Introduction to Linear Differential Equations

1. The charge f (t) at time t in a particular electric circuit can be modeled by the differential equation
f 00 + 2f 0 + 5f = 0.

(a) Find the general solution of f 00 + 2f 0 + 5f = 0 by first converting this to a continous linear
d~x
dynamical system = A~x.
dt
 
f (t) d~x
Solution. If ~x(t) = 0 , then the derivative is equal to
f (t) dt
 0 
d~x f (t)
= 00
dt f (t)

Since f 00 + 2f 0 + 5f = 0, we can rewrite this as:

f 0 (t)
 
=
−2f 0 (t) − 5f (t)
  
0 1 f (t)
=
−5 −2 f 0 (t)
 
0 1
= ~x(t)
−5 −2
 
0 1
The characteristic polynomial of is λ2 + 2λ + 5 = 0, which has roots −1 ± 2i. The
 −5 −2     
1 − 2i 1 1 1
(−1 + 2i)-eigenspace is ker = span , so ~v1 = is a basis of
−5 −1− 2i  −1 + 2i −1 + 2i
1
the (−1 + 2i)-eigenspace. Then, ~v2 = is a basis of the (−1 − 2i)-eigenspace, and the
−1 − 2i
d~x
general solution of = A~x is
dt
~x(t) = c1 e(−1+2i)t~v1 + c2 e(−1−2i)t~v2
   
1 1
= c1 e−t e2it + c2 e−t e−2it
−1 + 2i −1 − 2i
 
f (t)
Since ~x(t) = , this tells us that
f 0 (t)

f (t) = c1 e−t e2it + c2 e−t e−2it

Typically, we like to express the solutions in terms of real-valued functions, so let’s rewrite this a
bit more:

= c1 e−t (cos 2t + i sin 2t) + c2 e−t (cos 2t − i sin 2t)

= (c1 + c2 )(e−t cos 2t) + (ic1 − ic2 )(e−t sin 2t)

If we write a1 = c1 + c2 and a2 = ic1 − ic2 , then we can rewrite this as

= a1 e−t cos 2t + a2 e−t sin 2t

1

c Harvard Math 21b
 Note: It may seem like we’ve lost an i somewhere here, but we actually haven’t; if we are solving
the differential equation with real initial conditions, then a1 and a2 end up being real (see (b) for
a concrete example of this; if you were to solve for c1 and c2 , they wouldn’t be real). Practically
speaking, we’re really only interested in the solutions with real initial conditions (f (t) models
charge at time t here, and it wouldn’t make any physical sense for f (0) or f 0 (0) to be non-real).

(b) Find the solution of f 00 + 2f 0 + 5f = 0 satisfying the initial conditions f (0) = 1 and f 0 (0) = −5.
Solution. In (a), we found that the general solution of f 00 + 2f 0 + 5f = 0 can be written
as f (t) = a1 e−t cos 2t + a2 e−t sin 2t. So, we simply want to find the values of a1 and a2 that
will make this function satisfy the initial conditions f (0) = 1 and f 0 (0) = −5. The first initial
condition f (0) = 1 tells us that a1 e0 cos 0 + a2 e0 sin 0 = 1, so a1 = 1. To use the second
initial condition, we need to first evaluate f 0 (t); since we’ve already found that a1 = 1, we have
f 0 (t) = e−t (−2 sin 2t) − e−t cos 2t + a2 [e−t (2 cos 2t) − e−t sin 2t]. Therefore, f 0 (0) = −1 + 2a2 , so
the initial condition f 0 (0) = −5 tells us that a2 = −2, and f (t) = e−t cos 2t − 2e−t sin 2t .

2. Use the characteristic polynomial from Prep Video 26 to solve the following linear differential equations.

(a) f 00 + 4f 0 + 7f = 0.
Solution. From Prep Video 26, if we convert f 00 + 4f 0 + 7f = 0 into a continuous dynamical
d~x
system = A~x, the characteristic polynomial of A will simply be λ2 +4λ+7. This characteristic
dt √
−4 ± −12 √
polynomial has roots = −2 ± i 3. Since these are non-real, the general solution of
2 
−2t
√  √ 
00 0
f + 4f + 7f = 0 is c1 e cos 3t + c2 e−2t sin 3t . (Note: You may be more comfortable
√  √ 
writing the general solution as a1 e−2t cos 3t + a2 e−2t sin 3t , which is fine; we simply
√ 
mean that the solutions of the differential equation are linear combinations of e−2t cos 3t and
√ 
e−2t sin 3t .)

(b) f 00 + 4f 0 − 7f = 0.

Solution. By Prep Video 26, if we convert f 00 + 4f 0 − 7f = 0 into a continuous dynamical system

d~x
= A~x, the characteristic polynomial of A will simply be λ2 + 4λ − 7. This characteristic
dt √
−4 ± 44 √
polynomial has roots = −2 ± 11. Since these are distinct real roots, the general
2
√ √
solution of f 00 + 4f 0 − 7f = 0 is c1 e(−2+ 11)t
+ c2 e−(2− 11)t
.

(c) f 00 − 6f 0 + 5f = 0 with initial conditions f (0) = 3 and f 0 (0) = 7.

Solution. One way to do this would be to convert this into a continuous
 dynamical system
d~x f (t)
= A~x as we did in previous problems. Since ~x(t) = , we could write the initial
dt f 0 (t)
   
3 d~x 3
conditions as ~x(0) = . Then, we would be solving = A~x with ~x(0) = , which is
7 dt 7
something we certainly know how to do.
However, since we have a streamlined way to find the general solution of f 00 − 6f 0 + 5f = 0, it’s
actually more efficient to do that. If we convert f 00 − 6f 0 + 5f = 0 into a continuous dynamical

2

c Harvard Math 21b
 d~x
system = A~x, the characteristic polynomial of A will be λ2 − 6λ + 5 = (λ − 1)(λ − 5). This
dt
has roots 1 and 5, so the general solution of f 00 − 6f 0 + 5f = 0 is f (t) = c1 et + c2 e5t .
Now, we simply plug in the initial conditions to solve for c1 and c2 . The first initial condition is
f (0) = 3, so:

3 = f (0)
= c1 e0 + c2 e5·0
= c1 + c2

The second initial condition is about f 0 (0), so we first find f 0 (t): it is c1 et + 5c2 e5t . Then,

7 = f 0 (0)
= c1 e0 + 5c2 e5·0
= c1 + 5c2

So, we are trying to solve the linear system

(
c1 + c2 = 3
c1 + 5c2 = 7

We can do this using Gauss-Jordan elimination, and we get c1 = 2, c2 = 1. So, f (t) = 2et + e5t .

(d) f 000 = 8f . Hint: How can you express the difference of two cubes?
Solution. The characteristic polynomial of f 000 −8f = 0 is λ3 −8 which factors as (λ−2)(λ
√
2
+2λ+4)
2
since this is the difference
√ of two cubes.
√ The roots of λ + 2λ + 4 are λ = −1 ± 3i. So we have
distinct roots 2, −1 + 3i, and −1 − 3i. Thus the differential equation has a general solution of
√ √
f (t) = c1 e2t + c2 e−t cos( 3t) + c3 e−t sin( 3t).

3. (Weekly PSet 10, #1) In this problem, you’ll solve f 00 − 6f 0 + 9f = 0 by converting the problem to
a continuous dynamical system.
 
f (t) d~x
(a) If ~x(t) = , find a matrix A such that = A~x.
f 0 (t) dt
−1
The matrixA is not
 diagonalizable.
  However, by Weekly Problem
 3t Set3t8,
 #1, we can write A = SJS
3 1 1 0 e te
where J = and S = . It turns out that eJt = .
0 3 3 1 0 e3t
 3t
te3t

Jt e
(b) [Optional extra credit] Show that e = by using the power series definition of eJt .
0 e3t
( Hint: Start by computing the first several powers of J until you find a pattern.)
 
d~x a
(c) Solve the continuous dynamical system = A~x with initial condition ~x(0) = .
dt b

(d) Use your answer to (c) to find the general solution of f 00 − 6f 0 + 9f = 0.

4. Consider the homogeneous differential equation f (n) + an−1 f (n−1) + an−2 f (n−2) + · · · + a1 f 0 + a0 f = 0.

3

c Harvard Math 21b
 d~x
(a) In general, how would you convert this to a system = A~x? How big is the matrix A?
dt
(b) How many free variables will the general solution of the homogeneous differential equation have?

4

c Harvard Math 21b