[FieldTrip] Inconsistent results from ft_connectivityanalysis

[Maxime Fauvet]

Hi all,  

I'm trying to perform a granger causality analyse from EEG
and EMG signals but somehow i cannot achieve to get a complete
time-frequency map of the GC. The problem seems to be with the matrix
factorization step, the chol matlab function returning warning messages
for almost all time points.  

I use data from ft_freqanalysis and
initially, raw data contain 2 signals (one EEG and one EMG) sampled at
1000 Hz. The imput matrix size is 2*ntrials*ntime with ntrials ~= 20,
ntime ~= 3000. 

Do you have any idea why the matrix factorization
involved in GC computation is possible for some time points and is not
for others? 

Thank you in advance,  

Maxime.

  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20200831/9e728ff0/attachment.htm>