Consider a dynamical system as bellow that is subjected to an impulse input:

$\sum_{n=0}^N {a_n x^{(n)}}= f_0 \delta(t-t_0)$

the initial values are non-resting (all non-zero):

$x(t_0^-)=x_0\hspace{0.2cm} ;\hspace{0.2cm} x'(t_0^-)=x_1\hspace{0.2cm} ;\hspace{0.2cm} ... \hspace{0.2cm} ;\hspace{0.2cm} x^{(n-1)}(t_0^-)=x_{(n-1)} $.

How can I write the correct initial values for $t=t_0^+$? This is ambiguous since integrating the differential equation above, yields:

$ \sum_{n=1}^N a_n \big[x^{(n-1)} (t_0^+)-x^{(n-1)} (t_0^-)\big] = \sum_{n=1}^N a_n c_{n-1} = f_0$

where $c_i$s are the jumps in initial conditions. In literature, it is said that only the highest order will undergo a discontinuity, yet it's not clear for me why.

  • $\begingroup$ Welcome to Engineering Stack Exchange. This seems like an assumption mde by the book. If you assume a discontinuity in the highest term, then differentiate, all the differentials will have impulses - i.e. no visible discontinuity. Therefore, this is an easy system to analyze as a first order. Other solutions exist, but this is the easiest to model. $\endgroup$
    – Mark
    Jun 11 '18 at 22:19
  • $\begingroup$ How can a finite order continuous model describe a delay? $\endgroup$
    – Arash
    Jun 21 '18 at 4:45

I asked the question here and I found the answer myself. This is really a simple one really. Here is my answer for this question:

[...] I revisited this problem with another point of view which gives the answer right away. Since the $x$ is at least $N-1$ times differentiable, it's derivatives (up to order $N-2$) and itself must be continuous. So no shift until the last derivative. Using the my question's notation,

$c_n =0\hspace{1cm} \forall n<N-1$

$c_{N-1} =f_0/a_N$


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.