Reply by HardySpicer March 25, 20102010-03-25
On Mar 25, 9:11&#2013266080;am, "josef" <chuymv@n_o_s_p_a_m.gmail.com> wrote:
> I'm need your help > I'm trying to use the coherence function using AR method. But the values > obtained are greater than 1. I use the following equation: > > Cxy=abs(Pxy)^2/Pxx*Pyy > > where Pxx and Pyy are the obtained with the method AR. &#2013266080;Pxy is obtained as > follows: &#2013266080;first, I use the cross correlation between the two signals. The > result is evaluated with AR technique to obtain the Pxy > > I don't know if correct this. Someone can help me please > > Thank you for your attention
AR techniques - you mean solving the Yule-Walker equations I expect? (having found the correlation coefficients first). Yes should work but would be slow compared with FFT method (but more smooth). How much greater than 1? Hardy
Reply by josef March 24, 20102010-03-24
I'm need your help
I'm trying to use the coherence function using AR method. But the values
obtained are greater than 1. I use the following equation:

Cxy=abs(Pxy)^2/Pxx*Pyy

where Pxx and Pyy are the obtained with the method AR.  Pxy is obtained as
follows:  first, I use the cross correlation between the two signals. The
result is evaluated with AR technique to obtain the Pxy 

I don't know if correct this. Someone can help me please 

Thank you for your attention