Impulse Response: Understanding & Applying Maths

In summary: D^{\prime} to R...R^{\prime} and whose range is the set of all functions mapping D...D^{\prime} to R...R^{\prime}.
  • #1
Päällikkö
Homework Helper
519
11
This thread would've fitted in a lot of categories, but as it's the mathematics I can't quite grasp, I decided to fit it here. I'll be treating the discrete case, so I suppose it's precalculus maths.

Anyways, I know how to compute problems with the impulse response method, but I'm kinda baffled as to why to do it how it's done (as the proof we had in class was rather incomplete).

I looked up wikipedia's article on the matter: http://en.wikipedia.org/wiki/Impulse_response
This is the part I'm having problems understanding:
[tex]T\left[ \sum_{k}x\left[ k\right] \delta \left[ n-k\right] \right] = \sum_{k}x\left[ k\right] T\left[ \delta \left[ n-k\right] \right][/tex]
It's got something to do with linearity, apparently the homogeneity property. I just can't understand why the x[k] is being treated as a constant and the delta isn't.

Then the part of applying the mathematics:
Given a system, eg. y(k) + y(k-1) - y(k-2) = 2k, why am I supposed to write it out as
h(k) + h(k-1) - h(k-2) = delta(k) ?
That is, why does the solution of the latter equation give the impulse response of the system? Could someone show this mathematically?

Am I making any sense? :smile:
PS. Could someone recommend any good websites in the matter, I tried googling but found none?
 
Physics news on Phys.org
  • #2
Ok, I figured it out myself. The linearity stuff's due to the fact that T operates on n, thus x[k] is a constant.

For the second part I came up with a proof, but I'm not convinced it's water-proof :smile::
[tex]a_m h[n-m] + ... + a_0 h[n] = b_m \delta [n-m] + ... + b_0 \delta [n][/tex]

[tex]a_m h[n-m-j] + ... + a_0 h[n-j] = b_m \delta [n-m-j] + ... + b_0 \delta [n-j][/tex]

[tex]x[j] \left ( a_m h[n-m-j] + ... + a_0 h[n-j] \right ) = x[j] \left ( b_m \delta [n-m-j] + ... + b_0 \delta [n-j] \right )[/tex]

[tex] a_m x[j] T[ \delta [n-m-j] ] + ... + a_0 x[j] T[ \delta [n-j] ] = b_m x[j] \delta [n-m-j] + ... + b_0 x[j] \delta [n-j][/tex]

Adding sums, this part I'm not so sure of:
[tex]a_m T \left [ \sum _{j = -\infty} ^\infty x[j] \delta [n-m-j] \right ] + ... + a_0 T \left [ \sum _{j = -\infty} ^\infty x[j] \delta [n-j] \right ] = b_m \sum _{j = -\infty} ^\infty x[j] \delta [n-m-j] + ... + b_0 \sum _{j = -\infty} ^\infty x[j] \delta [n-j][/tex]

[tex]a_m y[n-m] + ... + a_0 y[n] = b_m x[n-m] + ... + b_0 x[n][/tex]

However, I don't think the proof works both ways due to the summation part?
 
Last edited:
  • #3
I thought about it again, and came up with the following (it's essentially the same as above, with the direction switched and with slight modifications to enable this):

[tex]
a_m y[n-m] + ... + a_0 y[n] &= b_m x[n-m] + ... + b_0 x[n]
[/tex]

[tex]
a_m T \left [ \sum _{j = -\infty} ^\infty x[j] \delta [n-m-j] \right ] &+ ... + a_0 T \left [ \sum _{j = -\infty} ^\infty x[j] \delta [n-j] \right ] = b_m \sum _{j = -\infty} ^\infty x[j] \delta [n-m-j] + ... + b_0 \sum _{j = -\infty} ^\infty x[j] \delta [n-j]
[/tex]

[tex]
a_m \sum _{j = -\infty} ^\infty x[j] h [n-m-j] + ... + a_0 & \sum _{j = -\infty} ^\infty x[j] h [n-j] = b_n \sum _{j = -\infty} ^\infty x[j] \delta [n-m-j] + ... + b_0 \sum _{j = -\infty} ^\infty x[j] \delta [n-j]
[/tex]

[tex]
\sum _{j = -\infty} ^\infty x[j] \left ( a_m h [n-m-j] + ... + a_0 h [n-j] \right ) = \sum _{j = -\infty} ^\infty x[j] \left ( b_m \delta [n-m-j] + ... + b_0 \delta [n-j] \right )
[/tex]

Moving all the terms to the left, under one sum:
[tex]
\sum _{j = -\infty} ^\infty x[j] \left ( a_m h [n-m-j] + ... + a_0 h [n-j] - \left ( b_m \delta [n-m-j] + ... + b_0 \delta [n-j] \right ) \right ) = 0
[/tex]

One solution'd be, that for all j:
[tex]
a_m h [n-m-j] + ... + a_0 h [n-j] = b_m \delta [n-m-j] + ... + b_0 \delta [n-j]
[/tex]

I'll shift the index by j, (or as the equations above are all equivalent, I'll choose j = 0):
[tex]
a_m h [n-m] + ... + a_0 h [n] = b_m \delta [n-m] + ... + b_0 \delta [n]
[/tex]

And there we have it. Could someone go though the above, and inform me of any mistakes?
 
Last edited:
  • #4
It appears you are confused about what a system is.

A system is a function in which the domain is a set of functions and the range is a set of functions. So if T is a system, and T(x) = y, then x is a function (not a number), and y is a function (not a number).

We can denote a system T using function notation:
[tex] T:[x:D \rightarrow R] \rightarrow [y:D^{\prime} \rightarrow R^{\prime}] [/tex]
This says that T is a system whose domain is the set of all functions mapping D into R and whose range is the set of all functions mapping D' into R'. Here D,D',R,R' are just abritrary sets.

In discrete time we take D and D' to be some discrete set, for example the Integers.

So if our system T is defined as:
[tex] T:[x:\mathbb{Z} \rightarrow \mathbb{R}] \rightarrow [y:\mathbb{Z} \rightarrow \mathbb{R}] [/tex]
Then T is a system (function) which takes as input a function mapping the integers into the reals, and returns a function mapping the integers into the reals.

Note that the Kronecker delta [itex] \delta [/itex] is an element of the domain of T, ie: it is a function mapping the integers into the reals. I emphasize that [itex] \delta [/itex] represents an entire function whereas [itex]\delta(n)[/itex] represents the value of [itex]\delta[/itex] evaluated at the integer n. We call [itex]\delta[/itex] an impulse.

We define the impulse response [itex]h[/itex] to be [itex]T(\delta)[/itex]. That is, [itex]h[/itex] is the function such that for every integer n, [itex]h(n) = T(\delta)(n)[/itex].

If T is linear and time invariant, then given a scalar [itex] \alpha [/itex] and fixed integer [itex] k [/itex] we have that [itex]T(\alpha\delta(n-k))= \alpha h(n-k)[/itex].

So in regards to your first question, using the language of linear algebra, for each k [itex]x[k][/itex] is a scalar (a real number), and [itex]\delta[n-k][/itex] is a vector (a function mapping the integers to the reals). Thus we can "pull out" the x[k]'s when we apply the linear operator T.

In regards to your second question consider a LTI system given by,
[tex] y(n) + a_1y(n-1) + ... + a_Ny(n-N) = b_0x(n) + b_1x(n-1) + ... + b_Mx(n-M)[/tex]

We can rewrite this as:
[tex] y(n) = -a_1y(n-1) - ... - a_Ny(n-N) + b_0x(n) + b_1x(n-1) + ... + b_Mx(n-M)[/tex]

What this says is that if we input a function x into the system, we get as output a function y such that for each n y(n) is given as above. Notice that for each n, y(n) depends on previous values of y, as well as x(n) and previous values of x.

To get the impulse response, we take as input [itex]\delta[/itex] rather than the arbitrary x. Since h is defined to be the function that is outputed when we input [itex]\delta[/itex] we replace y by h. There is nothing that really needs to be proved, it follows simply from the definition of h.
 
Last edited:
  • #5
nocturnal said:
It appears you are confused about what a system is.
Indeed, I hadn't seen the formal definition of a system before, although I did assume it'd look something like that.
Many thanks for the complete explanation.
 

1. What is an impulse response?

An impulse response is a mathematical concept that describes the output of a system when an impulse or sudden change is applied to the input of that system. It is represented by a graph or equation that shows how the system reacts to the impulse over time.

2. Why is understanding impulse response important?

Understanding impulse response is important because it allows us to analyze and predict the behavior of systems, such as electrical circuits, audio equipment, or even biological systems. By understanding how a system responds to an impulse, we can make informed decisions about how to design or improve that system.

3. How is impulse response used in mathematics?

In mathematics, impulse response is used in the field of signal processing, specifically in the area of linear time-invariant systems. It is also used in differential equations and Fourier analysis to study the behavior of systems over time.

4. What are some real-world applications of impulse response?

Impulse response has many real-world applications, including in audio engineering, where it is used to design and improve sound systems. It is also used in control systems for robotics and in medical imaging to analyze and improve the performance of medical equipment.

5. How can I apply impulse response in my own research or work?

If you are working in a field that involves systems analysis or signal processing, understanding and applying impulse response can be extremely useful. It can help you to analyze and improve the performance of systems, as well as make predictions about how they will behave under different conditions. You can also use impulse response in mathematical modeling to simulate and understand complex systems.

Similar threads

  • Engineering and Comp Sci Homework Help
Replies
2
Views
1K
  • Calculus and Beyond Homework Help
Replies
7
Views
133
  • Calculus and Beyond Homework Help
Replies
8
Views
116
  • Precalculus Mathematics Homework Help
Replies
8
Views
153
  • Precalculus Mathematics Homework Help
Replies
3
Views
1K
  • Calculus and Beyond Homework Help
Replies
0
Views
65
Replies
5
Views
4K
  • Advanced Physics Homework Help
Replies
14
Views
737
  • Precalculus Mathematics Homework Help
Replies
3
Views
1K
  • Calculus and Beyond Homework Help
Replies
1
Views
898
Back
Top