(11 intermediate revisions by 4 users not shown)
Line 1: Line 1:
= Question about computing the inverse z-transform=
+
[[Category:problem solving]]
 +
[[Category:z-transform]]
 +
<center><font size= 4>
 +
'''[[Digital_signal_processing_practice_problems_list|Practice Question on "Digital Signal Processing"]]'''
 +
</font size>
  
I'm not sure if I've done this correctly in the first place. I would like your input before I actually attempt to inverse it. 0
+
Topic: Computing an inverse z-transform
'''I tried the LaTeX but it failed miserably. ''' . Maybe it would be easier if you write it out while you are reading my LaTeX fail trying to help me, otherwise I can come in and see you on Monday.
+
 
:I fixed it. Now it works. For future references, [http://meta.wikimedia.org/wiki/Help:Displaying_a_formula here] is a good page to bookmark: it explains how to use latex in mediawiki, and gives many examples that you can cut and paste. -pm
+
</center>
 
----
 
----
== Computation of inverse z-transform by a student (with question about how to obtain ROC)=
+
==Question from a student==
Take <math>x[n] = a^n(u[n-2]+u[n]) </math>. We then have
+
Take <math>x[n] = a^n\left( u[n-2]+u[n]\right) </math>. We then have
 +
 
 
<math>
 
<math>
 
\begin{align}
 
\begin{align}
Line 26: Line 31:
 
:~ksoong
 
:~ksoong
 
----
 
----
 +
 
==Comments/corrections from [[user:mboutin|Prof. Mimi]]==
 
==Comments/corrections from [[user:mboutin|Prof. Mimi]]==
 
Take <math>x[n] = a^n(u[n-2]+u[n]) </math>. We then have
 
Take <math>x[n] = a^n(u[n-2]+u[n]) </math>. We then have
Line 33: Line 39:
 
&= \sum_{n=-\infty}^\infty a^n(u[n-2]+u[n])z^{-n},  {\color{OliveGreen}\surd}  \\
 
&= \sum_{n=-\infty}^\infty a^n(u[n-2]+u[n])z^{-n},  {\color{OliveGreen}\surd}  \\
 
&= \sum_{n=2}^\infty a^n(z^{-n}) + \sum_{n=0}^\infty  a^n(z^{-n}). {\color{OliveGreen}\surd} \\
 
&= \sum_{n=2}^\infty a^n(z^{-n}) + \sum_{n=0}^\infty  a^n(z^{-n}). {\color{OliveGreen}\surd} \\
\text{Now let }k=-n, \\
+
\text{Now let }k=-n,& {\color{red}\text{This change of variable is not useful, unfortunately.}} \\
\Rightarrow X(z) &= \sum_{k=-2}^\infty (a/z)^{\color{red}n} + \sum_{k=0}^\infty  (a/z)^{\color{red}n} ,{\color{red}\text{Oops! The terms inside the summation contain n, but the summation is over k.}} \\
+
\Rightarrow X(z) &= \sum_{k=-2}^{\color{red}-\infty} (a/z)^{\color{red}n} + \sum_{k=0}^\infty  (a/z)^{\color{red}n} ,{\color{red}\text{The terms inside the summation contain n, but the summation is over k.}} \\
&=\sum_{k=0}^\infty   \left( (a/z)^n + 2)\right) + \sum_{k=0}^\infty \left( \frac{a}{z}\right)^n \\
+
&=\sum_{k=0}^{\color{red}-\infty\left( (a/z)^n {\color{red} -(a/z)^{-2}-(a/z)^{-1}} )\right) + \sum_{k=0}^\infty \left( \frac{a}{z}\right)^n \\
& = \left(\frac{1}{1-a/z}+2\right) + \left(\frac{1}{1-a/z}\right), \\
+
& = \left(\frac{1}{1-a/z}+{\color{red} -(a/z)^{-2}-(a/z)^{-1}} )\right) + \left(\frac{1}{1-a/z}\right), {\color{red}\text{For this last step, you need to assume } \left| \frac{a}{z}\right|<1, \text{ else both sums diverge.}}  
& =  \frac{z}{z-a}+2 + \frac{z}{z-a}\\
+
& = \frac{z}{z-a}+2\frac{z-a}{z-a} + \frac{z}{z-a} , \\
+
& = \frac{4z-2a}{z-a}\\
+
& = \frac{4-2a/z}{1-a/z}, \text{ for } |z| < a ???
+
 
\end{align}
 
\end{align}
 
</math>
 
</math>
  
To answer your initial question ("if I end up with something that says 1/1-(1/z), I am confused. does it converge when |z|>a or when |z|<a?"),   
+
<span style="color:green"> The answer to your initial question ("if I end up with something that says 1/1-(1/z), I am confused. does it converge when |z|>a or when |z|<a?") is in the last step. As you can see from this step, X(z) only converges if
 +
|a|<|z|. Note that, since a could be a complex number, it is important not to say a< |z|</span>
 +
----
 +
 
 +
A good way to check your answer is to use the Z-transform table. You can use the time shift property on the first term (a^n *u[n-2]) and the second term (a^n *u[n]) can be directly converted using the table. Your final answer should match up with what you ended with above. -[[User:sbiddand|Sbiddand]]
  
 
----
 
----
  
Answer from Prof. MImi
+
Anybody see anything else? Do you have more questions? Comments? Please feel free to add below.
 
+
*In the step where you replaced -n by k, you forgot to replace the n inside the summation. Also, the first sum should then go from -2 to -infinity, instead of infinity.
+
*Actually, I do not see why you replaced -n by k in both sums. IN the first sum, you should have set k=n-2. In the second sum, you did not need to make any change of variable.
+
*The arrow in the middle of your computations, and the one towards the end should both be replaced by equal signs.
+
*The simplification of the first summation following the arrow is incorrect: you would need to add two terms instead of just one.
+
*The equality following the arrow is only valid when |z|&gt;|a|.You must write this next to the equality!
+
*This explanation would be much clearer if you had typed in your answer: this way I could make notes directly inside the computations and cross-out and replace stuff using different colors.
+
  
 
----
 
----
 +
What if we find a mistake in our computation of the Z transform in the last homework? Do we work with the wrong Z transform and end up with something that isn't the x[n] we chose for that question on homework 2 or do we take the correct Z transform for that x[n] that we chose in homework 2 and then calculate the inverse Z transform?
  
Anybody sees anything else? Do you have more questions?
+
VG
  
 +
:No, use the correct answer instead. (I recommend checking all your answers using  a table of z-transforms). --[[User:Mboutin|Mboutin]] 19:26, 14 September 2010 (UTC)
 
----
 
----
 +
[[2010_Fall_ECE_438_Boutin|Back to ECE438, Fall 2010, Prof. Boutin]]
 +
 +
[[ECE438|Back to ECE438]]

Latest revision as of 12:44, 26 November 2013

Practice Question on "Digital Signal Processing"

Topic: Computing an inverse z-transform


Question from a student

Take $ x[n] = a^n\left( u[n-2]+u[n]\right) $. We then have

$ \begin{align} X(z) &= \sum_{n=-\infty}^\infty x[n]z^{-n} \text{ (by definition of the z-transform)},\\ &= \sum_{n=-\infty}^\infty a^n(u[n-2]+u[n])z^{-n}, \\ &= \sum_{n=2}^\infty a^n(z^{-n}) + \sum_{n=0}^\infty a^n(z^{-n}). \\ \text{Now let }k=-n, \\ \Rightarrow X(z) &= \sum_{k=-2}^\infty (a/z)^n + \sum_{k=0}^\infty (a/z)^n ,\\ &=\sum_{k=0}^\infty \left( (a/z)^n + 2)\right) + \sum_{k=0}^\infty \left( \frac{a}{z}\right)^n \\ & = \left(\frac{1}{1-a/z}+2\right) + \left(\frac{1}{1-a/z}\right), \\ & = \frac{z}{z-a}+2 + \frac{z}{z-a}, \\ & = \frac{z}{z-a}+2\frac{z-a}{z-a} + \frac{z}{z-a} , \\ & = \frac{4z-2a}{z-a}, \\ & = \frac{4-2a/z}{1-a/z}, \text{ for } |z|<a ??? \end{align} $

So if I end up with something that says 1/1-(1/z), I am confused. does it converge when |z|>a or when |z|<a?

~ksoong

Comments/corrections from Prof. Mimi

Take $ x[n] = a^n(u[n-2]+u[n]) $. We then have $ \begin{align} X(z) &= \sum_{n=-\infty}^\infty x[n]z^{-n} \text{ (by definition of the z-transform)}, {\color{OliveGreen}\surd}\\ &= \sum_{n=-\infty}^\infty a^n(u[n-2]+u[n])z^{-n}, {\color{OliveGreen}\surd} \\ &= \sum_{n=2}^\infty a^n(z^{-n}) + \sum_{n=0}^\infty a^n(z^{-n}). {\color{OliveGreen}\surd} \\ \text{Now let }k=-n,& {\color{red}\text{This change of variable is not useful, unfortunately.}} \\ \Rightarrow X(z) &= \sum_{k=-2}^{\color{red}-\infty} (a/z)^{\color{red}n} + \sum_{k=0}^\infty (a/z)^{\color{red}n} ,{\color{red}\text{The terms inside the summation contain n, but the summation is over k.}} \\ &=\sum_{k=0}^{\color{red}-\infty} \left( (a/z)^n {\color{red} -(a/z)^{-2}-(a/z)^{-1}} )\right) + \sum_{k=0}^\infty \left( \frac{a}{z}\right)^n \\ & = \left(\frac{1}{1-a/z}+{\color{red} -(a/z)^{-2}-(a/z)^{-1}} )\right) + \left(\frac{1}{1-a/z}\right), {\color{red}\text{For this last step, you need to assume } \left| \frac{a}{z}\right|<1, \text{ else both sums diverge.}} \end{align} $

The answer to your initial question ("if I end up with something that says 1/1-(1/z), I am confused. does it converge when |z|>a or when |z|<a?") is in the last step. As you can see from this step, X(z) only converges if |a|<|z|. Note that, since a could be a complex number, it is important not to say a< |z|


A good way to check your answer is to use the Z-transform table. You can use the time shift property on the first term (a^n *u[n-2]) and the second term (a^n *u[n]) can be directly converted using the table. Your final answer should match up with what you ended with above. -Sbiddand


Anybody see anything else? Do you have more questions? Comments? Please feel free to add below.


What if we find a mistake in our computation of the Z transform in the last homework? Do we work with the wrong Z transform and end up with something that isn't the x[n] we chose for that question on homework 2 or do we take the correct Z transform for that x[n] that we chose in homework 2 and then calculate the inverse Z transform?

VG

No, use the correct answer instead. (I recommend checking all your answers using a table of z-transforms). --Mboutin 19:26, 14 September 2010 (UTC)

Back to ECE438, Fall 2010, Prof. Boutin

Back to ECE438

Alumni Liaison

Ph.D. 2007, working on developing cool imaging technologies for digital cameras, camera phones, and video surveillance cameras.

Buyue Zhang