Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No.

(1): 2020

Solving Some Types of the Second and Third

Order Spectral Linear Ordinary Differential
Equations
Aryan Ali Mohammed
Department of Mathematical Sciences, College of Basic Education, University of
Sulaimani, Sulaimaniyah, Iraq
[email protected], [email protected].

ARTICLE INFO
Submission date: 29 /12/ 2019
Acceptance date: 17 / 2 / 2020
Publication date: 31/ 3 / 2020

Abstract
Our article is for finding complete primitive for two different kinds of linear ordinary differential
equations in a spectral type. First kind is consisting of two different types of coefficients of order two;
one of them is polynomials, where as the other type is continuous functions and both are of real types.
The second kind is for a third order, and here the coefficients of this kind are also several real
polynomials, and in both kinds of spectral linear ordinary differential equations, the coefficients are
converted to the constants via varying the independent variable to a new one. We gave examples to
explain our mechanism.
Keywords: Order of differential equations, real functions and polynomials, complete primitive, variation
of variables.

1. Introduction
The second and third order linear ordinary differential equations (L.O.D.E) are
widely used of many problems in the distinct fields as: Engineering, mechanic, and ...
etc. The known methods like operator method, undetermined coefficients, reduction of
order and variation of parameters are studied to find the complete primitive of second
and third linear differential equations with constant and variable coefficients. The third
and second-order differential equations have been accurately inspected by many
authors. In [12] Moore, studied the behavior of their solutions, and their waverly
feature studied via Coppel [18], Marini [7], Hochtadt [3] and Hartman [9, 10]; and in
[6, 13, 16, 17] the study of the of limited and asymptotic behavior were considered,
finally in [15], Richard calculated the solutions numerically.
A lot of writers examined the asymptotic behavior for eigenvalues and identical
eigenfunctions to various kinds of cases for the spectral type [5, 14].
In [4], Karwan and Aryan studied the boundedness of non-zero solutions to the
spectral L.O.D.E. of second order. In [8], Marini and Zezza they studied asymptotically
formula of eigenfunctions L.D.E. of order two, further they specified needful and
appropriate terms to the integrals of their defined D.E.
In [11], Johnson, Busawon, and Barbot presented a substitutional procedure for
resolving the public non-homogeneous second order L.O.D.E.
In [2], Arficho derived a modern procedure for finding particular integral of
L.O.D.E. of order two by means of one given integral to the related homogeneous D.E.,

© Journal of University of Babylon for Pure and Applied Sciences (JUBPAS) by University of Babylon is licensed under a Creative
Commons Attribution 4. 0 International License, 2020.

12
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

further, he constructed another integral of the related homogeneous D.E. of this modern
procedure. Furthermore, he found complete integral of the defined L.O.D.E of order
two without using the known procedure for finding the particular integral.
In [1], Aryan constructed a complete integral to the L.O.D.E. of order 𝑛 by the
procedure for variation of parameter, as well as, he determined the non-zero solutions
of second order spectral L.O.D.E. with eigenparameter dependent boundary conditions.
Moreover, he specified the limitedness of non-zero solutions. Finally, they studied the
asymptotically formula for eigenvalues to the defined D.E.
In this paper, we study the solution of the second and third order spectral L.O.D.E.
with variable coefficients in which the coefficients of the first one consist of two distinct
kinds, one of them are polynomials, where as the other kind are continuous functions
and both of them are of real types. However, the coefficients of third order are
polynomials of real type.
Firstly, we examine two types of spectral L.O.D.E. of order two of the forms:
𝑑2𝑦 𝑑𝑦
𝑎𝑜 (𝑥) 2 + 𝑎1 (𝑥) + 𝑎2 (𝑥)𝑦(𝑥) = 𝜆2 𝑎0 (𝑥)𝑦(𝑥), 𝑥 ∈ (𝑎, 𝑏) (1)
𝑑𝑥 𝑑𝑥
and
𝑑2𝑦 𝑑𝑦
𝑎𝑜 (𝑥) 2 + 𝑎1 (𝑥) = 𝑘𝜆2 𝑦(𝑥), 𝑥 ∈ (𝑎, 𝑏) (2)
𝑑𝑥 𝑑𝑥
where 𝜆 ≠ 0 is a spectral parameter, 𝑎 > 0, and 𝑎 < 𝑏, and 𝑎𝑖 (𝑥), 𝑖 = 0,1,2, are
some real polynomials and 𝑎𝑜 (𝑥) ≠ 0, and 𝑘 is non-zero arbitrary constant, provided
that 𝑦(𝑥) and their derivatives are defined and continuous on the given interval. We
reduce equations (1) to an equation with constant coefficients by the substitution
𝑦(𝑥) = 𝑢(𝑥)𝑔(𝑥), where
1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥 )
𝑔(𝑥) = 𝑒 2 2 𝑎0 (𝑥) ,
and 𝑢(𝑥) is a new unknown function should be specified, and 𝑏 is a constant.
Finally, we convert equation (2) to the spectral linear differential equation by making
𝑎 (𝑥)
− ∫ 1 𝑑𝑥
change of variable 𝑡 = ∫ 𝑒 𝑎0(𝑥) 𝑑𝑥, to a form
2
𝑑 𝑦
+ 𝑘𝜆2 𝑦(𝑡) = 0.
𝑑𝑡 2
Secondly, we investigate spectral L.O.D.E. of order three of the type
𝑑3𝑦 𝑑2𝑦 𝑑𝑦
𝑎𝑜 (𝑥) 3 + 𝑎1 (𝑥) 2 + 𝑎2 (𝑥) = −𝜆3 𝑎3 (𝑥)𝑦(𝑥), 𝑥 ∈ (𝑎, 𝑏) (3)
𝑑𝑥 𝑑𝑥 𝑑𝑥
where λ ≠ 0 is a spectral parameter, 𝑎 > 0 and 𝑎 < 𝑏, where 𝑎𝑖 (𝑥), 𝑖 = 0,1,2,3 are
some real polynomials and 𝑎𝑜 (𝑥) ≠ 0, ∀𝑥 ∈ (𝑎, 𝑏). By making change of variable
1
𝑎 (𝑥) 3
𝑠=𝑘 ∫ (𝑎3 (𝑥)) 𝑑𝑥, equation (3) may reduce to the constant coefficients.
0
Note: Throughout the paper, L.O.D.E. stands for linear ordinary differential equations.

2. Studying the different kinds of spectral L.O.D.E. of order two

In this section, we consider the spectral L.O.D.E. of order two with distinct types of
coefficients through two kinds of transformations, and examples are given to explain
our technique.

Theorem 1. The spectral differential equation of the form

𝑑2𝑦 𝑑𝑦
𝑎𝑜 (𝑥) 2 + 𝑎1 (𝑥) + 𝑎2 (𝑥)𝑦(𝑥) = 𝜆2 𝑎0 (𝑥)𝑦(𝑥), 𝑥 ∈ (𝑎, 𝑏) (1)
𝑑𝑥 𝑑𝑥

13
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

where 𝜆, 𝑎0 (𝑥), 𝑎2 (𝑥) ≠ 0 and 𝑎𝑖 (𝑥), 𝑖 = 0,1,2 are real polynomial functions, reduced
to an equation with constant coefficients by the substitution 𝑦(𝑥) = 𝑢(𝑥) 𝑔(𝑥), where
1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥 )
𝑔(𝑥) = 𝑒 2 2 𝑎0 (𝑥) , if
2 (𝑥)
𝑏 2
1 𝑎1 1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎1 (𝑥)𝑎0′ (𝑥) 𝑎2 (𝑥)
( − ( 2 )) − ( 2 )+ = 𝐴 = constant,
4 4 𝑎0 (𝑥) 2 𝑎0 (𝑥) 𝑎0 (𝑥)
where 𝑏 is an arbitrary constant, and 𝑢(𝑥) is a new unknown function to be
determined.
Proof: At the beginning we find the first and second derivative by above
substitutions.
𝑑𝑦
= 𝑢(𝑥)𝑔′ (𝑥) + 𝑢′ (𝑥)𝑔(𝑥)
𝑑𝑥
𝑑𝑦 1 1 𝑎1 (𝑥) 1 1 𝑎 (𝑥) 1 1 𝑎1 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥) ′ (𝑥) (𝑏2𝑥−2 ∫𝑎0 (𝑥) 𝑑𝑥)
= 𝑢(𝑥) (𝑏 − ( )) 𝑒 2 2 𝑎0 (𝑥) +𝑢 𝑒 ,
𝑑𝑥 2 2 𝑎0 (𝑥)
2
𝑑2𝑦 1 1 𝑎1 (𝑥) 1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1
2 2 𝑎0 (𝑥)
𝑑𝑥)
= 𝑢(𝑥) (𝑏 − ( )) 𝑒
𝑑𝑥 2 2 2 𝑎0 (𝑥)
1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎1 (𝑥)𝑎0′ (𝑥) (𝑏12𝑥−12 ∫𝑎𝑎1 (𝑥) 𝑑𝑥)
+𝑢(𝑥) (− )𝑒 0 (𝑥)
2
2 𝑎0 (𝑥)
1 1 𝑎1 (𝑥) 1 1 𝑎 (𝑥) 1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥) (𝑏 𝑥− ∫ 1 𝑑𝑥)
+𝑢′ (𝑥) (𝑏 − ( )) 𝑒 2 2 𝑎0 (𝑥) + 𝑢′′ (𝑥)𝑒 2 2 𝑎0 (𝑥)
2 2 𝑎0 (𝑥)
1 1 𝑎1 (𝑥) 1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥 )
+ 𝑢′ (𝑥) (𝑏 − ( )) 𝑒 2 2 𝑎0 (𝑥) ,
2 2 𝑎0 (𝑥)
𝑑2𝑦 1 1 𝑎1 (𝑥)
′′ (𝑥) (𝑏2𝑥−2 ∫𝑎0 (𝑥) 𝑑𝑥) ′ (𝑥)
1 1 𝑎1 (𝑥) 1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥)
=𝑢 𝑒 + 2𝑢 (𝑏 − ( )) 𝑒 2 2 𝑎0 (𝑥) +
𝑑𝑥 2 2 2 𝑎0 (𝑥)
2
1 1 𝑎 (𝑥) 1 1 𝑎1 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥)
𝑢(𝑥)𝑒 2 2 𝑎0 (𝑥) ((𝑏 − ( ))
2 2 𝑎0 (𝑥)

1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎1 (𝑥)𝑎0′ (𝑥)

− ( )) .
2 𝑎02 (𝑥)
𝑑𝑦 𝑑2 𝑦
Setting and in the given differential equation, yields
𝑑𝑥 𝑑𝑥 2
1 1 𝑎 (𝑥) 1 1 𝑎1 (𝑥) 𝑥 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥) (𝑏 − ∫ 1 𝑑𝑥)
𝑎0 (𝑥) ( 𝑢′′ (𝑥) 𝑒 2 2 𝑎0 (𝑥) + 2𝑢′ (𝑥) (𝑏 − ( )) 𝑒 2 2 𝑎0 (𝑥) +
2 2 𝑎0 (𝑥)
2
1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥) 1 1 𝑎1 (𝑥) 1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎1 (𝑥)𝑎0′ (𝑥)
𝑢(𝑥)𝑒 2 2 𝑎0 (𝑥) ((𝑏 − ( )) − ( )) )
2 2 𝑎0 (𝑥) 2 𝑎02 (𝑥)

1 1 𝑎1 (𝑥) 1 1 𝑎 (𝑥) 1 1 𝑎 (𝑥)

(𝑏 𝑥− ∫ 1 𝑑𝑥) (𝑏 𝑥− ∫ 1 𝑑𝑥)
+ 𝑎1 (𝑥) (𝑢(𝑥) (𝑏 − ( )) 𝑒 2 2 𝑎0 (𝑥) + 𝑢′ (𝑥) 𝑒 2 2 𝑎0 (𝑥) )
2 2 𝑎0 (𝑥)
1 1 𝑎 (𝑥) 1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 𝑑𝑥) (𝑏 𝑥− ∫ 1 𝑑𝑥)
+𝑎2 (𝑥) 𝑢(𝑥) 𝑒 2 2 𝑎0 (𝑥) = 𝜆2 𝑎0 (𝑥)𝑢(𝑥) 𝑒 2 2 𝑎0 (𝑥) .

14
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

1 1 𝑎 (𝑥)
(𝑏 𝑥− ∫ 1 (𝑥) 𝑑𝑥)
Dividing both sides by 𝑎0 (𝑥) and 𝑒 2 2 𝑎 0 and after simplification we deduce

′′ (𝑥) ′ (𝑥)
𝑏 2 1 𝑎12 (𝑥) 1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎(𝑥)𝑎0′ (𝑥) 𝑎2 (𝑥)
𝑢 +𝑏𝑢 + (( − ( 2 )) − ( ) + )
4 4 𝑎0 (𝑥) 2 𝑎02 (𝑥) 𝑎0 (𝑥)
𝑢(𝑥) = 𝜆2 𝑢(𝑥).
𝑏2 1 𝑎2 (𝑥) 1 𝑎1′ (𝑥)𝑎0 (𝑥)−𝑎(𝑥)𝑎0′ (𝑥) 𝑎 (𝑥)
If ( 4 − 4 (𝑎12 (𝑥))) − 2 ( ) + 𝑎2 (𝑥) = 𝐴 = constant, then the last
0 𝑎02 (𝑥) 0

equation reduces to
𝑢′′ (𝑥) + 𝑏 𝑢′ (𝑥) + 𝐴 𝑢(𝑥) = 𝜆2 𝑢(𝑥),
is a spectral L.O.D.E. with constant coefficients in variable 𝑢 and 𝑥, therefore 𝑢(𝑥)
can be found with known methods and hence the solution 𝑦(𝑥) is specified from
𝑦(𝑥) = 𝑢(𝑥)𝑔(𝑥).

Example: Solve the spectral differential equation of the form

2
𝑑2𝑦 𝑑𝑦
𝑥 + 2𝑥 (𝑥 + 2 ) + 2 (𝑥 + 1)2 𝑦 = 𝑥 2 𝜆2 𝑦.
𝑑𝑥 2 𝑑𝑥
Solution: Comparing the given equation with our differential form
𝑑2 𝑦 𝑑𝑦
𝑎0 (𝑥) 𝑑𝑥 2 + 𝑎1 (𝑥) 𝑑𝑥 + 𝑎2 (𝑥)𝑦 = 𝜆2 𝑎0 (𝑥)𝑦, we get that:
𝑎0 (𝑥) = 𝑥 2 , 𝑎1 (𝑥) = 2𝑥(𝑥 + 2), 𝑎2 (𝑥) = 2(𝑥 + 1)2 ,
𝑎0′ (𝑥) = 2𝑥, 𝑎1′ (𝑥) = 4𝑥 + 4, 𝑎12 (𝑥) = (2𝑥 2 + 4𝑥)2 , 𝑎02 (𝑥) = 𝑥 4 .
Now
1 1 𝑎 (𝑥) 𝑥 1 2𝑥 2 +4𝑥 𝑥
(𝑏 𝑥− ∫ 1 𝑑𝑥) (𝑏 𝑥− ∫ 𝑑𝑥) ∫ 𝑑𝑥−2 ∫ 𝑥 −1 𝑑𝑥)
𝑔(𝑥) = 𝑒 2 2 𝑎0 (𝑥) = 𝑒 2 2 𝑥2 = 𝑒 (𝑏2 −
𝑥 𝑥
= 𝑒 (𝑏2 −2−2𝐿𝑛𝑥) .
1
Let 𝑏 = 2, so 𝑔(𝑥) = 𝑥 2 .
𝑏 2 1 𝑎12 (𝑥) 1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎1 (𝑥)𝑎0′ (𝑥) 𝑎2 (𝑥)
( − ( 2 )) − ( ) +
4 4 𝑎0 (𝑥) 2 𝑎02 (𝑥) 𝑎0 (𝑥)
1 4𝑥 4 + 16𝑥 3 + 16𝑥 2 1 ((4𝑥 + 4)𝑥 2 − (2𝑥 2 + 4𝑥)2𝑥)
= (1 − ( )) − ( )
4 𝑥4 2 𝑥4
(2𝑥 2 + 4𝑥 + 2)
+
𝑥2
𝑥 − 𝑥 4 − 4𝑥 3 − 4𝑥 2
4
1 −4𝑥 2 (2𝑥 2 + 4𝑥 + 2)
=( ) − ( ) +
𝑥4 2 𝑥4 𝑥2
−4𝑥 − 4 2𝑥 2 + 4𝑥 + 4 2𝑥 2
= + = 2 = 2 = 𝐴 = constant.
𝑥2 𝑥2 𝑥
Thus the equation 𝑢′′ (𝑥) + 𝑑 𝑢′ (𝑥) + 𝐴 𝑢(𝑥) = 𝜆2 𝑢(𝑥) becomes
𝑢′′ (𝑥) + 2 𝑢′ (𝑥) + 2 𝑢(𝑥) = 𝜆2 𝑢(𝑥),
or
𝑢′′ (𝑥) + 2 𝑢′ (𝑥) + (2 − 𝜆2 ) 𝑢(𝑥) = 0,
is a spectral L.O.D.E. of order two, so the complete integral with respect to the
variables 𝑢 and 𝑥 can be found as follows:
𝑑
(𝐷2 + 2𝐷 − (2 − 𝜆2 ))𝑧(𝑥) = 0, where 𝐷 = .
𝑑𝑥
or

15
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

2 2)
−2 ∓ √4 − 4(2 − 𝜆2 ) −2 ∓ 2√𝜆2 − 1
𝑚 + 2𝑚 + (2 − 𝜆 = 0, 𝑚 = =
2 2
𝑚 = −1 ∓ √𝜆2 − 1.
2 2
𝑢(𝑥) = 𝑎1 𝑒 (√𝜆 −1−1)𝑥 + 𝑏1 𝑒 (−√𝜆 −1−1)𝑥 .
Thereby we find 𝑦(𝑥) from 𝑦(𝑥) = 𝑢(𝑥)𝑔(𝑥)
2 2
𝑦(𝑥) = 𝑥 −2 (𝑎1 𝑒 (√𝜆 −1−1)𝑥 + 𝑏1 𝑒 (−√𝜆 −1−1)𝑥 ),
where 𝑎1 and 𝑏1 are constants.

Example: Solve the following spectral D.E.

𝑑2 𝑦 𝑑𝑦
𝑥 2 2 − 𝑥(3𝑥 + 2) + (2𝑥 2 + 3𝑥 + 2)𝑦 = 𝜆2 𝑥 2 𝑦.
𝑑𝑥 𝑑𝑥
Solution: 𝑎0 = 𝑥 2 , 𝑎1 = −3𝑥 2 − 2𝑥, 𝑎2 = 2𝑥 2 + 3𝑥 + 2,
𝑎0′ = 2𝑥, 𝑎1′ = −6𝑥 − 2, 𝑎12 = (−3𝑥 2 − 2𝑥)2 , 𝑎02 (𝑥) = 𝑥 4 .
Now,
1 1 𝑎 (𝑥) 𝑥 1 −3𝑥 2 −2𝑥 𝑥 3
(𝑏 𝑥− ∫ 1 𝑑𝑥) (𝑏 − ∫ 𝑑𝑥) −1
𝑔(𝑥) = 𝑒 2 2 𝑎0 (𝑥) = 𝑒 2 2 𝑥2 = 𝑒 (𝑏2 + 2 ∫ 𝑑𝑥+∫ 𝑥 𝑑𝑥)
𝑥 3
𝑔(𝑥) = 𝑒 (𝑑2 +2𝑥+𝑙𝑛𝑥) .
At 𝑏 = −3, we have 𝑔(𝑥) = 𝑒 𝑙𝑛𝑥 = 𝑥.
Therefore
𝑏 2 1 𝑎12 (𝑥) 1 𝑎1′ (𝑥)𝑎0 (𝑥) − 𝑎1 (𝑥)𝑎0′ (𝑥) 𝑎2 (𝑥)
( − ( 2 )) − ( 2 )+ =
4 4 𝑎0 (𝑥) 2 𝑎0 (𝑥) 𝑎0 (𝑥)
9 1 9𝑥 4 + 12𝑥 3 + 4𝑥 2 1 ((−6𝑥 − 2)𝑥 2 − (−3𝑥 2 − 2𝑥)2𝑥)
( − ( )) − ( )
4 4 𝑥4 2 𝑥4
2𝑥 2 + 3𝑥 + 2
+
𝑥2
−12𝑥 3 − 4𝑥 2 1 2𝑥 2 2𝑥 2 + 3𝑥 + 2 8𝑥 4
= − ( ) + = = 2 = 𝐴 = constant.
4𝑥 4 2 𝑥4 𝑥2 4𝑥 4
Thus, the equation 𝑢′′ (𝑥) + 𝑏 𝑢′ (𝑥) + 𝐴 𝑢(𝑥) = 𝜆2 𝑢(𝑥) reduces to
𝑢′′ (𝑥) − 3𝑢′ (𝑥) + 2 𝑢(𝑥) = 𝜆2 𝑢(𝑥),
or
𝑢′′ (𝑥) − 3𝑢′ (𝑥) + (2 − 𝜆2 ) 𝑢(𝑥) = 0.
Thus the complete integral of the resultant equation can be found by:
𝑑
(𝐷2 − 3𝐷 + (2 − 𝜆2 ) 𝑧(𝑥) = 0, where 𝐷 =
𝑑𝑥
𝑚2 − 3𝑚 + (2 − 𝜆2 ) = 0,
3 ∓ √9 − 4(2 − 𝜆2 ) 3 √1 + 4𝜆2
𝑚= = ∓ ,
2 2 2
3 √1+4𝜆2 3 √1+4𝜆2
( + )𝑥 ( − )𝑥
2 2 2 2
𝑧(𝑥) = 𝑐3 𝑒 + 𝑐4 𝑒 ,
where 𝑐3 and 𝑐4 are arbitrary constants.
And since 𝑦 = 𝑔(𝑥)𝑢(𝑥), hence a complete integral to the given differential equation
is
3 √1+4𝜆2 3 √1+4𝜆2
( + )𝑥 ( − )𝑥
2 2 2 2
𝑦(𝑥) = 𝑥 ( 𝑐3 𝑒 + 𝑐4 𝑒 ).

16
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

Theorem 2. The spectral differential equation

𝑑2 𝑦 𝑑𝑦
𝑎0 (𝑥) 𝑑𝑥 2 + 𝑎1 (𝑥) 𝑑𝑥 = 𝑘 𝜆2 𝑦, 𝑥 ∈ (𝑎, 𝑏)
where 𝜆 , 𝑎0 (𝑥) ≠ 0 and 𝑎𝑖 (𝑥), 𝑖 = 0,1 are real valued continuous functions, can be
reduced to a form 𝑦 ′′ (𝑡) − 𝐾𝜆2 𝑦(𝑡) = 0 by making the change of variable
𝑎 (𝑥)
−∫ 1 𝑑𝑥
𝑡 = ∫𝑒 𝑎0 (𝑥) 𝑑𝑥,
1
𝑖𝑓
𝑎0 (𝑥)
𝑎 (𝑥)
−2 ∫ 1 𝑑𝑥
= 𝑘1 𝑒 𝑎0 (𝑥) , where 𝑘1 and 𝑘 are arbitrary non zero constants, and
𝐾 = 𝑘1 𝑘.
Proof: Consider the differential equation
𝑑2 𝑦 𝑑𝑦
𝑎0 (𝑥) 𝑑𝑥 2 + 𝑎1 (𝑥) 𝑑𝑥 = 𝑘 𝜆2 𝑦,
now,
𝑎 (𝑥) 𝑑𝑡 𝑎 (𝑥)
−∫ 1 𝑑𝑥 −∫ 1 𝑑𝑥
𝑡 = ∫𝑒 𝑎0 (𝑥) 𝑑𝑥, = 𝑒 𝑎0 (𝑥) .
𝑑𝑥
𝑑𝑦 𝑑𝑦 𝑑𝑡 𝑎1 (𝑥)
−∫ 𝑑𝑥 𝑑𝑦
= = 𝑒 𝑎0 (𝑥) ,
𝑑𝑥 𝑑𝑡 𝑑𝑥 𝑑𝑡
𝑑2𝑦 𝑑 𝑎
− ∫ 1 𝑑𝑥 𝑑𝑦 𝑑 𝑎
− ∫ 1 𝑑𝑥 𝑑𝑦
𝑎
− ∫ 1 𝑑𝑥 𝑑 𝑑𝑦 𝑑𝑡
= (𝑒 𝑎0 ) = ( 𝑒 𝑎0 ) + 𝑒 𝑎0 ( )
𝑑𝑥 2 𝑑𝑥 𝑑𝑡 𝑑𝑥 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑥
𝑎1 (𝑥) − ∫𝑎𝑎1 𝑑𝑥 𝑑𝑦 𝑎
− ∫ 1 𝑑𝑥 𝑑 𝑑𝑦 𝑑𝑡
=− 𝑒 0 + 𝑒 𝑎0 ( )
𝑎0 (𝑥) 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑥
𝑑2𝑦 𝑎
−2 ∫ 1 𝑑𝑥 𝑑 𝑦
2
𝑎1 (𝑥) − ∫𝑎𝑎1 𝑑𝑥 𝑑𝑦
= 𝑒 𝑎0 − 𝑒 0 .
𝑑𝑥 2 𝑑𝑡 2 𝑎0 (𝑥) 𝑑𝑡
𝑑𝑦 𝑑2 𝑦
Substituting and in the given differential equation, yields
𝑑𝑥 𝑑𝑥 2
𝑎 2 𝑎 𝑎
−2 ∫ 1 𝑑𝑥 𝑑 𝑦 − ∫ 1 𝑑𝑥 𝑑𝑦 − ∫ 1 𝑑𝑥 𝑑𝑦
𝑎0 (𝑥) 𝑒 𝑎0 − 𝑎 1 (𝑥) 𝑒 𝑎0 + 𝑎1 (𝑥) 𝑒 𝑎0 = 𝑘 𝜆2 𝑦,
𝑑𝑡 2 𝑑𝑡 𝑑𝑡
𝑜𝑟
𝑑2𝑦 1 𝑎
−2 ∫ 1 𝑑𝑥
− 𝑒 𝑎0 𝑘 𝜆2 𝑦 = 0.
𝑑𝑡 2 𝑎0 (𝑥)
1 𝑎
−2 ∫ 1 𝑑𝑥
If = 𝑘1 𝑒 𝑎0 , where 𝑘1 is a non zero arbitrary constant, so the last
𝑎0 (𝑥)
equation reduces to
𝑑2𝑦
− 𝐾 𝜆2 𝑦(𝑥) = 0, where 𝐾 = 𝑘1 𝑘,
𝑑𝑡 2
Thus, the proof is finished.

Example: Consider the spectral Ordinary differential equation

𝑑2 𝑦 𝑑𝑦
Sin2 4𝑥 𝑑𝑥 2 + 2 sin 8𝑥 𝑑𝑥 = 𝜆2 𝑦.
Solution: 𝑎0 = sin2 4𝑥, 𝑎1 = 2 sin 8𝑥
𝑎 sin2 4𝑥 4 sin 4𝑥 cos 4𝑥
− ∫ 1 𝑑𝑥 −∫ 𝑑𝑥 −∫ 𝑑𝑥
𝑡= ∫ 𝑒 𝑎0 𝑑𝑥 = ∫ 𝑒 2 sin 8𝑥 𝑑𝑥 = ∫ 𝑒 sin2 4𝑥 𝑑𝑥

17
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

−1 1
𝑡 = ∫ 𝑒 − ∫(𝑠𝑖𝑛4𝑥) (4 cos 4𝑥) 𝑑𝑥
𝑑𝑥 = ∫ 𝑒 −𝑙𝑛 𝑠𝑖𝑛4𝑥 𝑑𝑥 = ∫
sin 4𝑥
1
=∫ 𝑑𝑥
2𝑠𝑖𝑛2𝑥𝑐𝑜𝑠2𝑥
1 sec 2 2𝑥 1
𝑡= ∫ 𝑑𝑥 = 𝑙𝑛 |𝑡𝑎𝑛2𝑥|,
2 𝑡𝑎𝑛2𝑥 4
or
1 𝑑𝑡 1
𝑡 = 𝑙𝑛 𝑡𝑎𝑛2𝑥, = .
4 𝑑𝑥 𝑠𝑖𝑛4𝑥
𝑑𝑦 𝑑𝑦 𝑑𝑡 1 𝑑𝑦
= = ,
𝑑𝑥 𝑑𝑡 𝑑𝑥 𝑠𝑖𝑛4𝑥 𝑑𝑡
2
𝑑 𝑦 𝑑 1 𝑑𝑦 −4𝑐𝑜𝑠4𝑥 𝑑𝑦 1 𝑑2𝑦 1
= ( )= + ,
𝑑𝑥 2 𝑑𝑥 𝑠𝑖𝑛4𝑥 𝑑𝑡 sin2 4𝑥 𝑑𝑡 𝑠𝑖𝑛4𝑥 𝑑𝑡 2 𝑠𝑖𝑛4𝑥
𝑑2𝑦 1 𝑑 2 𝑦 4𝑐𝑜𝑠4𝑥 𝑑𝑦
= − .
𝑑𝑥 2 sin2 4𝑥 𝑑𝑡 2 sin2 4𝑥 𝑑𝑡
Since
sin2 4𝑥 𝑦 ′′ (𝑥) + 2 sin 8𝑥 𝑦 ′ (𝑥) = 𝜆2 𝑦(𝑥),
then
2
1 𝑑2 𝑦 2
4𝑐𝑜𝑠4𝑥 𝑑𝑦 1 𝑑𝑦
sin 4𝑥 2 − sin 4𝑥 + 2𝑠𝑖𝑛8𝑥 = 𝜆2 𝑦(𝑡),
sin 4𝑥 𝑑𝑡 2 sin2 4𝑥 𝑑𝑡 𝑠𝑖𝑛4𝑥 𝑑𝑡
𝑑2𝑦 𝑑𝑦 1 𝑑𝑦
2
− 4𝑐𝑜𝑠4𝑥 + 4𝑠𝑖𝑛4𝑥𝑐𝑜𝑠4𝑥 = 𝜆2 𝑦(𝑡),
𝑑𝑡 𝑑𝑡 𝑠𝑖𝑛4𝑥 𝑑𝑡
𝑑2𝑦
− 𝜆2 𝑦(𝑡) = 0,
𝑑𝑡 2
Thus, the complete integral with respect to the variables 𝑦 and 𝑡 is:
𝑚2 − 𝜆2 = 0, 𝑚 = ∓ 𝜆 .
𝑦(𝑡) = 𝑐5 𝑒 𝜆 𝑡 + 𝑐6 𝑒 − 𝜆 𝑡 ,
or
𝜆 𝜆
𝑦(𝑥) = 𝑐5 (𝑡𝑎𝑛2𝑥) 4 + 𝑐6 (𝑡𝑎𝑛2𝑥)−4 .

3. Solving spectral L.O.D.E. of order three

In the following theorem, we determine the complete integral of a defined spectral
L.O.D.E. of order three where its coefficients are some real polynomial.

Theorem 3. Consider the spectral ordinary differential equation

𝑑3 𝑦 𝑑2 𝑦 𝑑𝑦
𝑎𝑜 (𝑥) 𝑑𝑥 3 + 𝑎1 (𝑥) 𝑑𝑥 2 + 𝑎2 (𝑥) 𝑑𝑥 = −𝜆3 𝑎3 (𝑥)𝑦, 𝑥 ∈ (𝑎, 𝑏) (3)
where λ ≠ 0, 𝑎 > 0 and 𝑎 < 𝑏, where 𝑎𝑖 (𝑥), 𝑖 = 0,1,2,3 are some real polynomials
1
𝑎 (𝑥) 3
and 𝑎𝑜 (𝑥) ≠ 0, ∀𝑥 ∈ (𝑎, 𝑏). By making change of variable 𝑠 = 𝑘 ∫ (𝑎3 (𝑥)) 𝑑𝑥,
0
equation (3) may be reduced to the constant coefficients if
4
′
1 𝑎1 (𝑠) 𝑎03 (𝑠) 𝑎3 (𝑠)
ℎ1 (𝑠) = ( 2 1 + 4 ( ) ) = ℎ2 (𝑠)
𝑘 3 (𝑠)𝑎 3 (𝑠)
𝑎𝑜 (𝑠)
𝑎0 3 𝑎33 (𝑠)

18
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

1 1 𝑎3′′ (𝑠)𝑎𝑜 (𝑠) − 𝑎3′ (𝑠) 𝑎𝑜′ (𝑠)

= ( 5 ( 1 )
3𝑘 2 3 (𝑠) 3 (𝑠)
𝑎3 𝑎𝑜
1 𝑎3′ (𝑠)𝑎𝑜′ (𝑠) + 𝑎3 (𝑠) 𝑎𝑜′′ (𝑠)
− 5 ( 1 )
𝑎33 (𝑠) 𝑎𝑜3 (𝑠)
4 8 2
2 ′ 2 ′
𝑎03 (𝑠)( 𝑎𝑜′ (𝑠)) − 2 𝑎03 (𝑠) 𝑎3 (𝑠) 𝑎1 (𝑠)𝑎03 (𝑠) 𝑎3 (𝑠)
− 2 2 ( 8 ( ( ) ) )+ 5 ( )
3 𝑎𝑜 (𝑠) 𝑎𝑜 (𝑠)
𝑎33 (𝑠) 𝑎33 (𝑠) 𝑎33 (𝑠)

1 𝑎2 (𝑠)
+ ( 1 2 ) ) = constant, where 𝑘 is a non zero constant.
3
𝑎03 (𝑠) 3 (𝑠)
𝑎3
Proof: We rewrite equation (3) as follows
𝑑3 𝑦 𝑑2 𝑦 𝑑𝑦
𝑎𝑜 (𝑥) 𝑑𝑥 3 + 𝑎1 (𝑥) 𝑑𝑥 2 + 𝑎2 (𝑥) 𝑑𝑥 + 𝜆3 𝑎3 (𝑥)𝑦 = 0. (3)
We change the independent variable in equation (3) from 𝑥 to 𝑠 by using the chain
rule.
1
𝑎3 (𝑥) 3
Now, since 𝑠 = 𝑘 ∫ ( ) 𝑑𝑥, (4)
𝑎0 (𝑥)
then
1
𝑑𝑦 𝑑𝑦 𝑑𝑠 𝑎3 3 𝑑𝑦
= = 𝑘( ) ,
𝑑𝑥 𝑑𝑠 𝑑𝑥 𝑎𝑜 𝑑𝑠
1
𝑑2𝑦 𝑑 𝑑𝑦 𝑑 𝑎3 3 𝑑𝑦
= ( )= (𝑘 ( ) ),
𝑑𝑥 2 𝑑𝑥 𝑑𝑥 𝑑𝑥 𝑎𝑜 𝑑𝑠
1 1
𝑑2𝑦 𝑑 𝑎3 3 𝑑𝑦 𝑎3 3 𝑑 𝑑𝑦
= 𝑘 (𝑘 ( ) ) + 𝑘 ( ) ( ),
𝑑𝑥 2 𝑑𝑥 𝑎𝑜 𝑑𝑠 𝑎𝑜 𝑑𝑥 𝑑𝑠
2 1
𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 𝑎3 3 𝑑 𝑑𝑦 𝑑𝑠
= ( ) ( ) +𝑘( ) ( ) ,
3 𝑎𝑜 (𝑎𝑜 ) 2 𝑑𝑠 𝑎𝑜 𝑑𝑠 𝑑𝑠 𝑑𝑥
or
2 2
𝑑2𝑦 𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 2
𝑎3 3 𝑑 2 𝑦
= ( ) ( ) + 𝑘 ( ) .
𝑑𝑥 2 3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 𝑎𝑜 𝑑𝑠 2
𝑑3𝑦 𝑑 𝑑2𝑦
= ( )
𝑑𝑥 3 𝑑𝑥 𝑑𝑥 2
2 2
𝑑 𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 2
𝑎3 3 𝑑 2 𝑦
= ( ( ) ( ) + 𝑘 ( ) )
𝑑𝑥 3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 𝑎𝑜 𝑑𝑠 2
2 2
𝑑 𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 𝑑 2
𝑎3 3 𝑑 2 𝑦
= ( ( ) ( ) ) + (𝑘 ( ) )
𝑑𝑥 3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 𝑑𝑥 𝑎𝑜 𝑑𝑠 2
5 2
−2𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 𝑘 𝑎3 −3 𝑑 𝑑𝑦
=( ( ) ( ) + ( ) ( ))
9 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 3 𝑎𝑜 𝑑𝑥 𝑑𝑠

19
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

2
𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑘 𝑎3 −3 𝑑𝑦 𝑑 𝑎3′ 𝑎3 𝑎𝑜′ 2
𝑑2𝑦
( ) + ( ) ( ) ( − ) + 𝑘
(𝑎𝑜 )2 3 𝑎𝑜 𝑑𝑠 𝑑𝑥 𝑎0 𝑎0 2 𝑑𝑠 2
1 2
2 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 2
𝑎3 3 𝑑 𝑑 2 𝑦 𝑑𝑠
( ( ) ( )) + 𝑘 ( ) ( ) ,
3 𝑎𝑜 (𝑎𝑜 )2 𝑎𝑜 𝑑𝑠 𝑑𝑠 2 𝑑𝑥
5 2
𝑑3𝑦 −2𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 𝑘 𝑎3 −3 𝑑𝑠 𝑑 2 𝑦
= ( ( ) ( ) + ( ) ( ))
𝑑𝑥 3 9 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 3 𝑎𝑜 𝑑𝑥 𝑑𝑠 2
2
𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑘 𝑎3 −3 𝑑𝑦 𝑎3′′ 𝑎𝑜 − 𝑎3′ 𝑎𝑜′
( ) + ( ) ( ) (
(𝑎𝑜 )2 3 𝑎𝑜 𝑑𝑠 (𝑎𝑜 )2
1
(𝑎3′ 𝑎𝑜′ − 𝑎3 𝑎𝑜′′ )(𝑎𝑜 )2 − 2𝑎3 𝑎𝑜′ 𝑎𝑜 𝑎𝑜′ 2
𝑑 2 𝑦 2 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′
− ) + 𝑘 ( ( ) )
(𝑎𝑜 )4 𝑑𝑠 2 3 𝑎𝑜 (𝑎𝑜 )2
2 1
2
𝑎3 3 𝑑 3 𝑦 𝑎3 3
+𝑘 ( ) 3
𝑘 ( ) .
𝑎𝑜 𝑑𝑠 𝑎𝑜
After some minor algebraic operations, we get
1
𝑑3𝑦 3
𝑎3 𝑑 3 𝑦 2
𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑 2 𝑦
= 𝑘 ( ) ( ) + 𝑘 ( ) ( )
𝑑𝑥 3 𝑎𝑜 𝑑𝑠 3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 2
5 2
2𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 𝑘 𝑎3 −3 𝑎3′′ 𝑎𝑜 − 𝑎3′ 𝑎𝑜′ 𝑑𝑦
− ( ) ( ) + ( ) ( )
9 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠
2 2
𝑘 𝑎3 −3 (𝑎3′ 𝑎𝑜′ + 𝑎3 𝑎𝑜′′ ) 𝑑𝑦 2𝑘 𝑎3 −3 𝑎3 (𝑎𝑜′ )2 𝑑𝑦
− ( ) ( ) − ( ) ( ) .
3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 3 𝑎𝑜 (𝑎𝑜 )3 𝑑𝑠
dy d2 y 𝑑3 𝑦
putting the expressions for , dx2 and in equation (3), we get:
dx 𝑑𝑥 3
1
𝑎3 𝑑 3 𝑦 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑2 𝑦
𝑎𝑜 𝑘 ( ) ( 3 ) + 𝑘 2 𝑎𝑜 ( ) (
3
)
𝑎𝑜 𝑑𝑠 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 2
5 2
2𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 𝑘 𝑎3 −3 𝑎3′′ 𝑎𝑜 − 𝑎3′ 𝑎𝑜′ 𝑑𝑦
− ( ) ( ) + 𝑎 ( ) ( )
9 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 3 𝑜 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠
2 2
𝑘 𝑎3 −3 (𝑎3′ 𝑎𝑜′ + 𝑎3 𝑎𝑜′′ ) 𝑑𝑦 2𝑘 𝑎3 −3 𝑎3 (𝑎𝑜′ )2 𝑑𝑦
− 𝑎𝑜 ( ) ( ) − 𝑎 ( ) ( )
3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 3 𝑜 𝑎𝑜 (𝑎𝑜 )3 𝑑𝑠
2 2 1
𝑘 𝑎3 −3 𝑎3′ 𝑎𝑜 − 𝑎3 𝑎𝑜′ 𝑑𝑦 2
𝑎3 3 𝑑 2 𝑦 𝑎3 3 𝑑𝑦
+𝑎1 ( ) ( ) + 𝑘 𝑎1 ( ) + 𝑘 𝑎 2 ( )
3 𝑎𝑜 (𝑎𝑜 )2 𝑑𝑠 𝑎𝑜 𝑑𝑠 2 𝑎𝑜 𝑑𝑠
+𝜆3 𝑎3 𝑦 = 0,
or
4
2 2
3
𝑑 𝑦 3
𝑎3 3 𝑎𝑜3 𝑎3 ′ 𝑑 2 𝑦 𝑘 𝑎0 3 𝑎3′′ 𝑎𝑜 − 𝑎3′ 𝑎𝑜′
𝑘 𝑎3 + 𝑘 2 (𝑎1 ( ) + 1 ( )) 2+ ( ( ) ( )
𝑑𝑠 3 𝑎𝑜 𝑎𝑜 𝑑𝑠 3 𝑝3 𝑎𝑜
𝑎33
8
2 2
𝑎0 3 𝑎3′ 𝑎𝑜′ + 𝑎3 𝑎𝑜′′ 𝑎0 3 𝑎3 (𝑎𝑜′ )2 2 𝑎𝑜3 𝑎3 ′ 2
−( ) ( ) − 2( ) ( ) − ( 5 ) (( ) )
𝑎3 𝑎𝑜 𝑎3 (𝑎𝑜 )2 3 𝑎𝑜
𝑎33

20
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

2 1
𝑎0 3 𝑎3 ′ 1 𝑎3 3 𝑑𝑦
+𝑎1 ( ) ( ) + 𝑎2 ( ) ) + 𝜆3 𝑎3 𝑦 = 0.
𝑎3 𝑎𝑜 3 𝑎𝑜 𝑑𝑠
1
Multiplying the last equation by 3 , we obtain
𝑘 𝑎3
4
𝑑3𝑦 1 𝑎1 𝑎03 𝑎3 ′ 𝑑 2 𝑦 1 1 𝑎3′′ 𝑎𝑜 − 𝑎3′ 𝑎𝑜′
+ ( 2 1+ 4 ( )) + ( 5( 1 )
𝑑𝑠 3 𝑘 3 3 3
𝑎𝑜 𝑑𝑠 2 3𝑘 2 3 3 (𝑠)
𝑎0 𝑎3 𝑎3 𝑎3 𝑎𝑜
4 8
1 𝑎3′ 𝑎𝑜′ + 𝑎3 𝑎𝑜′′ 𝑎03 ( 𝑎𝑜′ )2 2 𝑎03 𝑎3 ′ 2
− 5 ( 1 ) − 2( 2 ) − ( 8 )( ( ) )
3 𝑎𝑜
𝑎33 𝑎𝑜3 𝑎33 𝑎33
2
𝑎1 𝑎03 𝑎3 ′ 1 𝑎2 𝑑𝑦 𝜆3
+ ( 5 ) ( ) + ( 1 2) ) + 𝑦 = 0.
3
𝑎𝑜 3 3 3
𝑑𝑠 𝑘 3
𝑎3 𝑎0 𝑎3
The last equation becomes
𝑑3𝑦 𝑑2 𝑦 𝑑𝑦 1 3
3
+ ℎ 1 (𝑠) 2
+ ℎ2 (𝑠) + 𝜆 𝑦(𝑠) = 0,
𝑑𝑠 𝑑𝑠 𝑑𝑠 𝑘 3
where
4
1 𝑎1 𝑎03 𝑎3 ′
ℎ1 (𝑠) = ( 2 1+ 4 ( ) ),
𝑘 𝑎𝑜
𝑎03 𝑎33 𝑎33
4
1 1 𝑎3′′ 𝑎𝑜 − 𝑎3′ 𝑎𝑜′ 1 𝑎3′ 𝑎𝑜′ + 𝑎3 𝑎𝑜′′ 𝑎03 ( 𝑎𝑜′ )2
ℎ2 (𝑠) = ( 5 ( 1 ) − 5 ( 1 ) − 2 2
3𝑘 2 3 (𝑠) 3
3
𝑎3 𝑎𝑜 3
𝑎3 𝑎𝑜 𝑎33
8 2
2 𝑎03 𝑎3 ′ 2 𝑎1 𝑎03 𝑎3 ′ 1 𝑎2
− ( 8 ) ( ( ) ) + 5 ( ) + ( 1 2 ) ).
3 𝑎𝑜 𝑎𝑜 3
𝑎33 𝑎33 𝑎03 𝑎33
If ℎ1 (𝑠) = ℎ2 (𝑠) = constant, then the resultant equation become L.O.D.E. with
constant coefficients.

Example: Determine the complete integral to the following spectral D.E.

𝑑3 𝑦 𝑑2 𝑦 𝑑𝑦
3𝑥 3 𝑑𝑥 3 + 9𝑥 2 𝑑𝑥 2 + 3𝑥 𝑑𝑥 = −24𝜆3 𝑦.
𝑑3 𝑦 𝑑2 𝑦 𝑑𝑦
Solution: 3𝑥 3 𝑑𝑥 3 + 9𝑥 2 𝑑𝑥 2 + 3𝑥 𝑑𝑥 + 24𝜆3 𝑦 = 0. (5)
If we compare equation (5) with equation (3), we get that:
𝑎0 = 3𝑥 3 , 𝑎1 = 9𝑥 2 , 𝑎2 = 3𝑥, and 𝑎3 = 24.
Now
1 1
𝑎3 3 24 3 1
𝑠 = 𝑘 ∫ ( ) 𝑑𝑥 = 𝑘 ∫ ( 3 ) 𝑑𝑥 = 2𝑘 ∫ 𝑑𝑥 = 2𝑘 𝑙𝑛|𝑥| + 𝑐.
𝑎0 3𝑥 𝑥
Let 𝑥 > 0, thus 𝑠 = 2𝑘 𝑙𝑛𝑥 + 𝑐, and setting 2𝑘 = 1, and 𝑐 = 0,
𝑑𝑥
𝑠 = 𝑙𝑛𝑥, then 𝑥 = 𝑒 𝑠 and = 𝑒𝑠.
𝑑𝑠

21
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

𝑑𝑦 𝑑𝑦 𝑑𝑠 1 𝑑𝑦
= = ( 𝑠) ,
𝑑𝑥 𝑑𝑠 𝑑𝑥 𝑒 𝑑𝑠
𝑑2𝑦 𝑑 𝑑𝑦 𝑑 1 𝑑𝑦 𝑑𝑠 1 −𝑠
𝑑𝑦 1 𝑑2𝑦
= ( ) = (( ) ) = (−𝑒 + ( ) ),
𝑑𝑥 2 𝑑𝑥 𝑑𝑥 𝑑𝑠 𝑒 𝑠 𝑑𝑠 𝑑𝑥 𝑒 𝑠 𝑑𝑠 𝑒 𝑠 𝑑𝑠 2
𝑑2 𝑦 1 𝑑2𝑦 1 𝑑𝑦
2
= ( 2𝑠
) 2
− ( 2𝑠 ) ,
𝑑𝑥 𝑒 𝑑𝑠 𝑒 𝑑𝑠
𝑑3𝑦 𝑑 𝑑2𝑦 𝑑 1 𝑑2 𝑦 1 𝑑𝑦 𝑑𝑠
3
= ( 2
) = (( 2𝑠
) 2
− ( 2𝑠 ) )
𝑑𝑥 𝑑𝑥 𝑑𝑥 𝑑𝑠 𝑒 𝑑𝑠 𝑒 𝑑𝑠 𝑑𝑥

𝑑 1 𝑑2𝑦 𝑑 1 𝑑𝑦 1
= ( (( 2𝑠 ) ) + ((− ) )) ,
𝑑𝑠 𝑒 𝑑𝑠 2 𝑑𝑠 𝑒 2𝑠 𝑑𝑠 𝑒𝑠

𝑑3𝑦 1 𝑑3 𝑦 −2𝑠
𝑑2𝑦 −2𝑠
𝑑𝑦 −2𝑠
𝑑2 𝑦 1
= (( 2𝑠 ) 3 − 2𝑒 + 2𝑒 −𝑒 ) ,
𝑑𝑥 3 𝑒 𝑑𝑠 𝑑𝑠 2 𝑑𝑠 𝑑𝑠 2 𝑒 𝑠
𝑑3𝑦 1 𝑑3𝑦 3 𝑑2𝑦 2 𝑑𝑦
= ( ) − ( ) + ( ) .
𝑑𝑥 3 𝑒 3𝑠 𝑑𝑠 3 𝑒 3𝑠 𝑑𝑠 2 𝑒 3𝑠 𝑑𝑠
By placing the first, second, and third derivatives in the given differential equations,
we obtain:
1 𝑑3𝑦 3 𝑑2𝑦 2 𝑑𝑦 1 𝑑2𝑦 1 𝑑𝑦
3 𝑒 3𝑠 (( 3𝑠 ) 3 − ( 3𝑠 ) 2 + ( 3𝑠 ) ) + 9𝑒 2𝑠 (( 2𝑠 ) 2
− ( 2𝑠 ) ) +
𝑒 𝑑𝑠 𝑒 𝑑𝑠 𝑒 𝑑𝑠 𝑒 𝑑𝑠 𝑒 𝑑𝑠
1 𝑑𝑦
3𝑒 𝑠 (( 𝑠 ) ) + 24𝜆3 𝑦 = 0,
𝑒 𝑑𝑠
3
𝑑 𝑦 𝑑2𝑦 𝑑𝑦 𝑑𝑦 𝑑 2 𝑦 𝑑𝑦
− 3 + 2 − 3 + 3 + + 8𝜆3 𝑦 = 0,
𝑑𝑠 3 𝑑𝑠 2 𝑑𝑠 𝑑𝑠 𝑑𝑠 2 𝑑𝑠
𝑑3𝑦
3
+ 8𝜆3 𝑦 = 0,
𝑑𝑠
which is a linear spectral ordinary equation of third order with constant equations in
variable 𝑦 and 𝑠, hence the general solution is:
𝑦(𝑠) = 𝑐1 𝑒 −2𝜆𝑠 + 𝑒 𝜆𝑠 (𝑐2 cos √3 𝜆𝑠 + 𝑐3 sin √𝑠 𝜆𝑠),
or
𝑦(𝑥) = 𝑐1 𝑒 −2𝜆 𝑙𝑛𝑥 + 𝑒 𝜆 𝑙𝑛𝑥 (𝑐2 cos √3 𝜆 𝑙𝑛𝑥 + 𝑐3 sin √𝑠 𝜆 𝑙𝑛𝑥),
𝑐1
𝑦(𝑥) = 2𝜆 + 𝑥 𝜆 (𝑐2 cos √3 𝜆 𝑙𝑛𝑥 + 𝑐3 sin √𝑠 𝜆 𝑙𝑛𝑥 ),
𝑥
where 𝑐𝑖 , 𝑖 = 1,2,3 are constants.

4. Conclusion
In this paper, we have presented some techniques for solving two types of spectral
linear ordinary differential equations which are second and third orders including two
distinct kinds of coefficients. The first one is several polynomials, while the other kind
is continuous functions and all these kinds are of real type. We have shown how these
kinds are converted to the constant coefficients by modifying the independent variable
to another one via the specified transformations. We brought examples to explain the
usefulness and applicability of our technique. The importance of this work is to opening
the way to use another techniques and ideas to solve other types of spectral linear
ordinary differential equations of higher order.

22
 Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020

Conflict of Interests.
There are non-conflicts of interest .

References
[1] Aryan Ali Mohammed, Eigenfunctions and Asymptotic Behavior of Eigenvalues to
The Given Boundary Value Problem with Eigenparameter in the Boundary
Conditions, Journal of Zankoy Sulaimani – Part A, vol. 18, no. 1, pp. 179-190,
2016.
[2] D. Arficho, Method for Solving Particular Solution of Linear Second Order
Ordinary Differential Equations, J Appl Computat Math, vol. 4, no. 2, pp.1-3,
2015.
[3] H. Hochtadt, On The Stability of Certain Second-Order Differential Equations, Sot.
Ind. Appl. Math 12, pp. 58-59, 1964.
[4] Karwan H. M. Jwamer and Aryan Ali M., Boundedness of Normalized
Eigenfunctions of The Spectral Problem in The Case of Weight Function
Satisfying The Lipschitz Condition, Journal of Zankoy Sulaimani – Part A, vol.
15, no. 1, pp. 79-94, 2013.
[5] Karwan H.M. Jwamer and Aryan Ali M., Study the Behavior of The Solution and
Asymptotic Behaviors of Eigenvalues of a Six Order Boundary Value
Problem,International Journal of Research and Reviews in Applied Sciences, vol.
13, no. 3, pp. 790-799, 2012.
[6] L. Cesari, Asymptotic Behavior and Stability Problems in Ordinary Differential
Equations, Springer-Verlag, Berlin, 1963.
[7] M. Marini, Criteri di Limitatezza per le Soluzioni dell’ Equazione Lineare de1
Second Ordine, Boll. U.M.I 4, pp. 154-165, 1975.
[8] M. Marini and P. Zezza, On the Asymptotic Behavior of the Solutions of a Class of
Second Order Linear Differential Equations, Journal of Differential Equations
28, pp. 1- 17, 1978.
[9] P. Hartman P and A. Wintner, On The Laplace-Fourier Transcendent, Amer. J.
Math71, pp. 367-372, 1949.
[10] P. Hartman, Ordinary Differential Equations, Wiley, New York, 1964.
[11] P. Johnson, K. Busawon and J. Barbot, Alternative Solution of The
Inhomogeneous Linear Differential Equation of order two, J. Math. Anal. Appl.
339, pp. 582–589, 2008.
[12] R. A. Moore, The Behavior of Solutions of a Linear Differential Equation of
Second Order, Pacific J. Math 5, pp. 125-145, 1955.
[13] R. Bellman, Stability Theory of Differential Equations, McGraw-Hill, New York,
1953.
[14] T. Yu. Gadzhieva, Analysis of Spectral Characteristics of One Nonself Adjoint
Problem with Smooth Coefficients, PhD thesis, Dagestan State University, South
of Russian, 2010.
[15] U. Richard, Serie Asintotiche per Una Classe di Equazioni Differenziali non
Oscillanti de1 Second Ordine, Uniw. Polit. Torino, Rend. Sent. Mat 23, pp. 171-
217, 1964.
[16] V. V. Nemytskii and V. V. Stepanov, Qualitative Theory of Differential
Equations, Princeton Univ. Press, Princeton, New Jersey, 1960.

23
 ‫‪Journal of University of Babylon for Pure and Applied Sciences, Vol. (28), No. (1): 2020‬‬

‫‪[17] W. A. Coppel, Stability and Asymptotic Behavior of Differential Equations, Heat,‬‬

‫‪Boston, Mass, 1965.‬‬
‫‪[18] W. A. Coppel, Disconjugacy, Springer-Verlag, Berlin, 1971.‬‬

‫الخالصة‬
‫عملنا هي عبارةعن إيجاد الحل العام لنوعين مختلفين من المعادالت التفاضلية العادية الخطية من النوع الطيفي‪ .‬النوع األول يتكون‬
‫من نوعين مختلفين من المعامالت من الرتبة الثانية‪ ،‬واحد منهم هو متعددات الحدود‪ ،‬حيث أن النوع اآلخر هو دوال مستمرة وكالهما‬
‫أيضا للعديد من المتعددات الحدود الحقيقية‪،‬‬
‫من أنواع حقيقية‪ .‬النوع الثاني للمعادالت هی من الرتبة الثالثة‪ ،‬وهنا المعامالت من هذا النوع ً‬
‫وفي كال النوعين من المعادالت التفاضلية العادية الخطية الطيفية ‪ ،‬يتم تحويل المعامالت إلى الثوابت عبر تغيير المتغير المستقل إلى‬
‫متغير جديد‪ .‬قدمنا أمثلة لشرح آليتنا‪.‬‬
‫الکلمات الدالة‪ :‬رتبة المعادالت التفاضلية‪ ،‬متعددات الحدود والدوال الحقيقية‪ ،‬الحل العام‪ ،‬تبديل المتغيرات‪.‬‬

‫‪24‬‬