Suppose that system has the input/output relation as follows

$$\sum_{k=0}^{N}a_k \frac{d^ky(t)}{dt^k} = \sum_{k=0}^{M}b_k \frac{d^kx(t)}{dt^k}$$

Where $a_k, b_k \in \mathbb{R}$.

Obviously we need additional conditions to determine completely the input-output relationship for the system and those auxiliary conditions specify whether the system is LTI and causal or not.

The definition of initial rest condition for a system in general is that if $x(t) = 0$ for $t\lt t_0$ then $y(t) = 0$ for $t\lt t_0$.

First of all is it true that the initial rest condition for the all systems implies $y(t_0) = 0$? Also I have seen expressions like $y({t_0}^+)$ and $y({t_0}^-)$ but I don't know how they relate to $y(t_0)$. Another question is that whether the initial rest condition is necessary and sufficient for the mentioned system to be causal and LTI? What's the rigorous proof, then? I searched on the Internet but didn't found a proof or a clear explanation.

  • $\begingroup$ Please refer to Allen oppenheim's book Signal and Systems, all these points are clearly explained there. $\endgroup$ Apr 18, 2020 at 6:45
  • $\begingroup$ @Dspguysam I've seen that but it doesn't contain a proof. $\endgroup$
    – S.H.W
    Apr 18, 2020 at 6:47
  • $\begingroup$ it is there, i have specified in the answer, please have a look $\endgroup$ Apr 18, 2020 at 7:42

1 Answer 1


The notion of initial rest and causality follows only for linear systems (time invariant not required).

For proof of Initial rest being sufficient condition for an LTI system to be causal. Please refer to Problem 1.44 of Allen Oppenheim's book on Signal and Systems. Try and solve it yourself (mostly a homework problem, there is also a solution manual just to let you know in case you are stuck), it doesn't require anything extra other than convolution sum.

For $y(t_o^+)$ and $y(t_o^-)$ relating to $y(t_o)$ post the quote you are looking for clarifications on, it depends on multiple context, generally that would be specifying an initial condition of a one sided Fourier transform or one sided sequence.

  • $\begingroup$ And is that condition necessary for the linear constant-coefficient differential equations? $\endgroup$
    – S.H.W
    Apr 18, 2020 at 7:47
  • $\begingroup$ Yes initial rest iis necessary for a linear system to be causal. Ofcourse the poles have to be on the left of s plane for the trnasform to be stable. $\endgroup$ Apr 18, 2020 at 7:50

Your Answer

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

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