Given the signals

\begin{align} x(t) &= A_1\sin(\omega t)\\ y(t) &= A_2\sin(\omega t) \end{align}

Do $A_1$ and $A_2$ play a role in the $C_{xy}$ coherence estimator? MATLAB's mscohere.m shows no differences for $A_1=A_2$ and $A_1\neq A_2$ at all.


From the definition of the (magnitude-squared) coherence


with the cross-spectral density $G_{xy}(f)$, and the power spectra $G_{xx}(f)$ and $G_{yy}(f)$, respectively, it is clear that scaling of $x(t)$ or $y(t)$ does not change the value of $C_{xy}(f)$, because the scaling constants appear in the numerator as well as in the denominator and cancel out.

The coherence is a normalized quantity (and hence scaling-invariant), and it always satisfies $0\le C_{xy}(f)\le 1$.

  • $\begingroup$ Thank you. I know it may sounds a naive question to make, however in a very popular paper [IF 29.2] that is clearly stated as in "Coherence mixes the effects of amplitude and phase in the interrelations between two signals". $\endgroup$
    – IDKreally
    Feb 1 '17 at 20:32
  • $\begingroup$ @IDKreally And now you do know really. :D $\endgroup$
    – Peter K.
    Feb 2 '17 at 0:11

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.