0
$\begingroup$

The following sequence is given which is supposed to be time-variant:

$$y[n] = \sum_{k=n_0}^n x[k]$$

I'm having difficulties proving the time-variance or finding a counterexample for it being time-invariant. My idea (which proves it being time-invariant?) is:

\begin{align}y_1[n] &= T\{x_1[n]\}\\x_2[n] &= x_1[n-n_0]\end{align}

check, whether $T\{x_1[n]\} = y_1[n-n_0]$

$$y_2[k] = \sum_{k=n_0}^n x_2[k] = \sum_{k=n_0}^n x_1[k - n_0] = \sum_{k=n_0-n_0}^{n-n_0} x_1[k] = y_1[n-n_0]$$

Where is my mistake? Working around the summation is giving me trouble.

EDIT: Following the advice in the comments for $n = n_0$ the left equation equals to 1 while the right equation equals to 0, making both equations unequal and therefore the sequence time-variant?

Second sequence

This sequence is supposed to be time-invariant.

\begin{align}y[n] &= \sum_{k=n-n_0}^{n+n_0} x[k]\\ y_2[k] &= \sum_{k=n-n_0}^{n+n_0} x_2[k] = \sum_{k=n-n_0}^{n+n_0} x_1[k - n_0] = \sum_{k=n_0-2n_0}^{n} x_1[k] = y_1[n-n_0]\end{align}

I followed the same steps but I'm still not sure about for the case $n = n_0$ since when added in left and right equation

$$\sum_{k=0}^{2n} = \sum_{k=-n}^{n}$$

They seem unequal to me?

Improve this question
$\endgroup$
10
  • $\begingroup$ What is the relationship connecting arrays $$ x_2[.] $$ and $$ x_1[.] $$ of your expressions? $\endgroup$
    – V.V.T
    Commented Jul 15, 2020 at 12:14
  • $\begingroup$ Added to original post. $x_2$ is a shifted $x_1$ $\endgroup$
    – Granolaboy
    Commented Jul 15, 2020 at 12:20
  • $\begingroup$ Does the last equation hold if n < n_0? (summation from zero to negative index) $\endgroup$
    – V.V.T
    Commented Jul 15, 2020 at 12:25
  • $\begingroup$ I also thought about that, then the summation equals to 0. However if $n < n_0$ then the left summation also equals to 0 and both the equations are equal? I could possibly find a pair of $n$ and $n_0$ for which the equations are not equal. $\endgroup$
    – Granolaboy
    Commented Jul 15, 2020 at 12:29
  • $\begingroup$ Let n = n_0. Then y[n_0] = x[n_0]. The x series left-shifted by n_0 indices is x[n - n_0]; and the summation is y_shifted[n_0] = x[0]. Right? $\endgroup$
    – V.V.T
    Commented Jul 15, 2020 at 12:54

1 Answer 1

Reset to default
2
$\begingroup$

Sequences that are "time-invariant" must be constants (not varying with time at all).

What the OP is actually asking about is a (discrete-time) system whose response $y$ to the input $x$ is defined to be $$y[n] = \begin{cases}\sum_{k=n_0}^n x[k], & n \geq n_0,\\??? & n < n_0\end{cases}$$ where the $???$ is there because according to some traditionalists, if $n < n_0$, then that sum is an empty sum which has value $0$ but more liberal-minded might treat that sum as representing $$x[n] + x[n+1] + x[n+2] + \cdots + x[n_0-1] + x[n_0]$$ (with more serious-minded folks even negating that sum in analogy with the standard result that for $a<b$, $\int_b^a = -\int_a^b$). So, for the traditional meaning of things and $n_0$ denoting a (finite) integer, the system is not a time-invariant system. It is time-invariant (with unit pulse response $h[n]$ being the unit step function $u[n]$) if $n_0 = -\infty$ as noted in a comment by Robert Bristow-Johnson on the main question.

Improve this answer
$\endgroup$
4
  • $\begingroup$ Would you please explain how the system becomes time-invariant when $n_0 = -\infty$? The left equation is $\sum_{k=\infty}^n x_2[k]$ and the right equation $\sum_{k=0}^{n-(-\infty)} x_1[k] = \sum_{k=0}^{\infty} x_1[k]$ ? $\endgroup$
    – Granolaboy
    Commented Jul 16, 2020 at 6:07
  • $\begingroup$ $- \infty$ on the left equation* $\endgroup$
    – Granolaboy
    Commented Jul 16, 2020 at 6:24
  • $\begingroup$ If $n_0=-\infty$, then \begin{align}y[n]&=\sum_{k=-\infty}^nx[k]\\&=\sum_{k=-\infty}^nu[n-k]x[k]\\&=\sum_{k=-\infty}^\infty u[n-k]x[k]\\&=u\star x\big\vert_n\end{align} where we have used the fact that $u[n-k]=1$ whenever $k$ is such that $n-k\geq 0$, i.e. $k \leq n$ and $u[n-k]=0$ whenever $k$ is such that $n-k<0$, i.e. $k>n$. So, we have expressed the output as a convolution of $x$ with a fixed impulse response (the unit step function) and so we have a time-invariant system, in fact, a linear time-invariant system. $\endgroup$
    – Dilip Sarwate
    Commented Jul 16, 2020 at 17:38
  • $\begingroup$ Thank you very much! $\endgroup$
    – Granolaboy
    Commented Jul 19, 2020 at 11:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.