Dear Burkhard,
The number of iterations is limited in code to 64 (see spm_nlsi_N line
284) . This is heuristic but from our experience it is sufficient to
achieve good fits. In some cases the algorithm converges before
reaching this number (see line 544 of spm_nlsi_N). The information
about the number of iterations and final dF is not stored in the DCM
struct, but you can get it from your logs. In any case, there
shouldn't be a problem with not converging before 64 since the changes
in free energy are negligible at that stage even if the formal
convergence criterion is not reached.
Best,
Vladimir
On Thu, Aug 4, 2011 at 1:42 PM, Burkhard Maess <[log in to unmask]> wrote:
> Dear SPM experts,
>
> I have conducted some model inversions using spm_dcm_erp using a matlab
> script.
> The results were stored in .mat files. The text output of matlab during
> inversion procedure was redirected
> into corresponding log-files.
> Now, I realize that a few of the inversions did not succeed: after 64
> iterations no convergence was achieved.
> While I can see this from my log-files, I wonder whether this is also stored
> in the SPM-DCM data structure.
>
> The number of passed iterations is certainly stored somewhere in the data
> structure, but I cannot find it.
> Would you please help in identifying the information within the DCM.XXX
> structure which describes
> details of the iterative inversion procedure? e.g.
> a) number of iterations needed
> b) stop criterion actually applied
> c) set of stop criterions from the last iteration (to see, whether another
> criterion were close to finish as well).
>
> many thanks for your help,
> Burkhard
>
> --
> ------------------------------------------------------------
> Dr. Burkhard Maess
> Max Planck Institute for Human Cognitive and Brain Sciences
> Stephanstr. 1a, P.O. Box 500355, D-04303 Leipzig
> Aussenstelle Bennewitz, phone/fax: +49(3425)887525-26/-11 mail: maess 'at'
> cbs.mpg.de, http://www.cbs.mpg.de
> ------------------------------------------------------------
> Please do not attach files which may contain executable code, e.g. .rtf is
> preferred over .doc(x)
>
|