Sie sind auf Seite 1von 4

Sec. 1.

Functionals 1

Continuous-Time Unit Impulse


Ilya Pollak
ECE 301 Signals and Systems Section 2, Fall 2010
Purdue University

 1 Functionals
As we have seen before, a signal operates on numbers and produces numbers. In
other words, any signal x is a mapping from a set of numbers (called the domain
of x) to another set of numbers (called the range of x), as shown in Fig. 1(a). For
example, a continuous-time complex-valued signal x is a mapping from the set of
real numbers to the set of complex numbers: given any real number t, the value
x(t) is a complex number—see Fig. 2(a).

 T  T  T  T
T
Domain of x, Range of x,T Domain of v,T Range of v,T
 D, a set of T signal x  R, a set of T  D, a set of T functional v R, a set of T
T numbers  T numbers  T signals  T numbers 
T  T  T  T 
T  T  T  T 
(a) (b)

Figure 1. (a) A signal is a mapping between two sets of numbers. (b) A functional is a
mapping between a set of signals and a set of numbers.

A functional operates on signals and produces numbers. In other words, any


functional v is a mapping from a set of signals (called the domain of v) to a set
of numbers (called the range of v), as illustrated in Fig. 1(b). We will consider
here complex-valued functionals that operate on continuous-time complex-valued
signals. For any such input signal x, a functional v produces a complex number
which we will denote v(x)—see Fig. 2(b).

input t, output x(t), input x, output v(x),


a number a number a signal a number
Signal x Functional v

(a) (b)

Figure 2. (a) A generic block diagram for a signal. (b) A generic block diagram for a functional.
2

Example. Here are some examples of functionals.


Z ∞
v1 (x) = x(t) cos(t)dt
−∞
Z ∞ Z ∞
v2 (x) = x(t)u(t)dt = x(t)dt
−∞ 0
v3 (x) = x(0)
v4 (x) = x(15)
Strictly speaking, in order to fully specify a functional, we must specify its domain.
Note that the domain of the functional v1 must be a subset of the set of all signals
x such that x(t) cos(t) is integrable. The domain of v2 can only contain signals
which are integrable on the interval [0, ∞). The domains of v3 and v4 can only
contain signals which are continuous at 0 and at 15, respectively.

 2 Functionals Induced by Functions


The functional vs induced by a function s is defined as follows:
Z ∞
vs (x) = x(t)s(t)dt (1)
−∞

The domain of such a functional can only contain signals x for which xs is inte-
grable. For example, the functional v2 defined above is induced by the unit step.
As another example, consider a short pulse δ∆ of duration ∆, illustrated in Fig. 3:

 0, t≤0
δ∆ (t) = 1/∆, 0 < t ≤ ∆ (2)
0, t>∆

This short pulse induces the following functional:


1 ∆
Z ∞ Z
δ∆ (t)x(t)dt = x(t)dt (3)
−∞ ∆ 0

 3 The Definition of the CT Unit Impulse


There are some functionals v that are not induced by any function—i.e., that
cannot be written in the form (1) for any function s. The continuous-time unit
impulse is such a functional.
The continuous-time unit impulse δ is a functional that for any signal x which
is continuous at zero, produces x(0)—in other words, it is the functional v3 defined
above.
Sec. 4. The CT Unit Impulse as the Limit of Short Rectangular Pulses 3

δ∆ (t) C

1

X

∆ t

Figure 3. Rectangular pulse of unit area and duration ∆.

 4 The CT Unit Impulse as the Limit of Short Rectangular Pulses


The textbook says that the CT unit impulse δ is the limiting case of the short
pulse δ∆ of unit area defined in Eq. (2), as the width of the pulse goes to zero:

lim δ∆ = δ. (4)
∆→0

This is correct; however, the limiting operation here is not the usual pointwise
limit.
As a matter of fact, the pointwise limit of δ∆ (t) for any t is zero. To see this,
consider the cases t ≤ 0 and t > 0 separately. For t ≤ 0, we have from Eq. (2) that
δ∆ (t) = 0 regardless of ∆. Therefore, for any t ≤ 0, the limit δ∆ (t) as ∆ → 0 is
zero. For t ≥ 0, as soon as ∆ is smaller than t, we also have δ∆ (t) = 0, according
to Eq. (2). Thus, no matter how small t is, as long as t is positive, the limiting
value of δ∆ (t) is zero as ∆ → 0. For all t we therefore have:

lim δ∆ (t) = 0.
∆→0

This shows that Eq. (4) cannot be interpreted pointwise. The correct interpre-
tation of Eq. (4) is in the sense of functionals: as ∆ → 0, what the functional
induced by δ∆ does to any signal approaches what δ does to the same signal. To
see this, we look at the output value of the functional induced by δ∆ for some
4

arbitrary signal x (see Eq. (3)), and take the limit as ∆ → 0:


1 ∆ F (∆) − F (0)
Z
lim x(t)dt = lim , (5)
∆→0 ∆ 0 ∆→0 ∆
where F is the antiderivative of x, i.e., F ′ (t) = x(t). The right-hand side of
Eq. (5) is the definition of F ′ (0), which is equal to x(0):
1 ∆ F (∆) − F (0)
Z
lim x(t)dt = lim
∆→0 ∆ 0 ∆→0 ∆

= F (0)
= x(0),
which is the same as the output value of the CT unit impulse δ when x is the
input. Thus, in the limit ∆ → 0, the functional induced by δ∆ produces the value
of a signal at zero—the same as δ. This is the sense in which Eq. (4) holds.

 5 Notation
It is possible to demonstrate that there does not exist a function s such that
Z ∞
s(t)x(t)dt = x(0)
−∞

for every signal x which is continuous at zero. Therefore, the CT unit impulse is
a functional that is not induced by any function. Despite this, the notation δ(t)
is used in much engineering literature, including the textbook. We will therefore
have to use it throughout the course. Another piece of really bad, confusing
notation is: Z ∞
δ(t)x(t)dt
−∞
This notation simply means x(0), nothing more! Unfortunately, this notation is
also used throughout the text, and we will therefore have to use it class.
If t0 is a fixed real number, the unit impulse shifted by t0 is denoted δt0 and
is defined as the following functional:
v(x) = x(t0 ).
Two more pieces of bad but widely used notation:
δt0 (t) = δ(t − t0 )
Z ∞
δ(t − t0 )x(t)dt = x(t0 )
−∞

Das könnte Ihnen auch gefallen