Hi,
I yesterday encountered an error estimating a single session 548 Scan regular
Blockdesign with 6 regressors (5+const).
------
Temporal non-sphericity (over voxels) : ...REML estimation
ReML Block - 1
ReML Iteration : 1 ...NaN
Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = -1.000000e+00.
> In spm_reml at 77
In spm_spm at 805
Warning: Matrix is close to singular or badly scaled.
Results may be inaccurate. RCOND = -1.000000e+00.
> In spm_reml at 82
In spm_spm at 805
Warning: Matrix is singular, close to singular or badly scaled.
Results may be inaccurate. RCOND = NaN.
> In spm_reml at 115
In spm_spm at 805
ReML Iteration : 2 ...NaN
...
-----
The error in similar form has been mentioned before, eg
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0510&L=SPM&P=R26407&I=-3
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0504&L=SPM&P=R48484&I=-3
but the suggested fix: "feature('jitallow','structs','off');"
did not work (well on anther data set it did)
What did work was downgrading from Matlab7.01 (R14 SP 3) to 7.0.4 (R14 SP2).
But I then found another dataset that gave me the same error with SP2. Funny
enough I can estimate this second dataset in SP3. (double dissociation)
There was a similar problem for DCM and a possible fix mentioned here:
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0510&L=SPM&P=R26407&I=-3
Any chance this fix might also work with SPM2.
Some System data:
Reproducible on 2 Linux workstations running debian sarge. P4 Proc, 1GB Mem,
with and without installed atlas3-sse libs. SPM2 with all available updates
including Volmar's Patch for the fprint problem.
http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind0508&L
I also recompiled the mexfiles ->same problem.
For now I mainly run R14 SP3 reloading SP2 if necessary but that's not a real
solution, is it?. I'd appreciate any suggestions.
Regards and thanks,
Roland
--
Dr. Roland Marcus Rutschmann <[log in to unmask]>
Institute for Experimental Psychology, University of Regensburg
Universitätsstraße 31, 93053 Regensburg, Germany
Tel: +49 941 943 2533, Fax: +49 941 943 3233
http://www.psychologie.uni-regensburg.de/Rutschmann
|