The topic of this question is the causal linear time invariant system (LTI system). I need some help to understand better how to go from this formula $$ y(t)=\int_{-\infty}^{+\infty} x(τ)h(t-τ)dτ\label{1}\tag{1}$$ to this one: $$y(t)=\int_{0}^{t} x(τ)h(t-τ)dτ.\label{2}\tag{2}$$

The system in question using a visual diagram representation is: $h(t)\longrightarrow T[x(t)]\longrightarrow y(t)$

I know that the causality criterion is: $h(t)=0$ for all $t<0$. In simple words, the output $y(t)$ is independent of all future input values.

In formula \eqref{1} do NOT swap positions of $x(t)$ and $h(t)$. That is, I do NOT want to see the analysis of the convolution integral $$\int_{-\infty}^{+\infty}h(τ)x(t-τ)dτ. $$ I already comprehend how to solve it (a.k.a. analyse it) in this manner, but i get confused when i see formula \eqref{1}.


Like you say, $h(t) = 0$ for all $t < 0$. This means that $h(t-\tau) = 0$ for all $t < \tau$ or, equivalently, for all $\tau > t$.

That is, in the range $\tau > t$, the integrand is zero, and so

$$\int_{-\infty}^\infty x(\tau) h(t-\tau) \, \mathrm{d}\tau = \int_{-\infty}^t x(\tau) h(t-\tau) \, \mathrm{d}\tau + \int_{t}^\infty 0 \, \mathrm{d}\tau = \int_{-\infty}^t x(\tau) h(t-\tau) \, \mathrm{d}\tau $$

This establishes the upper limit, so now what about the lower limit? To determine this, we must consider what defines the moment in time $t=0$? What separates it from any other arbitrary point of time? Why do we set the lower limit to 0 instead of any other arbitrary time?

$t=0$ is the time before which no input is present. That is, $x(t) = 0$ for all $t < 0$. Therefore, we can say $x(\tau) = 0$ for all $\tau < 0$, which means the integrand of the convolution goes to zero for negative $\tau$. That is

$$\int_{-\infty}^t x(\tau) h(t-\tau) \, \mathrm{d}\tau = \int_{-\infty}^0 0 \, \mathrm{d}\tau + \int_{0}^t x(\tau) h(t-\tau) \, \mathrm{d}\tau = \int_{0}^t x(\tau) h(t-\tau) \, \mathrm{d}\tau $$

So, what if there existed an input before $t=0$? What if $x(t) < 0$ for $t < -t_0$? Then the convolution integral would instead have become

$$\int_{-t_0}^t x(\tau) h(t-\tau) \, \mathrm{d}\tau $$


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.