: 2168-9679 Open Access

Approximate Solution to First-Order Integro-differential Equa-

tions Using Polynomial Collocation Approach
Ganiyu Ajileye* and F.A. Aminu
Department of Mathematics and Statistics, Federal University Wukari, Taraba, Nigeria

Abstract
In this study, power series and shifted Chebyshev polynomials are used as basis function for solving first order volterra integro-differential equations
using standard collocation method. An assumed approximate solution in terms of the constructed polynomial was substituted into the class of
integro-differential equation considered. The resulted equation was collocated at appropriate points within the interval of consideration [0,1] to
obtain a system of algebraic linear equations. Solving the system of equations, by inverse multiplication, the unknown coefficients involved in the
equations are obtained. The required approximate results are obtained when the values of the constant coefficients are substituted back into the
assumed approximate solution. Numerical example are presented to confirm the accuracy and efficiency of the method.
Keywords: Integro-differential equations • Collocation • Power series polynomial • Shifted chebyshev polynomial

Definition 1: Standard Collocation Method: This is the method partitioning

Introduction intervals into different points. For example, let say an interval [a,b], N be the
number of collocation points to get the desired collocation points, we use;
An Integro-differential equation is any type of equation combining an
unknown function involving both integral and derivative operators. It is b−a
xp =
a+ p, where p =
1, 2, … N
applicable in several fields, such as engineering and science. These equations N
are either Fredholm, Volterra, or Fredholm-Volterra integro-differential Definition 2: Power Series Polynomial: A power series is basically an
equations. As a result, a numerical method for solving such equations is infinite degree polynomial that represents some function. It is generally written
required. Some numerical approaches in literature for solving integro-differential ∞
in the form; y ( x) = ∑a n x n Where an is the coefficient of the polynomials which
equations include: Collocation method [1-5], Hybrid linear multistep method n =0

[6-8], Differential transform method [9], Perturbed Method [10], Homotopy are real numbers.
Perturbation [11], Bernoulli matrix method [12], The Mellin transform approach Definition 3: Shifted Chebyshev Polynomials: This is a special type of
[13] and Pseudospectral Method [14]. Shahooth et al. presented a numerical polynomial generated from Chebyshev polynomial. The shifted Chebyshev has
method for solving the linear Volterra-Fredholm integro-differential equations of interval [0,1], while Chebyshev has an interval of [-1,1]. The shifted Chebyshev
the second kind [15]. This method is called the Bernestein polynomial method. is generated using;
This technique transforms the integro-differential equations into the system
of algebraic equations. Some numerical results are presented to illustrate the Tn * ( x ) = Tn ( 2 x − 1) , n ≥ 0
efficiency and accuracy of this method. where Tn (x) is the shifted chebyshev term,while Tn (x) is the chebyshev
In this work, we consider Volterra Integro-differential equation of the form: term.
x Few terms of the shifted Chebyshev terms are;
′( x)
y= f ( x ) + λ ∫k ( x, t ) y ( t ) dt
0 T0*(x)=1
Subject to initial condition; T1*(x)=2x-1
T2*(x)=8x2-8x+1
( 0 ) a=
y j= j, j 0,1
T3*(x)=32x3-48x2+18x-1
Where y j = d j y / dy j y(x) is the unknown function, f(x)is a known
T4*(x)=128x4-256x3+160x2+32x+1
function, λ is a known constant and k(x,t) is the integral kernel function.
Basic Definitions
Methodology
This section defines the basic terms that will be frequently used during the
research work. In this section, combination of collocation method and power series
approximation is employed for the numerical solution of volterra integro-
*Address for Correspondence: Ganiyu Ajileye, Department of Mathematics and dfferential equations.
Statistics, Federal University Wukari, Taraba, Nigeria, Email: [email protected] Let the solution to (1) and (2) be approximated by (3)
Copyright: © 2022 Ajileye G, et al. This is an open-access article distributed N
under the terms of the Creative Commons Attribution License, which permits= y ( x ) ∑an x n , where n ≥ 0
unrestricted use, distribution, and reproduction in any medium, provided the n =0

original author and source are credited. Where an is the real coefficients to be determined, differenting (3) with
Date of Submission: 04 August, 2022, Manuscript No. jacm-22-73430; Editor respect to x, gives
N
assigned: 06 August, 2022, PreQC No. P-73430; Reviewed: 12 August, 2022, QC
=
No. Q-73430; Revised: 18 August, 2022, Manuscript No. R-73430; Published: 24 y ' ( x )
n
∑
nan x n −1 , where n ≥ 0
August, 2022, DOI: 10.37421/ 2168-9679.2022.11.486
 Ajileye G, et al. J Appl Computat Math, Volume 11:8, 2022

substituting (3) and (4) into (1), gives steps using shifted Chebyshev polynomial as the basis polynomial function, we
obtained the approximate solution.
N ( n −1)  N x n 
∑ n ( )  ∑ë ∫ant k ( x, t ) dt 
na x = f x + Numerical Examples: In this section, numerical examples with initial
 conditions are solved to confirm the efficiency and accuracy of the method. Let
 n   n =0 a  y_n (x) and y(x) be the approximate and exact solution respectively. Error_N
simplifying (5) gives, = |y_n (x)-y(x)|.
Example 1: Considering Volterra integro-differential equation
N
 ( n −1) x

∑ an  nx − ë ∫ t n
k ( x, t ) dt =f ( x)
y′ ( x ) + y ( x ) = f ( x ) − ∫ty ( t ) dt
x

n =0  a 
0
Hence,

N Where f ( x) = (x 2
+ 2 x + 1) e − x + 5 x + 8
∑a τ ( x ) = f ( x )
n =0
n n subject to initial condition y(0)=10
Exact solution: y(x)=10-xe-x
Where
Solution 1: Solving at N = 5 using power series polynomial. Substituting
 n −1 x
 approximate solution (3) into example 1 gives
 nx − ë ∫ t n
k ( x, t ) dt =τ n ( x)
   5 ( n −1) 
5
 5 x n +1 
∑ n  ∑ n ( )  ∑ ∫ant dt 
a n
na x + a x =f x −
Equation (6) can be written in the form = n 0=  n0 = n 0 0 
Collecting the like terms gives
τ n ( x) A = f ( x)
Where 5
 ( n −1) x

∑ an  nx + x n
+ ∫ t n +1dt  =f ( x)
[ a0 a1 … an ]
τ 0 ( x ) τ 1 ( x ) … τ n ( x )  , A =  
T
τn = n =0 0

Collocating equation (7) using the standard collocation points Equation (12) can be written in the form

τ ( x) A = f ( x)
b−a
Where x p= a + p, p= 0,1, 2 …….N , gives Where
N
5
 x

τ (=
x) ∑ nx(
n −1)
[ a0 a1 … an ]
T
+ x n + ∫t n +1dt  =
,A
n =0  0 

 1 2 3 4 
collocating at x5= 0 1 and substituting the initial
τ ( xp ) A = f ( xp ) conditions gives  5 5 5 5 

τ * ( xp ) A = f * ( xp )
f ( x p )  f ( x0 )
T
= f ( x1 ) … f ( xN )  Solving equation (14), we obtained the following approximate solution

y5 =9.9999999937 − 0.9999952318 x + 0.9987005900 x 2 − 0.4921341016 x 3 + 0.1487825625 x 4 − 0.0232281987 x 5

Considering the initial condition,
Using the above steps for solving shifted Chebyshev polynomial as the
basis polynomial function, we obtained the approximate solution;
y ( j)
( a =)
j a j ,=
j 0,1, 2 ….n − 1 y5 =9.9999999820 − 0.9999901422 x + 0.9976550264 x 2 − 0.4861629152 x3 + 0.1707549056 x 4 − 0.0425249792 x5

Example 2: Considering Volterra integro-differential equation

Substituting (3) into (9) we have, x
y′ ( x ) + y=
( x ) f ( x ) + ∫et − x y ( t ) dt
0

d ( j)  N 
j ∑ n
a xn  = a j Where f(x)=0
dx  n =0  subject to initial condition y(0)=1
Exact solution: y ( x ) = e cosh ( x )
−x
Using equation(10) and equation (8) gives
Solution 2: Solving at N = 5 using power series polynomial. Substituting
τ * ( xp ) A = f * ( xp ) approximate solution (3) into example 2 gives
Solving equation (11) using inverse multiplication to get the values of the  5 ( n −1) 
5
 5  x

unknown coefficient an, n=1, 2, …N and substituting the value of the coefficients  na
 n 0=
n ∑
x  +
 n0
an x n
=f ∑
( x ) +  an et − x t n dt  ∑∫
into the approximate solution gives the numerical results. Repeating the above= = n 0 0 

Page 2 of 4
Ajileye G, et al. J Appl Computat Math, Volume 11:8, 2022

Collecting the like terms gives 0.8 0.600948259 0.600915006 0.583318222

0.9 0.582649444 0.582637242 0.559142197
5
 ( n −1) 
x

∑a n  nx + x n − ∫et − x t n dt  =f ( x) 1 0.567667642 0.567675589 0.537414142

n =0  0 
Table 4. Absolute error of example 2.
Equation (15) can be written in the form
Xi Power SeriesN=5 Shifted ChebyshevN=5
τ ( x) A = f ( x) 0 6.24E-08 8.43E-08
0.1 2.42E-05 8.38E-05
Where 0.2 4.35E-05 4.65E-04
0.3 3.51E-05 1.34E-03
0.4 1.60E-05 2.89E-03
5
 x

τ (=
x) ∑ nx(
n −1)
[ a0 a1 … an ]
T
+ x n − ∫et − x t n dt  =
,A 0.5 7.14E-06 5.25E-03
n =0  0  0.6 1.56E-05 8.48E-03
0.7 3.08E-05 1.26E-02
 1 2 3 4  0.8 3.33E-05 1.76E-02
collocating at x5 = 0 1 and substituting the initial
conditions gives  5 5 5 5  0.9 1.22E-05 2.35E-02
1 7.95E-06 3.03E-02

τ * ( xp ) A = f * ( xp )
y5 = 0.9999999376 − 0.9999822616 + 0.9949659319 x 2 − 0.6356387354 x3 + 0.2600574315 x 4 − 0.0517267155 x5
Solving equation (17), we obtained the following approximate solution
(Tables 1-4).
Conclusion
Table 1: Exact and approximate values of example 1.
The numerical solution of the Volterra integro-differential equation is
Xi Exact Solution Power Seriesn=5 Shifted Chebyshevn=5
presented in this study and the results are compared to the exact solution.
0 10 9.999999994 9.999999982 Power series and shifted Chebyshev polynomials are used as the basis
0 9.909516258 9.909509989 9.909508005 functions.
0 9.836253849 9.836242518 9.836278452
0 9.777754534 9.77774555 9.777945271 The numerical results of the examples as shown in tables shows that the
0 9.731871982 9.731868389 9.732450172
results of power series approximation converges to the exact solution than the
results of shifted chebyshev approximation. Observation also shows that as
1 9.69673467 9.696733792 9.69799158
the value of increases i.e. (from 0 to 1), the results of the shifted chebyshev
1 9.670713018 9.670710096 9.67297361
diverges from the exact solutions.
1 9.652390287 9.652383354 9.655955045
1 9.640536829 9.640529449 9.645598296
1 9.634087306 9.634086223 9.640618379 Conflict of Interest
1 9.632120559 9.632125613 9.639731878
None.
Table 2. Absolute error for example 1.

Xi Power SeriesN=5 Shifted Chebyshevn=5 References

0 6.00E-09 1.70E-08
1. Olumuyiwa, A. Agbolade and Timothy A. Anak. "Solution of first order Volterra linear
0 6.27E-06 8.25E-06
integro-differential equations by collocation method." J Appl Math 1510267 (2017).
0 1.13E-05 2.46E-05
0 8.98E-06 1.91E-04 2. Adesanya, A. O., Yahaya Y. A., Ahmed B. and Onsachi R. O. "Numerical Solution
of Linear Integral and Integro-Differential Equations Using Boubakar Collocation
0 3.59E-06 5.78E-04
Method." Intern J Math Sci Optim Theory Appl 2019 (2019): 592-598.
1 8.78E-07 1.26E-03
1 2.92E-06 2.26E-03 3. Gegele, O. A., Evans O. P. and Akoh D. "Numerical Solution of Higher Order Linear
Fredholm–Integro–Differential Equations." Am J Eng Res 3 (2014): 243-247.
1 6.93E-06 3.56E-03
1 7.38E-06 5.06E-03 4. Nemati, S., Pedro Lima and Yadollah Ordokhani. "Numerical method for the mixed
1 1.08E-06 6.53E-03 Volterra-Fredholm integral equations using hybrid Legendre functions." Appl Math
2015 (2015): 184-193.
1 5.05E-06 7.61E-03
5. Abubakar, A. and Taiwo O. A. "Integral collocation approximation methods for
the numerical solution of high-orders linear Fredholm-Volterra integro-differential
Table 3. Exact and approximate values of example 2.
equations." J Comput Appl Math 4 (2014): 111-117.
Xi Exact Solution Power Seriesn=5 Shifted Chebyshevn=5 6. Ajileye, G., Amoo S. A. and Ogwumu O. D. "Hybrid Block Method Algorithms for
0 1 0.999999938 0.999999916 Solution of first order initial value problems in ordinary differential equations." J Appl
0.1 0.909365377 0.90934122 0.909281555 Computat Math 7 (2018): 2.
0.2 0.835160023 0.835116552 0.834695109 7. Mehdiyeva, Galina, Mehriban Imanova and Vagif Ibrahimov. "On the Construction
0.3 0.774405818 0.774370716 0.773063213 of the Multistep Methods to Solving the Initial-Value Problem for ODE and the
0.4 0.724664482 0.724648492 0.72176967 Volterra Integro-Differential Equations."
0.5 0.68393972 0.683932578 0.678688419 8. Ajileye, G., Okedayo G. T. and Aboiyar T. “Laguerre Collocation Approach for
0.6 0.650597106 0.650581523 0.642112506 Continuous Hybrid Block Method for Solving Initial Value Problems of Second
0.7 0.623298482 0.623267655 0.610683052 Order Ordinary Differential Equations.” Trends Food Sci Techn 2 (2017): 379- 383.

Page 3 of 4
Ajileye G, et al. J Appl Computat Math, Volume 11:8, 2022

9. Celik, Ercan and Khatereh Tabatabaei. "Solving a class of Volterra integral equation 14. El-Kady, M. and Biomy M. "Efficient Legendre pseudospectral method for solving
systems by the differential transform method." Int J Nonlinear Sci 16 (2013): 87-91. integral and integro-differential equations." Commun Nonlinear Sci Numer Simul 15
(2010): 1724-1739.
10. Uwaheren, O. A., Adebisi A. F. and Taiwo O. A. "Perturbed collocation method for
solving singular multi-order fractional differential equations of Lane-Emden type." J 15. Shahooth, M. K., Ahmad R., Salman U. and Swidan W., et al. "Approximation
Niger Soc Phys Sci (2020): 141-148. solution of solving linear Volterra Fredholm integro differential equations of the
second kind by using Berstein polynomial method." J Appl Computat Math 5 (2016):
11. Raftari, Behrouz. "Numerical solutions of the linear Volterra integro-differential 298.
equations: Homotopy perturbation method and finite difference method."
World Appl Sci J 9 (2010).
12. Bhrawy, Ali H., Emran Tohidi and Fazlollah Soleymani. "A new Bernoulli matrix
method for solving high-order linear and nonlinear Fredholm integro-differential
equations with piecewise intervals." Appl Math Comput 219 (2012): 482-497. How to cite this article: Ajileye, Ganiyu and Aminu F.A. “Approximate Solution
13. Fadugba, Sunday Emmanuel. "Solution of fractional order equations in the domain to First-Order Integro-differential Equations Using Polynomial Collocation
of the Mellin transform." J Niger Soc Phys Sci (2019): 138-142. Approach.” J Appl Computat Math 11 (2022): 486.

Page 4 of 4