[FieldTrip] Wilson Factorization

[Per Arnold Lysne]

Hi Mathieu,

    I have had a similar problem when trying to factor a spectral matrix generated from an average evoked response. In case you are trying to do the same thing, my solution has been to transform individual trials to the time/frequency domain and do the averaging there. I get usable results when factoring the resulting power spectral matrix.

    Hope that helps,

Per Lysne
University of New Mexico

________________________________________
From: fieldtrip-bounces at science.ru.nl <fieldtrip-bounces at science.ru.nl> on behalf of Schoffelen, J.M. (Jan Mathijs) <jan.schoffelen at fcdonders.ru.nl>
Sent: Monday, January 12, 2015 12:18 PM
To: FieldTrip discussion list
Subject: Re: [FieldTrip] Wilson Factorization

Mathieu,

Since your question is of relatively poor quality, I can only venture a poor quality guess: it’s likely that your data is rank deficient. The Wilson algorithm involves inversion of matrices, rank deficiency will quickly lead to nans.

Please consult the following link (and references therein) in order to optimize the probability of obtaining a useful answer, and to optimize the goodwill of the FT-community (especially the ‘Ten simple rules…’ are a must read).

http://fieldtrip.fcdonders.nl/discussion_list

Best wishes,
Jan-Mathijs


On Jan 12, 2015, at 4:44 PM, Mathieu Sitko <mathieu.sitko at wanadoo.fr> wrote:

> I have a problem with the convergence of spectral matrix factorization: with a tolerance of 1e-8, all my data (H,S,Z) are NaN values. How could you explain that?
> thank you
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip


_______________________________________________
fieldtrip mailing list
fieldtrip at donders.ru.nl
http://mailman.science.ru.nl/mailman/listinfo/fieldtrip