Hi,
> this is getting somewhat disturbing ... many people have seen this error
> now, and it seems to have to something with non-sphericity correction, that
> is apparently implemented with some bugs left in the code. The error happens
> without a clear reason, it does for one dataset and not for another
> apparently modelled the same way. I stumbled upon this error about 2 times,
> so the problem isn't occurring very often, but, as can be seen on the list,
> persistently.
I am sorry, I should have answered earlier. Did you catch this post:
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind04&L=SPM&P=R460237&I=-3
I think if it were me, I would add a line in spm_spm, after the
spm_svd call, with something like:
if size(s, 1) < nScan, error('Autocovariance estimation poorly
conditioned; reduce model or increase number of observations?'); end
In MarsBaR I put in an option to use Keith Worsley's simple AR
estimation - which is only useful for the fMRI first-level analyses,
but is more robust, and does not rely on a hard-coded estimate of the
autocorrelation. It would be fairly simple to port over to SPM, I
just haven't had the time. If anyone is interested, the code is here:
http:[log in to unmask]
The function above is just a modified version of spm_spm, and it may
be simple to see how to change the relevant bits of spm_spm to get the
same effect. You'll see that MarsBaR uses some unmodified copies of
spm functions, which have a 'pr_' prepended - such as 'pr_spm_reml.m'
which is identical to 'spm_reml.m'. The autocovariance changes in the
above function are especially lines 291 to 331. The actual estimation
is done by:
http:[log in to unmask]
Best,
Matthew
