Hi Guofan

I suspect your problem may be twofold. 1) I think those versions of
spm_peb_ppi and spm_regions are pretty old. Please update to the newest
versions (I'm not sure what the version numbers should be). 2) Some previous
versions of matlab produced this error for certain matrix operations. I
vaguely recall it may have been 7.0.4 but it could have been 6.5. 

Darren

> -----Original Message-----
> From: Guofan Xu [mailto:[log in to unmask]] 
> Sent: Monday, August 06, 2007 5:19 PM
> To: [log in to unmask]
> Subject: RE: asking for help with SPM-PPI problem?
> 
> Dear Dr. Gitelman:
>   Thanks for your email and appreciate your help.
> 
> The matlab version: Version 6.5.0.180913a Release 13
>   We are currently using spm5 to do the PPI analysis.
> spm_peb_ppi.m     2.10 Darren Gitelman 03/03/13
> spm_regions.m.      2.17 Karl Friston 03/03/03
> 
>   By the way, we ran the sample dataset from SPM website 
> "attention to visual motion" 
> (http://www.fil.ion.ucl.ac.uk/~wpenny/datasets/attention.html)
>  with SPM5_PPI and it gave us the same warning message.  
> 
> thanks,
> 
> guofan
> 8/6
> 
> 
> 
> >>> "d gitelman" <[log in to unmask]> 08/06/07 4:04 PM >>>
> Hi Guofan
> 
> I need a little more to know what is happening. The error 
> suggests there is a problem inverting the design matrix but 
> this has been seen with different versions of matlab and/or 
> earlier versions of the spm functions. 
> 
> What version of spm are you using? Also please send the 
> versions of spm_peb_ppi.m and spm_regions.m. What version of 
> matlab are you using.
> Single vs. multi-session should not matter expect with 
> multisession you have to run the PPI for each session, which 
> is a pain.
> 
> Darren
> 
> > -----Original Message-----
> > From: Guofan Xu [mailto:[log in to unmask]]
> > Sent: Friday, August 03, 2007 2:40 PM
> > To: [log in to unmask]
> > Subject: asking for help with SPM-PPI problem?
> > 
> > Hi, Darren:
> >   This is Guofan from UW-Madison Sterling Johnson's 
> Neuroimaging lab.  
> > I've been working on PPI analysis of fMRI data and my 
> colleagues told 
> > me that you are expert of the brain network & PPI.  I keep 
> getting a 
> > warning messege and maybe you could help me with it:
> > 
> > This is from Matlab window:
> > Warning:  Matrix is close to singular or badly scaled. 
> > Results may be inaccurate. RCOND=1.266417e-20.
> > 
> >    I initally got this error mesege from multiple session 
> fMRI dataset 
> > and though it related to the multi-session issues reported in SPM 
> > list.  However, the warning messege still exist when I change the 
> > original estimation to single session.  We actually went 
> through the 
> > whole PPI process and result looks interesting.  However, I just 
> > concern about the potential mistake related to the warning.
> > 
> > thanks,
> > 
> > Guofan Xu, MD, PhD
> > Postdoc fellow
> > Department of Medicine & Radiology
> > V.A Hosptial -GRECC
> > 2500 Overlook Terrace
> > Madison, WI, 53705
> > 
> 
> 
>