Digital Signal Processing Reference
In-Depth Information
20.3.1 Definitions
| 2 = zz this is
not analytic in the variable z (the presence of z makes it non-analytic [Churchill
and Brown, 1984]). So dh ( z ) /dz does not exist. However, imagine for a moment
that z and z denote two separate, independent, variables and define the two-
variable function
| 2 . Since
To motivate the idea consider the example h ( z )=
|
z
|
z
g ( z, z )= zz .
We can differentiate with respect to z by holding z constant, and vice versa.
That is, we can define the partial derivatives
∂g ( z, z )
∂z
∂g ( z, z )
∂z
= z
and
= z.
A subtle point about notation. We know that z and z are not independent;
knowing one we can find the other. However, given the original function h ( z ), if
we replace all occurrence of z with the notation z 1 and all occurrence of z with
z 2 , then the two-variable function g ( z 1 ,z 2 ) can certainly be differentiated with
respect to each variable separately. Instead of using the new notations z 1 and
z 2 we simply carry on with z and z . The fact that any h ( z ) can be expressed
uniquely as g ( z, z ) is explained in a footnote in Sec. 20.3.2.
The usefulness of this “brave” definition will become clear as soon as some
of its properties are established. For now, just consider another example: let
h ( z )=2Re[ z ]. This is not analytic because h ( z )= z + z .So dh ( z ) /dz does not
exist. Define the two-variable function
g ( z, z )= z + z .
Then
∂g ( z, z )
∂z
∂g ( z, z )
∂z
= 1
and
=1 .
Finally consider the simple example h ( z )= z. In this case g ( z, z )= z so that
∂g/∂z = 1 and ∂g/∂z =0 . Thus
∂z
∂z = 1
∂z
∂z
and
=0 .
(20 . 18)
Similarly,
∂z
∂z
∂z
∂z
= 0
and
=1 .
(20 . 19)
More generally,
if g ( z, z ) is free from z then ∂g/∂z =0 ,
(20 . 20)
if g ( z, z ) is free from z then ∂g/∂z =0 . (20 . 21)
In complex variable theory the derivative ∂z /∂z is undefined because z is not
analytic. Observe however that the meaning of the notation here is different.
 
Search WWH ::




Custom Search