6.5 Impulse Functions 343
6.5 Impulse Functions 343
6.5 Impulse Functions 343
(c) Compare the results of part (b) with those from Problem 20 and from Section 3.8 for
a sinusoidally forced oscillator.
23. Consider the initial value problem
where
n
!
f (t) = u0 (t) + 2 (−1)k u11k/4 (t).
k=1
Observe that this problem is identical to Problem 19 except that the frequency of the
forcing term has been increased somewhat.
(a) Find the solution of this initial value problem.
(b) Let n ≥ 33 and plot the solution for 0 ≤ t ≤ 90 or longer. Your plot should show a
clearly recognizable beat.
(c) From the graph in part (b), estimate the “slow period” and the “fast period” for this
oscillator.
(d) For a sinusoidally forced oscillator, it was shown in Section 3.8 that the “slow fre-
quency” is given by |ω − ω0 |/2, where ω0 is the natural frequency of the system and ω is
the forcing frequency. Similarly, the “fast frequency” is (ω + ω0 )/2. Use these expressions
to calculate the “fast period” and the “slow period” for the oscillator in this problem. How
well do the results compare with your estimates from part (c)?
1
2τ
τ τ t
FIGURE 6.5.1 Graph of y = dτ (t).
t
FIGURE 6.5.2 Graphs of y = dτ (t) as τ → 0+ .
Equations (5) and (6) can be used to define an idealized unit impulse function δ,
which imparts an impulse of magnitude one at t = 0 but is zero for all values of t
other than zero. That is, the “function” δ is defined to have the properties
δ(t) = 0, t ̸= 0; (7)
" ∞
δ(t) dt = 1. (8)
−∞
August 7, 2012 21:04 c06 Sheet number 37 Page number 345 cyan black
There is no ordinary function of the kind studied in elementary calculus that satisfies
both Eqs. (7) and (8). The “function” δ, defined by those equations, is an example of
what are known as generalized functions; it is usually called the Dirac3 delta function.
Since δ(t) corresponds to a unit impulse at t = 0, a unit impulse at an arbitrary point
t = t0 is given by δ(t − t0 ). From Eqs. (7) and (8), it follows that
δ(t − t0 ) = 0, t ̸ = t0 ; (9)
" ∞
δ(t − t0 ) dt = 1. (10)
−∞
The delta function does not satisfy the conditions of Theorem 6.1.2, but its Laplace
transform can nevertheless be formally defined. Since δ(t) is defined as the limit
of dτ (t) as τ → 0+ , it is natural to define the Laplace transform of δ as a similar
limit of the transform of dτ . In particular, we will assume that t0 > 0 and will define
L{δ(t − t0 )} by the equation
To evaluate the limit in Eq. (11), we first observe that if τ < t0 , which must eventually
be the case as τ → 0+ , then t0 − τ > 0. Since dτ (t − t0 ) is nonzero only in the interval
from t0 − τ to t0 + τ, we have
" ∞
L{dτ (t − t0 )} = e−st dτ (t − t0 ) dt
0
" t0 +τ
= e−st dτ (t − t0 ) dt.
t0 −τ
1 −st0 sτ
= e (e − e−sτ )
2sτ
or
sinh sτ −st0
L{dτ (t − t0 )} = e . (12)
sτ
3 Paul A. M. Dirac (1902–1984), English mathematical physicist, received his Ph.D. from Cambridge in 1926
and was professor of mathematics there until 1969. He was awarded the Nobel Prize for Physics in 1933
(with Erwin Schrödinger) for fundamental work in quantum mechanics. His most celebrated result was the
relativistic equation for the electron, published in 1928. From this equation he predicted the existence of
an “anti-electron,” or positron, which was first observed in 1932. Following his retirement from Cambridge,
Dirac moved to the United States and held a research professorship at Florida State University.
August 7, 2012 21:04 c06 Sheet number 38 Page number 346 cyan black
The quotient (sinh sτ)/sτ is indeterminate as τ → 0+ , but its limit can be evaluated
by L’Hôpital’s4 rule. We obtain
sinh sτ s cosh sτ
lim+ = lim+ = 1.
τ→0 sτ τ→0 s
Then from Eq. (11) it follows that
L{δ(t − t0 )} = e−st0 . (13)
Equation (13) defines L{δ(t − t0 )} for any t0 > 0. We extend this result, to allow t0 to
be zero, by letting t0 → 0+ on the right side of Eq. (13); thus
L{δ(t)} = lim e−st0 = 1. (14)
t0 →0+
In a similar way, it is possible to define the integral of the product of the delta
function and any continuous function f . We have
" ∞ " ∞
δ(t − t0 )f (t) dt = lim+ dτ (t − t0 )f (t) dt. (15)
−∞ τ→0 −∞
Using the definition (4) of dτ (t) and the mean value theorem for integrals, we find
that
" ∞ " t0 +τ
1
dτ (t − t0 )f (t) dt = f (t) dt
−∞ 2τ t0 −τ
1
= · 2τ · f (t ∗ ) = f (t ∗ ),
2τ
where t0 − τ < t ∗ < t0 + τ. Hence t ∗ → t0 as τ → 0+ , and it follows from Eq. (15) that
" ∞
δ(t − t0 )f (t) dt = f (t0 ). (16)
−∞
The following example illustrates the use of the delta function in solving an initial
value problem with an impulsive forcing function.
4 Marquis Guillaume de L’Hôpital (1661–1704) was a French nobleman with deep interest in mathematics.
For a time he employed Johann Bernoulli as his private tutor in calculus. L’Hôpital published the first
textbook on differential calculus in 1696; in it appears the differentiation rule that is named for him.
August 7, 2012 21:04 c06 Sheet number 39 Page number 347 cyan black
Thus
e−5s e−5s 1
Y(s) = = % & . (19)
2s2 +s+2 2 s+ 2+
1 15
4 16
0.3
0.2
0.1
5 10 15 20 t
–0.1
In dealing with problems that involve impulsive forcing, the use of the delta
function usually simplifies the mathematical calculations, often quite significantly.
August 7, 2012 21:04 c06 Sheet number 40 Page number 348 cyan black
However, if the actual excitation extends over a short, but nonzero, time interval,
then an error will be introduced by modeling the excitation as taking place instan-
taneously. This error may be negligible, but in a practical problem it should not be
dismissed without consideration. In Problem 16 you are asked to investigate this
issue for a simple harmonic oscillator.
13. Consider again the system in Example 1 of this section, in which an oscillation is excited
by a unit impulse at t = 5. Suppose that it is desired to bring the system to rest again after
exactly one cycle—that is, when the response first returns to equilibrium moving in the
positive direction.
(a) Determine the impulse kδ(t − t0 ) that should be applied to the system in order to
accomplish this objective. Note that k is the magnitude of the impulse and t0 is the time
of its application.
(b) Solve the resulting initial value problem, and plot its solution to confirm that it
behaves in the specified manner.
14. Consider the initial value problem
20
- 20
-
19. f (t) = δ(t − kπ/2) 20. f (t) = (−1)k+1 δ(t − kπ/2)
k=1 k=1
15
- 40
-
21. f (t) = δ[t − (2k − 1)π] 22. f (t) = (−1)k+1 δ(t − 11k/4)
k=1 k=1
23. The position of a certain lightly damped oscillator satisfies the initial value problem
20
!
y′′ + 0.1y′ + y = (−1)k+1 δ(t − kπ), y(0) = 0, y′ (0) = 0.
k=1
Observe that, except for the damping term, this problem is the same as Problem 18.
(a) Try to predict the nature of the solution without solving the problem.
(b) Test your prediction by finding the solution and drawing its graph.
(c) Determine what happens after the sequence of impulses ends.
24. Proceed as in Problem 23 for the oscillator satisfying
15
!
y′′ + 0.1y′ + y = δ[t − (2k − 1)π], y(0) = 0, y′ (0) = 0.
k=1
Observe that, except for the damping term, this problem is the same as Problem 21.
August 7, 2012 21:04 c06 Sheet number 42 Page number 350 cyan black
25. (a) By the method of variation of parameters, show that the solution of the initial value
problem
y′′ + 2y′ + 2y = f (t); y(0) = 0, y′ (0) = 0
is
" t
y= e−(t−τ) f (τ) sin(t − τ) dτ.
0
(b) Show that if f (t) = δ(t − π), then the solution of part (a) reduces to
(c) Use a Laplace transform to solve the given initial value problem with
f (t) = δ(t − π), and confirm that the solution agrees with the result of part (b).
Theorem 6.6.1 If F(s) = L{f (t)} and G(s) = L{g(t)} both exist for s > a ≥ 0, then
H(s) = F(s)G(s) = L{h(t)}, s > a, (1)
where " "
t t
h(t) = f (t − τ)g(τ) dτ = f (τ)g(t − τ) dτ. (2)
0 0
The function h is known as the convolution of f and g; the integrals in Eq. (2) are
called convolution integrals.
The equality of the two integrals in Eq. (2) follows by making the change of vari-
able t − τ = ξ in the first integral. Before giving the proof of this theorem, let us
make some observations about the convolution integral. According to this theorem,
the transform of the convolution of two functions, rather than the transform of their
ordinary product, is given by the product of the separate transforms. It is conven-
tional to emphasize that the convolution integral can be thought of as a “generalized
product” by writing
h(t) = (f ∗ g)(t). (3)