Dear Sun,
This is not the kind of error that can be fixed by changing
@file_array\subref.m. The main reason for it is probably the fact that
you have a very long time window of 2 sec. It's also not surprising
that group inversion requires more memory than individual inversions.
Interestingly, the error occurs after the inversion is already done at
the stage when results are displayed for each subject. So perhaps by
some smart commenting out of some lines in spm_eeg_invert you could
get the solutions saved without displaying them and then review the
subjects one-by-one after clearing the memory. But if you don't
understand the code sufficiently to do that you should either reduce
your time window or downsample your data or find a 64-bit machine with
enough memory.
Best,
Vladimir
On Mon, Apr 19, 2010 at 5:59 AM, Sun Delin <[log in to unmask]> wrote:
> Dear Vladimir,
>
> Would you please give me help again? An error message of "Out of memory" appeared when I was doing group inversion. The grandmean data input were from 13 subjects of 4 conditions, i.e. one trial for each condition per subject. In fact, errors also occurred when I was doing the inverse reconstruction for each subject. However, the individual inversions could be completed after restarting the matlab program and pressing the button of 'Invert' again. The memory information of my PC is as follows:
> Maximum possible array: 425 MB (4.457e+008 bytes) *
> Memory available for all arrays: 1059 MB (1.110e+009 bytes) **
> Memory used by MATLAB: 698 MB (7.318e+008 bytes)
> Physical Memory (RAM): 2031 MB (2.130e+009 bytes)
>
> I have changed the variable "mem" in the \spm8\@file_array\subref.m to "mem = 200*1024*1024", however, errors still occurred.
>
>
>
> Error messages:
> -------------------------------------------
> ..........
>
> Inverting subject 1
> ReML Iteration : 1 ...2.670066e+003
> ReML Iteration : 2 ...1.925378e+003
> ReML Iteration : 3 ...1.005712e+003
> ReML Iteration : 4 ...2.893591e+002
> ReML Iteration : 5 ...3.684136e+001
> ReML Iteration : 6 ...2.707117e+000
> ReML Iteration : 7 ...2.196782e-001
> ReML Iteration : 8 ...2.415073e-002
> ReML Iteration : 9 ...3.278310e-003
> Percent variance explained 95.34 (94.78)
> type: 'GS'
> modality: {'EEG'}
> smooth: 0.6000
> xyz: [0 0 0]
> rad: 128
> scale: 0.0075
> M: [8196x112 double]
> J: {[8196x16 double] [8196x16 double] [8196x16 double] [8196x16 double]}
> Y: {[112x16 double] [112x16 double] [112x16 double] [112x16 double]}
> L: [112x8196 double]
> qC: [8196x1 double]
> qV: [2001x2001 double]
> T: [2001x16 double]
> U: {[112x123 double]}
> Is: [8196x1 double]
> It: [1x2001 double]
> Ic: {[1x123 double]}
> Nd: 8196
> pst: [1x2001 double]
> dct: [1x193 double]
> F: 1.2926e+004
> ID: 121.8664
> R2: 95.3410
> VE: 0.9941
> woi: [-200 1800]
>
> ??? Out of memory. Type HELP MEMORY for your options.
>
> Error in ==> spm_eeg_invert_display at 62
> J = J{con}*T';
>
> Error in ==> spm_eeg_invert at 810
> spm_eeg_invert_display(D{i});
>
> Error in ==> spm_eeg_inv_group at 213
> D = spm_eeg_invert(D);
>
> Error in ==> sdl_SPM8_source_group_inv at 45
> spm_eeg_inv_group(S);
> -------------------------------------------
>
|