MATH 219

Fall 2017
Lecture 21
Lecture notes by Özgür Kişisel

Content: Laplace transform. Definition and first properties.

Suggested Problems:
§6.1: 1, 2, 10, 17, 23, 26, 27

1 Improper integrals and piecewise continuous func-

tions
Definition 1.1 Let a ∈ R. The improper integral from a to ∞ of f (t) is defined
to be Z ∞ Z M
f (t)dt = lim f (t)dt.
a M →∞ a
We say that the integral converges if the limit exists and diverges if it does not
exist.

Definition 1.2 Let f (t) be a function defined on [a, b]. We say that f (t) is piece-
wise continuous on [a, b] if there exists finitely many points ai having the property
a = a0 < a1 < a2 < . . . < an−1 < an = b such that

• f (t) is continuous on each interval (ai−1 , ai ),

• the right and left limits of f (t) at ai exist for each 0 < i < n, the right limit
of f (t) exists at a0 and the left limit of f (t) exists at an .

We say that f (t) is piecewise continuous on an infinite interval if it is piecewise

continuous on every finite subinterval.

1
Example 1.1 Suppose that
(
1, 2n ≤ t < 2n + 1,
f (t) =
0, 2n + 1 ≤ t < 2n + 2

for all n ∈ Z. Then f (t) is piecewise continuous on R.

Example 1.2 Say 

1 1
1,
 ≤t< ,
g(t) = 2n 2n − 1
1 1
0,
 ≤t<
2n + 1 2n
for n a positive integer, and g(0) = 0. Then g(t) is not piecewise continuous on
[0, 1] since it has infinitely many jump discontinuities in a finite interval.

The main result that we need about piecewise continuous functions is the following:

Theorem 1.1 Say f (t) is piecewise continuous on the interval [a, b]. Then f (t) is
integrable on [a, b].

We will also need the following comparison theorem:

Theorem 1.2 Suppose that

• f (t) and g(t) are piecewise continuous on [a, ∞),

• |f (t)| ≤ g(t) on [a, ∞),

R∞
• a g(t)dt converges.
R∞
Then a
f (t)dt is convergent.

2
2 Laplace transform
Laplace transform is an example of an integral transform. Let us first define integral
transforms in general.

Definition 2.1 Say K(s, t) is a given function of s and t. Then

Z β
F (s) = K(s, t)f (t)dt
α

is called an integral transform with kernel K(s, t).

An integral transform produces a function F (s) of s from a function f (t) of t. With

careful choices of K(s, t), one can hope that certain nice properties of f (t) transform
into some other nice properties of F (s).

Definition 2.2 Suppose that f (t) is piecewise continuous on [0, ∞). Then
Z ∞
L{f (t)} = F (s) = e−st f (t)dt
0

is called the Laplace transform of f (t).

The Laplace transform of f (t) typically converges for certain values of s and diverges
for others.

Example 2.1 Let us find the Laplace transform of f (t) = 1.

Z ∞
L(1) = e−st 1dt
0
Z M
= lim e−st dt
M →∞ 0
e−st M
= lim −|
M →∞ s 0
1 e−sM
= lim − .
M →∞ s s

3
 1
The limit exists and equals if s > 0. It does not exist if s < 0. For s = 0, the
s
integration step above is not valid. For s = 0 the integral instead gives limM →∞ M
which diverges. Hence,
1
L{1} =
s
and it converges if and only if s > 0.

Theorem 2.1 Suppose that f (t) is piecewise continuous on [0, ∞). Say that there
exist constants K, a such that |f (t)| ≤ Keat for all t. Then L{f (t)} converges for
s > a.

Proof Set g(t) = Keat . First, let us look at the Laplace transform of g(t):
Z ∞
L{g(t)} = e−st g(t)dt
0
Z M
= lim e−st Keat dt
M →∞ 0
Z M
= lim Ke(a−s)t dt
M →∞ 0
Ke(a−s)t M
= lim |0
M →∞ a − s

K Ke(a−s)M
= lim − .
M →∞ s − a s−a
The limit exists if s > a. Since |f (t)| ≤ g(t) by assumption, |f (t)e−st | ≤ g(t)e−st .
Now, apply the comparison theorem to f (t)e−st and g(t)e−st . This shows that
L{f (t)} converges for s > a. 

Remark 2.1 1. There exist functions for which it is impossible to find any such
t
K, a. For instance, consider the double exponential function f (t) = ee . Then
the Laplace transform of f (t) does not converge for any value of s.

2. Using the comparison theorem, one can show that the region of convergence
for the Laplace transform of any function is of the form s > a, or the empty
set or all of R.

4
 3. It turns out that it is more natural to consider the values of s in C rather
than R. In this case, the region of convergence in the theorem above becomes
Re(s) > a.

Let us now compute the Laplace transforms of some familiar functions.

Example 2.2 Suppose that f (t) = eat .

Z ∞
at
L{e } = e−st eat dt
0
Z M
= lim e(a−s)t dt
M →∞ 0
(a−s)t
e
= lim |M
0
M →∞ a − s

1 e(a−s)M
= lim − .
M →∞ s − a s−a
1
This limit exists when Re(s) > Re(a) and is equal to .
s−a

Lemma 2.1 Suppose that L{f (t)} = F (s) for Re(s) > a and L{g(t)} = G(s) for
Re(s) > b. Then
L{c1 f (t) + c2 g(t)} = c1 F (s) + c2 G(s)
for Re(s) > max(a, b).

Proof
Z ∞
L{c1 f (t) + c2 g(t)} = e−st (c1 f (t) + c2 g(t))dt
0
Z ∞ Z ∞
−st
= c1 e f (t)dt + c2 e−st g(t)dt
0 0
= c1 L{f (t)} + c2 L{g(t)}.

These operations are valid in the region where both summands are convergent,
namely for Re(s) > max(a, b). 

5
Example 2.3 Say f (t) = cos at. By Euler’s formula, eiat = cos at + i sin at. From
here, we deduce
1 iat
cos at = (e + e−iat )
2
1
L{cos at} = (L{eiat } + L{e−iat })
2 
1 1 1
= +
2 s − ia s + ia
s
= 2 .
s + a2

s
Since Re(ia) = Re(−ia) = 0, the Laplace transform of cos at converges to
s2 + a2
for Re(s) > 0.

Example 2.4 Say f (t) = sin at. As in the previous example, we can write
1 iat
sin at = (e − e−iat )
2i  
1 1 1
L{sin at} = −
2i s − ia s + ia
a
= 2 .
s + a2
Again, the Laplace transform converges for Re(s) > 0.

Example 2.5 Suppose now that f (t) = tn where n is a nonnegative integer. Denote
the Laplace transform of tn by Fn (s). When n = 0, f (t) = 1 therefore F0 (s) = 1s for
Re(s) > 0. Now let n > 0 and Re(s) > 0 :

6
 Z ∞
n
Fn (s) = L{t } = tn e−st dt
0
Z M
= lim tn e−st dt
M →∞ 0
M
tn e−st M 1
Z
= lim − | + ntn−1 e−st dt
M →∞ s 0 s 0
Z ∞
n
= tn−1 e−st dt
s 0
n
= Fn−1 (s)
s

The first limit on the third line above is zero by L’Hospital’s rule. We therefore
obtain the recursive relation Fn (s) = ns Fn−1 (s). Using this relation and the fact that
F0 (s) = 1s , one easily sees that
n!
L(tn ) = Fn (s) =
sn+1

The methods of computation used in two of the examples above can be significantly
generalized:

Proposition 2.1 Suppose that L{f (t)} = F (s) for Re(s) > k. Then

L{eat f (t)} = F (s − a)

for Re(s) > k + Re(a).

Proof
Z ∞
at
L{e f (t)} = e−st eat f (t)dt
Z0 ∞
= e−(s−a)t f (t)dt
0
= F (s − a).

Since the effect of this operation is to replace s by s − a, the latter integral converges
for Re(s − a) > k, namely Re(s) > k + Re(a). 

7
Proposition 2.2 Suppose that f (t) is piecewise continuous and |f (t)| ≤ Kekt for
all t, so that L{f (t)} = F (s) converges for Re(s) > k. Then

dF
L{tf (t)} = −
ds
for Re(s) > k.

Sketch of proof:
Z ∞
F (s) = e−st f (t)dt
0
d ∞ −st
Z
dF
− = − e f (t)dt
ds ds 0
Z ∞
d −st

= − e f (t) dt
ds
Z ∞0
= e−st tf (t)dt
0
= L{tf (t)}.

The reason that this is a sketch of proof rather than an honest proof is the step
d
at which we switched the places of the ds operation and the integral sign. Such
changes in two limiting processes are certainly not automatic in mathematics, one
can easily construct examples where they fail. The condition needed is that the
function in question decays “quickly enough”, which turns out to work here because
of the bound on f (t). We refer the reader to books of advanced calculus for a more
careful discussion of such matters. 

Example 2.6 Let us revisit the example of f (t) = tn . Setting Fn (s) = L{tn } as
before, the proposition above implies,
dFn−1
Fn (s) = L{tn } = L{t · tn−1 } = − .
ds
Indeed, one can directly check the equality
 
n! d (n − 1)!
=− .
sn+1 ds sn

8
 √
Example 2.7 Let us find the Laplace transform of f (t) = e−2t sin 3t. Since
a
L{sin at} = 2 , we have
s + a2
√
√ 3
L{sin 3t} = 2 .
s +3
Now, using the first proposition above,
√
−2t
√ 3
L{e sin 3t} = .
(s + 2)2 + 3

Example 2.8 Let us find the Laplace transform of f (t) = t2 cos 2t. First of all,
s
L{cos 2t} = 2 .
s +4
By using the second proposition above,
s2 − 4
 
d s
L{t cos 2t} = − =
ds s2 + 4 (s2 + 4)2
Using the proposition once again,
s2 − 4 2s3 − 24s
 
2 d
L{t cos 2t} = − = 2 .
ds (s2 + 4)2 (s + 4)3

Our final result in this lecture concerns the Laplace transform of the function ob-
tained from f (t) by scaling the variable t.

Proposition 2.3 Suppose that L{f (t)} = F (s) for Re(s) > k. Say c > 0 is a
constant. Then
1 s
L{f (ct)} = F
c c
for Re(s) > ck.

Proof: Making the substitution τ = ct in the first integral below, we get

Z ∞
L{f (ct)} = e−st f (ct)dt
Z0 ∞
s 1
= e− c τ f (τ ) dτ
0 c
1  s 
= F .
c c
9
The integral on the second line converges for Re(s/c) > k, namely Re(s) > ck. 
This result tells us the following: If we scale the domain of f (t) so that the graph
is contracted (c > 1 case), the opposite happens to the graph of F (s) which is
expanded along its domain. Furthermore, the amplitude of F (s) is scaled by 1/c.
Taking c < 1 has the reverse effect.

10