NANs in Time frequency analysis.

[Sameer Walawalkar]

Thanks Jan-Mathijs,

That was my expectation given how TFRs are calculated.

sameer

On Thu, 29 Mar 2007, Jan Mathijs Schoffelen wrote:

> Dear Sameer,
>
> A nan is given back at those time-frequency points at which no value can be
> estimated. This means that at the toi-foi pairs you specified in your
> configuration, the corresponding timewindow as specified in t_ftimwin cannot
> be completely filled with data. In your case, I therefore expect this
> 'problem' to be worse for the lower frequencies. It might seem a sorry
> answer, but this is just a thing you have to live with.
>
> Yours,
>
> Jan-Mathijs
>
> -----Original Message-----
> From: FieldTrip discussion list [mailto:FIELDTRIP at NIC.SURFNET.NL] On Behalf
> Of Sameer Walawalkar
> Sent: Thursday, March 29, 2007 4:21 PM
> To: FIELDTRIP at NIC.SURFNET.NL
> Subject: [FIELDTRIP] NANs in Time frequency analysis.
>
> Hi,
>
> I am using the following settng to run time frequency analysis.
>
> cfgTFR = [];
> cfgTFR.output       = 'pow';
> cfgTFR.channel      = 'MEG';
> cfgTFR.method       = 'mtmconvol';
> cfgTFR.foi          = 05 : 2 : 35;
> cfgTFR.t_ftimwin    = 5./cfgTFR.foi;
> cfgTFR.tapsmofrq    = 0.4 *cfgTFR.foi;
> cfgTFR.toi          = 1:0.05:2.5;
> cfgTFR.pad          = 'maxperlen';
>
> TFRmult = freqanalysis(cfgTFR, data);
>
> and find that the TFRplots for all channels (plottd using multiplot) are
> always deep blue in the high time end, especially at low frequencies.
>
> Inspection of TFR mult shows many NANs in the corresponding time regions.
>
> Does anyone has any ideas about this?
>
> thanks for your reply.
>
> best,
> sameer
>
>