You are on page 1of 12

1

The Laplace Transform

1.1 Introduction
As a discipline, mathematics encompasses a vast range of subjects. In pure
mathematics an important concept is the idea of an axiomatic system whereby
axioms are proposed and theorems are proved by invoking these axioms logi
caly. These activities are ofen of little interest to the applied mathematician to
whom the pure mathematics of algebraic structures will seem like tinkering with
axioms for hours in order to prove the obvious. To the engineer, this kind of pure
mathematics is even more of an anathema. The value of knowing about such
structures lies in the ability to generalise the "obvious" to other areas. These
generalisations are notoriously unpredictable and are ofen very surprising. In
deed, many say that there is no such thing as non-applicable mathematics, just
mathematics whose application has yet to be found.
The Laplace Tansform expresses the confict between pure and applied
mathematics splendidly. There is a temptation to begin a book such as this
on linear algebra outlining the theorems and properties of normed spaces. This
would indeed provide a sound basis for fture results. However most applied
mathematicians and all engineers would probably turn of. On the other had,
engineering texts present the Laplace Tansform as a toolkit of results with little
attention being paid to the underlying mathematical structure, regions of vaid
ity or restrictions. What has been decided here is to give a brief introduction to
the underlying pure mathematical structures, enough it is hoped for the pure
mathematician to appreciate what kind of creature the Laplace Tansform is,
whilst emphasising applications and giving plenty of exaples. The point of
view from which this book is written is therefore defnitely that of the applied
mathematician. However, pure mathematica aides, some of which can be quite
1
2 An I ntroduction to Laplace Transforms and Fourier Series
.{F(t)} = f(s)
t space
s space
Figure 1. 1: The Laplace Tansform a a mapping
extensive, will occur. It remains the view of this author that Laplace Tansforms
only come aive when they are used to solve real problems. Those who strongly
disagree with this will fnd pure mathematics textbooks on integral transforms
much more to their liking.
The main area of pure mathematics needed to understand the fndamental
properties of Laplace Tansforms is analysis and, to a lesser extent the normed
vector space. Anaysis, in particular integration, is needed from the start a
it governs the existence conditions for the Laplace Tansform itself; however
a is soon apparent, caculations involving Laplace Tansforms can take place
without explicit knowledge of analysis. Normed vector spaces and associated
linear algebra put the Laplace Tansform on a frm theoretical footing, but
can be left until a little later in a book aimed at second year undergraduate
mathematics students.
1. 2 The Laplace Tansform
The defnition of the Laplace Tansform could hardly be more straghtforward.
Given a suitable function F(t) the Laplace Tansform, written f(s) is defned
by
f(s) = 1
0
F(t)e
-s
t
dt.
This bald statement may satisfy most engineers, but not mathematicias. The
question of what constitutes a "suitable function" will now be addressed. The
integral on the right has infnite range and hence is what is called an improper
integral. This too needs careful handling. The notation .{F(t)} is used to
denote the Laplace Tansform of the function F(t) .
Another way of looking at the Laplace Tansform is as a mapping from points
in the t domain to points in the s domain. Pictorially, Figure 1. 1 indicates
this mapping process. The time doman t will contain all those functions F(t)
whose Laplace Tansform exists, whereas the frequency domain s contains al the
1 . The Laplace Transform 3
images . { F( t)}. Another apect of Laplace Tansforms that needs mentioning
at this stage is that the variable s often has to take complex values. This means
that f ( s) is a function of a complex variable, which in turn places restrictions on
the (real) function F(t) given that the improper integral must converge. Much
of the analysis involved in dealing with the image of the function F(t) in the s
plane is therefore complex analysis which may be quite new to some readers.
As ha been said earlier, engineers are quite happy to use Laplace Tans
forms to help solve a variety of problems without questioning the convergence of
the improper integrals. This goes for some applied mathematicians too. The ar
gument seems to be on the lines that if it gives what looks a reasonable answer,
then fne! In our view, this takes the engineer's maxim "if it ain't broke, don't fx
it" too far. This is primarily a mathematics textbook, therefore in this opening
chapter we shall be more mathematically explicit than is customary in books on
Laplace Tansforms. In Chapter 4 there is some more pure mathematics when
Fourier series are introduced. That is there for similar reaons. One mathe
matical question that ought to be aked concerns uniqueness. Given a function
F(t), its Laplace Tansform is surely unique from the well defned nature of the
improper integral. However, is it possible for two diferent functions to have the
same Laplace Tansform? To put the question a diferent but equivalent way,
is there a function N(t) , not identically zero, whose Laplace Tansform is zero?
For this function, called a null function, could be added to any suitable function
and the Laplace Tansform would remain unchanged. Null functions do exist,
but as long a we restrict ourselves to piecewise continuous functions this ceaes
to be a problem. Here is the defnition of piecewise continuous:
Defnition 1. 1 If a n i nterval [0, to ] say ca n be partitio ned i nto a fnite num
ber of subi ntervals [0, t i], [t
1
, t
2
], [t
2
, t
3
], . . . , [t
n
, to ] with 0, t
1
, t
2
, . . . , t
n
, to a n i n
creasi ng se
que nce of times a nd such that a give n fu nctio n f(t) is co nti nuous i n
each of these subi ntervals but not necessarily at the e nd poi nts themselves, the n
f(t) is piece wise co nti nuous i n the i nterval [0, to ].
Only functions that difer at a fnite number of points have the same Laplace
Tansform. I F
1
(t) = F(t) except at a fnite number of points where they difer
by fnite values then .{F
1
(t)} = .{F(t) }. We mention this again in the next
chapter when the inverse Laplace Tansform is defned.
In this section, we shall examine the conditions for the existence of the
Laplace Tansform in more detail than is usua. In engineering texts, the simple
defnition followed by an explanation of exponential order is all that is required.
Those that are satisfed with this can virtually skip the next few paragraphs and
go on study the elementary properties, Section 1.3. However, some may need
to know enough background in terms of the integrals, and so we devote a little
space to some fundamentals. We will need to introduce improper integras, but
let us frst defne the Riemann integral . It is the integral we know and love, and
is defned in terms of limits of sums. The strict defnition runs a follows:-
Let F(x) be a function which is defned and is bounded in the interval
a x b and suppose that m and M are respectively the lower and upper
4 An I ntroduction to Laplace Transforms and Fourier Series
bounds of F(x) in this interval (written
[
a, b] see Appendix C). Take a set of
points
Xo = a, x
1
, X
2
, , Xr
-
1
, Xr. , X
n
= b
and write
5r = Xr - Xr
-1
Let Mr, mr be the bounds of F(x) in the subinterval
(Xr
-
1
, Xr) and form the sums
n
s = Lmr5r.
r
= 1
These are called respectively the upper and lower Riemann sums corresponding
to the mode of subdivision. It is certainly clear that S ;:s. There are a variety
of ways that can be used to partition the interval (a, b) and each way will have
(in general) diferent Mr and mr leading to diferent S and s. Let M be the
minimum of all possible Mr and m be the maimum of all possible mr A lower
bound or supremum for the set S is therefore M(b - a) and an upper bound or
infmum for the set s is m( b - a)
.
These bounds are of course rough. There are
exact bounds for S and s, call them J and I respectively. I I = J, F(x) is said
to be Riemann integrable in (a, b) and the value of the integral is I or J and is
denoted by
I = J = 1
b
F(x)dx.
For the purist it turns out that the Riemann integral is not quite genera
enough, and the Stieltjes integral is actually required. However, we will not use
this concept which belongs securely in specialist fnal stage or graduate texts.
The improper integral is defned in the obvious way by taking the limit:
lim
1
R
F(x)dx =
{
o
F(x)dx
R-o a
Jo
provided F(x) is continuous in the interval a x R for every R, and the limit
on the lef exists.
This is enough of general theory, we now apply it to the Laplace Tansform.
The parameter x is defned to take the increasing values from 0 to o. The con
dition j F(x) j Meo
x
is termed " F(x) is of exponentia order" and is, speaking
loosely, quite a weak condition. All polynomial functions and (of course) expo
nential functions of the type e
k
x
(k
constant) are included a well a bounded
functions. Excluded functions are those that have singularities such a ln(x) or
1/(x - 1) and functions that have a growth rate more rapid than exponential,
for example e
x
2

Functions that have a fnite number of fnite discontinuities
are also included. These have a special role in the theory of Laplace Tansforms
(see Chapter 3) so we will not dwell on them here: sufce to say that a function
such as
{
1 2n < x < 2n + 1
F(x) =
0 2n + 1 < x < 2n + 2 where n = 0, 1, . . .
1 . The Laplace Transform
is one example. However, the function
F(x)
=
{

x rational
x irrational
5
is excluded because although all the discontinuities are fnite, there ae infnitely
many of them.
We shall now follow standard practice and use t (time) instead of x a the
dummy variable.
1. 3 Elementary Properties
The Laplace Tansform ha many interesting and useful properties, the most
fundamental of which is linearity. It is linearity that enables us to add results
together to deduce other more complicated ones ad is so baic that we state it
a a theorem and prove it frst.
Theorem 1.2 (Linearity) If Ft (t) and F
2
(t) are two fnctions whose Laplace
Tansform exists, then
where a and b are arbitrry constants.
Proof
.{ aFt (t) + bF
2
(t)} = 1
0
(aFt + bF
2
) e
-st
dt
= 1
0
( aFt e
-st
+ bF
2
e
-
st
)
dt
=
a 1
0
Ft e
-
st
dt + b 1
0
F
2
e
-
st
dt
= a.{F1 (t)} + b.{F
2
(t)}
where we have assumed that
so that
j aFt + bF
2
! :
j ai
!
Ft
l
+
l
b
l 1
F
2
l
:
(j a
!
Mt +
l
b
i
M
2
)e
as
t
where 0
3
= max{ Ot, 0
2
}. This proves the theorem.
0
In this section, we shall concentrate on those properties of the Laplace Tansform
that do not involve the calculus. The frst of these takes the form of another
theorem because of its generality.
6 An Introduction to Laplace Transforms and Fourier Series
Theorem 1.3 (First Shift Theorem) If it is possible to choose constants M
and a such that
I
F(t)
l
M
e
o
t
, that is F(t) is of exponential order, then
.{
e
-
b
t
F(t)}
=
f(s + b)
provided b a. (In practice if F(t) is of exponential order then the constant a
can be chosen such that this inequality holds.}
Proof The proof is straightforward and runs a follows:-
.{
e
-
b
t
F(t) } = lim
{
T
e-s
t
e-
b
t
F(t)dt
T-o J
o
= 1
0
e-s
t
e-
b
t
F(t)dt
(
a the limit
exists
)
1
0
e-
(
s
+b
)t
F(t)dt
f(s + b) .
This establishes the theorem.
0
We shall make considerable use of this once we have established a few elementary
Laplace Tansforms. This we shal now proceed to do.
Example 1.4 Find the Laplace Tansform of the function F(t)
= t.
Solution Using the defnition of Laplace Tansform,
.(t) = lim
{
T
t
e-s
t
dt.
T-o J
o
Now, we have that
this lat expression tends to
1
2
as T o.
s
Hence we have the result
1
.(t)
=
2
s
1. The Laplace Transform
We can generalise this result to deduce the following result:
Corollary
"
(
n
)
n!
. . .
}. t =
s
n
+
1
, n a positive mteger.
Proof The proof is straghtforward:
.(t
n
) = 1
0
t
n
e
-s
t
dt this time taking the limit straight away
[
t
n ]
0
1
o
t
n
-
1
-
-
e
-st
+
_
n
__
e
-s
t
dt
s
0
0
s
=

.(t
n
-
1
) .
s
I we put n = 2 in this recurrence relation we obtain
I we asume
then
This establishes that
by induction.
"
(
n
+
1
)
_ n
+
1 _ (n
+
1)!
}. t
- -
.
s s
n
+
1
s
n
+
2
7
0
Example 1. 5 Find the Laplace Tansform of .{te
at
} and deduce the value of
.{t
n
e
at
}, where a is a real constant and n a positive integer.
Solution Using the frst shift theorem with b =
-
a gives
so with
we get
Using F(t) = t
n
the formula
.{F(t)e
at
}
= f(s
-
a)
1
F(t) = t and f = 2
s
8 An I ntroduction to Laplace Transforms and Fourier Series
follows.
Later, we shall generalise this formula further, extending to the cae where
n is not an integer.
We move on to consider the Laplace Tansform of trigonometric functions.
Specifcally, we shall calculate .{ sin(t) } and .{ cos(t) }. It is unfortunate, but
the Laplace Tansform of the other common trigonometric functions tan, cot,
esc and sec do not exist a they all have singularities for fnite t. The condition
that the fnction F(t) ha to be of exponential order is not obeyed by any of
these singular trigonometric functions a can be seen, for example, by noting
that
je-
a
t
tan(t) l - o a t --n/2
and
je-
a
t
cot(t) l
-
+ o as t --0
for all values of the constant a. Similarly neither esc nor sec are of exponential
order.
In order to fnd the Laplace Tansform of sin(t) and cos(t) it is best to
determine .(e
it
) where i = J. The function e
it
is complex valued, but
it is both continuous and bounded for all t so its Laplace Tansform certainly
exists. Taking the Laplace Tansform,
Now,
.(
e
it
)
=
f
o e-
st
e
it
dt
=
f
o e
t ( i
-
s
)
dt
[
e
i
-
s
)
t ]
0
z - s 0
1
s - i
s
.
1
-+z-
s
2
+ 1 s
2
+
1

.(
e
it
) .(cos(t) + i sin(t) )
=
.(cos(t) ) + i.(sin(t)) .
Equating real and imaginary parts gives the two results
and
s
.(cos(t))
=

1
s +
.(sin(t))
=

1
.
s +
The linearity property ha been used here, and will be used in fture without
further comment.
1. The Laplace Transform 9
Given that the restriction on the type of function one can Laplace Tansform
is weak, i.e. it has to be of exponential order and have at most a fnite number
of fnite jumps, one can fnd the Laplace Tansform of ay polynomia, any
combination of polynomial with sinusoidal functions and combinations of these
with exponentials (provided the exponential functions grow at a rate
,
.
where a is a constant) . We can therefore approach the problem of calculating the
Laplace Tansform of power series. It is possible to take the Laplace Tansform
of a power series term by term a long a the series uniformly converges to a
piecewise continuous function. We shall investigate this further later in the text;
meanwhile let us look at the Laplace Tansform of functions that are not even
continuous.
Fnctions that are not continuous occur naturally in branches of electrical
and control engineering, and in the software industry. One only ha to think
of switches to realise how widespread discontinuous functions are throughout
electronics and computing.
Exaple 1. 6 Find the Laplace Tansform of the fnction represented by F(t)
where
F(t) = 2to - t to t 2to
{ t O < t < to
0 t > 2to
.
Solution This function is of the "saw-tooth" variety that is quite common in
electrical engineering. There is no question that it is of exponential order and
that
1
0
,

F(t)dt
exists and is well defned. F(t) is continuous but not diferentiable. This is not
troublesome. Carrying out the calculation is a little messy and the details can
be checked using computer algebra .
.(F(t)) =
1
0
,

F(t)dt
=

,
+

.
(2to t)
,

,
t
0
.
=

.
+
.
;
,

dt
+

2to
,
.

.
;
,

,
S
0 o
s

.
.

to
.

.
+
to

.
+

.
=
-
e -
-
e -e - e
s .
o
. .
.
=

,
.
_

, +
.

,

.
_
,

.
=
s

,,
. +
,

.
,
=
.

,
.
,

10 An I ntroduction to laplace Transforms and Fourier Series
I the next chapter we shall investigate in more detail the properties of discon
tinuous fnctions such a the Heaviside unit step function. As an introduction
to this, let us do the following example.
Example 1.7 Deterine the Laplace Tansfor of the step fnction F(t) de
fned by
F(t) =
{
0 0 t < to
a t?to.
Solution F(t) itself is bounded, so there is no question that it is also of expo
nential order. The Laplace Tansform of F(t) is therefore
C(F(t) ) =
=
e
-
s
t
F(t)dt
=
=
=
Here is another useful result.
lo ae-
s
t
dt
to
[
-

e
-st
]
o
S
to
a
-
s
t
o
-
e
s
d
Theorem 1. 8 If C(F(t)) = f (s) then C(tF(t)) =-
ds
f (s)
and in generl C(t
n
F(t)) = ( -
l
)
n
d

f (s).
s
n
C(F(t)) =
=
e
-
st
F(t)dt
and diferentiate this with respect to s to give
! =
=
e-st
F(t)dt
=
=
-
te-s
t
F(t)dt
asuming absolute convergence to justify interchanging diferentiation and (im
proper) integration. Hence
d
C(tF(t)) =-
ds
f (s) .
1. The Laplace Transform 11
One can now see how to progress by induction. Assume the result holds for n,
so that
C(t
n
F(t))
=
( -
l)
n
d

f(s)
s
n
and diferentiate both sides with respect to s (asuming all appropriate conver
gence properties) to give
or
So

+
1
C(t
n
+
1
F(t)) = (-1 )
n
+
1
ds
n
+
1
f(s)
which establishes the result by induction.
Example 1.9 Determine the Laplace Tansform of the fnction t sin(t) .
0
Solution To evaluate this Laplace Tansform we use Theorem 1. 8 with f(t)
=
sin(t) . This gives
. d
{
1
}
2s
C{t s m(t) }
=
-
ds 1
+
s
2
=
( 1
+
s
2
)
2
which is the required result.
1.4 Exercises
1. For each of the following functions, determine which ha a Laplace Tans
form. I it exists, fnd it; if it does not, say briefy why.
(a) ln(t) , (b) e
3t
, (c) et
2
, (d) e
1
1t
, (e) 1/t,
(f) f(t) =
{
1 f t s even
0 1f t 1s odd.
2. Determine from frst principles the Laplace Tansform of the follow
i
ng
functions:-
(a) e
k
t
, (b) t
2
, (c) cosh(t) .
3. Find the Laplace Tansforms of the following functions:-
12 An Introduction to laplace Transforms and Fourier Series
4. Find the Laplace Tansform of the function F(t) , where F(t) is given by
{ t O < t < l
F(t) = 2 - t 1 t < 2
0 otherwise.
5. Use the property of Theorem 1.8 to determine the following Laplace Tans
forms
(a) te
2
t
, (b) t cos(t), (c) t
2
cos(t) .
6. Find the Laplace Tansforms of the following functions:-
(a) sin(wt + ) , (b) e
5
t
cosh(6t) .
7. If G(at + b) = F(t) determine the Laplace Tansform of G in terms of
C{F} = /(s) and a fnite integral.
8. Prove the following change of scale result:-
C{F(at)} = ().
Hence evluate the Laplace Tansforms of the two functions
(a) t cos(6t), (b) t
2
cos(7t) .