1
$\begingroup$

I have a periodical function $f(x)$ that is sampled in time, at a rate of 1 millisecond.

Together with this, I also have a second function $t(x)$.

I want to find the correlations between periods of the first function and the second function $t(x)$.

How can I approach this issue?

$\endgroup$
  • $\begingroup$ There is no such predefined thing as "correlation between periods". You can compute a correlation function for your two functions. $\endgroup$ – mbaitoff Mar 25 '13 at 16:55
  • $\begingroup$ Do you have samples of the second function too? Are they sampled at the same rate? Are the functions' periods the same? $\endgroup$ – Jim Clay Mar 26 '13 at 11:55
  • $\begingroup$ yes, I have samples of the second function, and no, they have diff periods $\endgroup$ – Corovei Andrei Mar 26 '13 at 11:58
  • $\begingroup$ @CoroveiAndrei: is period of one of the functions known? $\endgroup$ – mbaitoff May 26 '13 at 5:27
  • $\begingroup$ I'm new to DSP, but what about wavelet coherence (or squared coherence)? $\endgroup$ – Jase May 26 '13 at 6:29
1
$\begingroup$

If you want to merely compare the periods of the functions $f_1$ and $f_2$, you could use the following "metric" (it might not technically be one if it disobeys the triangle inequality, but someone else will have to check for that):

$\Delta (f_1, f_2) \equiv \dfrac{\left| T_1 - T_2 \right|}{\max {T_1, T_2}}$

Then the similarity can simply be defined as $s(f_1, f_2) \equiv 1-\Delta(f_1, f_2)$

$\endgroup$
  • $\begingroup$ What are T1 and T2? $\endgroup$ – user13107 Aug 28 '13 at 4:57
0
$\begingroup$

Cross correlation is used for comparing functions

$\endgroup$

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.