P A M L

December
2014

PURE AND APPLIED

MATHEMATICS LETTERS
An International Journal

On the second order linear differential

equation
by
A. Hernández-Galeana, R. López-Vázquez and J. López-Bonilla

HCTM Technical Campus

P A M L December 2 0 1 4
 Pure and Applied Mathematics Letters 2(2014)31-34

Contents lists available at HCTM, Technical Campus

Pure and Applied Mathematics Letters

Journal homepage: www.pamletters.org
ISSN 2349-4956

On the second order linear differential equation

A. Hernández-Galeana1, R. López-Vázquez1, J. López-Bonilla 2
1
ESFM, Instituto Politécnico Nacional (IPN), Edif. 9, Col. Lindavista CP 07738, México DF
2
ESIME-Zacatenco, IPN, Edif. 5, 1er. Piso, Col. Lindavista CP 07738, México DF

Abstract
We consider the linear differential equation of second order and we accept to know one solution of the corresponding
homogeneous equation, then we show how to obtain one more solution for the homogeneous case and a particular solution for
the original equation.

Keywords: Linear differential equation, Abel-Liouville-Ostrogradski identity, Euler-Lagrange procedure.

AMS subject classification: 34A05, 34A30, 34C14

Article Information:
Received 18.06.2014 Revised 04.08.2014 Accepted 13.08.2014 Manuscript No. PAML-1801062014
Communicated by Prof. V. K. Kaushik

1 Introduction
Given a linear differential equation of second order, the first step is to obtain one solution of the corresponding homogeneous
equation (HE), after it is tradition to employ the Abel-Liouville-Ostrogradski identity to deduce the another solution of this HE.
Finally, the non-trivial ‘ansatz’ of Lagrange is applied to construct a particular solution of the original equation. Our technique
uses self-adjoint and exact operators to solve this differential problem, it is not necessary the Lagrange’s ansatz, in other words,
our approach justifies the variation of parameters method. Besides, the process here exhibited gives, in natural manner, a
simple deduction of the Lewis invariant associated to Ermakov-Milne-Pinney equation.
We accept that for the second order linear differential equation:
𝑝 𝑥 𝑦 ′′ + 𝑞 𝑥 𝑦 ′ + 𝑟 𝑥 𝑦 = 𝜙 𝑥 , (1)
it is known one solution 𝑦1 (𝑥) of the corresponding HE:

𝑝 𝑥 𝑦 ′′ + 𝑞 𝑥 𝑦 ′ + 𝑟 𝑥 𝑦 = 0, (2)
then (2) admits the solution [1, 5 and 19]:
𝑥 𝑊(𝜂 )
𝑦2 𝑥 = 𝑦1 𝑥 [𝑦1 𝜂 ]2
𝑑𝜂, (3)

with the expression of Abel-Liouville-Ostrogradski [5]:

___________________
Corresponding author: José Luis López-Bonilla , E-mail addresses: jlopezb@ipn.mx

31
 𝑥𝑞 𝜉
𝑊 𝑥 = exp − 𝑝 𝜉
𝑑𝜉 . (4)

Besides, a particular solution of (1) is given by:

𝑥 𝑦1 𝜉 𝜙(𝜉 ) 𝑥 𝑦2 𝜉 𝜙(𝜉)
𝑦𝑝 𝑥 = 𝑦2 𝑥 𝑑𝜉 − 𝑦1 𝑥 𝑑𝜉 , (5)
𝑝 𝜉 𝑊(𝜉 ) 𝑝 𝜉 𝑊(𝜉 )

obtained via the Variation of Parameters method [1, 5 and 19] introduced by Newton (Principia), Bernoulli [3], Euler (1741) and
Lagrange (1759, 1777).
In Sec. 2 we employ the concepts of exact and self-adjoint operators [7, 8, 12 and 16] to establish an alternative deduction of (3)
and (5). The Sec. 3 shows a simple manner to construct the Lewis invariant [13-15] which is a constant of motion for the
Ermakov [6]-Milne [17]-Pinney [18] equation.

2 Linear differential operator

The functions 𝑦𝑗 (𝑥) are solutions of (2), then 𝑦1 𝑝𝑦2 ′′ + 𝑞𝑦2 ′ + 𝑟𝑦2 − 𝑦2 𝑝𝑦1 ′′ + 𝑞𝑦1 ′ + 𝑟𝑦1 = 0, that is:
𝑑𝑊
𝑝 + 𝑞𝑊 =0, (6)
𝑑𝑥

where W is the wronskian [1, 5, 9 and 19] of 𝑦1 and 𝑦2 :

𝑊 = 𝑦1 𝑦2 ′ − 𝑦2 𝑦1 ′ . (7)
The integration of (6) and (7) gives (4) and (3), respectively.
The variation of parameters method uses the following ansatz to obtain a particular solution of (1):
𝑦𝑝 𝑥 = 𝑐2 𝑥 𝑦2 𝑥 + 𝑐1 𝑥 𝑦1 𝑥 , (8)

thus (1) implies:

𝑑𝐴
𝑝 + 𝑞 𝐴 + 𝑝 𝑐1 ′ 𝑦1 ′ + 𝑐2 ′ 𝑦2 ′ = 𝜙 , 𝐴 = 𝑐1 ′ 𝑦1 + 𝑐2 ′ 𝑦2 ,
𝑑𝑥
𝜙
where we accept that A = 0, therefore 𝑐1 ′ 𝑦1 + 𝑐2 ′ 𝑦2 = 0 and 𝑐1 ′ 𝑦1 ′ + 𝑐2 ′ 𝑦2 ′ = 𝑝
, with determinant = W ≠ 0, then
𝑦2 𝜙 𝑦1 𝜙
𝑐1 ′ = 𝑝𝑊
& 𝑐2 ′ = 𝑝𝑊
, and (5) is immediate from (8).

Now we shall employ the concepts of self-adjoint and exact operators [7, 8, 12 and 16] to exhibit an alternative approach to (3)
and (5). In fact, (1) can be written in the form:
𝑑2 𝑑
𝐿𝑦 = 𝜙 , 𝐿=𝑝 𝑑𝑥 2
+ 𝑞 𝑑𝑥
+ 𝑟, (9)

and it is known [7, 12] that 𝛾 𝑥 𝐿 = 𝑂 is a self-adjoint operator, 𝑂† = 𝑂 , for:

1 1 𝑥𝑞 𝜉
𝛾 𝑥 = 𝑝𝑊
= 𝑝
exp 𝑝 𝜉
𝑑𝜉 . (10)

Besides, 𝜌 𝑥 𝑂 is an exact operator [7]:

𝑑
𝑑𝑥
𝛼 𝑦′ + 𝛽 𝑦 = 𝜌 𝛾 𝜙 , 𝛼 = 𝜌𝛾𝑝, 𝛽 = 𝜌 𝛾 𝑞 − 𝛼′ , (11)

if ρ satisfies the differential equation:

𝛾 𝑝 𝜌′′ + 2 𝛾 𝑝 ′ − 𝛾 𝑞 𝜌′ + 𝛾𝑝 ′′
− 𝛾𝑞 ′+ 𝛾𝑟 𝜌=0,
where we can apply (10) to obtain the following constraint on ρ :
𝑝 𝜌′′ + 𝑞 𝜌′ + 𝑟 𝜌 = 0 , (12)
𝑦1
which coincides with the HE (2), then it is natural to take 𝜌 = 𝑦1 𝑥 . Therefore 𝜌 𝛾 = is an integrating factor for (1)
𝑝𝑊
because it adopts the structure (11) of simple resolution if only we know one solution of (2). From (11):
𝑦1 𝑦1 ′ 𝑑 𝑦1 2 𝑑 𝑦 𝑦1 𝜙
𝛼= 𝑊
, 𝛽= − 𝑊
, 𝑑𝑥 𝑊 𝑑𝑥 𝑦1
= 𝑝𝑊
, (13)

whose integration gives the general solution of (1):

32
 𝑥 𝑊
𝑦 𝑥 = 𝑎1 𝑦1 + 𝑎2 𝑦1 𝑦1 2
𝑑𝜉 + 𝑦𝑝 , (14)

such that:
𝑥 𝜂 𝑥 𝑥
𝑊(𝜂) 𝑦1 𝜉 𝜙(𝜉) 𝑦2 𝑦1 𝜙 𝑦2 𝜙
𝑦𝑝 = 𝑦1 𝑑𝜂 ∙ 𝑑𝜉 = 𝑦1 𝑑𝜉 − 𝑑𝜉 = 5 ,
𝑦1 2 (𝜂) 𝑝 𝜉 𝑊(𝜉) 𝑦1 𝑝𝑊 𝑝𝑊
where we employed integration by parts; thus (14) is in total harmony with (3) and (5). Therefore, the concepts of exact and self-
adjoint operators imply (13) and its resolution gives the particular solution of the variation of parameters method, but our
approach no need the ansatz (8).
In our process first we find γ(x) such that 𝛾𝐿 is a self-adjoint operator, and after it is multiplied by ρ(x) to obtain an exact
operator. Now we return to (1) multiplying by λ(x) to convert it to an exact form:
𝑑
𝑑𝑥
𝜇 𝑦′ + 𝜏 𝑦 = 𝜆 𝜙 , (15)

which is possible if λ(x) verifies the differential condition [7]:

𝑝 𝜆′′ + 2 𝑝′ − 𝑞 𝜆′ + 𝑝′′ − 𝑞 ′ + 𝑟 𝜆 = 0 , 𝜇 = 𝜆𝑝, 𝜏 = 𝜆 𝑞 − 𝜇′ . (16)

In several situations we may have the property [4]:

𝑝′′ − 𝑞 ′ + 𝑟 = 0 , (17)
then the differential equation (16) admits the solution λ = 1, therefore 𝜇 = 𝑝 & 𝜏 = 𝑞 − 𝑝′ , and (15) adopts the structure:
𝑑 𝑑 𝑦
𝑑𝑥
𝑝2 𝑊 𝑑𝑥 𝑝𝑊
= 𝜙, (18)

whose integration gives the complete solution of (1); in (18) we note the absence of 𝑦1 𝑥 . In resumé, if p, q, r verify (17) then
to employ (18); to use (13) if (17) is violated, and in both cases to utilize (4).

3 Ermakov-Milne-Pinney equation
Here we shall apply the analysis developed in Sec. 2 to give a simple deduction of the Lewis constant [13 -15] associated to
Ermakov-Milne-Pinney equation [6, 17 and 18]:

𝑦 ′′ + 𝑟 𝑦 = 𝑎 𝑦 −3 , 𝑎 > 0, (19)
which corresponds to (1) with p = 1, q = 0, W = 1, 𝜙 = 𝑎 𝑦 −3 , then from (5):
𝑥 𝑎𝑦1 𝑥 𝑎𝑦2
𝑦𝑝 = 𝑦2 𝑑𝜂 − 𝑦1 𝑑𝜂 , 𝑦𝑗 ′′ + 𝑟 𝑦𝑗 = 0, 𝑗 = 1, 2 (20)
𝑦𝑝 3 𝑦𝑝 3

It is easy to calculate the wronskian:

𝑥 𝑎𝑦2
𝑦𝑝 𝑦2 ′ − 𝑦2 𝑦𝑝 ′ = − 𝑦𝑝 3
𝑑𝜂 ,

because W = 1, thus:
𝑥 𝑥 2
𝑦2 𝑑 𝑦2 𝑦2 −3
1 𝑑 −3
= − 3 𝑎𝑦2 𝑦𝑝 𝑑𝜂 = − ( 𝑎𝑦2 𝑦𝑝 𝑑𝜂) ,
𝑦𝑝 𝑑𝑥 𝑦𝑝 𝑦𝑝 2𝑎 𝑑𝑥
therefore
𝑑 𝑦2 2 1 𝑑
( ) = − (𝑦 𝑦 ′ − 𝑦2 𝑦𝑝 ′ )2 ,
𝑑𝑥 𝑦𝑝 𝑎 𝑑𝑥 𝑝 2
𝑑𝐼
that is, = 0 where I is the Lewis invariant [13-15]:
𝑑𝑥
𝑦
𝐼 = 𝑎 (𝑦 2 )2 + (𝑦𝑝 𝑦2 ′ − 𝑦2 𝑦𝑝 ′ )2 (Constant of Motion ) (21)
𝑝

In [10 and 11] we find that:

𝑦𝑝 𝑦𝑝 𝑥 1
𝑦1 = 𝑎 1/4
sin 𝐵(𝑥) , 𝑦2 = 𝑎 1/4
cos 𝐵(𝑥) , 𝐵 𝑥 = − 𝑎 𝑦𝑝 2
𝑑𝜂 + 𝜑 , (22)

which verify (20), besides [18]:

33
 𝑦𝑝 = 𝑏1 𝑦1 2 + 2𝑏2 𝑦1 𝑦2 + 𝑏3 𝑦2 2 , 𝑏1 = 𝑏3 = 𝑎, 𝑏2 = 0 . (23)

In (21) we can employ (22) for 𝑦2 and thus to deduce the exact value of the Lewis constant, 𝐼 = 𝑎.

4 Conclusions
Our approach based in exact and self-adjoint operators justifies the ansatz utilized in the variation of parameters method.
Besides, our process has strong support in the concept of wronskian, which generates applications to important laws as the
Ermakov-Milne-Pinney equation. We consider that is easy to generalize our technique to linear differential equations of third
and fourth order [2]. The self-adjoint operators are important in quantum mechanics and in the construction of Green’s
functions [7 and 12], then there our process is interesting to solve the corresponding differential equations.

References
[1] Ahsan Z., Differential equations and their applications, Prentice-Hall, New Delhi, (2004).
[2] Ahsan Z., Cruz-Santiago R. and López-Bonilla J., Linear differential equations of third and fourth order, Aligarh Bull.
Maths., 13(1) (2012), pp. 5-7.
[3] Bernoulli J., Acta Eruditorum, (1697).
[4] Clegg J., A new factorization of a general second order differential equation, Int. J. Math. Educ. Sci. Techn., 37(1) (2006),
pp. 51-64.
[5] Elsgotz L., Differential equations and variational calculus, MIR, Moscow, (1983).
[6] Ermakov V. P., Second-order differential equations, Conditions of complete integrability, Univ. Izv. Kiev 20 (1880), pp. 1-
25
[7] Greenberg M. D., Application of Green’s function in science and engineering, Prentice-Hall, New Jersey, (1971).
[8] Hernández-Galeana A., López-Bonilla J. and Rivera-Rebolledo J., Second order linear differential equation in its exact
form, The SciTech, J. of Science & Technology, 2(1) (2013), pp. 34-36.
[9] Hoene Wronski J. M., Réfutation de la théorie des fonctions analytiques de Lagrange, Paris, (1812).
[10] Ioffe M. and Korsch H. J., Nonlinear supersymmetric (Darboux) covariance of the Ermakov-Milne-Pinney equation, Phys.
Lett., A 311 (2003), pp. 200-205.
[11] Kryachko E. S., Few sketches on connections between the Riccati and Ermakov-Milne-Pinney equations, Int. J. Quantum
Chem., 109(13) (2009), pp. 2897-2902.
[12] Lanczos C., Linear differential operators, Dover, New York, (1997).
[13] Leach P. G. L. and Andriopoulos K., The Ermakov equation: A commentary, Appl. Anal. Discrete Math., 2 (2008), pp. 146-
157.
[14] Lewis H.R., Classical and quantum systems with time-dependent harmonic oscillator-type Hamiltonians, Phys. Rev. Lett.,
18 (1967), pp. 510-512.
[15] Lewis H. R., Class of exact invariants for classical and quantum time-dependent harmonic oscillators, J. Math.Phys., 9(11)
(1968), pp. 1976-1986.
[16] López-Bonilla J., Yaljá Montiel J. and Zaldívar A., 2th order linear differential operator in its exact form, J. Vect. Rel., 5(1)
(2010), pp. 139-141.
[17] Milne W., The numerical determination of characteristic numbers, Phys. Rev., 35(7) (1930), pp. 863-867.
[18] Pinney E., The nonlinear differential equation y’’ + p(x)y +cy -3 = 0, Proc. Am. Math. Soc., 1(5) (1950), pp. 681-681.
[19] Soare M. V., Teodorescu P. and Toma I., Ordinary differential equations with applications to mechanics, Springer, Berlin,
(2007).

34