International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025

Print ISSN: 2053-2229 (Print),

Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK

A Two-Point Off-Grid Colocation Point for the

Direct Solution of Fourth Order Ordinary
Differential Equations
M. O.Alabi
Department of Physical Sciences, Chrisland University, Abeokuta, Ogun State.

M. T. Raji
Department of Mathematics, Federal University of Agriculture, Abeokuta, Ogun State

M. A.Kehinde
Department of Mathematics, Federal College of Education (Special), Oyo, Oyo State.

doi: https://doi.org/10.37745/ijmss.13/vol13n25374 Published June 25, 2025

Citation: Alabi M.O., Raji M.T., Kehinde M.A (2025) A Two-Point Off-Grid Colocation Point for the Direct Solution
of Fourth Order Ordinary Differential Equations, International Journal of Mathematics and Statistics Studies, 13 (2),
53-74

Abstract: The desire to find solutions to differential equations cannot be over emphasized based
on the importance of such equations. Many people have developed different initial value solvers
to handle various differential equations based on the order of the differential equations. In solving
differential equations of order greater than one, it is often the practiced to resolve such a
differential equation into system of first order ordinary differential equations and then an
appropriate method is applied. Also, in some cases the analytical solutions to some of the
differential equations are intractable, hence there is need to circumvent this hurdle, and this is
done by the introduction of approximate solution otherwise referred to as Numerical solution. This
presentation focuses on derivation and implementation of a direct method to solve directly the
fourth order ordinary differential equations by interpolating at some selected grid points and
collocating at both grid and off grid points. Also, the derived method shall be applied to solve
some fourth order ordinary differential equations to compare the level of accuracy of the method
with the analytical solution.
Key Words: grid point, collocation, interpolation, error constant, convergency, consistency

INTRODUCTION

The analytical solution to some differential equations may be highly intractable, then an
approximate solution is hereby sought for. Many of the approximate methods are either one step
53
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
method such as Runge – Kutta method, Eulers Method, Heuns Method, Picard iterative method
among others. Lack of accuracy of single step methods gave birth to Multistep methods which are
in the form of Predictor – Corrector mode. Some of the methods are Simpson’s methods, Adams
Method (Adams Moulton and Admas Bashford method) to mention just a few. [8, 13]

In terms of the level of accuracy of the results generated using predictor corrector methods, there
comes the introduction of Block method. One of the major setbacks of the predictor corrector
method is that the method depends on another information for starting values thereby altering the
accuracy of the results. The new method proposed named Block method has the advantage of self-
starting thereby improving upon the output of the result. [8,13]

This paper presents the derivation and application of Block Linear Multistep method for the direct
solution of fourth order initial value problems of the form

𝑦′′′′(𝑥) = 𝑓(𝑥, 𝑦(𝑥), 𝑦 ′ (𝑥), 𝑦′′(𝑥), 𝑦′′′(𝑥))

(1)

together with

𝑦(𝑥0 ) = 𝑦0 , 𝑦 ′ (𝑥0 ) = 𝑦1 , 𝑦 ′′ (𝑥0 ) = 𝑦2 , 𝑦 ′′′ (𝑥0 ) = 𝑦3 (2)

The method is based on interpolating the polynomial equation at some selected grid points while
collocation was done at some selected grid points and two off grid points. Many authors have
worked extensively on Linear Multistep Method to solve differential equations in which most of
them are Predictor – Corrector in nature which is an improvement over and above the known single
step methods [1, 2, 5, 6, 14]. These methods have some level of better accurate results over and
above any known single step method, but they equally suffer some setback in terms of accuracy
because they depend on single step methods in determining their starting value, also it is often
laborious to develop an appropriate corrector method.

To circumvent this hurdle of developing an appropriate predictor – corrector method, researchers

came up with an alternate method that has a better level of accuracy, and this method is given the

54
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
name Block Linear Multi- step method. This method has an advantage of self-starting which does
not depend on additional information in getting its starting value as against the Predictor –
Corrector method. Among researchers that have worked extensively on this method include [3, 7,
10, 11, 12] to mention but a few.

METHODOLOGY

In the course of developing initial value solvers for Ordinary Differential Equations, many
authors make use of power series method of the form

𝑦(𝑥) = ∑∞
𝑟=0 𝑎𝑟 𝑥
𝑟
(3)

where 𝑎𝑟 is an arbitrary constant, while some used either perturbation method, Canonical
polynomials, and some make use of Legendry polynomial or Hermite Polynomial as basic
functions. [1, 14].

In this presentation, Chebyshev polynomial is hereby used as basis function to derive the proposed
Linear Multistep method. The choice of Chebyshev polynomial is because it is the most accurate
monomial among all other monomials within the interval [−1, 1] in which some authors have used
this polynomial as basic functions in deriving initial value solvers for ordinary differential
equations. [4, 9].

Here, the linear multistep method

∑𝑘𝑟=0 𝛼𝑟 𝑦𝑛+𝑟 = ℎ𝑛 ∑𝑘𝑟=0 𝛽𝑟 𝑓𝑛+𝑟 + 𝛽𝑗 𝑓𝑛+𝑗 + 𝛽𝑣 𝑓𝑛+𝑣 (4)

is hereby proposed where 𝑛 is the order of the differential equation, both 𝛼 𝑎𝑛𝑑 𝛽 are arbitrary
constants and not necessary 𝛼𝑟 = 0, ℎ is the step length while 𝑗 𝑎𝑛𝑑 𝑣 are non-integer collocation
points. It is important to note that the smaller the value of ℎ, the better the accuracy of the method
[13].

DERIVATION OF THE METHOD

Consider the polynomial equation

55
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
𝑦(𝑥) = ∑𝑘𝑛=1 𝑎𝑛 𝑇𝑛 (𝑥) (5)

in which 𝑎𝑛 is an arbitrary constant and 𝑇𝑛 (𝑥) is the Chebyshev polynomials. Note that the
Chebyshev polynomial can be generated recessively using the relation

𝑇𝑛+1 = 2𝑥 𝑇𝑛 (𝑥) − 𝑇𝑛−1 (𝑥), 𝑛 = 0(1)𝑘 (6)

where 𝑇0 (𝑥) = 1 𝑎𝑛𝑑 𝑇1 (𝑥) = 𝑥 (Fox and Parker).

Here, a four step Block Linear Multistep method is hereby proposed with two off grid
collocation points. To achieve this, interpolation was done at four grid points of regular interval
𝑥 = 𝑥𝑛+𝑟 , 𝑟 = 0(1)3, and collocation was done at five grid points and two off grid points 𝑥𝑛+𝑟 ,
𝑟 = 0(1)4 , 𝑥𝑛+ 1, 𝑎𝑛𝑑 𝑥𝑛 + 3
2 2

In order to achieve this, equations (5) and (6) were made use of by letting 𝑘 = 10 which leads to
the polynomial equation

𝑦(𝑥) = 𝑎0+ 𝑎1 𝑥 + 𝑎2 (𝑥 2 − 1) + 𝑎3 (4𝑥3 − 3𝑥) + 𝑎4 (8𝑥 4 − 8𝑥 2 + 1) +

𝑎5 (16𝑥 5 − 20𝑥 3 + 5𝑥) + 𝑎6 (32𝑥 6 − 48𝑥 4 + 18𝑥 2 − 1) +

𝑎7 (64𝑥 7 − 112𝑥 5 + 56𝑥 3 − 7𝑥) +

𝑎8 (128𝑥 8 − 256𝑥 6 + 160𝑥 4 − 32𝑥1 + 1)

(7)

𝑎9 (256𝑥 9 − 576𝑥 7 + 432𝑥 5 − 120𝑥 3 + 9𝑥) +

𝑎10 (512𝑥10 − 1280𝑥 8 + 1120𝑥 6 − 400𝑥 4 + 50𝑥 2 − 1)

Using the shifted Chebyshev polynomial in which

2𝑥−𝑏−𝑎 𝑥−𝑘ℎ−2ℎ
𝑥= , 𝑥𝑛 ≤ 𝑥 ≤ 𝑥𝑛+4 , 𝑡ℎ𝑒𝑟𝑒 𝑏𝑦 𝑥 = , hence equation (7) becomes
𝑏−𝑎 2ℎ

𝑥−𝑘ℎ−2ℎ 𝑥−𝑘ℎ−2ℎ 2
𝑦(𝑥) = 𝑎0 + 𝑎1 ( ) + 𝑎2 {( ) − 1}
2ℎ 2ℎ

56
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
𝑥−𝑘ℎ−2ℎ 3 𝑥− 𝑘ℎ−2ℎ 𝑥−𝑘ℎ−2ℎ4 𝑥−𝑘ℎ−2ℎ4
𝑎3 {4 ( ) − 3( )} + 𝑎4 {8 ( )− 8( ) + 1} +
2ℎ 2ℎ 2ℎ 2ℎ

𝑥−𝑘ℎ−2ℎ 4 𝑥−𝑘ℎ−2ℎ 2 𝑥−𝑘ℎ−2ℎ

𝑎5 {16 ( ) − 20 ( ) + ( )} +
2ℎ 2ℎ 2ℎ

𝑥−𝑘ℎ−2ℎ 6 𝑥−𝑘ℎ−2ℎ 4 𝑥−𝑘ℎ−2ℎ 2

𝑎6 {32 ( ) − 48 ( ) + 18 ( ) − 1} +
2ℎ 2ℎ 2ℎ

𝑥−𝑘ℎ−2ℎ 7 𝑥−𝑘ℎ−2ℎ 5 𝑥−𝑘ℎ−2ℎ 3 𝑥−𝑘ℎ−2ℎ

𝑎7 {64 ( 2ℎ
) − 112 ( 2ℎ
) + 56 ( 2ℎ
) − 7( 2ℎ
)} +

𝑥−𝑘ℎ−2ℎ 8 𝑥−𝑘ℎ−2ℎ 6 𝑥−𝑘ℎ−2ℎ 4 𝑥−𝑘ℎ−2ℎ

𝑎8 {128 ( ) − 256 ( ) + 160 ( ) − 32 ( ) + 1}
2ℎ 2ℎ 2ℎ 2ℎ

𝑥−𝑘ℎ−2ℎ 9 𝑥−𝑘ℎ−2ℎ 7 𝑥−𝑘ℎ −2ℎ 5 𝑥−𝑘ℎ−2ℎ 3

𝑎9 {256 ( ) − 576 ( ) + 432 ( ) − 120 ( ) +
2ℎ 2ℎ 2ℎ 2ℎ

𝑥−𝑘ℎ−2ℎ
9( )} +
2ℎ

𝑥−𝑘ℎ−2ℎ 10 𝑥−𝑘ℎ−2ℎ 8 𝑥−𝑘ℎ−2ℎ 6

𝑎10 {512 ( ) − 1280 ( ) + 1120 ( ) −
2ℎ 2ℎ 2ℎ

𝑥−𝑘ℎ−2ℎ 4 𝑥−𝑘ℎ−2ℎ 2
400 ( ) + 50 ( ) − 1}
2ℎ 2ℎ

The fourth derivative of 𝑦(𝑥) yields

12𝑎4 120 𝑥−𝑘ℎ−2ℎ 72𝑎6 𝑥−𝑘ℎ−2ℎ 2

𝑦 𝑖𝑣 (𝑥) = + 𝑎5 { ℎ 4 ( )} + {10 ( ) −}+
ℎ4 2ℎ ℎ4 2ℎ

840𝑎7 𝑥−𝑘ℎ−2ℎ 3 𝑥−𝑘ℎ−2ℎ 240𝑎8 𝑥−𝑘ℎ−2ℎ 4 𝑥−𝑘ℎ−2ℎ2

{4 ( ) − }+ {56 ( ) − 24 ( ) + 1}
ℎ4 2ℎ 2ℎ ℎ4 2ℎ 2ℎ

216𝑎9 𝑥−𝑘ℎ−2ℎ 5 𝑥−𝑘ℎ−2ℎ 3 𝑥−𝑘ℎ−2ℎ

+ {224 ( ) − 140 ( ) + 15 ( )} +
ℎ4 2ℎ 2ℎ 2ℎ

120𝑎10 𝑥−𝑘ℎ−2ℎ 6 𝑥−𝑘ℎ−2ℎ 4 𝑥−𝑘ℎ−2ℎ 2

{1344 ( ) − 1120 ( ) + 210 ( ) − 5}
ℎ4 2ℎ 2ℎ 2ℎ

Interpolating 𝑦(𝑥) at 𝑥 = 𝑥𝑛+𝑘 , 𝑘 = 0(1)3 and collocating equation 𝑦 𝑖𝑣 (𝑥) at

57
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
𝑥 = 𝑥𝑛+𝑘 , 𝑘 = 0 (1)4 𝑎𝑛𝑑 𝑎𝑡 𝑥 = 𝑥𝑛+ 1, 𝑥𝑛+ 3 yields the linear equation in matrix form 𝐴𝑥 =
2 2

𝑏, in which

𝑥 = (𝑎0 𝑎1 𝑎2 𝑎3 𝑎4 𝑎5 𝑎6 𝑎7 𝑎8 𝑎9 𝑎10 )𝑇

𝑏
𝑇
= (𝑦𝑛 2𝑦𝑛+1 𝑦𝑛+2 2𝑦𝑛+3 ℎ4 𝑓𝑛 8ℎ4 𝑓𝑛+ 1 ℎ4 𝑓𝑛+1 8ℎ4 𝑓𝑛+ 3 ℎ4 𝑓𝑛+2 ℎ4 𝑓𝑛+3 ℎ4 𝑓𝑛+4 )
2 2

and

A = 1 -1 1 -1 1 -1 1 -1 1 -1 1

2 -1 -1 2 -1 -1 2 -1 -1 2 -1

1 0 -1 0 1 0 -1 0 1 0
-1

2 1 -1 -2 -1 1 2 1 -1 -2
-1

0 0 0 0 12 -120 648 -2520 7920 -21384

51480

0 0 0 0 96 -720 2664 -6300 10020 -9234 -1965

0 0 0 0 12 -60 108 0 - 360 648 -180

0 0 0 0 96 -240 -216 2520 -1080 -3078

3915

0 0 0 0 12 0 - 72 0 240 0 -
600

0 0 0 0 12 60 108 0 - 360 -648 -180

58
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
0 0 0 0 12 -120 648 -2520 7920 -21384
51480

Solving the matrix equation leads to

1
𝑎0 = {6048000𝑦𝑛+1 − 6048000𝑦𝑛+2 + 6048000𝑦𝑛+3 + 10571ℎ4 𝑓𝑛
6048000
− 76160ℎ4 𝑓𝑛+ 1 + 302172ℎ4 𝑓𝑛+1 − 106624ℎ4 𝑓𝑛+ 3 + 722378ℎ4 𝑓𝑛+2
2 2

+ 155564ℎ4 𝑓𝑛+3 + 99ℎ4 𝑓𝑛+4 }

1
𝑎1 = {−56448000𝑦𝑛 + 84672000𝑦𝑛+1 − 169344000𝑦𝑛+2 + 141120000𝑦𝑛+3
84672000
+ 284333ℎ4 𝑓𝑛 + 8817956ℎ4 𝑓𝑛+1 − 1782400ℎ4 𝑓𝑛+ 1 − 2452352ℎ4 𝑓𝑛+ 3
2 2

+ 19281094ℎ4 𝑓𝑛+2 + 4075092ℎ4 𝑓𝑛+3 + 277ℎ4 𝑓𝑛+4 }

1
𝑎2 = {1512000𝑦𝑛+1 − 3024000𝑦𝑛+2 + 1512000𝑦𝑛+3 + 3803ℎ4 𝑓𝑛
1512000
− 26880ℎ4 𝑓𝑛+ 1 + 111796ℎ4 𝑓𝑛+1 − 37632ℎ4 𝑓𝑛+ 3 + 266754ℎ4 𝑓𝑛+2
2 2

+ 60052ℎ4 𝑓𝑛+3 + 107ℎ4 𝑓𝑛+4 }

1
𝑎3 = {−16934400𝑦𝑛 + 50803200𝑦𝑛+1 − 50803200𝑦𝑛+2 + 16934400𝑦𝑛+3
50803200
+ 80311ℎ4 𝑓𝑛 − 534400ℎ4 𝑓𝑛+ 1 + 2442412ℎ4 𝑓𝑛+1 − 733824𝑓𝑛+ 3
2 2

+ 5783218𝑓𝑛+2 + 1424444ℎ4 𝑓𝑛+3 + 5039ℎ4 𝑓𝑛+4 }

1
𝑎4 = {109ℎ4 𝑓𝑛 − 640ℎ4 𝑓𝑛+ 1 + 3988ℎ4 𝑓𝑛+1 − 896ℎ4 𝑓𝑛+ 3 + 8662ℎ4 𝑓𝑛+2
168000 2 2

+ 2756ℎ4 𝑓𝑛+3 + 21ℎ4 𝑓𝑛+4 }

59
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
1
𝑎5 = {−511ℎ4 𝑓𝑛 − 3200ℎ4 𝑓𝑛+ 1 − 25452ℎ4 𝑓𝑛+1 − 18816ℎ4 𝑓𝑛+ 3 + 11102ℎ4 𝑓𝑛+2
4704000 2 2

+ 36036ℎ4 𝑓𝑛+3 + 841ℎ4 𝑓𝑛+4 }

1
𝑎6 = {−1071ℎ4 𝑓𝑛 + 12160ℎ4 𝑓𝑛+ 1 − 9772ℎ4 𝑓𝑛+1 + 17024ℎ4 𝑓𝑛+ 3
6048000 2 2

− 32578ℎ4 𝑓𝑛+2 + 13626ℎ4 𝑓𝑛+3 + 601ℎ4 𝑓𝑛+4 }

1
𝑎7 = {−609ℎ4 𝑓𝑛 + 3200ℎ4 𝑓𝑛+ 1 − 11508ℎ4 𝑓𝑛+1 + 18816ℎ4 𝑓𝑛+ 3 − 11102ℎ4 𝑓𝑛+2
5644800 2 2

+ 924ℎ4 𝑓𝑛+3 + 279ℎ4 𝑓𝑛+4 }

1
𝑎8 = {−49ℎ4 𝑓𝑛 + 640ℎ4 𝑓𝑛+ 1 − 1428ℎ4 𝑓𝑛+1 + 896ℎ4 𝑓𝑛+ 3 + 98ℎ4 𝑓𝑛+2
1209600 2 2

− 196ℎ4 𝑓𝑛+3 + 39ℎ4 𝑓𝑛+4 }

1
𝑎9 = {−511ℎ4 𝑓𝑛 − 3200ℎ4 𝑓𝑛+ 1 + 13748ℎ4 𝑓𝑛+1 − 18816ℎ4 𝑓𝑛+ 3
50803200 2 2

+ 11102ℎ4 𝑓𝑛+2 − 3164ℎ4 𝑓𝑛+3 + 841ℎ4 𝑓𝑛+4 }

1
𝑎10 = {109ℎ4 𝑓𝑛 − 64ℎ4 𝑓𝑛+ 1 + 119ℎ4 𝑓𝑛+1 − 896ℎ4 𝑓𝑛+ 3 + 262ℎ4 𝑓𝑛+2 − 44ℎ4 𝑓𝑛+3
6048000 2 2

+ 21ℎ4 𝑓𝑛+4 }

Substituting the value of 𝑎′𝑠 into 𝑦(𝑥) to generate the continuous scheme thus

60
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
𝑦(𝑥)
(−338688000𝑥 3 + 84672000𝑥)𝑦𝑛 +

(1016064000𝑥 3 + 508032000𝑥 2 − 508032000𝑥)𝑦𝑛+1 +

(−1016064000𝑥 3 − 1016064000𝑥 2 + 254016000𝑥 + 254016000)𝑦𝑛+2 +

(338688000𝑥 3 + 508032000𝑥 2 + 169344000𝑥)𝑦𝑛+3 +

10 9
(2343936𝑥 − 654080𝑥 − 7176960𝑥 8 + 3225600𝑥 7 + 6322176𝑥 6 −) ℎ4 𝑓 +
𝑛
2407104𝑥 + 3999380𝑥 3 − 292152𝑥 2 − 699867𝑥
5

−1376256𝑥10 − 4096000𝑥 9 + 20643840𝑥 8 + 18432000𝑥 7 −

( 53760000𝑥 6 − 26496000𝑥 5 − 9676800𝑥 4 + 2752000𝑥 3 + ) ℎ4 𝑓𝑛+ 1 +
2
3467520𝑥 2 + 652800𝑥 − 24192

1 2558976𝑥10 + 17597440𝑥 9 − 44782080𝑥 8 − 72737280𝑥 7 +

= (69233472𝑥 6 + 109686528𝑥 5 + 17954200𝑥 4 − 15888880𝑥 3 ) ℎ4 𝑓𝑛+1 +
254016000
−8216964𝑥 2 + 933408𝑥 + 44898

−19267584𝑥10 − 24084480𝑥 9 + 72253440𝑥 8 + 59607360𝑥 7

( −67436544𝑥 6 − 151732224𝑥 5 + 64350720𝑥 3 + ) ℎ4 𝑓𝑛+ 3 +
2
3161088𝑥 2 − 8203776𝑥

5634048𝑥10 + 14210560𝑥 9 − 11450880𝑥 8 − 63947520𝑥 7 −

(36728832𝑥 6 + 89526528𝑥 5 + 169344000𝑥 4 + 69035960𝑥 3 −) ℎ4 𝑓𝑛+2 +
39883536𝑥 2 − 21910728𝑥

−946176𝑥10 − 4049920𝑥 9 − 2903040𝑥 8 + 11773440𝑥 7 +

( 26780544𝑥 6 + 19643904𝑥 5 + 20160𝑥 4 − 6203120𝑥 3 − ) ℎ4 𝑓𝑛+3 +
−1633128𝑥 2 + 154896𝑥 + 420

( 451584𝑥10 + 1076480𝑥 9 − 80640𝑥 8 − 1627200𝑥 7 − ) ℎ4 𝑓𝑛+4

{ 301056𝑥 6 + 1137024𝑥 5 − 609020𝑥 3 + 18312𝑥 2 + 102276𝑥 }

Further simplification yields

1
𝛼0 (𝑡) = (𝑡 − 4𝑡 3 ) , 𝛼1 (𝑡) = 4𝑡 3 + 2𝑡 2 − 2𝑡, 𝛼2 (𝑡) = 3 + 3𝑡 − 4𝑡 2 − 4𝑡 3 and
3

1
𝛼3 (𝑡) = (2𝑡 + 6𝑡 2 + 4𝑡 3 ),
3

61
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
10
𝛽0 (𝑡) =
1
(93𝑡 − 26𝑡 9 − 285𝑡 8 + 128𝑡 7 + 215𝑡 6 − 96𝑡 5 − 89𝑡 4 + 159𝑡 3 −)
10080 12𝑡 2 − 28𝑡

𝛽1 (𝑡) =
1
(−55𝑡10 − 163𝑡 9 + 819𝑡 8 + 731𝑡 7 − 2133𝑡 6 − 1051𝑡 5 − 384𝑡 4 )
2 10080 109𝑡 3 + 1380𝑡 2 + 108𝑡

𝛽1 (𝑡) =
1
(102𝑡10 + 697𝑡 9 − 1777𝑡 8 − 2886𝑡 7 + 2747𝑡 6 + 4353𝑡 5 + )
10080 712𝑡 4 − 631𝑡 3 − 433𝑡 2 + 37𝑡 + 2

𝛽3 (𝑡) =
1
(−765𝑡10 − 956𝑡 9 + 2867𝑡 8 + 2365𝑡 7 − 2676𝑡 6 − 6021𝑡 5 )
2 10080 + 1937𝑡 4 + 2554𝑡 3 + 125𝑡 2 − 326𝑡
10
𝛽2 (𝑡) =
1
(224𝑡 + 564𝑡 9 − 454𝑡 8 − 2538𝑡 7 − 1458𝑡 6 + 3553𝑡 5 )
10080 + 6720𝑡 4 + 2740𝑡 3 − 1584𝑡 2 − 869𝑡
10
𝛽3 (𝑡) =
1
(−38𝑡 − 161𝑡 9 − 151𝑡 8 + 467𝑡 7 + 1063𝑡 6 + 780𝑡 5 +)
10080 𝑡 4 − 246𝑡 3 − 65𝑡 2 + 6𝑡
1
𝛽4 (𝑡) = (18𝑡10 + 43𝑡 9 − 3𝑡 8 − 65𝑡 7 − 12𝑡 6 + 45𝑡 5 − 24𝑡 3 + 𝑡 2 + 4𝑡)
10080

𝑥−𝑘ℎ−2ℎ
Let 𝑡 = , and evaluating 𝑥 𝑎𝑡 𝑥𝑛+4 , this makes 𝑡 = 1, then this leads to the discrete
𝑘ℎ

scheme of the form

ℎ4
𝑦𝑛+4 − 4𝑦𝑛+3 + 6𝑦𝑛+2 − 4𝑦𝑛+1 + 𝑦𝑛 = (7𝑓𝑛+4 + 1692𝑓𝑛+3 + 𝑓𝑛+2 −
10080

896 𝑓𝑛+ 3 + 2924𝑓𝑛+1 − 640𝑓𝑛+ 1 + 95𝑓𝑛 ) (8)

2 2

The first derivative of the continuous scheme yields

1 1
𝛼′0 (𝑡) = (−6𝑡 2 + 2)
3ℎ

1
𝛼 ′1 (𝑡) = (6𝑡 2 + 2𝑡 − 1)
ℎ

1
𝛼 ′ 2 (𝑡) = (−12𝑡 2 − 8𝑡 + 3)
2ℎ

1
𝛼 ′ 3 (𝑡) = (6𝑡 2 + 6𝑡 + 1)
3ℎ

62
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
1
𝛽 ′ 0 (𝑡) = (465𝑡 9 − 117𝑡 8 − 1140𝑡 7 + 448𝑡 6 + 753𝑡 5 − 240𝑡 4 − 178𝑡 3 +
10080ℎ

238.5𝑡 2 − 12𝑡 − 14)

1
𝛽 ′ 1 (𝑡) = (−275𝑡 9 − 733.5𝑡 8 + 3276𝑡 7 + 2558.5𝑡 6 − 6399𝑡 5 − 2627.5𝑡 4 −
2 10080ℎ

768𝑡 3 + 163.5𝑡 2 + 1380𝑡 + 54)

1
𝛽 ′1 (𝑡) = (510𝑡 9 + 3141𝑡 8 − 7108𝑡 7 − 10101𝑡 6 + 8241𝑡 5 + 10882.5𝑡 4 +
10080ℎ

1424𝑡 3 − 946.5𝑡 2 − 433𝑡 + 18.5)

1
𝛽 ′ 3 (𝑡) = (−3825𝑡 9 − 4302𝑡 8 + 11468𝑡 7 + 8277.5𝑡 6 − 8028𝑡 5 −
2 10080ℎ

15052.5𝑡 4 + 3874𝑡 3 + 3831𝑡 2 + 125𝑡 − 163)

1
𝛽 ′ 2 (𝑡) = (1120𝑡 9 + 2538𝑡 8 − 1816𝑡 7 − 8883𝑡 6 − 4374𝑡 5 + 8882.5𝑡 4 +
10080ℎ

13440𝑡 3 + 4110𝑡 2 − 1584𝑡 − 434.5)

1
𝛽 ′ 3 (𝑡) = (−190𝑡 9 − 724.5𝑡 8 − 460𝑡 7 + 1634.5𝑡 6 + 3189𝑡 5 + 1950𝑡 4 +
10080ℎ

2𝑡 3 − 369𝑡 2 − 65𝑡 + 3)

1
𝛽 ′ 4 (𝑡) = (90𝑡 9 + 193.5𝑡 8 − 12𝑡 7 − 227.5𝑡 6 − 36𝑡 5 + 112.5𝑡 4 − 36𝑡 2 + 𝑡 +
10080ℎ

2)

The second derivative of the continuous scheme yields

−2𝑡 1 1 1
𝛼 ′′ 0 (𝑡) = , 𝛼 ′′1 (𝑡) = (6𝑡 + 1), 𝛼 ′′ 2 (𝑡) = (−6𝑡 − 2), 𝛼 ′′ 3 (𝑡) = (2𝑡 + 3)
ℎ2 ℎ2 ℎ2 ℎ2

1
𝛽 ′′ 0 (𝑡) = (2092.5𝑡 8 − 468𝑡 7 − 3990𝑡 6 + 1344𝑡 5 + 1882.5𝑡 4 − 480𝑡 3 −
10080ℎ2

267𝑡 2 + 238.5𝑡 − 6)

1
𝛽 ′′ 1 (𝑡) = (−1237.5𝑡 8 − 2934𝑡 7 + 11466𝑡 6 + 7675.5𝑡 5 − 15997.5𝑡 4 −
2 10080ℎ2

5255𝑡 3 − 1152𝑡 2 + 163.5𝑡 + 690)

63
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
1
𝛽 ′′1 (𝑡) = (2295𝑡 8 + 12564𝑡 7 − 24878𝑡 6 − 30303𝑡 5 + 20602.5𝑡 4 +
10080ℎ2

21765𝑡 3 + 2136𝑡 2 − 946.5𝑡 − 216.5)

1
𝛽 ′′ 3 (𝑡) = (−17212.5𝑡 8 − 17208𝑡 7 + 40138𝑡 6 + 24832.5𝑡 5 − 30105𝑡 4 −
2 10080ℎ2

30105𝑡 3 + 5811𝑡 2 + 3831𝑡 + 62.5)

1
𝛽 ′′ 2 (𝑡) = (5040𝑡 8 + 10152𝑡 7 − 6356𝑡 6 − 26649𝑡 5 − 10935𝑡 4 + 17765𝑡 3 +
10080ℎ2

20160𝑡 2 + 4110𝑡 − 792)

1
𝛽 ′′ 3 (𝑡) = (−855𝑡 8 − 2898𝑡 7 − 1610𝑡 6 + 4903.5𝑡 5 + 7972.5𝑡 4 + 3900𝑡 3 +
10080ℎ2

3𝑡 2 − 369𝑡 − 32.5)

1
𝛽 ′′ 4 (𝑡) = (405𝑡 8 + 774𝑡 7 − 42𝑡 6 − 682.5𝑡 5 − 90𝑡 4 + 225𝑡 3 − 36𝑡 + 0.5)
10080ℎ2

The third derivative of the continuous scheme leads to

−1 3 −3 1
𝛼 ′′′ 0 (𝑡) = , 𝛼 ′′′1 (𝑡) = , 𝛼 ′′′ 2 (𝑡) = , 𝛼 ′′′ 3 (𝑡) =
ℎ3 ℎ3 ℎ3 ℎ3

1
𝛽 ′′′ 0 (𝑡) = (8370𝑡 7 − 1638𝑡 6 − 11970𝑡 5 + 3360𝑡 4 + 3765𝑡 3 − 720𝑡 2 −
10080ℎ3

267𝑡 + 119.25)

1
𝛽 ′′′ 1 (𝑡) = (−4950𝑡 7 − 10269𝑡 6 + 34398𝑡 5 + 19188.75𝑡 4 − 31995𝑡 3 −
2 10080ℎ3

7882.5𝑡 2 − 1152𝑡 + 81.75)

1
𝛽 ′′′1 (𝑡) = (9180𝑡 7 + 43974𝑡 6 − 74634𝑡 5 − 75757.5𝑡 4 + 41205𝑡 3 +
10080ℎ3

32647.5𝑡 2 + 2136𝑡 − 473.25)

1
𝛽 ′′′ 3 (𝑡) = (−68850𝑡 7 − 60228𝑡 6 + 120414𝑡 5 + 62081.25𝑡 4 − 40140𝑡 3 −
2 10080ℎ3

45157.5𝑡 2 + 5811𝑡 + 1915.5)

64
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
1
𝛽 ′′′ 2 (𝑡) = (20160𝑡 7 + 35532𝑡 6 − 19068𝑡 5 − 66622.5𝑡 4 − 21870𝑡 3 +
10080ℎ3

26647.5𝑡 2 + 20160𝑡 + 2055)

1
𝛽 ′′′ 3 (𝑡) = (−3420𝑡 7 − 10143𝑡 6 − 4830𝑡 5 + 12258.75𝑡 4 + 15945𝑡 3 +
10080ℎ3

5850𝑡 2 + 3𝑡 − 184.5)

1
𝛽 ′′′ 4 (𝑡) = (1620𝑡 7 + 2709𝑡 6 − 126𝑡 5 − 1706.25𝑡 4 − 180𝑡 3 + 337.5𝑡 2 − 18)
10080ℎ3

Evaluating the first derivative of the continues scheme at 𝑥𝑛+𝑘 , 𝑘 = 0(1)4 leads to the following
discrete schemes.

11 1 1 ℎ3
ℎ𝑦′𝑛 + 𝑦𝑛 − 3𝑦𝑛+1 + 𝑦 − 𝑦 = (427.5𝑓𝑛 + 2201𝑓𝑛+ 1 +
6 2 𝑛+2 3 𝑛+3 10080 2

360.5𝑓𝑛+1 − 11032𝑓𝑛+ 3 − 573𝑓𝑛+2 + 18𝑓𝑛+3 + 1.5𝑓𝑛+4 )

2

1 1 1 ℎ3
ℎ𝑦′𝑛+1 + 𝑦 + 2 𝑦𝑛+1 − 2𝑦𝑛+2 + 𝑦 = (50𝑓𝑛 − 451𝑓𝑛+ 1 + 152𝑓𝑛+1 −
3 𝑛 6 𝑛+3 10080 2

411𝑓𝑛+ 3 + 280𝑓𝑛+2 − 15𝑓𝑛+3 − 2𝑓𝑛+4 )

2

1 3 1 ℎ3
ℎ𝑦′𝑛+2 − 𝑦 + 𝑦𝑛+1 − 𝑦 − 𝑦 = (−14𝑓𝑛 + 54𝑓𝑛+ 1 + 18.5𝑓𝑛+1 −
6 𝑛 2 𝑛+2 3 𝑛+3 10080 2

163𝑓𝑛+ 3 + 434.5𝑓𝑛+2 + 3𝑓𝑛+3 + 2𝑓𝑛+4 )

2

1 3 11 ℎ3
ℎ𝑦′𝑛+3 + 𝑦 − 2 𝑦𝑛+1 + 2𝑦𝑛+2 − 𝑦𝑛+3 = (24𝑓𝑛 + 387𝑓𝑛+ 1 + 481𝑓𝑛+1 +
3 𝑛 6 10080 2

344𝑓𝑛+ 3 + 1759𝑓𝑛+2 + 119𝑓𝑛+3 − 3𝑓𝑛+4 )

2

11 17 13 ℎ3
ℎ𝑦′𝑛+4 + 𝑦𝑛 − 7𝑦𝑛+1 + 𝑦𝑛+2 − 𝑦𝑛+3 = (203.5𝑓𝑛 + 3371𝑓𝑛+ 1 +
6 2 3 10080 2

5628𝑓𝑛+1 − 3795𝑓𝑛+ 3 + 12999𝑓𝑛+2 + 4970𝑓𝑛+3 + 87.5𝑓𝑛+4 )

2

65
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
In like manner, the evaluation of the second derivative of the continuous scheme at
𝑥 = 𝑥𝑛+𝑘 , 𝑘 = 0(1)4 yields

ℎ2
ℎ2 𝑦′′𝑛 − 2𝑦𝑛 + 5𝑦𝑛+1 − 4𝑦𝑛+2 + 𝑦𝑛+3 = (−922.5𝑓𝑛 − 5881𝑓𝑛+ 1 −
10080 2

3140.5𝑓𝑛+1 + 27378.5𝑓𝑛+3 + 1739𝑓𝑛+2 − 58.5𝑓𝑛+3 − 7𝑓𝑛+4 )

2

ℎ2
ℎ2 𝑦′′𝑛+1 − 𝑦𝑛 + 2𝑦𝑛+1 − 𝑦𝑛+2 = 10080
(−107𝑓𝑛 + 98𝑓𝑛+ 1 + 21𝑓𝑛+1 + 2027𝑓𝑛+3 −
2 2

37𝑓𝑛+2 + 4𝑓𝑛+3 + 𝑓𝑛+4 )

ℎ2
ℎ2 𝑦′′𝑛+2 − 𝑦𝑛+1 + 2𝑦𝑛+2 − 𝑦𝑛+3 = (−6𝑓𝑛 + 690𝑓𝑛+ 1 + 216.5𝑓𝑛+1 +
10080 2

62.5𝑓𝑛+3 + 792𝑓𝑛+2 − 32.5𝑓𝑛+3 + 0.5𝑓𝑛+4 )

2

ℎ2
ℎ2 𝑦′′𝑛+3 + 𝑦𝑛 − 4𝑦𝑛+1 + 5𝑦𝑛+2 − 2 𝑦𝑛+3 = (96𝑓𝑛 − 782𝑓𝑛+ 1 + 2624𝑓𝑛+1 −
10080 2

385𝑓𝑛+3 + 7007𝑓𝑛+2 + 872𝑓𝑛+3 − 9𝑓𝑛+4 )

2

ℎ2
ℎ2 𝑦′′𝑛+4 + 2𝑦𝑛 − 7𝑦𝑛+1 + 8𝑦𝑛+2 − 3 𝑦𝑛+3 = (346.5𝑓𝑛 − 6581𝑓𝑛+ 1 +
10080 2

3018.5𝑓𝑛+1 − 9920.5𝑓𝑛+3 + 12495𝑓𝑛+2 + 11014.5𝑓𝑛+3 + 554𝑓𝑛+4 )

2

So also, the evaluation of the third derivative of the continuous scheme at 𝑥=

𝑥𝑛+𝑘 , 𝑘 = 0(1)4 result to the following schemes

ℎ
ℎ3 𝑦′′′𝑛 + 𝑦𝑛 − 3𝑦𝑛+1 + 3𝑦𝑛+2 − 𝑦𝑛+3 = (1223𝑓𝑛 + 4818𝑓𝑛+ 1 + 22504𝑓𝑛+1 −
10080 2

58624𝑓𝑛+ 3 − 39906𝑓𝑛+2 + 83𝑓𝑛+3 + 8𝑓𝑛+4 )

2

66
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
ℎ
ℎ3 𝑦′′′𝑛+1 + 𝑦𝑛 − 3𝑦𝑛+1 + 3𝑦𝑛+2 − 𝑦𝑛+3 = (95𝑓𝑛 + 2689𝑓𝑛+ 1 − 317𝑓𝑛+1 −
10080 2

7548𝑓𝑛+ 3 − 1800𝑓𝑛+2 + 69𝑓𝑛+3 + 16𝑓𝑛+4 )

2

ℎ
ℎ3 𝑦′′′𝑛+2 + 𝑦𝑛 − 3𝑦𝑛+1 + 3𝑦𝑛+2 − 𝑦𝑛+3 = (119.25𝑓𝑛 + 81.75𝑓𝑛+ 1 −
10080 2

473.25𝑓𝑛+1 + 1915.5𝑓𝑛+ 3 + 2055𝑓𝑛+2 − 184.5𝑓𝑛+3 − 18𝑓𝑛+4 )

2

ℎ
ℎ3 𝑦′′′𝑛+3 + 𝑦𝑛 − 3𝑦𝑛+1 + 3𝑦𝑛+2 − 𝑦𝑛+3 = (152𝑓𝑛 − 4389𝑓𝑛+ 1 + 7599𝑓𝑛+1 −
10080 2

5322𝑓𝑛+ 3 + 12016𝑓𝑛+2 + 3703𝑓𝑛+3 − 12𝑓𝑛+4 )

2

ℎ
ℎ3 𝑦′′′𝑛+4 + 𝑦𝑛 − 3𝑦𝑛+1 + 3𝑦𝑛+2 − 𝑦𝑛+3 = (1019.25𝑓𝑛 − 2580𝑓𝑛+ 1 −
10080 2

21722.25𝑓𝑛+1 − 24153.75𝑓𝑛+ 3 − 3006𝑓𝑛+2 + 15479.25𝑓𝑛+3 + 2636.25𝑓𝑛+4 )

2

ANALYSIS OF THE METHOD

At this juncture, some basic analysis of the method shall be carried out such as order, consistency
and error constant of the derived scheme as well as to test for zero stability of the derived
method.

Theorem 1: The necessary and sufficient condition for a Linear Multistep method to converge is
that it must be both consistent and zero stable. [13]

It is important to state that consistency controls the magnitude of the local truncation error while
zero stability concerns the manner of error committed at each stage of computation. So also, for a
Linear Multistep method to be consistent, it must have order 𝑝 ≥ 1 while a Linear Multistep
method is said to be zero stable if no root of the first characteristics polynomial 𝜌(𝑟) has
modulus greater than one and if every root with modulus one is simple.[13]

67
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
The Linear Multistep method is said to be of order 𝑝 if 𝐶0 = 𝐶1 = 𝐶2 . . . 𝐶𝑝+3 = 0 while
𝐶𝑝+4 ≠ 0 and 𝐶𝑝+4 is referred to as error constant. To determine the order and error constant of
the derived method, the method hereunder is applied.

𝐶0 = ∑𝑘𝑗=0 𝛼𝑗 , 𝐶1 = ∑𝑘𝑗=1 𝑗𝛼𝑗 , 𝐶2 = ∑𝑘𝑗=1 𝑗 2 𝛼𝑗 , 𝐶3 = ∑𝑘𝑗=1 𝑗 3 𝛼𝑗

1
𝐶4 = ∑𝑘𝑗=1 𝑗 4 𝛼𝑗 − ∑𝑘𝑗=0 𝛽𝑗 − 𝛽1 − 𝛽3 and
4! 2 2

1 1 1 𝑞−4 3 𝑞−4
𝐶𝑞 = ∑𝑘𝑗=1 𝑗 𝑞 𝛼𝑗 − ∑𝑘 𝑗 𝑞−4 𝛽𝑗 + (2) 𝛽1 + (2) 𝛽3 , 𝑞 = 5(1)𝑘
𝑞! (𝑞−4)! 𝑗=1 2 2

Applying the above method, it was discovered that the derived method is of order 8 in which
𝐶12 = −2.09983𝐸10− 3

Also to determine the zero stability of the method, the roots of the first characteristics
polynomial of the derived method were sought for, that is

𝑟 4 − 4𝑟 3 + 6𝑟 2 − 4𝑟 + 1 = 0 (9)

Solving equation (9) yields 𝑟 = 1 four times and this shows that the method is both consistent
and zero stable, hence the method satisfies the condition for convergence. At this juncture, some
numerical experiments shall be carried out using the derived method to solve some fourth order
ordinary differential equations and the output shall be compared with the analytical solution to
see the level of convergency of the method derived.

SOME NUMERICAL COMPUTATIONS

Illustration I. Determine the solution to the differential equation

𝑦 𝑖𝑣 − 5𝑦 ′′ + 4𝑦 = 0: 𝑦(0) = 3, 𝑦 ′ (0) = 5, 𝑦 ′′ (0) = −2, 𝑦 ′′′ (0) = 4

The analytical solution is

𝑦(𝑥) = 9.5𝑒 𝑥 − 2.5𝑒 −𝑥 − 2.5𝑒 2𝑥 − 1.5𝑒 −2𝑥

The numerical computation to Illustration I is as shown in Table 1 below

68
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
Table 1: Numerical Computation to Illustration I

x y- Exact y- Computed Absolute Error

0 3.0000000000 3.0000000000 0.0000000000
0.1 3.9554271520 3.9554267810 3.70611345E – 7
0.2 4.8214575070 4.8214579123 4.05629981E – 7
0.3 5.5930986650 5.5930986817 1.67000000E – 8
0.4 6.2586887450 6.2586894524 7.07400000E – 7
0.5 6.7990016890 6.7990019135 2.24035153E – 7
0.6 7.1860158886 7.1860158992 1.04349130E – 8
0.7 7.3812920978 7.3812929607 8.62233099E – 7
0.8 7.3338905724 7.3338991064 8.533994393E – 6
0.9 6.9777384132 6.9777319206 6.492674771E – 6
1.0 6.2283355947 6.2283359043 3.09049219E – 7

Illustration II: Solve completely the differential equation

𝑦 𝑖𝑣 (𝑥) − 4𝑦 ′′′ (𝑥) + 5𝑦 ′′ (𝑥) − 4𝑦 ′ (𝑥) + 4𝑦(𝑥) = 0

Subject to

𝑦(0) = 4, 𝑦 ′ (0) = 6, 𝑦 ′′ (0) = 8 𝑎𝑛𝑑 𝑦 ′′′ (0) = −6

The theoretical solution is

𝑦(𝑥) = 6.24𝑒 2𝑥 − 4.8𝑥𝑒 2𝑥 − 2.24 cos 𝑥 − 1.68 sin 𝑥

Table 2 below shows the numerical output to Illustration II

69
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
Table 2: The Output to Numerical Illustration II

x y- Exact y- Computed Absolute Error

0 4.0000000000 3.9971824673 2.8175327E – 3
0.1 4.6387504168 4.6387569032 6.486386876E – 6
0.2 5.3477207926 5.3477814017 6.060825418E – 5
0.3 6.1097425587 6.1097429702 4.11157979E – 7
0.4 6.8969373685 6.8969301864 7.183042454E – 6
0.5 7.6669823776 7.6669829011 5.23306771E – 7
0.6 8.3582417278 8.3582499013 8.173606549E – 6
0.7 8.8834436851 8.8834409334 2.75703608E – 6
0.8 9.1214965568 9.1214969062 3.49400000E – 7
0.9 8.9069275947 8.9069279904 3.95827600E – 7
1.0 8.0162923625 8.0162928705 5.07661766E – 7

Numerical Experimentation III

Determine the complementary solution to the differential equation

𝑦 𝑖𝑣 + 3𝑦 ′′ + 2𝑦 = 0 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑦(0) = 4, 𝑦 ′ (0) = 3, 𝑦 ′′ (0) = 5, 𝑦 ′′′ (0) = 9

The analytical solution is

𝑦(𝑥) = 7 cos 𝑥 + 7 sin 𝑥 − 3 cos 2𝑥 − 2 sin 2𝑥

The numerical experimentation to Illustration III is as shown in Table 3 below

70
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
Table 3: Output to Numerical Experimentation of Illustration III

x y- Exact y- Computed Absolute Error

0 4.0000000000 3.9999856721 1.43279E – 5
0.1 4.3263246776 4.3263299503 5.2719398685E – 6
0.2 4.7091316937 4.7091376027 5.908871837E – 6
0.3 5.1507050786 5.1507009762 4.102789517E – 6
0.4 5.6485230437 5.6485299112 6.8668597878E – 6
0.5 6.1952078158 6.1952009227 6.893541819E – 6
0.6 6.7786951827 6.7786990172 3.834431474E – 6
0.7 7.3826182329 7.3826190584 8.25422386E – 7
0.8 7.9868909626 7.9868919675 1.004952328E – 6
0.9 8.5684691677 8.5684690832 8.44099000E – 8
1.0 9.1022586906 9.1022580943 5.9642232E – 7

Illustration IV: Solve completely the differential equation

𝑦 𝑖𝑣 − 7𝑦 ′′ − 18𝑦 = 0, subject to

𝑦(0) = 1, 𝑦 ′ (0) = 3, 𝑦 ′′ (0) = −4, 𝑦 ′′′ (0) = 2

The theoretical solution is

𝑦(𝑥) = 0.7 cos √2 𝑥 + 1.7 sin √2𝑥 + 0.3 cos 3𝑥 + 0.2 sin 3𝑥

The output to Illustration IV is as shown in Table 4 below

71
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
Table 4: The Computational Result to Illustration IV

x y- Exact y- Computed Absolute Error

0 1.0000000000 0.9998976435 1.023565E – 4
0.1 0.41444878516 0.4144489764 1.911914776E – 7
0.2 0.18544967057 0.1854489027 7.67870000E – 7
0.3 -0.02361051641 -0.02361097833 4.619182019E – 7
0.4 -0.19621158857 -0.19621170912 1.205287269E – 7
0.5 -0.33366275639 -0.33366272305 3.33723102E – 8
0.6 -0.44284711285 -0.44284719021 7.73066337E – 8
0.7 -0.53039610736 -0.53039609542 1.19508374E – 8
0.8 -0.60052532177 -0.60052535801 3.62437427E – 8
0.9 -0.65417021927 -0.65417007853 1.407662188E – 7
1.0 -0.68883259955 -0.68883250974 8.9909047E – 8
DISCUSSION OF RESULTS

At this juncture, a critical analysis of the results generated from the numerical illustrations as
presented in Tables 1 to 4 shall be analyzed.

First and foremost, the derived method satisfied the conditions for which a Multistep Method will
converged as earlier enumerated. In this presentation, four critical illustrations were carried out to
compare the level of accuracy of results generated using the derived method with the analytical
method. From the table of results, that is from Tables 1 to 4, it was observed that the column of y-
computed which shows the results generated using the derived method compares favorably well
with the results of analytical solutions to the differential equations under consideration, as shown
in the column of y - exact. The absolute error is very minimal in all illustrations used.

In like manner, as the Grid Block Multistep Methods can be used to solve ordinary differential
equations, it is hereby clearly shown that Off – Grid Block Linear Multistep Method can equally

72
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
be used to solve differential equations as shown in the tables of results as presented in Tables 1 to
4.

It is not an overemphasized thing to say that this method is noble because the method is self-
starting whereby it requires no additional information to get its starting values thereby reducing
the error in the computation unlike the Predictor - Corrector methods. Also, the method affords the
users to make use of the method in solving a fourth order ordinary differential equations directly
without necessarily resolving such a differential equation to system of first order ordinary
differential equations before solving which is often laborious.

At this juncture, this method is hereby unequivocally recommended to be used for finding
numerical solutions to Fourth Order Ordinary Differential Equations without any need of resolving
such differential equations to system of first order ordinary differential equations.

REFERENCES

1 Aboiyar, T., Luga, T., Iyorter, B. V. (2015) “Derivation of Continuous Linear Multistep
Methods Using Hermite Polynomials as Basis Functions” American Journal of Applied
Mathematics and Statistics Vol.3 No. 6, pp 220 – 225
2 Adeniyi, R. B. and Alabi, M. O. (2006) “Derivation of Continuous Multistep Methods
Using Chebyshev Polynomial Basis Function”. Abacus Vol. 33. Pp. 351 – 361.
3 Adesanya, A. O., Momoh, A. A., Alkali, M. A., and Tahir, A. (2012) “Five Steps Block
Method for the Solution of Fourth Order ODEs”. International Journal of Engineering
Research and Application, Vol. 2(5) pp 991 – 998.
4 Alabi, M. O. (2008) “A Continuous Formulation of Initial Value Solvers with Chebyshev
Basis Function in a Multistep Collocation Technique” Ph. D. Thesis, University of Ilorin,
Ilorin Kwara State.
5 Alabi, M.O., Oladipo, A.T, Adesanya A.O, Okedoye, M.A, Babatunde,O.Z, (2007),
“Formulation of Some Linear Multistep Schemes for Solving First Order Initial Value
Problems Using Canonical Polynomials as Basis Functions.” Journal of Modern
Mathematics and Statistics. 1(1) Pp 3 - 7.
6 Alabi, M.O., Oladipo, A.T., and Adesanya, A.O. (2008), “Initial Value Solvers for
Second Order Ordinary Differential Equations using Chebyshev Polynomial as Basis
Functions”. Journal of Modern Mathematics and Statistics 2(1) Pp.18 – 27.
7 Alabi, M. O., Raji, M. T., and Alabi, I. Y. (2025) “Off Grid Initial Value Solver for First
Order Ordinary Differential Equations in Block Form Using Chebyshev Polynomial as
Basis Function” International Journal of Mathematics and Statistics Studies Vol. 13 (1),
pp. 13 – 35.

73
 International Journal of Mathematics and Statistics Studies, 13 (2), 53-74, 2025
Print ISSN: 2053-2229 (Print),
Online ISSN: 2053-2210 (Online)
Website: https://www.eajournals.org/
Publication of the European Centre for Research Training and Development -UK
8 Butcher, J. C. (2003) “Numerical Methods for Ordinary Differential Equations”. John
Waley and Sons, London.
9 Fox, L. and Parker, I. B., (1968) “Chebyshev Polynomials in Numerical Analysis”
Oxford University Press, London.
10 James, A. A., Adesanya, A. O., and Sunday, J. (2013) “Continuous Block Method for the
Solution of Second Order Initial Value Problems of ODEs.” International Journals of
Pure and Applied Mathematics. Vol 83(3) pp 405 - 416
11 James, A. A., Odesanya, A. O., Sunday, J. and Yakubu, D. G. (2013) “Half- Step
Continuous Block Method for the Solution of Modeled Problems of ODEs. “American
Journal of Computational Mathematics, Vol.3, pp 261 – 269.
12 Jator, S. N. (2007) “A sixth Order Linear Multistep Method for the Direct Solution of
𝑦 ′′ = (𝑥, 𝑦, 𝑦 ′ )” International Journal of Pure and Applied Mathematics 40, pp. 457 –
472.
13 Lambert, J. D. (1938) “Computational Methods in Ordinary Differential Equations.
JohnWaley and Sons, London.
14 Omolehin, J.O., Ibiejugba, M. A, Alabi, M.O. and D.J. Evans. (2003), “A new Class
of Adams Bashforth Schemes for ODES” Intern. J. Computer Math. Vol. 80(5), Pp
629 – 638.

74