[Eeglablist] Bug in pop_runica.m... fastica() called with out-of-order args

Arnaud Delorme arno at salk.edu
Fri Nov 4 17:42:05 PST 2005


Yes, this is correct and this has been fixed quite a while ago (although 
it is not yet reflected in the EEGLAB version avaiable on the internet). 
This will be made avaiable in the new release. Testing fastica, it 
remains that, in my personal experience, it fails to converge on very 
big datasets with more thatn 300 000 sample points.

Best,

Arno

Dan Keith wrote:

> Hi,
>
> (I sent this msg about two weeks ago, but it never showed up on the list)
>
> I was trying to use the fastica algorithm via pop_runica.m within 
> EEGLAB. I was getting horrible results on my decompositions. Looking 
> at the code in pop_runica.m, I noticed that the order of the result 
> arguments to fastica on line 512 was inconsistent with the documented 
> order of arguments in fastica.
>
> Specifically, the fastica function expects to return [icasig,A,W] as 
> the components, mixing, and unmixing matrices, respectively. However, 
> the call used in pop_runica.m is:
>     [ ICAComp, EEG.icaweights, EEG.icawinv ] = fastica(...)
> which seems to be backwards.
>
> If I swap the last two result args:
>    [ ICAComp, EEG.icawinv, EEG.icaweights ] = fastica(...)
> then fastica gives the correct results when executed via pop_runica.m
>
> Please fix this bug in EEGLAB. I believe it may give a distorted 
> impression of the suitability of fastica. Effectively, fastica is 
> unusable via pop_runica.m in the current EEGLAB implemenetation. Note 
> that ICAComp is ignored in pop_runica; the components are computed in 
> pop_runica.m from the unmixing matrix.
>
> thanks for your attention,
>
> dan keith
> Neuroinformatics Center
> University of Oregon
>
>-- 
>  
>
> Dan "Bud" Keith - dkeith at cs.uoregon.edu
>
>------------------------------------------------------------------------
>
>_______________________________________________
>eeglablist mailing list eeglablist at sccn.ucsd.edu
>http://sccn.ucsd.edu/mailman/listinfo/eeglablist
>Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
>


-- 

*Arnaud Delorme, Ph.D.*
Swartz Center for Computational Neuroscience, INC, University of San 
Diego California
La Jolla, CA92093-0961, USA

*Tel* :/(+1)-858-458-1927 ext 15/
*Fax* :/(+1)-858-458-1847/
*Web page*: sccn.ucsd.edu/~arno <http://www.sccn.ucsd.edu/%7Earno>
*To think upon*:

    What we want is to become ourselves. Nothing that is should be
    suppressed; nothing is superfluous.

        /Friedrich Nietzche/  

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20051104/cf1e8d33/attachment.htm


More information about the eeglablist mailing list