Mathematically speaking, when I try to use some signal to disprove a system is invertible, can I use the signal like $\delta(t+\infty)$ ($\delta$ representing the Dirac distribution)? For example, the two inputs $0$ and $\delta(t+\infty)$ are distinct, but their outputs from the system $\int_{-\infty}^{t}e^{\tau}x(\tau)d\tau$ will then both be $0$. In this way, can I say the system is invertible?

  • 5
    $\begingroup$ Personally, I'd say that you can't do that. In any case, it wouldn't be useful, since the system might be invertible for all signals that one actually cares about. $\endgroup$
    – MBaz
    Feb 9 '19 at 0:49

It's possible, in mathematics, to complete real numbers with "infinite" values, with sound topological properties; for instance non-standard analysis or the Extended real number line (discussion at Math StackExchange).

However, in this context, for any standard real number $t$, the rule is to set $t\pm \infty = \pm \infty$ (see Arithmetic operations). So, using your notations, for every $t$, you would have [CAUTION ADVISED] $\delta(t+\infty)=\delta(\infty)$, to which I am not able to give a meaning, else than something being a constant evererywhere equal to $\delta(\infty)$. Of course, $\delta$ is a distribution, and should not be treated as a function, but defined as a functional operator, or via limits of functions.

However, so far, I have never seen

  • a (serious and useful) use of the extended real system in signal processing,
  • a reference trying to define such a Dirac at infinity.

For the latter, I fear (just intuition) that constructing this with test functions can be troublesome, since two different limits are required (location and amplitude). This is just an analogy, but $\delta(0)\times \delta(0)$ is, as far as I know, not defined.

  • 1
    $\begingroup$ The only sensible use I could see of this would be to use it as a test function, e.g. to have something to convolve a system with to get the value it converges to at $\infty$; but that would essentially require us to define that the convolution with $\delta(t+\infty)$ does that; it would feel consistent, but exactly as you state, there's nothing inherently meaningful about $\delta(\infty)$ $\endgroup$ Feb 9 '19 at 13:30
  • $\begingroup$ Yes, I feel that one would need a double limit, one classical on the amplitude, and one on the ordinal variable, which could lead to ambiguities $\endgroup$ Feb 9 '19 at 14:43
  • $\begingroup$ As far as I can remember when I used then, that mostly simplifies proofs and notationt $\endgroup$ Feb 9 '19 at 20:14

The Dirac delta generalized function is only defined under an integral

$$ Dirac(t-a) = \int_{-\infty}^\infty f(t)\delta(t-a)dt = f(a)\, $$

Typically there's also a function under the integral.


$$ Dirac(t-\infty) = f(-\infty)\, $$

if one considers $-\infty$ to be a point - not to mention problems evaluating the $Dirac$ at -$\infty$.


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.