[FieldTrip] Very short Morlet wavelets - non-zero mean correction missing ?

[Michael Meg]

Dear colleagues and friends,

looking at the current Fieldtrip version (20141016), I could not find the
"low-cycle correction" that you need for less than 5 cycles in Morlet
wavelet analysis. On the other hand a gwidth (Gaussian width, I suppose)
parameter defaults to 3 instead of five. Maybe it would be good to
implement the correction by subtracting exp( (2*pi*freqoi(ifreqoi))2 / 2)
from the complex exponential after its created? (The correction deals with
the fact that the complex Morlet wavelet has a non-zero mean at the low
cycle numbers. See for example: The Illustrated Wavelet Transform Handbook:
Introductory Theory and Applications in Science, Egnineering, Medicine and
Finance. Paul S Addison. page 35) Please disregard this email if this has
been done in the meantime. Best, Michael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20141017/9ae7df93/attachment-0001.html>