[Eeglablist] sphering problem

[Mikhail Zvyagintsev]

Dear EEGLAB users,
Could you, please, tell me what would be the possible reasons in ICA 
processing (runica.m) that
after calculating of 'sqrtm' from covariance matrix of data without mean
(string in runica.m(line 748 in version eeglab4.515) - sqrtm(cov(data')) 
I always get NaN matrix. 'cov' returns 'normal' covariance matrix.
Thank you,
Mikhail Zvyagintsev