You are on page 1of 22

Math246 Exercises

3/16/16, 10:23 PM

Higher-Order Linear Ordinary Differential Equations


Homogeneous Equations: General Methods and Theory
Find the general solution to the following dierential equations.

Exercise 1
Consider the second-order dierential equation x
all solutions to it. Is that weird? Why (or why not)?

= x . The four functions cosh(u) , sinh(u) , eu , and eu are

Solution
It is not weird, because we can express two of these functions in terms of the other two, for instance using the
identities

eu + eu
cosh(u) =
2

eu eu
sinh(u) =
.
2

and

So there are only two linearly independent functions in play here.

Exercise 2
Consider the fourth-order dierential equation y (4) = y . Check that the functions y 1 (t) = cos(t) and
y 2 (t) = sin(t) both are solutions to it. There should be two other fundamentally dierent solutions; can you find
them by inspection?

Solution
n
0
1
The following table lists derivatives of sine and cosine:
2
3
4

d
dt n

cos(t)

cos(t)
sin(t)
cos(t)
sin(t)
cos(t)

d
dt n

sin(t)

sin(t)
cos(t)
sin(t)
cos(t)
sin(t)

Indeed the fourth

derivatives match the functions themselves, so they are solutions. One further solution comes from e (if it is its
own derivative then it is its own fourth derivative). The last solution is et .
Another possibility you may have guessed after doing the previous problem is sinh(t) and cosh(t) which are also
fundamentally dierent from sin(t) and cos(t) , and are independent solutions.
http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 1 of 22

Math246 Exercises

sin(t)

3/16/16, 10:23 PM

cos(t)

Exercise 3
3z

4z

are solutions to w 7w + 12w = 0 . Then find c 1 and c 2 so that


W (z) = c 1 e3z + c 2 e4z is a solution to the dierential equation satisfying the initial conditions W (0)
W (0) = 0 .
Check that e

and e

=2,

Solution
The check is routine. To find the constants, note that W (0)
constants come from the system of equations

= c 1 + c 2 and W (0) = 3c 1 + 4c 2 , so the desired

c1 + c2 = 2
3c 1 + 4c 2 = 0,
whose solution is c 1

= 8 , c 2 = 6 . (So W (z) = 8e3z 6e4z .)

Exercise 4
Check that ex sin(x) and ex cos(x) are solutions to y 2y + 2y = 0 . Then find c 1 and c 2 so that
Y (x) = c 1 ex sin(x) + c 2 ex cos(x) is a solution to the dierential equation satisfying the initial conditions
Y (0) = 3 , Y (0) = 2 .

Solution
Well check f(x)

= ex sin(x) , and the other one will be similar. The product rule tells us
f (x) = ex sin(x) + ex cos(x)

and

f (x) = (ex sin(x) + ex cos(x)) + (ex cos(x) ex sin(x)) = 2ex cos(x).


Therefore

f 2f + 2f = 2ex cos(x) 2(ex sin(x) + ex cos(x)) + 2ex sin(x) = 0,


so it checks out.
To find the constants, note that Y (0)

= c 2 since sin(0) = 0 . Immediately we get c 2 = 3 . Next

Y (x) = c 1 ex sin(x) + c 1 ex cos(x) + c 2 ex cos(x) c 1 ex sin(x),

(0) =

(0) = 2

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 2 of 22

Math246 Exercises

3/16/16, 10:23 PM

and plugging in x =
we have c 1 = 5 .

0 destroys the terms with sine, leaving Y (0) = c 1 + c 2 . Since Y (0) = 2 and c 2 = 3 ,

The solution is Y (x)

= 5ex sin(x) + 3ex cos(x).

Exercise 5
a. Check that e and x2
x

+ 2x + 2 are solutions to the equation


xy (x + 2)y + 2y = 0.

b. Find c 1 and c 2 so that Y (x) = c 1 e + c 2 (x2 + 2x + 2) is a solution to the dierential equation satisfying

the initial conditions Y (1) = e , Y (1) = e + 2 .


x

c. On what interval(s) is the function Y you found in part (b) a solution to the equation?

Solution
x

a. All the derivatives of e are equal to e , so when we plug it in we get

xex (x + 2)ex + 2ex = 0.


The polynomial x2 + 2x + 2 is slightly more interesting. Its derivative is 2x + 2 and its second derivative is
2 , so putting it into the equation gives

x(2) (x + 2)(2x + 2) + 2(x2 + 2x + 2) = 2x (2x2 + 6x + 4) + (2x2 + 4x + 4) = 0,


so everything checks out.
b. Putting x

= 1 into Y and Y leads to the system of equations


ec 1 + 5c 2 = e
ec 1 + 4c 2 = e + 2.

Subtracting second equation from the first brings us to c 2 = 2 , and then putting that into the first
10
equation and rearranging gives us the unflattering other constant c 1 = e + 1. (So the solution is
x
2
( 10
e + 1)e 2(x + 2x + 2) .)

c. In normal form the equation were working with is

y
whose coecients have a discontinuity at x

x+2
2
y + y = 0,
x
x

= 0. So the solution is valid on the half-line (0, ) .

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 3 of 22

Math246 Exercises

3/16/16, 10:23 PM

Exercise 6
Check that w , w 3 , and w 5 are solutions to the third-order dierential equation

w 3 z 6w 2 z + 15wz 15z = 0 .
Then find constants c 1 , c 2 , and c 3 so that Z(w) = c 1 w + c 2 w 3
(0, ) satisfying Z(2) = 0 , Z (2) = 0 , and Z (2) = 1 .

+ c 3 w 5 are solutions to the equation on

Solution
The check is straightforward, though the numbers get large. For w , only the last two terms matter, and we get
15w 15w = 0 . For w 3 we get w 3 (6 6 + 15 15) = 0 . For w 5 , we get
w 5 (60 6 20 + 15 5 15) = 0.
The initial conditions lead to the following system of equations:

2c 1 + 8c 2 + 32c 3 = 0
c 1 + 12c 2 + 80c 3 = 0
12c 2 + 160c 3 = 1.
Solving the system can be done in a variety of ways; one option is to subtract the third equation from the second
to get c 1 = 80c 3 1 , then plug that into the first equation to get 8c 2 + 192c 3 = 2 , then multiply it by three and
1
the third equation by two, and then subtract those. That should give 256c 3 = 4 , or c 3 = 64
. From there c 1 = 14
and c 2

1
follow painlessly. The equation desired is
8

w3
w5
w

+
.
4
8
64

Exercise 7
2y 8y = 0 is {e2w , e4w }. Find a solution

which satisfies the general initial conditions Y (0) = y 0 , Y (0) = y 1 .

One fundamental set of solutions for the dierential equation y

Y = c 1 e2w + c 2 e4w

Solution
The system of equations that c 1 and c 2 satisfy is

c1 + c2 = y0 ,
2c 1 + 4c 2 = y 1 .

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 4 of 22

Math246 Exercises

3/16/16, 10:23 PM

By solving the first equation for c 1 we find c 1


c 2 = 13 y 0 + 16 y 1 . Therefore we obtain

= y 0 c 2 . By plugging this result into the second equation we find


2
1
y0 y1 ,
3
6
1
1
c2 = y0 + y1 .
3
6
c1 =

Exercise 8
One fundamental set of solutions for the dierential equation y + 9y = 0 is {cos(3t), sin(3t)}. Find a solution
Y = c 1 cos(3t) + c 2 sin(3t) that satisfies the general initial conditions Y ( 2 ) = y 0 , Y ( 2 ) = y 1 .

Solution

Because cos(3 2 )

= 0 and sin(3 2 ) = 1 the system of equations that c 1 and c 2 satisfy is


y 0 = c 2 ,
y 1 = 3c 1 .

Therefore c 1

1
3

y 1 and c 2 = y 0 .

Exercise 9
One fundamental set of solutions for the dierential equation u2 x + 4ux = 0 is {1, u3 } . Find a solution
X(u) = c 1 + c 2 u3 that satisfies the general initial conditions X(1) = x0 , X (1) = x1 .

Solution
The system of equations that c 1 and c 2 satisfy is

x0 = c 1 + c 2 ,
x1 = 3c 2 .
From this we see that c 2

1
3

x1 and so c 1 = x0 +

1
3

x1 . (Note that the interval of definition of the solution is


(0, ) because the normal form of the dierential equation has coecients that are undefined at u = 0.)

Exercise 10

=0

{1,

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 5 of 22

Math246 Exercises

3/16/16, 10:23 PM

x1 y = 0 is {1, x2 } . Find a solution


Y (x) = c 1 + c 2 x2 that satisfies the general initial conditions Y (2) = y 0 , Y (2) = y 1 .

a. One fundamental set of solutions for the dierential equation y


b. What is the interval of definition for your solution in (a)?
c. What happens if you try to solve for the initial conditions y(0)

= 1 , y (0) = 2?

Solution
a. The system of equations we have to solve is

y 0 = c 1 + 4c 2 ,
y 1 = 4c 2 .
Adding these equations gives c 1

= y 0 + y 1 , and of course c 2 =

1
4

y1 .

b. Because the coecient 1/x of the dierential equation is undefined at x = 0 and continuous elsewhere,
while the initial point is x = 2, the interval of definition of the solution is (, 0) .
c. The dierential equation is undefined at x = 0 so that x = 0 is not a valid initial point. Therefore there is no
such solution. Had you not seen this fact and tried to solve for c 1 and c 2 the system of equations that
results is

1= c 1 ,
2= 0 .
The second equation shows that you have made a mistake.

Exercise 11
Compute the Wronskian of the functions W1 (z)

= z and W2 (z) = cos(z) . Is this defined for all points z?

Solution
W [W1 , W2 ](z) = det (

z
1

cos(z)
) = z sin(z) cos(z).
sin(z)

Yes, the Wronskian of W1 and W2 is defined for all real z.

Exercise 12
Compute the Wronskian of the functions X 1 (u)
also explains what was going on in Exercise #1.

= cosh(u) , X 2 (u) = sinh(u) , and X 3 (u) = eu . Notice this

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 6 of 22

Math246 Exercises

3/16/16, 10:23 PM

Solution
cosh(u)
W [X 1 , X 2 , X 3 ](u) = det sinh(u)
cosh(u)

sinh(u)
cosh(u)
sinh(u)

eu
eu = 0,
eu

since this matrix has two copies of the same row.


If you didnt notice this, we can get to 0 the long way too. Well need the following: cosh(u)

sinh(u) =
get

e u e u
, and
2

e u +e u
,
2

cosh 2 (u) sinh 2 (u) = 1 . If you expand the determinant out by using the first row, you

cosh(u) (eu cosh(u) eu sinh(u)) sinh(u) (eu sinh(u) eu cosh(u)) + eu (sinh 2 (u) cosh 2 (u)) .
= eu and sinh(u) cosh(u) = eu . Using that in the
2
2
first two terms of the sum and using sinh (u) cosh(u) = 1 in the third one, we simplify all the way down to
The identities above tell us that cosh(u) sinh(u)

cosh(u) + sinh(u) eu ,
and indeed cosh(u) + sinh(u)

= eu , so everything cancels.

Exercise 13
(Continuation of Exercise #3) Show that {e3t , e4t } is a fundamental set of solutions for the second-order
dierential equation y

7y + 12y = 0 .

Solution
Compute the Wronskian:

W [e , e ](t) = det [
3t

4t

e3t
3e

3t

e4t
4e

4t

] = 4e7t 3e7t = e7t .

Since this is not identically zero, they form a fundamental set.

Exercise 14
(Continuation of Exercise #4) Show that {ex sin(x), ex
order dierential equation y 2y + 2y = 0 .

cos(x)} is a fundamental set of solutions for the second-

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 7 of 22

Math246 Exercises

3/16/16, 10:23 PM

Solution
Compute the Wronskian:

ex sin(x)
W [e sin(x), e cos(x)](x)= det [ x
e sin(x) + ex cos(x)
x

ex cos(x)
]
ex cos(x) ex sin(x)

= e2 x (sin(x) cos(x) sin2 x) e2x (sin(x) cos(x) + cos2 (x)) = e2x .


Since this is not identically zero, they form a fundamental set.

Exercise 15
(Continuation of Exercise #6) Show that {x, x

, x5 } is a fundamental set of solutions for the third-order

dierential equation

x3 y 6x2 y + 15xy 15y = 0.

Solution
Another Wronskian calculation:

x
3
5
W [x, x , x ](x) = det 1
0

x3
3x2
6x

x5
5x4 .
20x3

To calculate this determinant, well expand down the first column to exploit the zero. Doing so gives us

x (60x5 30x5 ) 1 (20x6 6x6 ) + 0 = 30x6 14x6 = 16x6 0,


so these functions do form a fundamental set.

Exercise 16
t

2t

Show that the functions e , e , and e

3t

are linearly independent on (, ).

Solution
Approach one: suppose that c 1 e + c 2 e2t + c 3 e3t = 0 for all t , and plug in three values for t . Ill choose t
t = ln(2) , and t = ln(3) , but of course any numbers will work. This gives the system
t

= 0,

0
http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 8 of 22

Math246 Exercises

3/16/16, 10:23 PM

0= c 1 + c 2 + c 3
0= 2c 1 + 4c 2 + 8c 3
0= 3c 1 + 9c 2 + 27c 3 .
Now the first equation says c 1 = c 2 c 3 , so the second equation says c 2 = 3c 3 , so the third equation says
6c 3 = 0. Then c 2 = 0 , so c 1 = 0 , so theyre all zero. This proves linear independence.
Approach two: Compute the Wronskian:

e
t
t 2t 3t
W [e , e , e ](t)= det e
et
t

e2t
2e2t
4e2t

e3t
3e3t
9e9t

= et (18e5t 12e5t ) e2t (9e4t 3e4t ) + e3t (4e3t 2e3t ) = 2e6t .

Since this is not identically zero, they are linearly independent.

Exercise 17
Show that log(z) , log(5z) , and 1 are linearly dependent over (0, ) .

Solution
The linear dependence of {log(z), log(5z), 1} is seen from the fact that log(5z)
shows that

= log(5) + log(z), which

log(z) log(5z) + ln(5) 1 = 0 .


The Wronskian of {log(z), log(5z), 1} is zero, but you cannot conclude that these functions are linearly
dependent from this fact alone.

Exercise 18
(Continuation of Exercise #3) Find a natural fundamental set for the dierential equation y
associated with the initial time 0 .

7y + 12y = 0

Solution
3t

4t

Earlier we saw that a fundamental set of solutions for this equation is {e , e }. If we seek a solution
Y (t) = c 1 e3t + c 2 e4t that satisfies the general initial conditions Y (0) = y 0 , Y (0) = y 1 then we obtain the
system of equations

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 9 of 22

Math246 Exercises

3/16/16, 10:23 PM

c1 + c2 = y0 ,
3c 1 + 4c 2 = y 1 .
These have the solution

c 1 = 4y 0 y 1 ,
c 2 = 3y 0 + y 1 .
The solution is thereby

Y (t) = (4y 0 y 1 )e3t + (3y 0 + y 1 )e4t = (4e3t 3e4t )y 0 + (e4t e3t )y 1 .


From this we can read o that the natural fundamental set of solutions for t
N 1 (t) = e4t e3t .
Notice that N 0 (0)

= 0 is N 0 (t) = 4e3t 3e4t and

= 1 , N 0 (0) = 12 12 = 0 , N 1 (0) = 0 , and N 1 (0) = 3 + 4 = 1 .

Exercise 19
(Continuation of Exercise #9) Find a natural fundamental set for the dierential equation w 2 z
associated with the initial time 1 .

+ 4wz = 0

Solution
Well use the generic initial conditions from a previous problem to get the desired functions. In said Problem 9 we
w

found that the general initial conditions Z(1) = z0 , Z (1) = z1 are satisfied by the constants c 1 = z0 + 3 z1
w

= 3 z1 . Tallying up the contributions to z0 and z1 , we obtain the functions N 0 (w) = 1 and


w
w
N 1 (w) = 3 3 w 3 .
and c 2

Exercise 20
(Continuation of Exercise #8) Find a natural fundamental set for the dierential equation y

with the initial time 2 .

+ 9y = 0 associated

Solution

In Problem 8 we found that the general initial conditions Y ( 2 )


y

= y 0 , Y ( 2 ) = y 1 are satisfied by the constants

c 1 = 31 and c 2 = y 0 . These contributions of y 0 and y 1 imply that the functions were looking for are
N 0 (t) = sin(3t) and N 1 (t) = 13 cos(3t) .

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 10 of 22

Math246 Exercises

3/16/16, 10:23 PM

Exercise 21
Check that e is a solution to w
z

2w + w = 0 , and then use reduction of order to find a fundamental set.

Solution
The check is straightforward, so lets dive in with the reduction of order. Let the other solution be w(z)
for an unknown function u . Now we dierentiate:

= ez u(z)

w (z) = ez u(z) + ez u (z)


and

w (z) = ez u(z) + 2ez u (z) + ez u (z).


Adding everything up, we get

0 = w 2w + w = ez u(z) + 2ez u (z) + ez u (z) 2ez u(z) 2ez u (z) + ez u(z) = ez u (z).
Now e is never zero, so we can divide it away, leaving us with u (z) = 0 . If we set k = u , then the dierential
equation is k = 0 , which has the solution k = 1 (any constant will do). Integrating both sides tells us that u = z
(up to a constant), so the other solution is w(z) = ez u(z) = zez , a fundamental set of solutions is {ez , zez },1
and the general solution is W (z) = c 1 ez + c 2 zez .
z

If you love constants, you might have said k = c 1 is the solution to k = 0 . If you did, thats fine; youll get
u(z) = c 1 z + c 2 , and that will give you the complete solution W (z) straightaway.

1. Check the Wronskian if you dont believe me.

Exercise 22
Its clear that Y1 (t)

= 1 solves the dierential equation ty + y = 0 . Find another solution by reduction of order.

Solution
Let the other solution be y(t) = Y1 (t)u(t) = u(t) . Plugging y into the equation gives tu + u = 0 . Make up a
new variable w = u , and put the equation in normal form to arrive at w + 1t w = 0. Now the first-order recipe
leads us to the solution w

1 1
t , and from there we get

1. The integrating factor is (t)

u = log(t) . (So a general solution is c 1 + c 2 log(t) .)

= t , which then says

d
dt

[tw] = 0 , so tw = c , and w = c

1
t .

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 11 of 22

Math246 Exercises

3/16/16, 10:23 PM

Exercise 23
If n is a positive integer, then Y1 (t) = 1 is also a solution to y
equation. [Hint. Your answer will have an n in it somewhere.]

n
x

y = 0 . Complete a fundamental set for this

Solution
= Y1 (x)u(x) = u(x). Plugging y into the equation gives u nx u = 0 . Make
n
up a new variable w = u , and put the equation in normal form to arrive at w x w = 0. Now the first-order
d
n
recipe tells us to use the integrating factor (x) = xn , which gives the equation dt [wxn ] = 0 , or w = cx .
c
Therefore u = n+1 xn+1 . A candidate for the other slot in a fundamental set is xn+1 .
Let the other solution be y(x)

Exercise 24
Check that Z 1 (u) = e2u cos(u) is a solution to D
complete a fundamental set for it.

z 4Dz + 5z = 0 . Then use reduction of order to

Solution
You should calculate that Z1 (u) = 2e2u
check out when those are plugged in.

cos(u) e2u sin(u) and Z1 (u) = 3e2u cos(u) 4e2u sin(u) . Things

To reduce order, let z(u) = Z 1 (u)q(u) be the second solution, with q a to-be-determined function. We have
z(u) = Z1 q + Z 1 q and z(u) = Z1 q + 2Z1 q + Z 1 q , and using the derivatives from the previous paragraph
we get the following:

z(u)= e2u cos(u)q(u)


z(u)= (2e2u cos(u) e2u sin(u))q(u) + e2u cos(u)q (u)
z(u)= (3e2u cos(u) 4e2u sin(u))q(u) + (4e2u cos(u) 2e2u sin(u))q (u) + e2u cos(u)q (u) .
Plugging this into the initial equation will wipe out the first term in each line when taken together. When the dust
settles, we wind up with

0 = e2u cos(u)q (u) 2e2u sin(u)q (u) ,


which is a first-order equation in
never zero, bringing us to

q . Lets use a new letter for that, say w = q . We can divide by e2u because its
(u) 2 sin(u)w(u) = 0 ,
cos(u)w

or in normal form

(u) 2 tan(u)w(u) = 0 .

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 12 of 22

Math246 Exercises

3/16/16, 10:23 PM

(u) 2 tan(u)w(u) = 0 .
w
= exp(2 ln(cos(u))), because tan(u) du = ln(cos(u)) (perform a
substitution to get this). This simplifies to q(u) = cos2 (u) . Upon multiplying through by the integrating factor, the
The proper integrating factor is q(u)
equation becomes

D[w(u) cos2 (u)] = 0 ,


so w(u) = c sec2 (u) . Now we integrate both sides of this to get q(u) ; an antiderivative of sec2 (u) is given by
tan(u) . So up to choice of constants, q(u) = tan(u) is a fine example, and so
z(u) = Z 1 (u)q(u) = e2u cos(u) tan(u) = e2u sin(u) is the other element of the fundamental set.

Exercise 25

Check that Z 1 (w) = w is a solution to (w 1)(w 2) z


second linearly independent solution.

wz + z = 0. Then use reduction of order to find a

Solution
is 1 .
The check isnt hardit boils down to the fact that the coecient on z is +1 and the coecient on wz
Lets worry more about the reduction of order.
We suppose the second solution is z(w) = wu(w) with u a function to be discovered. Then we have
(w) + u(w) and z(w) = wu
(w) + 2u
(w) , so when we plug into the equation we get
z(w) = wu

+ 2u
) w(wu
+ u) + wu = 0 .
(w 1)(w 2)(wu
As usual the terms with u(w) drop out. If we expand what remains we get

+ (w 2 6w + 4)u
= 0,
(w 3 3w 2 + 2w)u

a first-order dierential equation in u . Set k(w)

(w) , so that this equation is (in normal form)


=u

w 2 6w + 4

k+ 3
k = 0.
w 3w 2 + 2w
Now to get our integrating factor we need to be able to integrate this rational function of w . Thankfully we were
given a factorization of the denominatorits w(w 1)(w 2) . The resulting partial fractions decomposition
looks like this:

w 2 6w + 4
2
1
2
=
+
+
.
3
2
w
w1
w2
w 3w + 2w
(Ill spare you the details.) When we antidierentiate, we get

6w + 4

dw = 2 log(w) + log(w 1) 2 log(w 2) .

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 13 of 22

Math246 Exercises

3/16/16, 10:23 PM

w 2 6w + 4
dw = 2 log(w) + log(w 1) 2 log(w 2) .
w 3 3w 2 + 2w

Therefore, our integrating factor is

(w) = exp (2 log(w) + log(w 1) 2 log(w 2)) = exp (log (

w 2 (w 1)
(w 2)2

)) =

w 2 (w 1)
(w 2)2

Multiplying our dierential equation by (w) and using the first-order recipe, we get dw [ku] = 0, so
k(w) = c for some constant c . Then we integrate both sides to arrive at u(w) , which means our unknown
u(w)

function is given by the integral

(w 2)2
u(w) = c 2
dw .
w (w 1)
The constant c is just going to get passed around, and because any solution linearly independent from w will
satisfy the problem, lets just ignore it for now. Its partial fractions time again!1 This time the decomposition is

(w 2)2
1
4
=

.
w1
w2
w 2 (w 1)
1

(There could have been a w term in the expansion, but it wound up having a coecient 0 .) So, finally, we have
determined

u(w) = log(w 1) +

4
,
w

so the other solution we were looking for is

z(w) = wu(w) = w log(w 1) + 4 .


If you are like the person writing this solution, you will have to check this to believe that it works, but it does.

1. Try to look enthused.

Exercise 26
As suggested in the text, give a proof of Abels theorem for the third-order case. That is to say, suppose Y1 (t) ,
Y2 (t) , and Y3 (t) are three solutions to the dierential equation

y (t) + a1 (t)y (t) + a2 (t)y (t) + a3 (t)y(t) = 0,


then show that their Wronskian W (t)

= W [Y1 , Y2 , Y3 ](t) satisfies the first-order dierential equation

1 (t)W

= 0.

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 14 of 22

Math246 Exercises

3/16/16, 10:23 PM

W + a1 (t)W = 0.

Solution
We start by writing down what the Wronskian is:

W (t) = Y1 (Y2 Y3 Y2 Y3 ) Y2 (Y1 Y3 Y1 Y3 ) + Y3 (Y1 Y2 Y1 Y2 ) ,


which for simplicity Ill expand at this juncture. Ill also try to improve legibility by putting the subscripts in
increasing order in every term:

W (t) = Y1 Y2 Y3 Y1 Y2 Y3 Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 Y1 Y2 Y3 .
Now hold your breath and dierentiate this. Each term will spawn three terms when we do this, and Ill write each
on its own line.

W (t)= (Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 )
(Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 )
(Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 )
+ (Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 )
+ (Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 )
(Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 ) .
Twelve terms match up and cancel, leaving only six to contend with, which Ill rearrange a little bit:

W (t) = Y1 Y2 Y3 Y1 Y2 Y3 Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 Y1 Y2 Y3 .
The rearrangement was so that I could group terms as follows:

W (t) = Y1 (Y2 Y3 Y2 Y3 ) Y2 (Y1 Y3 Y1 Y3 ) + Y3 (Y1 Y2 Y1 Y2 ),


which is starting to look like the original line for W at the top of the solution. Notice that each term that survived
has a third derivative in it somewhere. We know from the dierential equation that, for i = 1, 2 , and 3 ,

Yi (t) = a1 (t)Yi (t) a2 (t)Yi (t) a3 (t)Yi (t).


Plug this fact in for all three of the third derivatives. We arrive at

W (t)= (a1 Y1 a2 Y1 a3 Y1 )(Y2 Y3 Y2 Y3 )

(a1 Y2 a2 Y2 a3 Y2 )(Y1 Y3 Y1 Y3 )
+(a1 Y3 a2 Y3 a3 Y3 )(Y1 Y2 Y1 Y2 ).

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 15 of 22

Math246 Exercises

3/16/16, 10:23 PM

Now expand all of this. All the terms with an a2 or an a3 pair up and cancel, while all of the terms with an a1 have
no mates. Therefore we can pull a1 (t) out from the entire expression. What remains is

W (t) = a1 (t)(Y1 Y2 Y3 Y1 Y2 Y3 Y1 Y2 Y3 + Y1 Y2 Y3 + Y1 Y2 Y3 Y1 Y2 Y3 ),
and the expression in the big parentheses is up to rearrangement exactly the W that we started with. So weve

shown that W = a1 W , or W + a1 W = 0 , as Abels theorem predicted.

Exercise 27
Suppose we have a second-order homogeneous dierential equation and we happen to know one of the solutions.
Then the method of reduction of order will always give us a first-order dierential equation whose solution is a
linearly independent solution to the equation. In the problems above, the first-order dierential equation is solvable,
but this doesnt happen in generaloften we wind up with an integral that we cant solve. This is not to say all is
lost; having an integral (or even just a dierential equation in the first place) opens up the possibility of determining
values of the function by use of numerical methods, after all.
Consider the dierential equation

y 2(2x2 + 1)y = 0.
x2

One of the homogeneous solutions is Y1 (x) = e . Check this. Then use the method of reduction of order to
come up with an expression for another solution to the homogeneous equation, linearly independent from Y1 . [You
should guess from the paragraph preceeding this that you probably will get stuck at an integral.]

Solution
We start with the check. We have Y1 (x)

= ex , Y1 (x) = 2xex , and, as we hoped,

x2

Y1 (x) = 2e + 4x2 e

x2

= (4x2 + 2)e

Now suppose the other solution to the equation is Y2 (x)


determined. We calculate:
x2

x2

x2

= 2(2x2 + 1)e .
2

= u(x)Y1 (x) = u(x)ex , where u(x) is to be

Y2 (x)= u (x)e + 2xu(x)e

x2

x2

x2

x2

Y2 (x)= (4x2 + 2)u(x)e + 4xu (x)e + u (x)e .


We therefore have
x2

x2

0 = Y2 (x) 2(2x2 + 1)Y2 = u (x)e + 4xu (x)e .

+ 4x

w=0

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 16 of 22

Math246 Exercises

3/16/16, 10:23 PM

x2

x2

x2

Writing w(x) = u (x), we see that w satisfies the first-order equation e w + 4xe w = 0 . The e just
clutters things up; as it is never zero, theres no harm in dividing it away. The dierential equation we need to solve
is w + 4xw = 0 , which doesnt look too scary. The coecient on w is a(x) = 4x which has as an
x2
antiderivative A(x) = 2x2 , so an appropriate integrating factor is (x) = e2 . We then get

d
2
[we2x ] = 0,
dx
and a member of the solution family is w(x)
we can say for our friend u is that

= e2x . This function doesnt have a nice antiderivative, so the best


u(x) = e2x dx,
2

and Y2 (x)

= ex e2x dx is our other solution.

Exercise 28
Show that {x

, x 2 } is a fundamental set of solutions for the second-order dierential equation


3

2x2 y + xy 3y = 0 . Make sure to check to see if the Wronskian of x1 and x2 is defined everywhere. (Note:
3
1
you will need first, to verify that each of the two functions {x , x 2 } satisfies the dierential equation and
second, to show that they form a fundamental set of solutions.)

Solution
3

We first check that Y1 = x1 and Y2 = x 2 satisfy the dierential equation (this involves just taking the first and
second-order derivatives and checking that they satisfy the above dierential equation).
Second, to verify whether Y1 and Y2 form a fundamental set of solutions, let us compute the Wronskian of the
two:

x1
W [Y1 , Y2 ](x) = det
x2

x2
1
2
2

=
2
x
=
,
1
3 2
(x)
x
2
3

which is never zero for any nonzero x. Therefore, Y1 (x) and Y2 (x) do indeed form a fundamental set of solutions.

Exercise 29
Without solving, determine the Wronskian of two solutions evaluated at t
equation:
5

y = 0 , y(1) = 5 ,

= 1 for the following dierential

= 10.

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 17 of 22

Math246 Exercises

3/16/16, 10:23 PM

t 5 y 2t 2 y t 7 y = 0 , y(1) = 5 , y = 10.
What is the interval of definition of this particular solution to the initial-value problem?

Solution
We first need to put the second-order, homogeneous linear equation into normal form. The normal form of the
equation looks like:

y 2t 3 y t 2 y = 0.
Now, using Abels Theorem, we obtain that the Wronskian is:

W (t) = ce

W (t) = ce
Then W (1)

dt

t3

t 2

=ce.

Also note that the only point where a solution to y 2t 3 y t 2 y = 0 would attain a singularity is at t = 0 . The
interval of definition of this particular solution to the initial-value problem is (0, +) , because the initial time point
t = 1 falls into that time interval.

Exercise 30
Without solving, determine the Wronskian of two solutions evaluated at u
equation:

= 4 for the following dierential

2u2 y + uy 3y = 0.
Is the Wronskian defined for all u ?

Solution
We first need to put the second-order, homogeneous linear equation into normal form. The normal form of the
equation looks like:

y +

1
3
2 = 0.
2u
2u

Now, using Abels Theorem, we obtain that the Wronskian is:


1

W (u) = ce 2u

dt

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

W (u) = c

log(u)

log

=c

=c

Page 18 of 22

Math246 Exercises

3/16/16, 10:23 PM

W (u) = ce
The Wronskian is defined for all u

12 log(u)

log

= ce

1
( u)

=c

1
.
(u)

> 0. Also, W (4) = 2c.

Exercise 31
The following problem is an application of the Method of Linear Superposition, Theorem 2.1 . Assume that
cos(x) and x are both solutions of the equation p(D)w = q(x) , for a certain polynomial p(x) and a certain
function q(x) .
(a) Write down a nonzero solution of the equation p(D)w
(b) Write down a solution w(x) of p(D)w

= 0.

= q(x) such that w(0) = 2.

Remark We havent discussed how to solve nonhomogeneous second-order linear dierential equations yet,
but we dont need to know that yet!

Solution
(a) By linearity, p(D) = (cos(x) x) = p(D) cos(x) p(D)x
the homogeneous equation could be w(x) = cos(x) x .

= q(x) q(x) = 0 . So a possible solution to

(b) All we need to do is to ensure that a solution to the dierential equation satisfies w(0) = 2. For example, we
can ensure this by considering w(x) = 2 cos(x) , or using linear suposition, w(x) = 2 cos(x) ax , where a is
any real parameter.

Exercise 32
Prove the following statement: ``If f(t) and g(t) are linearly independent solutions of a second-order linear
homogeneous dierential equation on an interval I , then f and g cannot have a maximum at the same location in
I ."
(Hint : This is a more theoretical argument than what youve seen before. Think of an argument by contradiction.
Your proof should include complete sentences.)

Solution
We will approach this problem through an argument via contradiction.
Suppose not, i.e. suppose that both f(t) and g(t) attain their maximum at the same point (call it t in I ), and let
us study what happens in this case. The Wronskian evaluated at t = t then becomes

W ( ) = det (

) = det (

) = f( ) 0 0 g( ) = 0,Page 19 of 22

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Math246 Exercises

3/16/16, 10:23 PM

f(t ) g(t )
f(t ) g(t )
W (t ) = det (
) = f(t ) 0 0 g(t ) = 0,
) = det (
f (t ) g (t )
0
0

since t maximum for both f and g implies that it is also a critical point, i.e. f

(t ) = g (t ) = 0.

Thus, we have obtained that W (t ) = 0 and since t is in I , Theorem 2.3 guarantees that W (t) = 0
everywhere in I . However, this is a contradiction with the fact that the Wronskian of 2 linearly independent
solutions of a second-order linear homogeneous dierential equation cannot equal zero, see Theorem 2.6 . This
implies that our initial assumption, that f and g can indeed have a maximum at the same location inside I , is false.

Exercise 33
1

= 1 and y 2 (x) = x 2 are solutions to the dierential equation yy + (y )2 = 0 for x > 0.


1
Then show that c 1 + c 2 x 2 is not in general a solution of this equation. Can you explain why this result doesnt
contradict the Method of Linear Superposition in Theorem 2.1 ?
Verify that y 1 (x)

Solution
Plugging in the derivatives of y 1 (x) =
solutions to the dierential equation.
Now, the derivatives of y ( x)

1 and y 2 (x) = x 2 into the dierential equation should yield that both are
1

= c 1 + c 2 x 2 are y =

dierential equation yields

1
2

c 2 x 2 and y =

yy + (y )2 =

1
4

c 2 x 2 . Plugging this into the

3
1
c 1 c 2 x 2 .
4

For the right hand side to equal 0 , we need to set either c 1 or c 2 to 0 . Thus, the dierential equation is not
satisfied in general for any linear combination of the solutions y 1 and y 2 .
2

This does not contradict the method of linear superposition, because the dierential equation yy + (y ) = 0
for x > 0 is not a linear homogeneous dierrential equation, which is a requirement for Theorem 2.1 to hold!

Exercise 34
More exploration of the Wronskian :)
If the functions w 1 and w 2 are linearly independent solutions of w + p(u)w + q(u)w = 0 , determine what the
necessary and sucient conditions are such that the functions w 3 = w 1 + w 2 and w 4 = w 1 + w 2 also
form a linearly independent set of solutions.

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 20 of 22

Math246 Exercises

3/16/16, 10:23 PM

Solution
We know that w 1 and w 2 form a linearly set of solutions of the second-order homogeneous dierential equation in
the problem. Therefore by Theorem 2.6 , the Wronkian of the two functions is nonzero, i.e.

W (w 1 , w 2 )(t) = w 1 w 2 w 1 w 2 0,
for all t in the interval where the solution to the dierential equation is defined.
Let us now study what the Wronskian of w 3 and w 4 looks like.

W (w 3 , w 4 )(t) = w 3 w 4 w 3 w 4 = (w 1 + w 2 )(w 1 + w 2 ) (w 1 + w 2 )(w 1 + w 2 )


= ( )w 1 w 2 + ( )w 2 w 1 = ( )(w 1 w 2 w 2 w 1 ) = ( )W (w 1 , w 2 )(t).
The second term is nonzero, so in order to ensure that W (w 3 , w 4 )(t)
condition, that is nonzero.

0 , we need to impose the following

Then, by Theorem 2.6 , we obtain that if and only if is nonzero, can w 3 and w 4 defined as in the
problem form a linearly indepedent set of solutions.

Exercise 35
A clever use of Abel s Theorem :
(a) Consider the dierential equation z
eau .

+ 2az + a2 z = 0 . Show that one of the solutions of the equation is

(b)Use Abels Formula to show that the Wronskian of any two solutions of the given equaton is

W (u) = z1 (u)z2 (u) z1 (u)z2 (u) = ce2au ,


where c is a constant.
(c) Consider z1 (u) = eau from part (a) and use the result in (b) to obtain a dierential equation satisfied by the
second solution z2 (u) . Then solve this equation to show that z2 (u) = ueau .

Solution
(a) The characteristic equation is r2 + 2ar + a2 = 0 . The characteristic root is a with multiplicity 2 . Thus, one
possible solution to the dierential equation would look like z1 (u) = c 1 eau (you can also plug in this solution to
verify that it indeed satisfies the dierential equation).

(b) Abels Formula states that the Wronskian satisfies the dierential equation W + 2aW = 0 , where
W [z1 , z2 ](u) = c. Solving this dierential equation, we obtain that W (u) = ce 2adu = ce2au .

1 (u)

au

W (u) =

1 (u)

(u)

(u) 2 (u) = c

2au ,

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

2 (u)

Page 21 of 22

Math246 Exercises

3/16/16, 10:23 PM

= eau into W (u) = z1 (u)z2 (u) z1 (u)z2 (u) = ce2au , we obtain that z2 (u) must
satisfy z2 + az2 = ceau . Using the method of integrating factors from Chapter 1 Section 8 , we obtain
that z2 (u) = ueau + cueau , for any constant c . Setting c = 0 , we obtain that a possible solution to the
dierential equation is z2 (u) = ueau . This solution is precisely the solution that would yield from our usual way
(c) Plugging in z1 (u)

of solving (a). This was just a detour!

Remark : When we learn about the methods of Key Identity Evaluations and Undetermined Coecients, we
will be able to solve z2 + az2 = ceau through methods other than just using integrating factors!

Exercise 36
Use the method of order reduction to find a second solution of the dierential equation
+ 2zw
2w = 0 , z > 0 , w 1 (z) = z .
z2 w

Solution
First of all, we should verify that w 1 (z)
calculation.

= z indeed satisfies the dierential-equation. This should be a quick

Second, using the suggestion in the problem to utilize the method of order reduction, let us now suppose there is
another solution to the dierential equation of the form w 2 (z) = zv(z) . Substituting this into the dierential
equation, we obtain that

z 2 (vz + 2v) + 2z(vz + z) 2zv = z 3 v + 4z 2 z.


= u , the equation becomes z 3 u
+ 4z 2 u = 0 . Upon simplification, we obtain that zu
+ 4u = 0. This
Setting v
4
is a separable equation which we can solve to obtain that u = cz
. Alternatively, we could also use the method
of integrating factors to get to the same result.
= u = cz 4 meaning that by integrating we obtain that v = c3 z 3 + k . We are not done yet - the
second solution to the dierential equation in the problem is thus w 2 (z) = zv(z) = z 2 , by taking c = 3 and
+ 2zw
2w = 0 to verify that w 2 is
k = 0 in the equation above. You can also plug in w 2 (z) = z 2 into z 2 w

Thus, v

indeed another solution of the dierential equation.

http://math246web.math.umd.edu/math246-exercises/print.php?ch=Homogeneous+Equations%3A+General+Methods+and+Theory&print=1

Page 22 of 22

You might also like