[FieldTrip] Fourier coefficients across tapers

[Eelke Spaak]

Dear Jose,

> It would be wrong to average across the tappers the Fourier coefficients?

Yes.

> Otherwise, it would be ok to average across the tapers the values that result from computing, for instance, coherence ? It that sensitive?

Yes.

> I mean, do the first 15 columns correspond to the Fourier coefficients that result from applying the 15 tapers to the first trial?

>From the top of my head: yes. There is also the 'cumtapcnt' field,
which I think reflects the number of tapers applied per trial.

Finally, you might want to have a look at ft_connectivityanalysis,
which does all of this bookkeeping for you. Alternatively, if you
still conclude it's more convenient to work in custom code, you could
have a look at low-level FT code for inspiration. E.g. ft_checkdata
subfunction fixcsd, which converts from univariate fourierspctrm to
bivariate cross-spectral density
(https://github.com/fieldtrip/fieldtrip/blob/a0e1edb20223eb51c97978f1947a6983c3ce7f23/utilities/ft_checkdata.m#L737),
and ft_connectivity_corr
(https://github.com/fieldtrip/fieldtrip/blob/master/connectivity/ft_connectivity_corr.m)
which computes coherence from cross-spectral density. Note that the
latter
interprets the leading dimension simply as "repetitions" (i.e. you can
average across trials and tapers in one go).

Cheers,
Eelke

>
> I would really appreciate some directions here,
>
> best,
> Jose
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://doi.org/10.1371/journal.pcbi.1002202