# Convolution of shifted signal [closed]

If $$y(t) = x(t)*h(t)$$, then what is the expression for $$y(t+a)$$?

Is it $$x(t+a)*h(t+a)$$ or $$x(t+a)*h(t)$$?

## closed as unclear what you're asking by MBaz, Laurent Duval, Matt L., lennon310, Dan BoschenOct 13 '18 at 21:38

Please clarify your specific problem or add additional details to highlight exactly what you need. As it's currently written, it’s hard to tell exactly what you're asking. See the How to Ask page for help clarifying this question. If this question can be reworded to fit the rules in the help center, please edit the question.

• Please show what work you have done in your own attempts to prove this – Dan Boschen Oct 12 '18 at 12:55
• It is x(t+a)*h(t) or x(t)*h(t+a) hint: x(t+a)*h(t+b)=y(t+a+b). – Ch.Siva Ram Kishore Oct 12 '18 at 23:43
• The convolution of $y(t+a)$ by what? – Laurent Duval Oct 13 '18 at 0:14
• If $y(t)$ is the output of an LTI system with impulse response $h(t)$ whose input is $x(t)$, then $y(t+a)$ is just $y(t)$ advanced in time by $a$ and so, apply time invariance property to figure out what the input signal needs to be in order to produce $y(t+a)$ as the output. – Dilip Sarwate Oct 13 '18 at 3:59

From your confusion of $$x(t+a) \star h(t+a)$$ vs $$h(t) \star x(t+a)$$ I guess that a little help on the argument manipulations on functions and convolutions could be appropriate here, moving in simple examples:

First let us express the usual simplistic case. Consider the relation: $$y(t) = h(t) \cdot x(t) + g(t) \tag{1}$$

then manipulations on the argument $$t$$ is applied to all functions on both sides

$$y(t+a) = h(t+a) \cdot x(t+a) + g(t+a)$$

or an arbitrary transform on $$t$$ would similary be: $$y(\phi(t)) = h(\phi(t)) \cdot x(\phi(t)) + g(\phi(t))$$

Now consider the case where two functions convolved to produce the third: $$y(t) = \int_{-\infty}^{\infty} h(\tau) x(t-\tau) d\tau$$

which is abbreviated as $$y(t) = h(t) \star x(t) \tag{2}$$

Now be careful to interpret the case-2. The variable $$t$$ shows in all functions as an argument but you may not apply the transform on $$t$$ as you did with the case-1, so assume you have an arbitrary transform on $$t$$ as $$\phi(t)$$ then

$$y(\phi(t)) \neq h(\phi(t)) \star x(\phi(t)) \tag{3}$$

For example, as in your case, if $$\phi(t) = t+a$$ then you get

$$y(t+a) \neq h(t+a) \star x(t+a)$$ but $$y(t+a) = h(t) \star x(t+a) = h(t+a) \star x(t)$$

The justification of this can (only) be seen when you consider the integral definition of the convolution operator:

\begin{align} y(t) &= h(t) \star x(t) \\ & = \int_{-\infty}^{\infty} h(\tau) x(t-\tau)d\tau \\ y(t+a) & = \int_{-\infty}^{\infty} h(\tau) x((t+a)-\tau)d\tau \\ & = h(t) \star x(t+a) \\ \end{align}

Note that since the live variable inside the integral only happens in one function ($$x(t-\tau)$$ in this case) then a change in $$t$$ will only affect one of them and you get:

$$y(t+a) = h(t) \star x(t+a)$$ or from commutativity of convolution you get $$y(t+a) = h(t+a) \star x(t)$$

So this provides the answer you were looking for. However, it's not over. Because the following case represents an exception:

$$y(-t) \neq h(t) \star x(-t)$$ but $$y(-t) = h(-t) \star x(-t) \tag{4}$$

So how to see this case-4. Again, using the integral definition :

Assuming that $$y(t) = h(t) \star x(t)$$, then compute the convolution between two new signals $$g(t)=h(-t)$$ and $$z(t)=x(-t)$$ as: \begin{align} w(t) &= g(t) \star z(t) \\ & = \int_{-\infty}^{\infty} g(\tau) z(t-\tau)d\tau &g(\tau)=h(-\tau),z(t-\tau)=x(-(t-\tau)) \\ & = \int_{-\infty}^{\infty} h(-\tau) x(-(t-\tau))d\tau &\text{ let } \tau'=-\tau \\ & = -\int_{\infty}^{-\infty} h(\tau') x(-(t+\tau'))d\tau' &\text{ replace } \tau' \text{ with } \tau \\ & = \int_{-\infty}^{\infty} h(\tau) x(-t-\tau) d\tau \\ & = y(-t) \\ \end{align}

hence we conclude that $$h(-t) \star x(-t) = w(t) = y(-t)$$. As stated before, you must always consult to the (explicit) integral definition to decide on the correct functions used in the convolution operator.