Вы находитесь на странице: 1из 47

Ordinary Differential Equations

Chapter1 Classification of Differential Equations and Their Solutions

I. Differential equations

A differential equation is an equation which involves differentials or derivatives.

Ex. 1
y  cos x, (1.1)
y + 4y  0, xdx + ydy  0, (1.2)
 2u 2  u
2
c 0 (1.3)
t 2 x 2

II. Classification of differential equations

1. Ordinary differential equations: Differential equations, in which there is a single independent


variable, are known as ordinary differential equations.
Equations (1.1) and (1.2) are ordinary differential equations.

2. Partial differential equations: If an equation involves more than one independent variable, so
that partial derivatives enter, it is known as a partial differential
equation. Equation (1.3) is a partial differential equation.

3. Linear differential equation: A linear differential equation is that there is no products or


nonlinear functions of dependent variable and its derivatives.

4. Nonlinear differential equation: A differential equation containing products or nonlinear


functions of dependent variable and its derivatives is a
nonlinear differential equation.

5. Quasi-linear differential equation: A nonlinear differential equation is quasi-linear when it is


linear in its highest derivative.

O. D. E. 1
Ex. 2
d2y
x2 + 2y  2 is linear. (1.4)
dx 2
dy
 x + y2 is nonlinear. (1.5)
dx
d2y
dx 2 + siny  0 is nonlinear. (1.6)
d2y
dx 2 + y2  0 is quasi-linear. (1.7)

III. The order and degree of differential equations

1. Order: The order of a differential equation is the order of the highest derivative.

2. Degree: The degree of a differential equation is the degree of the highest derivative entering,
when the equation has been rationalized and cleared of fractions.

Ex. 3
y + ytanx  sin2x is first order and first degree. (1.8)
(y  xy)2  k2(1 + y2) is first order and second degree. (1.9)
y2  (1 + y2)3 is second order and second degree. (1.10)

IV. Solutions

A relation, connecting the independent and dependent variables, which satisfies the differential
equation identically, is called a solution of a differential equation.

1. General solution: A solution involving the maximum number of arbitrary constants is called the
general solution.

2. Particular solution: A solution which is derivable from the general solution by assigning fixed
values to the arbitrary constants is called a particular solution.

3. Singular solution: A solution which cannot be obtained from the general solution is called a
singular solution. A singular solution can occur only in the solution of
nonlinear differential equations.

O. D. E. 2
Chapter2 Ordinary Differential Equations of the First Order and
First Degree

General form: 1. M(x, y)dx + N(x, y)dy  0 (2.1a)


2. y  f(x, y) (2.1b)

I. Separable Differential equations

Form: M1(x)N1(y)dx + M2(x)N2(y)dy  0

M 1 ( x) N ( y)
M 2 ( x)
dx   2
N1 ( y )
dy  C

Ex. 1
9yy + 4x  0.
Solution: 9ydy + 4xdx  0  9y2 + 4x2  C

Ex. 2
y  1 + y2.
dy
Solution:  dx  tan1y  x + C  y  tan(x + C)
1  y2

Ex. 3
y + 5x4y2  0, y(0)  1.
dy 1
Solution: 2
 5 x 4 dx     x 5  C
y y
1
y(0)  1  1  C  The solution is y  x5 + 1

Ex. 4
y  2xy, y(0)  1.
dy
Solution:  2xdx  lny  x2 + C  y  e  x  C  y  C e  x
2 2

y
y(0)  1  1  Ce0  C  1 The solution is y  e  x
2

II. Reducible to separable differential equations

O. D. E. 3
1. Homogeneous equation

Definition: If f (tx, ty)  trf (x, y), then f (x, y) is a homogeneous function of degree r.
Substituting t  1/x into f (tx, ty)  trf (x, y), we have
y 1 y y
f (1, )  r f ( x, y )  f ( x, y )  x r f (1, )  x r F ( ) .
x x x x
y
In particular r  0, then f (x, y)  F( )
x
If M(x, y)dx + N(x, y)dy  0, when M(x, y) and N(x, y) are of the same degree in x and y,
M/N is a homogeneous function of degree zero, and the differential equation can be written
as
dy M y (2.2)
  F( )
dx N x .
Let y  ux, equation (2.2) becomes
du dx du
ux  F (u )   is separable.
dx x F (u )  u

Ex. 5
2xyy  y2 + x2  0.
Solution: Let y  ux, the equation becomes
2xux(u + ux)  u2x2 + x2  0  2u(u + xu)  u2 + 1  0  2xuu + u2 + 1  0
2udu dx
  ln(1 + u2)  lnx + C  1 + u2  C/x
1 u 2
x
2

1+     x2 + y2  Cx
y C
x x

Ex. 6
y y
(x + ycos )dx  xcos dy  0.
x x
dx
Solution: Let y  ux  (x + uxcosu)dx  xcosu(udx + xdu)  0  cosudu 
x
y
sinu  lnx + C  sin  lnx  C
x

2. M(x, y) and N(x, y) are linear in x and y

O. D. E. 4
Form: (a1x + b1y + c1)dx + (a2x + b2y + c2)dy  0 (2.3)

a1 b1
(1) If  , let x  X + , y  Y + , equation (2.3) becomes
a 2 b2
[(a1X + b1Y) + (a1 + b1 + c1)]dX + [(a2X + b2Y) + (a2 + b2 + c2)]dY  0. (2.4)
 a1  b1  c1  0
We choose 
a 2   b2  c2  0
Then equation (2.4) reduces to
(a1X + b1Y) dX + (a2X + b2Y) dY  0
is a homogeneous equation.

a1 b1 c
(2) If   k  1 , equation (2.3) becomes
a 2 b2 c2
[k (a2 x + b2 y) + c1] dx + (a2 x + b2 y + c2) dy  0. (2.5)
dv  a 2 dx
Let v  a2x + b2y  dy  , equation (2.5) becomes
b2
dv  a 2 dx
(kv + c1)dx + (v + c2) 0
b2
a v  c2
[kv  c1  2 ( v  c2 )]dx  dv  0 is a separable one.
b2 b2

a1 b1 c1
(3) If    k, equation (2.3) becomes
a 2 b2 c2
k(a2x + b2y + c2)dx + (a2x + b2y + c2)dy  0. (2.6)
a. If a2x + b2y + c2  0, we get only a trivial solution.
b. If a2x + b2y + c2  0, equation (2.6) reduces to kdx + dy  0  y  kx + C.

Ex. 7

O. D. E. 5
(4x + 3y + 1)dx + (x + y + 1)dy 0.
4 x  3 y  1  0
Solution:   x  2, y  3
 x  y 1  0
Let X  x  2, Y y + 3, we have
(4X + 3Y)dX + (X + Y)dY  0
Let Y  uX  (4X + 3uX)dX + (X + uX)(udX + Xdu)  0
dX 1 u
( 4  4u  u 2 )dX  X (1  u )du  0   du  0
X (u  2) 2
dX  1 1  1
  2
du  0  ln X  ln(u  2)  C
X  u  2 ( u  2)  u2
1 x2
ln[ X (u  2)]   C  ln(2 x  y  1)  C
u2 2x  y  1

Ex. 8
(2x  4y + 5)y + x  2y + 3  0.
1  u
Solution : Let u  x  2 y  u   1  2 y   y  
2
1  u
The equation becomes ( 2u  5) u30
2
( 2u  5)(1  u )  2u  6  0  4u  11  ( 2u  5)u   0
2u  5
( 4u  11)dx  ( 2u  5)du  0  dx  du  0
4u  11
1 1/ 2 1 1/ 2
dx  (  )du  0   dx   (  )du  C1
2 4u  11 2 4u  11
u 1
x   ln(4u  11)  C1  8 x  4u  ln( 4u  11)  8C1
2 8
8 x  4( x  2 y )  ln[4( x  2 y )  11]  8C1  4 x  8 y  ln( 4 x  8 y  11)  C

Ex. 9
(x  y  1)dx + (2x  2y  2)dy  0.
Solution: Dividing by x  y  1  dx + 2dy  0  x + 2y  C

3. Type of y  f (ax + by + c) (2.7)

O. D. E. 6
du
a
Let u  ax + by + c  u  a + by  y  dx , equation (2.7) becomes
b
du
a
du du du
dx  f (u )   a  bf (u )   a  bf (u )   dx is separable.
b dx dx a  bf (u )

Ex. 10
y  (x + y  7)2. (2.8)
Solution: Let u  x + y  7  u  1 + y, equation (2.8) has the result
du 1
u  1  u2  2  dx  tan u  x + C  x + y  7  tan(x + C)
1 u

[Exercises] 1. (2x + y  2)y  4


2. y  sin2(x  y)
[Answers] 1. y  2ln(2x + y) + C 2. tan(x  y)  x + C

4. Isobaric equations

O. D. E. 7
Definition: If f (tx, tmy, tm1y)  trf (x, y, y), we say that f (x, y, y) is an isobaric function of weight
r.
In particular, if t  1/x, there follows
y y 1
f (1, m , m1 )  r f ( x, y , y ) or
x x x
y y y y
f ( x, y, y )  x r f (1, m , m1 )  x r F ( m , m 1 ). ( 2.9)
x x x x
y y
A differential equation can be put into the form (2.9), then F ( m , m1 )  0.
x x
y
Solving this for m 1 , and multiplying by xm1, we have
x
y
y   x m1( ) (2.10)
xm
y dy du
Let u  m
 y  ux m   x m  muxm1 , (2.10) is reducible to
x dx dx
du du dx
x m  muxm 1  x m1(u )   is separable.
dx  (u)  mu x
[Note] If the weights m and m  1 are assigned to y and y respectively, the term xaybyc has the
weight a + bm + c(m  1). If an expression is to be isobaric, all its terms must be isobaric
and of the same weight.

Ex. 11
2x3y  1 + 1  4 x 2 y
.
1
Solution: The weights of each term are 3 + (m  1), 0, (0,2 + m), if m  2, every term has the
2
same weight. Let y  ux2  y  ux2  2ux3, the equation becomes
2x3(x2u  2ux3)  1 + 1  4u  2xu  4u  1 + 1  4u
du dx du dx d ( 1  4u  1) dx
    
1  4u  1  4u 2 x 1  4u ( 1  4u  1) 2 x 2( 1  4u  1) 2 x
1  4x2 y  1
ln( 1  4u  1)  ln x  C   C
x

Ex. 12
y2 + (1 + xy)y  0.

O. D. E. 8
Solution: weight 2m,m  , 1 + m + m  , so 2m  m    m  1.
Let y  ux1  y  ux1  ux2
x2u2 + (1 + u) (x1u  x2u)  0  x2u2 + x1u  x2u + x1uu  x2u2  0
(1  u )du dx 1  dx
(1  u )u   x 1u      1du   ln u  u  ln x  C
u x  u  x
u
ln  u  C  ln y  xy  C
x

Ex. 13
x3y  x2y + y2  0.
Solution: weight 3 + m  1, 2 + m, 2m, so m + 2  2m  m  2, Let y  ux2,  y  x2u + 2xu
x3(x2u + 2xu)  x2ux2  ux2   xu + 2u  u  u2  
du dx 1 1  dx ux
 0  du   0  ln  C
u  u2 x  u u  1  x u 1
y/x xy
C C
y / x 1
2
y  x2

[Exercises] 1. 2x2y  x2y2 + 2xy + 1  0


2. x3y + 4x2y + 1  0
3. (x + 2x2y)y + 2y + 3xy2  0
4. 2xyy  y2 + y 4  4x 2
[Answers] 1. x + x2y  C(1  xy) 2. x2 + 2x4y  C 3. x2y + x3y2  C 4. Cy2  x2  C2

III. Exact differential equations

1. Exact differential equations


O. D. E. 9
A first order differential equations M(x, y)dx + N(x, y)dy  0 is called exact if its left side is
u u
du   x dx   y dy , its solution is u  C
M N
[Condition]  y   x
u u  2u  2u
Since M  ,N  , by the assumption of continuity  , we have
x y  x y  y x
M N
 .
y x
The solution can be obtained from the following :
u  Mdx  k ( y )
x

u
and from  N , we can determine k ( y ).
y

Ex. 14
(x3 + 3xy2) dx + (3x2y + y3) dy  0.
 ( x 3  3 xy 2 )  (3 x 2 y  y 3 )
Solution: Since   6xy, it is exact, and
y x
u   x (x3 + 3xy2) dx + k(y)  x4/4 + 3x2y2/2 + k(y)
u
y
 N  3x2y + k(y)  3x2y + y3,  k(y)  y4/4

the solution is x4/4 + 3x2y2/2 + y4/4  C

Ex. 15
(sinx coshy) dx  (cosx sinhy) dy  0, y(0)  0.
 (sin x cosh y )  (  cos x sinh y )
Solution: Since   sinxsinhy, it is exact, and
y x
u   x sinx coshy dx + k(y)  cosx coshy + k(y)
u
y
 N  cosx sinhy + k(y)  cosx sinhy  k(y)  C  u  cosx coshy  C

y(0)  0  cos0 cosh0  C  C  1, The solution is cosx coshy  1

dx 1 x 
[ Exercises ] 1.   dy  0
x y
2 2  y y x  y2
2 
 
( x  y )dx  ( x  y )dy
2. 0
x2  y2
dx 2 x 
3.    dy  0

xy  y y3 

O. D. E. 10
1 x x
[Answers] 1. y2 + 2Cx  C 2 2. log x  y  tan y  C 3.  log y  C
2 2

2. Integrating factors

If M(x, y) dx + N(x, y) dy  0 is not exact, we multiply by (x, y),  M(x, y) dx +  N(x, y)dy 
0 is exact. The function (x, y) is called an integrating factor.

[Case 1]If (x, y)  (x), from the condition of exactness, we get


 M  N M N M N 
   N    N     
y x y x  y x 
M N

d y x
 dx
 N
M N

y x
If only depends on x , we have
N
M N

y x
 dx
( x )  e N

[Case 2] If (x, y)  (y), from the condition of exactness, we get


M N

 M  N M N d y x
   M     dy
y x y x  M
M N

y x
If depends only on y , we have
M
M N

y x
 dy
( y )  e M

Ex. 16
ydx  xdy  0 (2.11)

O. D. E. 11
M N
Solution : M  y , N   x   1,  1
y x
M N

y x 1  ( 1) 2
 
N x x
2
  x dx 2
e  e 2 ln x  e ln x  x 2
u   Mdx  f ( y )   x 2 ydx  f ( y )
x x
1
  x y  f ( y)
u
 N   x 1  f ( y )  x  2  (  x )  f ( y )  0  f ( y )  0
y
y
The solution is  x 1 y  C1   C
x

Ex. 17
2xydx + (4y + 3x2) dy  0, y(0.2)  1.5.
M / y  N / x 2x  6x 2
Solution: Since   depends only on y, integrating factor is
M  2 xy y
2
y y dy  y2 and u   x 2xy3dx + k(y)  x2y3 + k(y)
e
u
y
 N  3x2y2 + k(y)  y(4y + 3x2)  k(y)  y, and the general solution is

x2y3 + y  C. Substituting y(0.2)  1.5, we have


(0.2)2(1.5)3 + (1.5)4  C  C  4.9275  The solution is x2y3 + y  4.9275

[Case 3] Form: xrys(mydx + nxdy)0 (2.12)


Since d(xayb)  xa1yb1(aydx + bxdy), (2.12) has an integrating factor xy,
where +r  m  1,  + s  n  1.
More generally, suppose xy is the integrating factor of the equation

O. D. E. 12
xrys(mydx + nxdy) + xpyq(uydx + vxdy)  0. (2.13)
Multiplying (2.13) by xy and rearranging the terms, we have
(mx+ry+s+1 + ux+py+q+1) dx + (nx+r+1y+s+vx+p+1y+q) dy  0.
M N
This is exact, provided  y   x , i.e.,

m( + s + 1)x+ry+s + u( + q + 1)x+py+q  n( + r + 1)x+ry+s + v( + p + 1)x+py+q


for all values of x and y, we have
m( + s + 1)  n( + r + 1), u( + q + 1)  v( + p + 1).
The two equations can be solved for  and .

Ex. 18
x4y(3ydx + 2xdy) + x2(4ydx + 3xdy)  0.
Solution: (3x4y2 + 4x2y)dx + (2x5y + 3x3)dy  0
Multiplying xy
(3x+4y+2 + 4x+2y+1)dx + (2x+5y+1 + 3x+3y)dy  0
M  3x+4y+2 + 4x+2y+1, N  2x+5y+1 + 3x+3y
M N
To be exact 
y x
3( + 2)x+4y+1 + 4( + 1)x+2y  2( + 5)x+4y+1 + 3( + 3)x+2y
3( + 2)  2( + 5), 4( + 1)  3( + 3)    1,  2
1 6 4 4 3
u  xMdx + f (y)  x(3x5y4 + 4x3y3)dx + f (y)  x y + x y + f (y)
2
u
 N  2x6y3 + 3x4y2 + f (y)  2x6y3 + 3x4y2  f (y)  0  f (y)  0
y
1 6 4 4 3
The solution is x y +x y C
2

Ex. 19
2ydx + xdy + xy(3ydx + 2xdy)  0.
Solution: Multiplying xy and rearranging the terms, we get
(2xy+1 + 3x+1y+2)dx + (x+1y + 2x+2y+1)dy  0
For exactness 2( + 1)xy + 3( + 2)x+1y+1  ( + 1)xy + 2( + 2)x+1y+1
O. D. E. 13
2( + 1)   + 1, 3( + 2)  2( + 2)    1,  0
u   x (2xy + 3x2y2)dx + k(y)  x2y + x3y2 + k(y)
u
y
 N  x2 + 2x3y + k(y)  x2 + 2x3y  k(y)  C

The general solution is x2y + x3y2  C

IV. Linear differential equations of first order

1. Form: y + p(x)y  r(x) (2.14)

If r(x)  0, (2.14) is said to be homogeneous, otherwise it is said to be nonhomogeneous.


(2.14) is equivalent to
d
[F(x)y]  F(x)r (2.15)
dx
F
Fy + Fy  Fr  y + yr
F
Comparing with (2.14), we have
F dF
 pdx  F  e 
pdx
 p
F F
Integrating (2.15), we get the solution:
1
F 
Fy   Frdx + C  y  [ Frdx  C ]

Ex. 20
y  y  e2x.
Solution: F  e dx  ex
1 1
y [  Frdx  C ]   x [  e  x  e 2 x dx  C ]  e x [  e x dx  C ]
F e
 e ( e  C )  e  Ce x
x x 2x

O. D. E. 14
Ex. 21
y + 2y  ex(3sin2x + 2cos2x).
Solution: F  e 2dx  e2x
1 1
y [  Frdx  C ]  2 x [  e 2 x  e x (3 sin 2 x  2 cos 2 x )dx  C ]
F e
 e  2 x [  e 3 x (3 sin 2 x  2 cos 2 x )dx  C ]  e  2 x ( e 3 x sin 2 x  C )
 e x sin 2 x  Ce  2 x

Ex. 22
y + ytanx  sin2x, y(0)  1.
Solution: F  e tanxdx  elnsecx  secx
1 1
[  Frdx  C ] 
sec x 
y [ sec x  sin 2 xdx  C ]
F
1
 cos x[   2 sin x cos xdx  C ]  cos x[  2 sin xdx  C ]
cos x
 cos x ( 2 cos x  C )  2 cos 2 x  C cos x
y(0)  1  1  2cos2(0) + Ccos(0)  C  3
y  2cos2x + 3cosx

2. Reducible to linear form: Bernoulli equations

Form: y + p(x)y  g(x)ya (a is any real number) (2.16)


(2.16)  yay + p(x)y1a  g(x) (2.17)
Let u  y1a  u  (1  a)yay, (2.17) becomes
1
u + pu  g(x)  u + (1  a)pu  (1  a)g(x)
1 a

O. D. E. 15
This is linear in u.

Ex. 23
y  Ay  By2 (A, B are constants).
Solution: y2y  Ay1  B
Let u  y1  u  y2y, there follows
B
u   Au  B  F  e  Adx  e Ax  u  e  Ax (  e Ax Bdx  C )   Ce  Ax
A
1
y
B / A  Ce  Ax

Ex. 24
(ey + x)y  1.
Solution: The equation can be written in the form
dx
 x  ey which is linear in x.
dy
F  e dy  ey,  x  ey[ ey eydy + C]  ey(y + C)

O. D. E. 16
Chapter 3 Ordinary Differential Equations of the First Order and Higher
Degree

I. Equations Solvable for y

Definition: f (x, y, y) can be replaced by a product of factors, rational in x and y, each of lower
degree in y, then f (x, y, y) is said to be reducible, otherwise, it is said to be irreducible.

Ex. 1
y2 + (x + y)y + xy  0.
Solution: (y + x) (y + y)  0
y + x  0 or y + y  0
x2
y + C or y  Cex
2

Ex. 2
y2 + y2  1.
Solution : y    1  y 2
y  1  y 2 or y    1  y 2

sin1y  x + C or cos1y  x + C
y  sin(x + C) or y  cos(x + C)
The two solutions are the same.

[Exercises] 1. xyy2 + (x2  y2) y  xy  0


2. (2xy  y)2  8x3

[Answers] 1. y  Cx, x2 + y2  C 2. y2  2x (x + C)2

II. Equation solvable for y  f (x, y) (3.1)

O. D. E. 17
Differentiating (3.1), we obtain
 f  f dy 
y   x   y  dx

Since y is no longer present, it may be looked upon as an equation of the first order in the variables
x and y. It may happen that we can integrate this equation. Suppose its solution to be

w(x, y, C)  0 (3.2)


Eliminating y between (3.1) and (3.2), we have the general solution.

Ex. 3
4xy2 + 2xy  y  0.
dy  dy 
Solution : Differentiating  ( 4 y 2  8 xy  )  (2 y   2 x )  y  0
dx dx
dy  dy  dy  dy 
( 4 y  2  8 xy  )  ( y   2 x )  0  4 y ( y   2 x )  ( y  2x )0
dx dx dx dx
dy 
( 4 y   1)( y   2 x )0
dx
dy  dx dy  C
1. y   2 x 0 2  0  ln x  2 ln y   C1  xy 2  C  y   
dx x y x
Substituting into the original equation, we have
4C  2 Cx  y  0  (y  4C)2  4Cx  (y  k)2  kx
2. 4y + 1  0  y  1/4, Substitute into the original equation, there follows
x  x
     y  0  x + 4y  0
4  2
This is the singular solution.

Ex. 4

O. D. E. 18
y + xy  x4y2.
dy  dy 
Solution : Differenti ating  y   y   x  4 x 3 y 2  2 x 4 y 
dx dx
dy 
(1  2 x 3 y )( 2 y   x)0
dx
dy  2dx dy  C
1. 2 y   x 0   0  x 2 y  C  y  2
dx x y x
Substituting into the original equation, we have
y + x(C/x2)  x4(C/x2)2  xy  C2x  C
2. 1  2x3y  0  y  1/2x3, The singular solution is
y + x(1/2x3)  x4(1/2x3)2  y + 1/2x2  1/4x2  4x2y + 1  0

[Exercises] Find the general solutions of the following equations:


1. xy2  2yy  x  0
2. 4x2y  2x3y + ay2  0
[Answers] 1. x2  2Cy  C 2  0 2. y  Cx2 + aC 2  0

III. Equation solvable for x  f (y, y) (3.3)

Differentiating (3.3) with respect to y, we obtain


dx 1 f  f dy 
  
dy y   y  y  dy
We may look upon this equation as the first order in y and y. If we can integrate it, we obtain
w(y, y, C)  0 (3.4)
Eliminating y between (3.3) and (3.4), we get the general solution.

Ex. 5

O. D. E. 19
2yy2  2xy + y  0.
2 yy  2  y y
Solution : x   x  yy  
2 y 2 y
dx dy  1 y
 ( y  y )  ( 
dy dy 2 y 2 y
1 dy  1 y
 ( y  y )  ( 
y dy 2 y 2 y 2
dy  1 dy 
( y  y )  ( y   y )
dy 2 y 2 dy
dy  dy dy 
1. y   y  0   
dy y y
Substituting into the original equation, we have
2
C C
2 y    2 x    y  0  2C 2  2Cx  y 2  0
 y  y
1 1
2. 1   0  y    , substituting into the original equation, we have
2 y 2
2
1  1 1
2 y    2 x    y 0 y x0
2  2  2

Ex. 6
y2y2  3xy + y  0.

O. D. E. 20
y 2 y 2  y 1 2 yy 2  1  y 2 y  dy 
Solution : x       2 
3y y 3y  3 3 y   dy
2 yy  2  2 yy  2  1 dy  y y  2  1  y dy  
0  y   2    0
3y  3y  2
dy 3  y dy  
y dy  C
1. 2   0  y 2 y   C  y   2 , the general solution is
y  dy y
2
C C
y 2  2   3x 2   y  0  C 2  3Cx  y 3  0
y  y 
yy  2  1 1
2.  0  y    , the singular solution is
3 y
1  1  1
y    3x    y  0  2 y  3x  0
2

 y  y  y

[Exercises] Find the general solutions of the following equations:


1. a2yy2  2xy + y  0
2. y3  4xyy + 8y2  0
3. 2yy3  3xy + 2y  0
[Answers] 1. y2  2Cy + a2C 2  0 2. y  C(x  C)2 3. 4y3  C(3x  2C)2

IV. Clairaut equation: y  xy  f (y) (3.5)

Differentiating (3.5) with respect to x, we get


dy   f dy    f  dy 
y  y  x    x   0
dx  y  dx   y   dx
f dy 
The vanish of the first term x   y   0 deduce the singular solution, the second  0 , has
dx
y  C
Hence the general solution is

O. D. E. 21
y  Cx  f (C)

Ex. 7
xy2  yy + 3  0.
Solution: The equation can be written as
3
y  xy  
y
This is Clairaut equation, the general solution is
C 2x  Cy + 3  0

Ex. 8
x2y2  2(xy  4)y + y2  0.
Solution: The equation can be written as
x2y2  2xyy + 8y + y2  0
(xy  y)2 + 8y  0
This is Clairaut equation, the general solution is
C 2x2  2C(xy  4) + y2  0

O. D. E. 22
Chapter 4 Ordinary Differential Equations of Higher Order

I. Linear differential equations of higher order

Form: pn(x)y(n)(x) + pn1(x)y(n1)(x) + pn2(x)y(n2)(x) +…+ p1(x)y(x) + p0(x)y(x)  r(x) (4.1)


or Ly(x)  r(x),
dn d n 1 d n 2 d
where L  p n n  p n 1 n 1  p n 2 n 2    p1  p0 is called a differential operator.
dx dx dx dx

1. Linear dependence
Definition: A linear combination of n functions y1(x), y2(x),…, yn(x), is an expression of the form
k1y1(x) + k2y2(x) +…+ knyn(x),
where ki’s are constants.

Definition: If the equation


k1y1(x) + k2y2(x) +…+ knyn(x)  0 (4.2)
holds on an interval I for some k1, k2,…, kn not all zero, the function y1(x), y2(x),…, yn(x) are
said to be linearly dependent. If y1(x), y2(x),…, yn(x) are linearly dependent, there exists at least
one constant, say ki, not zero, and the corresponding function yi(x) can be expressed as a linear
combination of the other functions, i.e.,
1
yi(x)   k [k1y1(x) + k2y2(x) +…+ ki1yi1(x) + ki+1yi+1(x) +…+ knyn(x)] (4.3)
i

Definition: If (4.2) holds on interval I only for k1  k2 … kn  0, then y1(x), y2(x),…, yn(x) are
said to be linearly independent.

Assume that each of a set of n functions y1(x), y2(x),…, yn(x) possesses n finite derivatives at
all points of an interval I. Then
dy dy dy
k1 1  k 2 2    k n n  0
dx dx dx
2
d y1 2
d y2 d 2 yn
k1  k 2    k n 0
dx 2 dx 2 dx 2
 (4.4)
d n 1 y1 d n 1 y 2 d n 1 y n
k1  k 2    k n 0
dx n 1 dx n 1 dx n 1
If the functions y1(x), y2(x),…, yn(x) are linearly dependent, i.e., there exists nontrivial solution of
ki’s, the determinant

O. D. E. 23
y1 y2  yn
y1 y 2  y n
W(y1, y2 ,…, yn; x)  (4.5)
   
y1( n 1) y 2( n 1)  y n( n 1)
vanishes identically over I. This determinant is called the Wronskian of the functions y1(x), y2(x),
…, yn(x). If the Wronskian of y1(x), y2(x),…, yn(x) is not identically zero over I., then the
functions y1(x), y2(x),…, yn(x) are linearly independent.
[Note] The vanishing of the Wronskian is necessary but not sufficient condition for linear
dependence of a set of functions.

Ex. 1
Show that ex, e2x are linearly independent on all interval.
ex e2 x
Solution: W(e , e ; x)  x
x 2x
 exe2x  0
e 2e 2 x

Ex. 2
Show that x, x2 are linearly independent on {x xR, x  0}.
x x2
Solution: W(x, x2; x)  1 2x
 x2  0 if x  0.

Principle of superposition: If y1(x), y2(x) are the solutions of the homogeneous linear differential
equation
pn(x)y (x) + pn1(x)y(n1)(x) +…+ p1(x)y(x) + p0(x)y(x)  0
(n)
(4.6)
then k1y1(x) + k2y2(x) is also a solution of (4.6), where k1, k2 are arbitrary constants.
An nth order linear differential equation (4.6) has n linearly independent solutions y1(x),
y2(x),…, yn(x), and its general solution can be expressed as
yh(x)  k1y1(x) + k2y2(x) +…+ knyn(x)
where ki’s are arbitrary constants, and yh(x) is called homogeneous solution or complementary
solution of (4.1). Now suppose that one particular solution of (4.1), say yp(x), then the complete
solution of (4.1) is
y(x)  yh(x) + yp(x)

2. Linear differential equation with constant coefficients

O. D. E. 24
Form: any(n)(x) + an1y(n1)(x) +…+ a1y(x) + a0y(x)  r(x) (4.7)
or Ly  r(x)
where a0, a1, …, an are constants, and
dn d n 1 d n 2 d
L  an n  a n1 n 1  a n 2 n 2    a1  a0 .
dx dx dx dx

(1) Homogeneous solution:

Let yh(x)  ex, substitute into the homogeneous equation of (4.7), i.e., r(x)  0, we have
(ann + an1n1 +…+ a1 + a0)ex  0
Since ex  0, we get
ann + an1n1 +…+ a1 + a0  0 (4.8)
Equation (4.8) is called the characteristic equation of the homogeneous equation of (4.7).

[Case 1] If (4.8) has n distinct real roots, say 1, 2,…, n, then
yh(x)  k1e 1x  k 2 e  2 x  k n e  n x

[Case 2] If (4.8) has a real double root, say 1  2,


∵Lex  an(  1)2 (  3) (  4)…(  n)ex
x
∴ Le    0, e 1x is a solution of Ly  0.
1

and


(Lex)  2(  1)(  3)(  4)…(  n)ex

+ (  1)2 (  4)(  5)…(n)ex


+ (  1)2 (  3)(  5)…(  n)ex
+…
+ (  1)2 (  3)(  4)…(  n1)ex




L e x  0
 1

   e x 
but


L e x    1
 L
  
  L( xe  1x )  0
    1

xe 1x is also a solution of Ly  0. By a simple extension of this argument, it can be


shown that the part of the homogeneous solution corresponding to an m-fold root 1
is
(k1 + k2x + k3x2 +…+ kmxm1) e 1x

Ex. 3
y  2y  y + 2y  0.

O. D. E. 25
Solution: The characteristic equation is
3  22   + 2  0  ( + 1) (  1) (  2)  0    1, 1, 2
The general solution is y  k1ex + k2ex + k3e2x

Ex. 4
y(5)  3y(4) + 3y(3)  y  0.
Solution: The characteristic equation is
5  34 + 33  2  0  2 (  1)3  0    0, 0, 1, 1, 1
The general solution is y  k1 + k2x + (k3 + k4x + k5x2)ex

[Case 3] If (4.8) has simple complex root, say 1    i, then solution
k1e(+i)x + k2e(i)x  k1ex(cosx + isinx) + k2ex(cosx  isinx) 
ex[(k1 + k2)cosx + i(k1  k2)sinx]  ex(C1cosx + C2sinx)
where C1  k1 + k2, C2  i(k1  k2)
Hence the solution can be expressed as ex(k1cosx + k2sinx)

[Case 4] If (4.8) has complex multiple roots, the solution is


ex[(k1 + k2x +…+ kmxm1)cosx + (km+1 + km+2x +…+ k2mxm1)sinx]

Ex. 5
y  2y + 2y  0, y(0)  0.5, y(0)  1, y(0)  2.
Solution: The characteristic equation is
3  22 + 2  0   (2  2 + 2)  0    0, 1  i
The general solution is y(x)  k1 + ex(k2cosx + k3sinx)
y(0)  0.5  0.5  k1 + k2
y(0)  1  1  k2 + k3  k1  2.5, k2  2, k3  1
y(0)  2  2  2k3
∴y  2.5 + ex(2cosx + sinx)

Ex. 6
y(7) + 18y(5) + 81y(3)  0.
O. D. E. 26
Solution: The characteristic equation is
7 + 185 + 813  0  3 (2 + 9)2  0    0, 0, 0, 3i, 3i
The general solution is y(x)  k1 + k2x + k3x2 + (k4 + k5x)cos3x + (k6 + k7x)sin3x

(2) Particular solution

a. Method of undetermined coefficients:

Term in r(x) Choice for yp(x)


xn cnxn + cn  1xn  1 +…+ c1x + c0
epx cepx
cosqx Acosqx + Bsinqx
sinqx Acosqx + Bsinqx

Ex. 7
y(4)  y  4.5e2x.
Solution: The characteristic equation is
4  1  0  (2  1) (2 + 1)  0    1, i
yh(x)  c1ex + c2ex + c3cosx + c4sinx
Let yp(x)  Ae2x  yp  2Ae2x  yp  4Ae2x  yp  8Ae2x  yp(4)  16Ae2x
Substitute in the given equation, there follows
(16A  A)e2x  4.5e2x  A  0.3
The complete solution is y(x)  yh(x) + yp(x)  c1ex + c2ex + c3cosx + c4sinx + 0.3e2x

Ex. 8
y  3y + 3y  y  30ex.

O. D. E. 27
Solution: The characteristic equation is
3  32 + 3  1  0  (  1)3  0    1, 1, 1
yh(x)  (k1 + k2x + k3x2)ex
Let yp(x)  Ax3ex  yp  A(x3 + 3x2)ex, yp  A(x3 + 6x2 + 6x)ex,
yp  A(x3 + 9x2 + 18x + 6)ex
Substituting into the problem, we have
A[(x3 + 9x2 + 18x + 6)  3(x3 + 6x2 + 6x) + 3(x3 + 3x2)  x3]ex  30ex  A  5
The complete solution is y(x)  yh(x) + yp(x)  (k1 + k2x + k3x2 + 5x3)ex

Ex. 9
y  2y + y  x2ex.
Solution: The characteristic equation is
2  2 + 1  0  (  1)2  0    1, 1
yh(x)  (k1 + k2x)ex
Let yp(x)  (Ax4 + Bx3 + Cx2)ex  yp  [Ax4 + (4A + B)x3 + (3B + C)x2 + 2Cx]ex,
yp  [Ax4 + (8A + B)x3 + (12A + 6B + C)x2 + (6B + 4C)x + 2C]ex,
Substituting into the problem, we have
[(A  2A + A)x4 + (8A + B  8A  2B + B)x3 + (12A + 6B + C  6B  2C + C)x2
+ (6B + 4C  4C)x + 2C]ex  x2ex  A  1/12, B  0, C  0
The complete solution is y(x)  yh(x) + yp(x)  (k1 + k2x + 1/12x4)ex

Ex. 10
y + 4y  x2 + cosx.
Solution: The characteristic equation is
2 + 4  0    2i  yh(x)  k1cos2x + k2sin2x
Let yp(x)  (Ax2 + Bx + C) + (Dcosx + Esinx)
yp 2Ax + B  Dsinx + Ecosx, yp  2A  Dcosx  Esinx
Substituting into the problem, we have
4Ax2 + 4Bx + (2A + 4C) + (D + 4D)cosx + (E + 4E)sinx  x2 + cosx
A  1/4, B  0, C  1/8, D  1/3, E  0
1 1 1
The complete solution is y(x)  yh(x) + yp(x)  k1cos2x + k2sin2x + x   cosx
2

4 8 3

b. Method of variation of parameters

O. D. E. 28
The homogeneous solution of the equation (4.7) is
n
yh(x)  k1y1(x) + k2y2(x) +…+ knyn(x)   k i y i ( x )
i 1

We replace the constants ki, i  1, 2,…, n, by ui(x) to be determined, so that


n n n
y p ( x )   ui ( x ) yi ( x )  y p   ui yi   ui yi
i 1 i 1 i 1
n
We choose  u y
i 1
i i 0 (4.9)
n n
then y p   ui yi   ui yi
i 1 i 1
n
Choose  u y  0
i 1
i i (4.10)
n
In a similar procedure, we have  u y
i 1
i
( n 2 )
i 0 (4.11)
n n
and y (pn )   ui yi( n )   ui yi( n 1)
i 1 i 1

By substituting yp, and its derivatives into (4.7), we have


n n
 n n n
a n  ui yi( n )   ui y i( n 1)   a n 1  ui yi( n 1)    a1  ui y i  a0  ui y i  r
 i 1 i 1  i 1 i 1 i 1
n n

 u [a
i 1
i
(n)
n yi  a n 1 yi( n 1)    a1 yi  a0 yi ]  a n  ui y i( n 1)  r
i 1
n

 u y
i 1
i
( n 1)
i  r / an (4.12)

The equations (4.9), (4.10),…,(4.12) gives a system of n equations for the unknown functions
u1, u2,…, un, where we can find u1, u2,…, un.

Ex. 11

O. D. E. 29
y + y  tanx.
Solution:2 + 1  0    i  yh(x)  k1cosx + k2sinx
Let yp(x)  u1cosx + u2sinx, then
yp  (u1sinx + u2cosx) + (u1cosx + u2sinx)
Set u1cosx + u2sinx  0…………
yp  (u1cosx  u2sinx) + (u1sinx + u2cosx)
Substitute in the given equation, there follows
[(u1cosx  u2sinx) + (u1sinx + u2cosx)] + (u1cosx + u2sinx)  tanx
u1sinx + u2cosx  tanx…………
From  and  we get
0 sin x
u1  tan x cos x  sinx tanx
W(cos x , sin x )
u1   sinx tanx dx   tanx dcosx  cosx tanx   cosx sec2x dx  sinx  lnsecx + tanx
cos x 0
u2   sin x tan x  sinx  u2   sinx dx  cosx
W(cos x, sin x )
y(x)  yh(x) + yp(x)  k1cosx + k2sinx + (sinx  ln secx + tanx  )cosx + ( cosx) sinx
 (k1  ln secx + tanx  )cosx + k2sinx

Ex. 12
y + y  cotx.
Solution: 3 +   0    0, i  yh(x)  k1 + k2cosx + k3sinx
Let yp(x)  u1 + u2cosx + u3sinx, then
u1 + u2cosx + u3sinx  0
u2sinx + u3cosx  0
u2cosx  u3sinx  cotx
u1  cotx, u2  cotxcosx, u3  cosx
u1  lnsinx, u2  cosx  lncscx  cotx, u3  sinx
y(x)  yh(x) + yp(x)  k1 + k2cosx + k3sinx  1  cosx lncscx  cotx + lnsinx

3.Euler-Cauchy equations (Equidimensional linear differential equations)

O. D. E. 30
Form: anxny(n)(x) + an1xn1y(n1)(x) +…+ a1xy(x) + a0y(x)  r(x) (4.13)
d dt d 1 d
Let x  et    t
dx dx dt e dt
d d
x 
dx dt
d2 t 1 d 1 d
2
d d  2 1 d  1 d 
x2 2  x2  x t  t   e  t 2  t 
dx dx  dx  e dt  e dt   e dt e dt 
2
d d d d 
 2     1
dt dt dt  dt 

d3 3t 1 d  1 d  1 d  2t d  1  1 d 1 d  2t d  1 d 1 d
2 2
x3 3
 e t  t  t   e 
 t t 2  t

   e 
 2t 2
 2 t 
dx e dt  e dt  e dt  dt  e  e dt e dt  dt  e dt e dt 
 1 d3 1 d2 1 d2 1 d  d3 d2 d
 e 2 t  2 t 3  2 2t 2  2 t 2  2 2 t   3  3 2  2
 e dt e dt e dt e dt  dt dt dt
d d  d 
   1   2 
dt  dt  dt 
………………………………………………………
dn d d  d  d  d 
xn n
   1   2    3     n  1
dx dt  dt  dt  dt   dt 
The transformed equation thus becomes linear with constant coefficients, and y then can be
determined in terms of t.

Ex. 13
x2y  3xy + 4y  x2lnx.
Solution: Let x  et  t  lnx, we get
d d  dy d2y dy
  1  y  3  4 y  te 2t
 2
4  4 y  te 2 t 
dt  dt  dt dt dt
The characteristic equation is 2  4 + 4  0    2, 2  yh  (k1 + k2t)e2t
Assume particular solution is yp  (At3 + Bt2) e2t
yp  [2At3 + (3A + 2B)t2 + 2Bt]e2t
yp  [4At3 + (12A + 4B)t2 + (6A + 8B)t + 2B]e2t
  [(4A  8A + 4A)t3 + (12A + 4B  12A  8B + 4B)t2 + (6A + 8B  8B)t + 2B]e2t  te2t
A  1/6, B  0
1 1
y  yh + yp  (k1 + k2t + t3)e2t  [k1 + k2lnx + (lnx)3]x2
6 6

Ex. 14
x2y  4xy + 6y  7x4sinx.
Solution: Let x  et  t  lnx, we get

O. D. E. 31
d2y dy
2
 5  6 y  7e 4 t sin e t  2  5 + 6  0    2, 3  yh  k1e2t + k2e3t
dt dt
Let yp  u1e2t + u2e3t
u1e2t + u2e3t  0
2u1e2t + 3u2e3t  7e4tsinet
0 e 3t
u1   7e 4 t sin e t 3e 3t  7e2tsinet  u1  7etcoset + 7sinet
W(e 2 t , e 3t )
e 2t 0
u2  2e 2t
 7e sin e t  7etsinet  u2  7coset
4t

W(e 2 t , e 3t )
y  yh + yp  k1e2t + k2e3t + (7etcoset + 7sinet)e2t + 7e3tcoset
 k1e2t + k2e3t + 7e2tsinet  k1x2 + k2x3 + 7x2sinx

4. Simultaneous linear differential equations

Ex. 15

O. D. E. 32
 dx dy
 dt  2 x  dt  6 y  2e
t

 dx dy
2  3x  3  8 y  1
 dt dt
d ( D  2) x  ( D  6) y  2e t  (i)
Solution: Let D   
dt ( 2 D  3) x  (3D  8) y  1 (ii)
(i)  2  (ii)  x + (D + 4)y  4et + 1………(iii)
(i)  (D + 2)(iii)  [(D + 6)  (D + 2)(D + 4)]y  2et (D + 2)(4et + 1)
(D2  D  2)y  2et  (4et + 8et + 2)  (D2  D  2)y  10et  2………(iv)
The characteristic equation of (iv) is 2    2  0    1, 2  yh  k1et + k2e2t
Assume yp  Aet + B  yp  Aet  yp  Aet
(iv)  (A  A  2A)et  2B  10et  2  A  5, B  1
y  yh + yp  k1et + k2e2t + 5et + 1
Substituting into (iii), we get
x  (k1  4k1)et + (2k2  4k2)e2t + (5  20)et  4 + 4et + 1  5k1et  2k2e2t  11et  3
x  5 t  2  2 t  11 t  3
   k1   e  k 2   e   e   
 y 1  1   5  1 

Ex. 16

O. D. E. 33
 dx dy
 dt  x  dt  y  sec t
 dx dy
4  5 x  5  4 y  5 sec t
 dt dt
d ( D  1) x  ( D  1) y  sec t  (i)
Solution: Let D  
dt ( 4 D  5) x  (5D  4) y  5 sec t  (ii)
(ii)  (i)  4  x + Dy  sect………(iii)
(D + 1)(iii)  (i)  (D2 + 1)y  sect tant………(iv)
The characteristic equation of (iv) is 2 + 1  0    i  yh  k1cost + k2sint
Assume yp  u1cost + u2sint
u1cost + u2sint  0
u1sint + u2cost  sect tant
0 sin t
u1    tan 2 t  u1   tan t  t
sec t tan t cos t
cos t 0
u 2   tan t  u 2  ln | sec t |
 sin t sec t tan t

y  yh + yp  k1cost + k2sint  sint + tcost + sint lnsect


Substituting into (iii), we get
x  k1sint  k2cost + tsint  cost lnsect  sint tant + sect
x  sin t   cos t  t sin t  cos t ln | sec t |  sin t tan t  sec t 
   k1    k2   
 y cos t   sin t    sin t  t cos t  sin t ln | sec t | 

5. Linear differential equations with variable coefficients

O. D. E. 34
pn(x)y(n)(x) + pn1(x)y(n1)(x) +…+ p1(x)y(x) + p0(x)y(x)  r(x) (4.14)
(1) Exact differential equations
Consider a simple case of equation (4.14):
p3y + p2y + p1y + p0y  (p3y)  p3y + p2y + p1y + p0y  (p3y) + (p2  p3)y + p1y + p0y
 (p3y) + [(p2  p3)y]  (p2  p3)y + p1y + p0y
 (p3y) + [(p2  p3)y] + (p1  p2 + p3)y + p0y
 (p3y) + [(p2  p3)y] + [(p1  p2 + p3)y]  (p1  p2 + p3)y + p0y
 (p3y) + [(p2  p3)y] + [(p1  p2 + p3)y] + (p0  p1 + p2  p3)y
If the coefficient of y is identically zero, the other three terms can be integrated, and we have a
new equation of less order. In a similar way, equation (4.1) is said to be exact, if
n

 ( 1)
i 0
i
pi( i ) ( x )  0

Ex. 17
(x  1)y + (x + 1)y + y  2x.
Solution: Since p0  p1 + p2  0, the equation is exact.
[(x  1)y]  y + (x + 1)y + y  2x
[(x  1)y] + xy + y  2x
[(x  1)y] + (xy)  2x  (x  1)y + xy  x2 + C1
x x 2  C1
y  y
x 1 x 1
This is a linear first order equation.
x
h  x  1 dx  x  ln( x  1)
 x 2  C1 h
y  e h  

e dx  C 2  
ex
 e ( x
x 2
 C1 )dx  C 2 
 x 1  x 1
x
e
 [ x 2  2 x  2  C1 )e x  C 2 ]
x 1

(2) Reduction of order

One of the important properties of linear differential equations is the fact that, if one
O. D. E. 35
homogeneous solution of an equation of order n is known, a new linear differential equation of
order n  1, determining the remainder of the solution, can be obtained.
Suppose u(x) is the homogeneous solution, substituting y  v(x) u(x) into the original
equation, we have a new equation of less order. We can try emx, cosmx, sinmx, xm for the
homogeneous solution.

Ex. 18
xy + (x  1)y  y  0.
Solution: Since ex is a homogeneous solution, let y  vex
y  vex  vex, y  vex  2vex + vex
x(v  2v + v)ex + (x  1)(v  v)ex  vex  0
xv  (x + 1)v  0
This is a first order equation in v
x 1
v  C e  x
dx
 C1xex
1

v   C1xex + C2  C1(x  1)ex + C2


y  C1(x  1) + C2ex

[Exercises] 1. xy  (2x + 1)y + (x + 1)y  x2  x  1


2. (1 + x2)y + 2xy  2y  3x
3. (x  1)y  xy + y  1
4. (1  x)y + xy  y  (1  x)2

[Answers] 1. y  C1ex + C2x2ex + x 2. y  C1x + C2(1 + xtan1x) + x ln 1  x 2


3. y  C1x + C2ex + 1 4. y  C1x + C2ex + x2 + 1

II. Nonlinear differential equations of higher order

1. Dependent variable absent: f(x, y, y,…, y(n))  0

O. D. E. 36
Let p  y, the resulting equation is of order n  1 in p.

Ex. 19
(1 + x2)y + y2 + 1  0.
dp dp dx
Solution: Let p  y  (1  x )  p2 1  0  2  0
2

dx p 1 1 x2
px C x
tan 1 p  tan 1 xC  C1  p  1
1  px C1 x  1
C1  x  1 C1  1 / C1 
y dx  C2     dx  C 2
C1 x  1  C1 C1 x  1 
x 1
  (C1  1 / C1 ) ln(C1 x  1)  C 2
C1 C1

3 2
d2y  dy  d 2 y  dy 
[Exercises] 1.  x  2.   0
dx 2  dx  dx 2  dx 

[Answers] 1. x  C1sin(y  C2) 2. y  ln(x + C1) + C2

2. Independent variable absent: f(y, y, y,…, y(n))  0

dp dp dy dp
Let p  y, y   dx  dy dx  p dy ,…, the equation can be reduce to an (n  1) order, where p is

the new dependent variable and y is the new independent variable.

Ex. 20
yy + y2  y  0.
dp
Solution: Let p  y, y   p dy , the equation reduces to

O. D. E. 37
dp dp dy
yp  p2  p  0    0  ln( p  1)  ln y  C 
dy p 1 y
dy ydy  C1 
( p  1) y  C1  y  y  C1   dx  1  dy  dx
dx C1  y  C1  y 
y  C1ln(y + C1)  x + C2

2 2
d 2 y  dy  d 2 y  dy 
[Exercises] 1. y   2.   0
dx 2  dx  dx 2  dx 

[Answers] 1. y  C 2 e C1x 2. y  ln(C1x + C2)

3.Equations homogeneous in y and its derivatives

Let y  ez  y  ezz, y  ez(z + z2),…


f(x, y, y, y,…, y(n))  0  f(x, z, z,…, z(n))  0 is the kind of dependent variable absent
type.

Ex. 21
yy  y2  6xy2  0.
Solution: Let y  ez  y  ezz, y  ez(z + z2)
e2z(z + z2)  z2e2z  6xe2z  0  z  6x  0  z  x3 + C1x + C2
3
y  e x C1x C2

2 2
d 2 y  dy  d 2 y  dy 
[Exercises ] 1. y    2 . y   0
dx 2  dx  dx 2  dx 
[Answers] 1. y  C 2 e C1x 2. y 2  C 2 x  C1

4. Isobaric equations

If a number m exists such that the result of replacing x by tx , y by tmy, y by tm1y, y by


tm2y,…, y(n) by tmny(n), is the original function multiplied by tr, let y  uxm, x  et, we can simplify

O. D. E. 38
the original equation.

Ex. 22
x2yy  3xyy + x2y2 + 2y2 + x2  0.
Solution: weight: 2 + m + (m  2), 1 + m + (m  1), 2 + 2(m  1), 2m, 2  m  1
dy dy dt  du dx  du
Let y  ux, x  e t  y      x  u e  t  u
dx dt dx  dt dt  dt
 dx  d u du  t
2
d  du
y     u    2  e
dt  dt  dt  dt dt 
2
 d 2 u du   du   du 
e ue  2  e t  3e t ue t   u   e 2 t   u   2u 2 e 2 t  e 2 t  0
2t t

 dt dt   dt   dt 
2
d 2 u  du  d  du  du
u 2
    1  0   u   1  0  u  t  C1
dt  dt  dt  dt  dt
u 2
(C  t ) 2
udu  (C1  t )dt   1  C2
2 2
2
1 y (C1  ln x ) 2
     C2
2 x 2

[Exercises] 1. Show that the substitution y  u/[Q(x)u], where u  du/dx, reduces the nonlinear
first-order equation
y + P(x)y + Q(x)y2  R(x)
to the second-order equation
d 2u  Q   du
2
  P    RQu  0
dx  Q  dx
and also that the substitution y  y1 + 1/v, where y1(x) is any known solution of the
given equation, leads to the linear first-order equation
v  (P + 2Qy1)v  Q.
(The nonlinear form is known as Riccati’s equation.)
2. Use the procedures suggested above to obtain the general solution of the equation
x2y + xy + x2y2  1 in the form (x2  k)/(x2 + k), where k is an arbitrary constant.

O. D. E. 39
Chapter 5 Series Solutions of Differential Equations

I. Power series method

Definition: An expression of the form


A
m 0
m ( x  x0 ) m (5.1)

is called a power series, where A0, A1, A2,…, are constants, called the coefficients of the series. x0 is
a constant, called the center of the series, x is a variable.

Ex1. y + y  0

Solution: Assume y   Am x
m
, then
m 0
 

 m(m  1) Am x m 2   Am x m  0
m0 m 0
 

 m(m  1) A
m2
m x m  2   Am  2 x m  2  0
m2

 [m(m  1) A
m2
m  Am  2 ) x m  2  0
m(m  1) Am  Am  2  0 m  2,3,4, 
A A A A A
A2   0 , A4   2  0 , A6   4   0 , 
2! 4  3 4! 65 6!
A1 A A A A
A3   , A5   3  1 , A7   5   1 ,
3! 5  4 5! 76 7!
 x 2
x 4
x 6
  x 3
x5 x7 
y  A0 1       A1  x       A0 cos x  A1 sin x
 2! 4! 6!   3! 5! 7! 

Ex2. y + xy + y  0

O. D. E. 40

Solution: Assume y  Am x
m
, then
m 0
  

 m(m  1) A
m0
m x m2   mA
m0
m xm  A
m0
m xm  0
  

 m(m  1) Am x m  2 
m0
 (m  2) Am 2 x m 2 
m2
A
m2
m2 x m2  0

 [m(m  1) A
m2
m  ( m  2  1) Am  2 ] x m  2  0
Am  2
Am   m  2,3,4, 
m
A A A A A0
A2   0 , A4   2  0 , A6   4   ,,
2 4 24 6 246
A0 A
A2 k  ( 1) k  ( 1) k k 0
2  4  6   ( 2k ) 2 k!
A A A A A1
A3   1 , A5   3  1 , A7   5   ,,
3 5 35 7 357
A1 2  4  6   (2k ) A1 2 k k! A1
A2 k 1  (1) k  (1) k  (1) k
3  5  7   (2k  1) (2k  1)! (2k  1)!
 
(1) k
  
(1) 2 k! 2 k 1 
k k
y  A0 1   k x 2 k   A1  x   x 
 k 1 2 k !   k 1 ( 2 k  1)! 

(1) k 2 k 
(1) k 2 k k! 2 k 1
 A0  k x  A1  x
k  0 2 k! k  0 ( 2k  1)!

[Exercises] Obtain the general solutions of the following differential equations in terms of power
series around x  0.
1. y  xy  0 2. y + xy  1  x + x2

 
x 3m   
x 3m 1 
[Answers]1. y  A0 1     A1 x   
 m 1 2  3  5  6   (3m  1)(3m)   m 1 3  4  6  7   (3m)(3m  1) 
x2

2. y  x  1  Ae 2

II. Singular points of linear second-order differential equations

O. D. E. 41
A homogeneous second-order linear differential equation
d2y dy
2
 a1 ( x )  a0 ( x) y  0 (5.2)
dx dx
1. If a1(x), a0(x) are both analytic at point x  x0, the point x  x0 is said to be an ordinary point of
the differential equation, then the equation possesses two linearly independent solutions which

are both expressible in the form A
m 0
m ( x  x0 ) m .

2. Although a1(x) or a0(x) are not analytic at point x  x0, (x  x0) a1(x) and (x  x0)2 a0(x) are both
analytic at point x  x0, the point x  x0 is said to be a regular singular point, the equation has at

least one solution of the form ( x  x 0 )  Am ( x  x 0 ) .


r m

m 0

3. If a1(x) and a0(x) are not the case mentioned above, the equation is said to be irregular singular
at point x  x0, and a nontrivial solution of this type may or may not exist.

III. Frobenius method

In place of reducing a second-order linear differential to the standard form (5.2), it is


frequently more convenient to use a form which is, to some extent, cleared of fractions, particularly
if a1(x) or a0(x) is the ratio of two polynomials. For this reason, to investigate the nature of
solutions valid near x  0, we suppose that the equation has been put in the form
d2y dy
x 2 R( x ) 2  xP ( x )  Q ( x ) y  0 or L y  0 (5.3)
dx dx
d2 d
where L  x 2 R ( x ) 2  xP( x )  Q ( x )
dx dx
The fact that x  0 is a regular singular point of equation (5.3) means P(x)/R(x) and Q(x)/R(x)
have finite limits as x  0, and are analytic at x  0. Then we may write
P(x)  P0 + P1x + P2x2 +…
Q(x)  Q0 + Q1x + Q2x2 +…
R(x)  R0 + R1x + R2x2 +…
In the neighborhood of x  0, we can approximate (5.3) as
R0x2y + P0y + Q0y  0
which is a Cauchy-Euler equation, and its solution is xr for some r. In general, some of Rm, Pm, and
Qm, m  1, are not zero. It is reasonable to expect the equation to have the solution that behaves
like xr in the neighborhood of x  0. Thus we assume the solution of the form

y  x r  Am x m (5.4)
m 0

Substitution into the equation (5.3) then gives

O. D. E. 42
(R0 + R1x + R2x2 +…)  [r(r  1)A0xr + (r + 1)rA1xr + 1 + (r + 2)(r + 1)A2xr + 2 +…]
+ (P0 + P1x + P2x2 +…)  [rA0xr + (r + 1)A1xr + 1 + (r + 2)A2xr + 2 +…]
+ (Q0 + Q1x + Q2x2 +…)  [A0xr + A1xr + 1 + A2xr + 2 +…] +… 0
after multiplying and collecting the coefficients of successive power of x,
[R0r(r  1) + P0r + Q0]A0xr
+{[R0(r + 1)r + P0(r + 1) + Q0]A1 + [R1r(r  1) + P1r + Q1]A0}xr + 1
+{[R0(r + 2)(r + 1) + P0(r + 2) + Q0]A2 + [R1(r + 1)r + P1(r + 1) + Q1]A1
+ [R2r(r  1) + P2r + Q2]A0}xr + 2 +… 0 (5.5)
In order to abbreviate this relation, we next define the functions
fi(r + m)  Ri(r + m)(r + m  1) + Pi(r + m) + Qi
(5.5) becomes
r r+1 r+2
f0(r)A0x + [f0(r + 1)A1 + f1(r)A0]x + [f0(r + 2)A2 + f1(r + 1)A1 + f2(r)A0]x +… 0
 m
f0(r)A0x + [ f 0 ( r  m ) Am   f i ( r  m  i ) Am i ] x
r m
r 0
m 1 i 1

The coefficients of all powers of x must vanish independently. The vanishing of the lowest power xr
gives the requirement
f0(r)  0 (5.6)
This equation determines two values of r and is called the indicial equation. The vanishing of the
coefficient of xr + m gives the recurrence formula
m
f 0 ( r  m ) Am    f i ( r  m  i ) Ami (5.7)
i 1

which determines each Am in terms of the preceding A’s, and hence in terms of A0, if for each
positive m the quantity f0(r + m) is not zero.
The root of the indicial equation has the following three cases:

1. Distinct roots not differing by an integer (r  r1, r2), two independent solutions are

y1 ( x )  x r1  Am x m
m 0

y2 ( x)  x r2
 A x
m 0
m
m

2. Double roots r  r1  r2, only one solution can be obtained. From recurrence formula (5.7), (5.4)
can be written as

y ( x, r )  x r  Am ( r ) x m (5.8)
m 0

and
Ly(x,r)  f(r)A0xr  (r  r1)2A0xr (5.9)
The right-hand member of (5.9) vanishes when r  r1, and the first solution is

y1(x)  y(x,r1)  x
r1
A
m 0
m ( r1 ) x m

O. D. E. 43
Since r  r1 is a repeated zero of the right-hand member of (5.9), it follows
     
  r L y ( x, r )  0  L y ( x, r ) 0
  r  r1  r  r  r1
Hence a second solution is
  
y2 ( x)   y ( x, r )
 r  r  r1

When substituting into (5.8), we have


  
y 2 ( x )  x r1 ln x  Am ( r1 ) x m  x r1  Am ( r1 ) x m  y1 ( x ) ln x  x r1  Bm x m
m 0 m 1 m 1

3. Roots differing by an integer (r1  r2  k, k is an integer), we have


Ly(x,r)  f(r)A0xr  (r  r1)(r  r2)A0xr (5.10)
From the recurrence formula (5.7), the expression for the coefficients Ak(r), Ak+1(r), Ak+2(r),
… now all have a factor r  r2 in denominator, and hence will not approach finite limit as r  r2. If
we consider the product (r  r2)y(x,r), we see that as r  r2 terms with coefficients Am for which
m < k, will vanish and the remaining terms will approach finite limits, thus giving rise to an infinite
series of powers of x starting with a term involving
xr + k  xr . Thus the limiting series must be proportional to the series for which r  r1. In this case,
2 1

we have
L[(r  r2)y(x,r)]  f(r)A0xr  (r  r1)(r  r2)2A0xr (5.11)
and
     
 [( r  r2 ) L y ( x, r )]  0  L [( r  r2 ) y ( x, r )] 0
 r  r  r2  r  r  r2
Hence a second solution is
  
y2 ( x)   [( r  r2 ) y ( x, r )]
 r  r  r2

When substituting into (5.8), we have


 
  
d 
y 2 ( x )  ( r  r2 ) x r  Am ( r ) x m  ln x   x r  [( r  r2 ) Am ( r )] x m 
 m 0  r  r2  m 0 dr  r  r2

 Ak ( r2 ) y1 ( x ) ln x  x r2  C m x m
m 0

Ex3. 4xy + 2y + y  0



Solution: Assume y  x  Am x
r m

m 0
  
4  ( m  r )( m  r  1) Am x m  r 1  2  ( m  r ) Am x m  r 1   Am x m  r  0
m 0 m 0 m 0
  
4  ( m  r )( m  r  1) Am x m  r 1  2  ( m  r ) Am x m  r 1   Am 1 x m  r 1  0
m 0 m 0 m 1

[4 r ( r  1)  2 r ] A0 x r 1  {4( m  r )( m  r  1)  2( m  r )] Am  Am 1 }x m  r 1  0
m 0

O. D. E. 44
The indicial equation is
4r(r  1) + 2r  0  r  1/2, 0
The recurrence formula is
Am 1
Am  , m  1,2,3,…
2( m  r )( 2m  2 r  1)
r1  1/2
Am 1
Am   , m  1, 2, 3,
( 2m )( 2m  1)
A A A A A0
A1   0 , A2   1  0 , A3   0 ,, Am  ( 1) m
3! 5 4 5! 7! ( 2m  1)!

xm
y1 ( x )  x  ( 1) m
m 0 ( 2m  1)!
r2  0
Am 1
Am   , m  1, 2, 3, 
( 2m  1)(2m )
A A A A A0
A1   0 , A2   1  0 , A3   0 ,  , Am  ( 1) m
2! 43 4! 6! ( 2m )!

xm
y2 ( x)   ( 1) m
m 0 ( 2m )!
 
xm xm
The solution is y  C1 y1  C 2 y 2  C1 x  ( 1) m  C 2  ( 1) m
m 0 ( 2m  1)! m 0 ( 2m )!

Ex4. x(x  1)y + (3x  1)y + y  0



Solution: Assume y  x  Am x
r m

m 0

O. D. E. 45
 

 ( m  r )( m  r  1) A
m 0
m x m  r   ( m  r )( m  r  1) Am x m  r 1
m 0
  
 3 ( m  r ) Am x m  r   ( m  r ) Am x m  r 1   Am x m  r  0
m0 m 0 m 0
 

 ( m  r  1)( m  r  2) A
m 1
m 1 x m  r 1   ( m  r )( m  r  1) Am x m  r 1
m 0
  
 3 ( m  r  1) Am 1 x m  r 1   ( m  r ) Am x m  r 1   Am 1 x m  r 1  0
m 1 m 0 m 1

[  r ( r  1)  r ] A0 x r 1
  [( m  r ) Am 1  ( m  r ) Am ] x m  r 1  0
2 2

m 1

The indicial equation is


r(r  1)  r  0  r  0, 0
The recurrence formula is
Am  Am1, m  1, 2, 3,…

1
y1(x)   x 
m

m 0 1 x
Let y2  uy1  y2  uy1 + uy1 y2  uy1 + 2uy1 + uy1
Substitute into the given equation
x(x  1)(uy1 + 2uy1 + uy1) + (3x  1)(uy1 + uy1) + uy1  0
[x(x  1)y1 + (3x  1)y1 + y1]u + x(x  1)y1u + [2x(x  1)y1 + (3x  1)y1]u  0
du  dx
xu   u   0    0  u x  1  u  ln x
u x
ln x
y2 
1 x
C C ln x
The Solution is y  C1 y1  C 2 y 2  1  2
1 x 1 x

Ex5. (x2  x)y  xy + y  0


Solution: Assume y  x  Am x
r m

m 0

O. D. E. 46
 

 (m  r )(m  r  1) A
m 0
m x m  r   ( m  r )( m  r  1) Am x m  r 1
m 0
 
  ( m  r ) Am x m  r   Am x m  r  0
m 0 m 0
 

 (m  r  1)(m  r  2) A
m 1
m 1 x m  r 1   ( m  r )( m  r  1) Am x m  r 1
m 0
 
  ( m  r  1) Am 1 x m  r 1   Am 1 x m  r 1  0
m 1 m 1

 r ( r  1) A0 x r 1   [( m  r  2) 2 Am 1  ( m  r )( m  r  1) Am ] x m  r 1  0
m 1

The indicial equation is


r(r  1)  0  r  0, 1
The recurrence formula is
( m  r  2) 2
Am  Am 1 , m  1,2,3,
( m  r )( m  r  1)
r1  1
( m  1) 2
Am  Am 1 , m  1,2,3,
m( m  1)
A1  0, A2  0, A3  0,…
y1(x)x
Let y2  uy1  y2  xu + u, y2  xu + 2u, we have
(x2  x)(xu + 2u)  x(xu + u) + ux  0
x 1 x 1 1 1 1
x(x  1)u + (x  2)u  0  lnu  ln 2  u   2   2  u  ln x 
x x x x x
y2  xlnx + 1
The solution is y  C1y1 + C2y2  C1x + C2(xlnx + 1)

O. D. E. 47