[Eeglablist] Re: timef/crossf 'padratio'

[Scott Makeig]

In timef() and crossf(), the parameter 'padratio' does what it says. It 
pads a tapered data window of length winsize with (padratio-1)*winsize 
zeros. Delivered to an FFT, this gieves padratio times as many 
frequencies - though without actually increasing the freq resolution, in 
the sense that two very nearby frequencies could now be resolved. It 
does allow localization of the center of a single frequency peak, and 
makes prettier figures...

Scott

PS A 'padratio' of 16 is high - try 4 and 1 to compare.

PPS A more common problem is not knowing how to get low frequencies from 
timef(). Using the modified wavelet option ('cycles', [3 0.5]), an 
optional argument 'winsize', round(EEG.srate), applied to 3-sec epochs, 
will give a low frequency of 3 Hz. This is the default we use.


naeem at sbox.tugraz.at wrote:

>Hi!
>I am using eeglab for data analysis.
>There is a parameter 'padratio'used in plot channel time
>frequency---pop_timef().
> I am not able to understand what this parameter does?
>In tutorial this value is set to be 16-what is the criterion for a particular
>value of this parameter?
>I will appreciate any help
>naeem
>  
>