5
$\begingroup$

If I have two functions in a convolution like

$$X*Y=1$$

$$X*Z=1$$

then it means (trivially) $Y=Z$.

Is this correct or are there subtleties in the convolution theorem where $Y=Z$ isn't always true?

$\endgroup$
2
$\begingroup$

Here is a counter example. The functions $X,Y,Z$ below will satisfy that $X * Y = X * Z = 1$ but $ Y \not= Z$:

$$ X(t) := 1, \forall \,t \in \mathbb{R},$$

$$ Y(t) := \begin{cases} \frac{1}{C_Y} e^{-\frac{1}{1-t^2}} \quad \text{if $-1 \leq t\leq 1$}\\ 0 \quad \text{otherwise,} \end{cases} $$ where $C_Y$ is a (normalization) constant such that the integral of $Y$ over $[-1,1]$ equals 1, and $$ Z(t) := \begin{cases} \frac{1}{C_Z} e^{-\frac{1}{2^2-t^2}} \quad \text{if $-2 \leq t\leq 2$}\\ 0 \quad \text{otherwise,} \end{cases} $$ where, similar to $C_Y$, $C_Z$ is a (normalization) constant such that the integral of $Z$ over $[-2,2]$ equals 1.

Those are concrete functions as an explicit example. Any normalized (integral equals 1) function $f$ that has compact support will satisfy that $X * f = 1$. From there you can choose any $Y$ and $Z$ you want as long as they are different.

$\endgroup$
1
$\begingroup$

Good Question :

Here we can use the concept of Laplace Transformation

$$X(s) Y(s) = X(s) Z(s)$$

$$Y(s) = Z(s)$$

We know that "Laplace Transform of two signals can be same but their ROC will be definitely different"

Here is a example :

$$e^{-at} u(t) \rightleftharpoons \frac{1}{s+a}$$

$$-e^{-at} u(-t) \rightleftharpoons \frac{1}{s+a}$$

HENCE now it is proved that it is not necessary that

$$Y = Z$$

$\endgroup$

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.