N Thorderlinearnonhomogeneousdifferentialequations PDF
N Thorderlinearnonhomogeneousdifferentialequations PDF
N Thorderlinearnonhomogeneousdifferentialequations PDF
net/publication/329861238
CITATION READS
1 5,629
1 author:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Complex Computer Networks of Optimum Performance, Combinatorial Optimization and Graph Enumeration. View project
Numerical Algorithms that Acquire the Status of Full Proofs. View project
All content following this page was uploaded by Carlos Enrique Frasser on 22 December 2018.
Carlos E. Frasser
1. Preliminaries.
where P, Q and R are given functions of x. If R(x) 0, the DE is called linear nonhomogeneous.
If R(x) = 0, then (1) has the form
Theorem 1. If f(x) and g(x) are two particular solutions of the second order linear homogeneous
DE (2), then f(x) + g(x) is also a solution of (2).
𝑓 ′′ + 𝑃𝑓 ′ + 𝑄𝑓 = 0
𝑔′′ + 𝑃𝑔′ + 𝑄𝑔 = 0.
Substituting the sum f + g into (2) and keeping in mind the last two identities, we obtain
Theorem 2. If f is a particular solution of (2) and C is a constant, then the product Cf is also a
solution of (2). □
1
Two solutions f and g of (2) are said to be linearly independent (LI) on the segment [a, b] if their
ratio is not a constant number, that is,
𝑓
≠ 𝑐𝑜𝑛𝑠𝑡.
𝑔
Otherwise, the solutions are said to be linearly dependent (LD.) Thus, two solutions f and g are
said to be LD if there exists a constant number λ such that f /g = λ (or f = λg) on the segment
a ≤ x ≤ b.
𝑦 = 𝐶1 𝑓 + 𝐶2 𝑔, (3)
where C1 and C2 are arbitrary constants (sometimes called parameters), is the general solution of
(2). □
Theorem 4. Let f(x) and p(x) be a particular solution of (2) and a particular solution of (1),
respectively. Then
𝑓 ′′ + 𝑃𝑓 ′ + 𝑄𝑓 = 0.
𝑝′′ + 𝑃𝑝′ + 𝑄𝑝 = 𝑅.
Remark 1: Generalizing the result of Theorem 4, we can say that if C1f(x) + C2g(x) is the
general solution of (2) and p(x) is a particular solution of (1), then
2
𝑒 − ∫ 𝑃(𝑥)𝑑𝑥
𝑦 = 𝑓(𝑥) ∫ 𝑑𝑥 (6)
[𝑓(𝑥)]2
Proof: If f(x) is a solution of (2), then, by Theorem 2, y = Cf(x) is also a solution of (2). By using
the method of variation of parameters, let us assume that C is a function of x, say u(x). Therefore
𝑦 = 𝑢(𝑥)𝑓(𝑥) (7)
𝑦 ′ = 𝑢′ 𝑓 + 𝑢𝑓 ′ (8)
𝑧 = 𝑢′ 𝑧 ′ = 𝑢′′ .
Therefore,
𝑧 ′ 𝑓 + 𝑧(2𝑓 ′ + 𝑃𝑓) = 0
′
2𝑓 ′ + 𝑃𝑓
𝑧 + ( ) 𝑧 = 0,
𝑓
2𝑓 ′ +𝑃𝑓 2𝑓 ′ +𝑃𝑓
−∫ 𝑑𝑥 −∫ 𝑑𝑥
𝑧 = 𝐶1 𝑒 𝑓 = 𝑒 𝑓
𝑧 = 𝑒 −2 ln 𝑓 𝑒 − ∫ 𝑃𝑑𝑥
−2
𝑧 = 𝑒 𝑙𝑛𝑓 𝑒 − ∫ 𝑃𝑑𝑥
1 − ∫ 𝑃𝑑𝑥
𝑧= 𝑒
𝑓2
3
1 − ∫ 𝑃𝑑𝑥
𝑢′ = 𝑒
𝑓2
𝑒 − ∫ 𝑃(𝑥)𝑑𝑥
∫ 𝑑𝑢 = ∫ 𝑑𝑥 + 𝐶2
[𝑓(𝑥)]2
𝑒 − ∫ 𝑃(𝑥)𝑑𝑥
𝑢 =∫ 𝑑𝑥
[𝑓(𝑥)]2
𝑦 𝑒 − ∫ 𝑃(𝑥)𝑑𝑥
=∫ 𝑑𝑥
𝑓(𝑥) [𝑓(𝑥)]2
𝑒 − ∫ 𝑃(𝑥)𝑑𝑥
𝑦 = 𝑓(𝑥) ∫ 𝑑𝑥. □
[𝑓(𝑥)]2
𝑑 𝑑𝑦 𝑑2 𝑦
𝐷2 𝑦 = 𝐷(𝐷𝑦) = ( )= ,
𝑑𝑥 𝑑𝑥 𝑑𝑥 2
3 2
𝑑 𝑑2𝑦 𝑑3 𝑦
𝐷 𝑦 = 𝐷(𝐷 𝑦) = ( )=
𝑑𝑥 𝑑𝑥 2 𝑑𝑥 3
and so on. A polynomial in D must be interpreted as an operator that once applied to y gives a
linear combination of y and its successive derivatives. For instance,
𝑑 2 𝑦 𝑑𝑦
(𝐷2 + 𝐷 − 2)𝑦 = 𝐷2 𝑦 + 𝐷𝑦 − 2𝑦 = + − 2𝑦 = 𝑦 ′′ + 𝑦 ′ − 2𝑦
𝑑𝑥 2 𝑑𝑥
Such a polynomial in D is said to be a linear differential operator and we can designate it by the
letter L. If L1 and L2 are two linear differential operators, their sum and product are determined
by
Those linear differential operators that are polynomials in D with constant coefficients satisfy the
fundamental laws of algebraic computation, which makes possible deal with them as ordinary
polynomials concerning addition, multiplication and also factoring.
4
A second order linear homogeneous DE with constant coefficients has the form
𝑦 ′′ + 𝑎𝑦 ′ + 𝑏𝑦 = 0. (11)
Let us consider the algebraic equation (called characteristic equation) associated with (12)
(𝑟 2 + 𝑎𝑟 + 𝑏)𝑦 = 0 𝑜𝑟 𝑟 2 + 𝑎𝑟 + 𝑏 = 0. (13)
(𝑟 − 𝑟1 )(𝑟 − 𝑟2 ) = 0,
where
−𝑎 + √𝑎2 − 4𝑏 −𝑎 − √𝑎2 − 4𝑏
𝑟1 = , 𝑟2 = .
2 2
So from (12), we get
(𝐷 − 𝑟1 )(𝐷 − 𝑟2 )𝑦 = 0. (14)
𝑢 = (𝐷 − 𝑟2 )𝑦. (15)
It follows that
(𝐷 − 𝑟1 )𝑢 = 0
𝐷𝑢 − 𝑟1 𝑢 = 0
𝑢′ − 𝑟1 𝑢 = 0
whose solution is
𝑢 = 𝐶1 𝑒 − ∫ −𝑟1 𝑑𝑥 𝑜𝑟 𝑢 = 𝐶1 𝑒 𝑟1 𝑥 . (16)
𝑦 ′ + 𝑃(𝑥)𝑦 = 𝑄(𝑥)
5
then substituting (16) into (15), we obtain
(𝐷 − 𝑟2 )𝑦 = 𝐶1 𝑒 𝑟1 𝑥
𝐷𝑦 − 𝑟2 𝑦 = 𝐶1 𝑒 𝑟1 𝑥
𝑦 ′ − 𝑟2 𝑦 = 𝐶1 𝑒 𝑟1 𝑥 ,
which means
𝑐) 𝑒 ∫ 𝑃(𝑥)𝑑𝑥 = 𝑒 −𝑟2 𝑥 .
Therefore, the general solution of the second order linear homogeneous DE with constant
coefficients (11) is
Case I. Let us consider a2 - 4b > 0. This means that r1 r2 and r1, r2 R. Under these
conditions, (17) takes the following form
𝐶1
𝑦ℎ = 𝑒 𝑟2 𝑥 [ ∫(𝑟1 − 𝑟2 )𝑒 (𝑟1 − 𝑟2 )𝑥 𝑑𝑥 + 𝐶2 ]
𝑟1 − 𝑟2
𝐶1
𝑦ℎ = 𝑒 𝑟2 𝑥 [ 𝑒 (𝑟1 − 𝑟2 )𝑥 + 𝐶2 ]
𝑟1 − 𝑟2
𝐶1
𝑦ℎ = 𝑒 𝑟1 𝑥 + 𝐶2 𝑒 𝑟2 𝑥
𝑟1 − 𝑟2
𝐶1
𝑦ℎ = 𝐴𝑒 𝑟1 𝑥 + 𝐵𝑒 𝑟2 𝑥 ; 𝐴= , 𝐵 = 𝐶2 . (18)
𝑟1 − 𝑟2
Case II. Assume now that a2 - 4b = 0. This means that r1 = r2 = r, r R. Then (17) takes the
following form
𝑦ℎ = 𝑒 𝑟𝑥 [𝐶1 ∫ 𝑒 (𝑟−𝑟)𝑥 𝑑𝑥 + 𝐶2 ]
6
𝑦ℎ = 𝑒 𝑟𝑥 (𝐶1 ∫ 𝑑𝑥 + 𝐶2 )
𝑦ℎ = 𝑒 𝑟𝑥 (𝐶1 𝑥 + 𝐶2 )
𝑦ℎ = 𝐶1 𝑥𝑒 𝑟𝑥 + 𝐶2 𝑒 𝑟𝑥 (19)
Case III. Lastly, assume that a2 - 4b < 0. This means that r1, r2 C and
𝑎 √4𝑏 − 𝑎2 𝑎 √4𝑏 − 𝑎2
𝑟1 = − + 𝑖 , 𝑟2 = − − 𝑖 . (20)
2 2 2 2
If we make
𝑎 √4𝑏 − 𝑎2
𝛼̅ = − , ̅ =
2 2
then
𝑟1 = 𝛼̅ + 𝑖 ̅ , 𝑟2 = 𝛼̅ − 𝑖 ̅ . (21)
𝑦ℎ = 𝐴𝑒 𝑟1 𝑥 + 𝐵𝑒 𝑟2 𝑥
̅ ̅
𝑦ℎ = 𝐴𝑒 (𝛼̅ + 𝑖 )𝑥 + 𝐵𝑒 (𝛼̅ − 𝑖 )𝑥
̅ ̅
𝑦ℎ = 𝑒 𝛼̅𝑥 (𝐴𝑒 𝑖 𝑥 + 𝐵𝑒 −𝑖 𝑥 ) . (22)
Finally, if we make
𝐶1 = 𝐴 + 𝐵, 𝐶2 = 𝑖 (𝐴 − 𝐵),
7
then (24) takes the form
Remark 2: The arbitrary constants C1 and C2 of (25) are real as long as the constants A and B of
(24) are conjugate imaginaries:
1 1
𝐴= (𝐶 − 𝑖 𝐶2 ), 𝐵= (𝐶 + 𝑖 𝐶2 ). (26)
2 1 2 1
Under conditions (26);
1 𝑖 1 𝑖
𝐴+𝐵 = 𝐶1 − 𝐶2 + 𝐶1 + 𝐶2 = 𝐶1
2 2 2 2
and
1 𝑖 1 𝑖
𝑖 (𝐴 − 𝐵) = 𝑖 ( 𝐶1 − 𝐶2 − 𝐶1 − 𝐶2 ) = 𝑖 (−𝑖 𝐶2 ) = −𝑖 2 𝐶2 = 𝐶2 ,
2 2 2 2
which is true according to what we assumed previously.
where C1, C2 are constants and functions u1, u2 are determined according to the algebraic sign of
the discriminant as follows:
√𝑑
(a) If 𝑑 > 0, then 𝑢1 (𝑥) = 𝑒 𝑘𝑥 , 𝑢2 (𝑥) = 𝑒 −𝑘𝑥 , 𝑘 = .
2
(b) If 𝑑 = 0, then 𝑢1 (𝑥) = 1, 𝑢2 (𝑥) = 𝑥.
√−𝑑
(c) If 𝑑 < 0, then 𝑢1 (𝑥) = cos ̅ 𝑥 , 𝑢2 (𝑥) = sin ̅ 𝑥 , ̅ = .□
2
𝑢 (𝑥) 𝑢2 (𝑥)
𝑊(𝑥) = | 1 | = 𝑢1 (𝑥)𝑢2′ (𝑥) − 𝑢2 (𝑥) 𝑢1 ′(𝑥)
𝑢1 ′(𝑥) 𝑢2 ′(𝑥)
A second order linear nonhomogeneous DE with constant coefficients has the form
𝑦 ′′ + 𝑎𝑦 ′ + 𝑏𝑦 = 𝑅(𝑥). (27)
8
Its general solution (see Remark 1) is written in the form
𝑦 = 𝑦ℎ + 𝑦𝑝 ,
The following theorem allows us to find yp using the method of variation of parameters.
where
𝑅(𝑥)
𝑀(𝑥) = − ∫ 𝑔(𝑥) 𝑑𝑥,
𝑊(𝑥)
𝑅(𝑥)
𝑁(𝑥) = ∫ 𝑓(𝑥) 𝑑𝑥.
𝑊(𝑥)
𝑦𝑝 = 𝑀𝑓 + 𝑁𝑔 (28)
𝑓 ′′ + 𝑎𝑓 ′ + 𝑏𝑓 = 0, 𝑔′′ + 𝑎𝑔′ + 𝑏𝑔 = 0.
9
𝑀(𝑓 ′′ + 𝑎𝑓 ′ + 𝑏𝑓) + 𝑁(𝑔′′ + 𝑎𝑔′ + 𝑏𝑔) + (𝑀′ 𝑓 ′ + 𝑁 ′ 𝑔′ ) + (𝑀′ 𝑓 + 𝑁 ′ 𝑔)′ + 𝑎(𝑀′ 𝑓 +
𝑁 ′ 𝑔) = 𝑅,
Making 𝑀′ 𝑓 + 𝑁 ′ 𝑔 = 0,
then 𝑀′ 𝑓 ′ + 𝑁 ′ 𝑔′ = 𝑅,
0 𝑔
| |
′ 𝑅 𝑔′ 𝑅𝑔 𝑅
𝑀 = = − ⇒ 𝑀 = − ∫ 𝑔 𝑑𝑥
𝑓 𝑔 𝑊 𝑊
| |
𝑓′ 𝑔′
𝑓 0
| ′ |
′ 𝑓 𝑅 𝑓𝑅 𝑅
𝑁 = = ⇒ 𝑁 = ∫ 𝑓 𝑑𝑥. □
𝑓 𝑔 𝑊 𝑊
| ′ ′ |
𝑓 𝑔
A DE of the form
where coefficients a1, a2,…,an are real constants, is called an nth-order linear nonhomogeneous
DE with constant coefficients. As in the case of the second order DE, if R(x) = 0, (31) takes the
form
Let us assume that 𝑟1 , 𝑟2 , … , 𝑟𝑛 are the roots of (33). Then the following cases may be
considered:
(a) 𝑟1 , 𝑟2 , … , 𝑟𝑛 are distinct real numbers. In this case, the fundamental system of solutions
of (32) takes the form
10
𝑒 𝑟1 𝑥 , 𝑒 𝑟2 𝑥 , … , 𝑒 𝑟𝑛𝑥
𝑦ℎ = 𝐶1 𝑒 𝑟1 𝑥 + 𝐶2 𝑒 𝑟2 𝑥 + ⋯ + 𝐶𝑛 𝑒 𝑟𝑛𝑥 . (34)
(b) The roots of the characteristic equation are real, but some of them are repeated. Assume
that 𝑟1 = 𝑟2 = ⋯ = 𝑟𝑘 = 𝑟 so that r is a root of (33) that has multiplicity k while the
n – k remaining ones are distinct roots. In this case, the fundamental system of solutions
takes the form
(c) Some of the roots of the characteristic equation are complex. Assume that
𝑟1 = 𝛼̅ + 𝑖 ̅ , 𝑟2 = 𝛼̅ − 𝑖 ̅ , 𝑟3 = 𝛾̅ + 𝑖 𝛿̅ , 𝑟4 = 𝛾̅ − 𝑖 𝛿̅ , ̅ , 𝛿̅ ≠ 0,
and the remaining roots are real and different. In this case, the fundamental system of
solutions takes the form
Just like in the case of the second order linear DE, the general solution of (31) is the sum of the
general solution yh of the corresponding homogeneous DE and a particular solution yp of the
given nonhomogeneous DE:
𝑦 = 𝑦ℎ + 𝑦𝑝 .
When R(x) has a special form, a particular solution yp of (31) can be found by using the so-called
Method of Undetermined Coefficients.
11
𝑛
𝛼𝑥
𝑦𝑝 = 𝑒 ∑ 𝐵𝑘 𝑥 𝑘 . (38)
𝑘=0
𝑦𝑝 = 𝑥 𝜇 𝑒 𝛼𝑥 ∑ 𝐵𝑘 𝑥 𝑘 . (39)
𝑘=0
𝑞 𝑞
𝑞 𝑞
𝜇 𝛼𝑥
𝑦𝑝 = 𝑥 𝑒 (cos 𝛽𝑥 ∑ 𝐶𝑘 𝑥 + sin 𝛽𝑥 ∑ 𝐷𝑘 𝑥 𝑘 )
𝑘
(42)
𝑘=0 𝑘=0
When R(x) is of the form either (37) or (40), particular solutions of (31) are given by formulas
(39), (41) and (42) using the method of undetermined coefficients. However, when R(x) does not
have any of these two given forms, the method of undetermined coefficients is not suitable
anymore and the method of variation of parameters now offers an alternative technique to find a
particular solution of (31).
where
12
𝑅(𝑥)[𝑓2 (𝑥)𝑓3′ (𝑥) − 𝑓2 ′(𝑥)𝑓3 (𝑥)]
𝑀1 (𝑥) = ∫ 𝑑𝑥,
𝑊(𝑥)
Proof: According to our assumption, f1, f2, f3 are solutions of 𝑦 ′′′ + 𝑎𝑦 ′′ + 𝑏𝑦 ′ + 𝑐𝑦 = 0. Then
𝑦𝑝 = 𝑀1 𝑓1 + 𝑀2 𝑓2 + 𝑀3 𝑓3
𝑦𝑝′′ = (𝑀1 𝑓1 ′′ + 𝑀2 𝑓2 ′′ + 𝑀3 𝑓3 ′′) + 2(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′) + (𝑀1 ′′𝑓1 + 𝑀2 ′′𝑓2 + 𝑀3 ′′𝑓3 )
𝑦𝑝′′′ = 𝑀1 𝑓1′′′ + 𝑀1′ 𝑓1 ′′ + 𝑀2 𝑓2′′′ + 𝑀2 ′𝑓2 ′′ + 𝑀3 𝑓3′′′ + 𝑀3 ′𝑓3 ′′ + 2(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′)′
+ (𝑀1 ′′𝑓1 + 𝑀2 ′′𝑓2 + 𝑀3 ′′𝑓3 )′.
𝑀1 𝑓1′′′ + 𝑀1′ 𝑓1 ′′ + 𝑀2 𝑓2′′′ + 𝑀2 ′𝑓2 ′′ + 𝑀3 𝑓3′′′ + 𝑀3 ′𝑓3 ′′ + 2(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′)′
+ (𝑀1 ′′𝑓1 + 𝑀2 ′′𝑓2 + 𝑀3 ′′𝑓3 )′ + 𝑎𝑀1 𝑓1 ′′ + 𝑎𝑀2 𝑓2 ′′ + 𝑎𝑀3 𝑓3 ′′
+ 2𝑎(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′) + 𝑎(𝑀1 ′′𝑓1 + 𝑀2 ′′𝑓2 + 𝑀3 ′′𝑓3 ) + 𝑏𝑀1 𝑓1 ′
+ 𝑏𝑀1 ′𝑓1 + 𝑏𝑀2 𝑓2 ′ + 𝑏𝑀2 ′𝑓2 + 𝑏𝑀3 𝑓3 ′ + 𝑏𝑀3 ′𝑓3 + 𝑐𝑀1 𝑓1 + 𝑐𝑀2 𝑓2 + 𝑐𝑀3 𝑓3
=𝑅
which means
𝑀1 (𝑓1′′′ + 𝑎𝑓1′′ + 𝑏𝑓1′ + 𝑐𝑓1 ) + 𝑀2 (𝑓2′′′ + 𝑎𝑓2′′ + 𝑏𝑓2′ + 𝑐𝑓2 ) + 𝑀3 (𝑓3′′′ + 𝑎𝑓3′′ + 𝑏𝑓3′ + 𝑐𝑓3 )
+ (𝑀1 ′𝑓1 ′′ + 𝑀2 ′𝑓2 ′′ + 𝑀3 ′𝑓3 ′′) + 2(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′)′
+ (𝑀1 ′′𝑓1 + 𝑀2 ′′𝑓2 + 𝑀3 ′′𝑓3 )′ + 2𝑎(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′)
+ 𝑎(𝑀1 ′′𝑓1 + 𝑀2 ′′𝑓2 + 𝑀3 ′′𝑓3 ) + 𝑏(𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 ) = 𝑅
13
Using the identities established at the beginning of this proof and after making some
transformations, we get from the last DE
(𝑀1 ′𝑓1 ′′ + 𝑀2 ′𝑓2 ′′ + 𝑀3 ′𝑓3 ′′) + (𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′)′ + (𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 )′′
+ 𝑎(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′) + 𝑎(𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 )′
+ 𝑏(𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 ) = 𝑅.
0 𝑓2 𝑓3
| 0 𝑓2 ′ 𝑓3 ′ |
𝑅 𝑓2 ′′ 𝑓3 ′′ 𝑅(𝑓2 𝑓3′ − 𝑓2 ′𝑓3 ) 𝑅(𝑓2 𝑓3′ − 𝑓2 ′𝑓3 )
𝑀1′ = = ⇒ 𝑀1 = ∫ 𝑑𝑥,
𝑓1 𝑓2 𝑓3 𝑊 𝑊
| 𝑓1 ′ 𝑓2 ′ 𝑓3 ′ |
𝑓1 ′′ 𝑓2 ′′ 𝑓3 ′′
𝑓1 0 𝑓3
| 𝑓1 ′ 0 𝑓3 ′ |
𝑓1 ′′ 𝑅 𝑓3 ′′ 𝑅(𝑓1 𝑓3′ − 𝑓1 ′𝑓3 ) 𝑅(𝑓1 𝑓3′ − 𝑓1 ′𝑓3 )
𝑀2′ = = − ⇒ 𝑀2 = − ∫ 𝑑𝑥,
𝑓1 𝑓2 𝑓3 𝑊 𝑊
| 𝑓1 ′ 𝑓2 ′ 𝑓3 ′ |
𝑓1 ′′ 𝑓2 ′′ 𝑓3 ′′
𝑓1 𝑓2 0
| 𝑓1 ′ 𝑓2 ′ 0 |
𝑓1 ′′ 𝑓2 ′′ 𝑅 𝑅(𝑓1 𝑓2′ − 𝑓1 ′𝑓2 ) 𝑅(𝑓1 𝑓2′ − 𝑓1 ′𝑓2 )
𝑀3′ = = ⇒ 𝑀3 = ∫ 𝑑𝑥. □
𝑓1 𝑓2 𝑓3 𝑊 𝑊
| 𝑓1 ′ 𝑓2 ′ 𝑓3 ′ |
𝑓1 ′′ 𝑓2 ′′ 𝑓3 ′′
14
Theorem 9. Let f1, f2, f3, f4 be LI solutions of 𝑦 (4) + 𝑎𝑦 ′′′ + 𝑏𝑦 ′′ + 𝑐𝑦 ′ + 𝑑𝑦 = 0. Let W be the
Wronskian of functions f1, f2, f3, f4. Then 𝑦 (4) + 𝑎𝑦 ′′′ + 𝑏𝑦 ′′ + 𝑐𝑦 ′ + 𝑑𝑦 = 𝑅(𝑥) has a particular
solution yp given by
𝑦𝑝 = 𝑀1 𝑓1 + 𝑀2 𝑓2 + 𝑀3 𝑓3 + 𝑀4 𝑓4
where
Proof. Assuming that f1, f2, f3, f4 are solutions of the fourth order homogeneous DE and
(4)
substituting the values of 𝑦𝑝 , 𝑦𝑝′ , 𝑦𝑝′′ , 𝑦𝑝′′′ , 𝑦𝑝 into the fourth order nonhomogeneous DE, we
get the DE
(𝑀1 ′𝑓1 ′′′ + 𝑀2 ′𝑓2 ′′′ + 𝑀3 ′𝑓3 ′′′ + 𝑀4 ′𝑓4 ′′′) + (𝑀1 ′𝑓1 ′′ + 𝑀2 ′𝑓2 ′′ + 𝑀3 ′𝑓3 ′′ + 𝑀4 ′𝑓4 ′′)′
+ (𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′ + 𝑀4 ′𝑓4 ′)′ ′ + (𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 + 𝑀4 ′𝑓4 )′′ ′
+ 𝑎(𝑀1 ′𝑓1 ′′ + 𝑀2 ′𝑓2 ′′ + 𝑀3 ′𝑓3 ′′ + 𝑀4 ′𝑓4 ′′)
+ 𝑎(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′ + 𝑀4 ′𝑓4 ′)′ + 𝑎(𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 + 𝑀4 ′𝑓4 )′ ′
+ 𝑏(𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + 𝑀3 ′𝑓3 ′ + 𝑀4 ′𝑓4 ′) + 𝑏(𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 + 𝑀4 ′𝑓4 )′
+ 𝑐(𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + 𝑀3 ′𝑓3 + 𝑀4 ′𝑓4 ) = 𝑅.
From this DE, we get the system of equations whose first three equations are zero and last
equation is R (in a similar way as was done when we proved Theorem 8). Its solution
corresponds to the formulas for M1, M2, M3, M4 established above. □
Theorem 10: Let f1, f2,…, fn be LI solutions of (32). Let W be the Wronskian of functions f1,
f2,…, fn. Then (31) has a particular solution yp given by
𝑦𝑝 = 𝑀1 𝑓1 + 𝑀2 𝑓2 + ⋯ + 𝑀𝑛 𝑓𝑛 . (43)
15
Generalizing, we can say that (43) and its successive derivatives satisfy the DE (as was done
when we proved the previous two theorems)
(𝑛−1) (𝑛−1) (𝑛−1) (𝑛−2) (𝑛−2) (𝑛−2)
{[𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 ] + [𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 ]′ +
(𝑛−2) (𝑛−2) (𝑛−2)
⋯ + (𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 )(𝑛−1) } + {𝑎1 [𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 ]+
(𝑛−3) (𝑛−3) (𝑛−3) ′
𝑎1 [𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯+ 𝑀𝑛 ′𝑓𝑛 ] + ⋯ + 𝑎1 (𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 )(𝑛−2) } +
⋯ + [𝑎𝑛−2 (𝑀1 ′𝑓1 ′ + 𝑀2 ′𝑓2 ′ + ⋯ + 𝑀𝑛 ′𝑓𝑛 ′) + 𝑎𝑛−2 (𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 )′] +
𝑎𝑛−1 (𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 ) = 𝑅 (44)
which allows us to obtain the values of M1, M2,…, Mn by forming from (44) the system of
equations
…
(𝑛−2) (𝑛−2) (𝑛−2)
𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 =0
(𝑛−1) (𝑛−1) (𝑛−1)
𝑀1 ′𝑓1 + 𝑀2 ′𝑓2 + ⋯ + 𝑀𝑛 ′𝑓𝑛 = 𝑅. □
3. Appendix.
𝑦 ′′′ − 7𝑦 ′ + 6𝑦 = 2 sin 𝑥.
(𝐷3 − 7𝐷 + 6)𝑦 = 0
𝑟 3 − 7𝑟 + 6 = 0
(𝑟 + 3)(𝑟 − 1)(𝑟 − 2) = 0
𝑟1 = −3, 𝑟2 = 1, 𝑟3 = 2.
Since 𝑟1 ≠ 𝑟2 ≠ 𝑟3 , then
𝑦ℎ = 𝐶1 𝑒 −3𝑥 + 𝐶2 𝑒 𝑥 + 𝐶3 𝑒 2𝑥 .
16
0 0
𝑦𝑝 = 𝑒 𝛼𝑥 (cos 𝛽𝑥 ∑ 𝐶𝑘 𝑥 𝑘 + sin 𝛽𝑥 ∑ 𝐷𝑘 𝑥 𝑘 )
𝑘=0 𝑘=0
𝑦𝑝 = 𝐶0 cos 𝑥 + 𝐷0 sin 𝑥.
It follows that
Substituting the last four identities into the nonhomogeneous DE, we get
8𝐶0 + 6𝐷0 = 2
6𝐶0 − 8𝐷0 = 0
4 3
whose solution is 𝐶0 = , 𝐷0 = 25 . So the particular solution of the given nonhomogeneous
25
DE is
4cos 𝑥 + 3 sin 𝑥
𝑦𝑝 = .
25
2. Finding yp by using the method of variation of parameters.
𝑓1 = 𝑒 −3𝑥 , 𝑓2 = 𝑒 𝑥 , 𝑓3 = 𝑒 2𝑥
𝑒 −3𝑥 𝑒 𝑥 𝑒 2𝑥
𝑊 = |−3𝑒 −3𝑥 𝑒 𝑥 2𝑒 2𝑥 | = −𝑒 𝑥 (−12 − 18)𝑒 −𝑥 + 𝑒 𝑥 (4 − 9) 𝑒 −𝑥 − 𝑒 𝑥 (2 + 3)𝑒 −𝑥
9𝑒 −3𝑥 𝑒 𝑥 4𝑒 2𝑥
= 30 − 5 − 5 = 20.
17
𝑅(𝑓2 𝑓3′ − 𝑓2 ′𝑓3 ) 2 sin 𝑥[𝑒 𝑥 (2𝑒 2𝑥 ) − 𝑒 𝑥 (𝑒 2𝑥 )]
𝑀1 = ∫ 𝑑𝑥 = ∫ 𝑑𝑥
𝑊 20
1 3𝑥
1 𝑒 3𝑥 (3 sin 𝑥 − cos 𝑥)
= ∫ 𝑒 sin 𝑥 𝑑𝑥 = [ ]
10 10 10
1 3𝑥
= 𝑒 (3 sin 𝑥 − cos 𝑥),
100
𝑅(𝑓1 𝑓3′ − 𝑓1 ′𝑓3 ) 2 sin 𝑥[𝑒 −3𝑥 (2𝑒 2𝑥 ) − (−3𝑒 −3𝑥 )(𝑒 2𝑥 )]
𝑀2 = − ∫ 𝑑𝑥 = − ∫ 𝑑𝑥
𝑊 20
1 1 𝑒 −𝑥 (sin 𝑥 + cos 𝑥)
= − ∫ 𝑒 −𝑥 sin 𝑥 𝑑𝑥 = − [− ]
2 2 2
1
= 𝑒 −𝑥 (sin 𝑥 + cos 𝑥),
4
𝑦𝑝 = 𝑀1 𝑓1 + 𝑀2 𝑓2 + 𝑀3 𝑓3
1 3𝑥 1
= [ 𝑒 (3 sin 𝑥 − cos 𝑥)] 𝑒 −3𝑥 + [ 𝑒 −𝑥 (sin 𝑥 + cos 𝑥)] 𝑒 𝑥
100 4
2 −2𝑥
+ [ − 𝑒 (2 sin 𝑥 +cos 𝑥)] 𝑒 2𝑥
25
3 sin 𝑥 − cos 𝑥 sin 𝑥 + cos 𝑥 4 sin 𝑥 + 2 cos 𝑥 3 sin 𝑥 + 4 cos 𝑥
= + − = .
100 4 25 25
Finally, the general solution is
3 sin 𝑥 + 4 cos 𝑥
𝑦 = 𝑦ℎ + 𝑦𝑝 = 𝐶1 𝑒 −3𝑥 + 𝐶2 𝑒 𝑥 + 𝐶3 𝑒 2𝑥 + .
25
References
Kreider, D.L., Kuller R.G., Ostberg D.R., Elementary Differential Equations, Addison-Wesley
Publishing Company, Inc., Reading, MA, 1968.
18