0
$\begingroup$

Given a $\mathcal{Z}$ transformed function $E(z)=\frac{1}{z+4}$.

I know there are several ways to get the inverse $\mathcal{Z}$ transform of this function :

Using partial fraction

$$E(z)=\frac{1}{z+4}$$

$$\frac{E(z)}{z}=\frac{1}{z(z+4)}$$

apply partial fraction here,

$$\frac{E(z)}{z}= \tfrac14 \left(\frac{1}{z} - \frac{1}{z+4} \right)$$

so $E(z)$ is,

$$E(z)= \tfrac14 \left(1 - \frac{z}{z+4} \right)$$

as you know, it is easy to use inverse Z-transform here.

$$ e[n] = \tfrac14 ( \delta[n] - (-4)^n ) $$

inversion formula method

$E(z)$ has simple pole at $z=-4$, the residue is evaluated as

$$\begin{split} \text{residue}_{z=-4} &= ((z+4)\times E(z)\times z^{k-1})_{z=-4}\\ &= (1\times z^{k-1})_{z=-4}\\ &= (-4)^{k-1}\\ &= -(1/4)\times (-4)^k \end{split}$$


The problem is that inverse $\mathcal{Z}$ transform of the same function $E(z)$has 2 different answers. As you see, there is no delta function when I used inversion formula method.

At first, I thought it was because a simple pole, $z=-4$, is outside of the unit circle, but same things happen when I used a simple pole which is inside of the unit circle.

Why does this difference happen? What is the real answer?

$\endgroup$
  • 1
    $\begingroup$ can you learn $\LaTeX$ and do the math with that? $\endgroup$ – robert bristow-johnson Sep 26 '17 at 4:36
  • $\begingroup$ You cannot invert the z-transform without knowing the region of convergence. The function on its own without the ROC lacks important information. $\endgroup$ – Jazzmaniac Dec 13 '17 at 12:36
1
$\begingroup$

Here's a much simpler method. We have

$$E (z) := \frac{1}{z+4} = z^{-1} \left(\frac{1}{1 - (-4) \, z^{-1}}\right)$$

which we assume to be the $\mathcal Z$ transform of a causal discrete-time signal. Consulting a table of $\mathcal Z$ transforms, we conclude that

$$\mathcal Z^{-1} \left\{ \frac{1}{1 - (-4) \, z^{-1}} \right\} = (-4)^n \, u (n)$$

where $u$ is the discrete-time step function. Hence, the inverse $\mathcal Z$ transform of $E$ is simply the discrete-time signal above after a unit delay, i.e.,

$$e (n) = (-4)^{n-1} \, u (n-1)$$

which is zero at $n=0$. Hence, your 1st candidate answer is (almost) correct. Your 2nd candidate answer is only correct if $n \geq 1$. When working with functions, do not forget what their domains are.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.