[Eeglablist] newtimef freqs output

Thu Aug 2 17:53:12 PDT 2012

Dear Katharina,

> I double checked it and the results are identical for both using 'padratio'
> 4 and 'nfreqs' 47. I use log-spaced frequencies, so both settings lead to
> the same 47 log-spaced frequencies from 4 to 50 Hz (and all the other input
> settings stay the same).

Ok.
Please tell me how many frequencies you obtained for the case of
'padratio', 4. I don't think the numbers are the same as the case of
'nfreqs', 47 (or could be... in that case, that is just an
coincident). Anyway, now I understand your intention, it is
meaningless to use 'nfreqs', 47. It could be 30, 20, or 50 as you
like.

> Changing the padratio to 2 and 8 changes the frequency resolution (and that
> is what it should do, right?).
> Using a padratio of 2 results in 23 log-spaced frequencies and a padratio of
> 8 results in 94 log-spaced frequencies.

This looks good.

Makoto

2012/8/2 Katharina Limbach <klim366 at aucklanduni.ac.nz>:
> Dear Makoto,
>
> I double checked it and the results are identical for both using 'padratio'
> 4 and 'nfreqs' 47. I use log-spaced frequencies, so both settings lead to
> the same 47 log-spaced frequencies from 4 to 50 Hz (and all the other input
> settings stay the same).
>
>
> [ersp,itc,powbase,times,freqs] = newtimef(EEG.data (83,:,:), 350, [-400
> 996], 250, [1 0.5],'baseline',[NaN],'freqs', [4 50], 'freqscale', 'log',
> 'padratio', 4);
>
> [ersp1,itc1,powbase1,times1,freqs1] = newtimef(EEG.data (83,:,:), 350, [-400
> 996], 250, [1 0.5],'baseline',[NaN],'freqs', [4 50], 'nfreqs', 47,
> 'freqscale', 'log');
>
> Changing the padratio to 2 and 8 changes the frequency resolution (and that
> is what it should do, right?).
> Using a padratio of 2 results in 23 log-spaced frequencies and a padratio of
> 8 results in 94 log-spaced frequencies.
>
> Thanks a lot, Katharina
>
>
>
> On 3 August 2012 11:56, Makoto Miyakoshi <mmiyakoshi at ucsd.edu> wrote:
>>
>> Dear Katharina,
>>
>> > I tried using 'nfreqs' 47 instead of 'padratio' 4 as you suggested. The
>> > results are identical.
>>
>> I'm not sure if this could be true. Could you check it again?
>> Also, please try 'padratio', 2 and 8 to see if it changes results.
>>
>> > Why is it better to define the number of frequencies
>> > than to use the padratio?
>>
>> I misunderstood you. I thought you were saying you want to obtain
>> linear scale from 4 to 50 Hz with 1 Hz interval. Please forget it.
>>
>> Makoto
>>
>> 2012/8/2 Katharina Limbach <klim366 at aucklanduni.ac.nz>:
>> > Dear Makoto,
>> >
>> > thanks a lot for your response.
>> > I think I figured out, that I was making a really stupid mistake. I
>> > thought
>> > that the output parameters were optional and that their order did not
>> > matter
>> > (I used [ersp,times,freqs]) . So the output that was given for 'freqs'
>> > was
>> > actually the 'powbase' output (and likewise the 'times' output was the
>> > itc...).
>> >
>> > So I changed my code to:
>> >
>> > [ersp,itc,powbase,times,freqs] = newtimef(EEG.data (83,:,:), 350, [-400
>> > 996], 250, [1 0.5],'baseline',[NaN],'freqs', [4 50], 'freqscale', 'log',
>> > 'padratio', 4);
>> >
>> > and I am now getting more sensible results (times from -260 to 860 and
>> > frequencies from 4 to 50).
>> > I am really sorry for this!
>> >
>> > I tried using 'nfreqs' 47 instead of 'padratio' 4 as you suggested. The
>> > results are identical. Why is it better to define the number of
>> > frequencies
>> > than to use the padratio?
>> >
>> > I have one more question about the powbase. I do not want to baseline my
>> > wavelets (and I set 'baseline' [NaN]), but I am getting a powbase
>> > output.
>> > Does that mean that the baseline power is computed but not used for the
>> > wavelets?
>> >
>> > Thanks a lot, Katharina
>> >
>> >
>> >
>> > On 3 August 2012 08:06, Makoto Miyakoshi <mmiyakoshi at ucsd.edu> wrote:
>> >>
>> >> Dear Katharina,
>> >>
>> >> Sorry for slow response.
>> >>
>> >> > The values range from 1.6608914e+04 to 3.0225791e+05.
>> >>
>> >> This means your frequency range is between 1.6 MHz to 30.2 MHz, which
>> >> is like an old generation CPU clock speed. Something is seriously
>> >> wrong. However, I don't know what causes this from your code.
>> >>
>> >> By the way, if you want something close to 4,5,6,... 50, you should
>> >> use 'nfreq', 47 instead of 'padratio', 4. Could you try it to see what
>> >> happens?
>> >>
>> >> Makoto
>> >>
>> >> 2012/7/22 Katharina Limbach <klim366 at aucklanduni.ac.nz>:
>> >> > Dear all,
>> >> >
>> >> > I have a question regarding the output of the newtimef function.
>> >> > I am using the following command (input data: 1 participant, 1
>> >> > condition,
>> >> > electrode # 67. My sampling rate is 250 Hz and the epochs go from
>> >> > -400
>> >> > to
>> >> > 996 ms (350 frames)):
>> >> >
>> >> > [ersp,times,freqs] = newtimef(EEG.data (67,:,:), 350, [-400 996],
>> >> > 250,
>> >> > [1
>> >> > 0.5],'baseline',[NaN], 'freqs', [4 50], 'freqscale', 'log',
>> >> > 'padratio',
>> >> > 4,
>> >> > 'plotphase', 'off', 'plotitc', 'off');
>> >> >
>> >> > The resulting plot is showing frequencies from 4 to 50 Hz (which is
>> >> > what
>> >> > I
>> >> > was looking for) but I do not understand the output I get for
>> >> > 'freqs':
>> >> > The values range from 1.6608914e+04 to 3.0225791e+05.
>> >> > Shouldn't they simply be in Hz and go from 4 to 50 Hz? Do I have to
>> >> > convert
>> >> > them to get the correct values or is there something else that I am
>> >> > doing
>> >> > wrong?
>> >> >
>> >> > Also my 'times' output is not a vector of output times in ms, but a
>> >> > 47
>> >> > x
>> >> > 200 martix with complex numbers.
>> >> >
>> >> >
>> >> > Thanks a lot for your help,
>> >> > Katharina
>> >> >
>> >> > --
>> >> > Katharina Limbach
>> >> > PhD Candidate
>> >> > Department of Psychology
>> >> > University of Auckland
>> >> > Private Bag 92019
>> >> > Auckland
>> >> > New Zealand
>> >> > tel: + 64 9 3737 599 ext 85969
>> >> >
>> >> >
>> >> > _______________________________________________
>> >> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> >> > To unsubscribe, send an empty email to
>> >> > eeglablist-unsubscribe at sccn.ucsd.edu
>> >> > For digest mode, send an email with the subject "set digest mime" to
>> >> > eeglablist-request at sccn.ucsd.edu
>> >>
>> >>
>> >>
>> >> --
>> >> Makoto Miyakoshi
>> >> JSPS Postdoctral Fellow for Research Abroad
>> >> Swartz Center for Computational Neuroscience
>> >> Institute for Neural Computation, University of California San Diego
>>
>>
>>
>> --
>> Makoto Miyakoshi
>> JSPS Postdoctral Fellow for Research Abroad
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>
>
>
>
> --
> Katharina Limbach
> PhD Candidate
> Department of Psychology
> University of Auckland
> Private Bag 92019
> Auckland
> New Zealand
> tel: + 64 9 3737 599 ext 85969
>

-- 
Makoto Miyakoshi
JSPS Postdoctral Fellow for Research Abroad
Swartz Center for Computational Neuroscience
Institute for Neural Computation, University of California San Diego