3dDeconvolve: ERROR: symeigval_double error code = 1

AFNI Message Board
1

I am getting a 3dDeconvolve error when i run my script, kindly suggest me the issue here.

3dDeconvolve -input /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/Opt_ii_Despike/despiked_sprl_shft_al_brain_disdaq_mcf.nii -ortvec /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/extra_regressors/covar_design.1D lll -nobucket -polort A -errts /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/2_MotionFiltered/despiked_sprl_shft_al_brain_disdaq_mcf_mregress
++ 3dDeconvolve extending num_stimts from 0 to 10 due to -ortvec
++ 3dDeconvolve: AFNI version=AFNI_19.2.26 (Sep 24 2019) [64-bit]
++ Authored by: B. Douglas Ward, et al.
++ current memory malloc-ated = 596,152 bytes (about 596 thousand)
++ loading dataset /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/Opt_ii_Despike/despiked_sprl_shft_al_brain_disdaq_mcf.nii
++ current memory malloc-ated = 596,152 bytes (about 596 thousand)
++ STAT automask has 162006 voxels (out of 673920 = 24.0%)
++ Skipping check for initial transients
++ Imaging duration=2390.0 s; Automatic polort=16
++ Number of time points: 1195 (no censoring)

  • Number of parameters: 27 [27 baseline ; 0 signal]
    ++ Memory required for output bricks = 3,224,033,280 bytes (about 3.2 billion)
    ++ Wrote matrix values to file Decon.xmat.1D
    ++ ========= Things you can do with the matrix file =========
    ++ (a) Linear regression with ARMA(1,1) modeling of serial correlation:

3dREMLfit -matrix Decon.xmat.1D -input /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/Opt_ii_Despike/despiked_sprl_shft_al_brain_disdaq_mcf.nii
-Rerrts /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/2_MotionFiltered/despiked_sprl_shft_al_brain_disdaq_mcf_mregress_REML -verb

[left]3dREMLfit -matrix Decon.xmat.1D -input /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/Opt_ii_Despike/despiked_sprl_shft_al_brain_disdaq_mcf.nii
-Rerrts /BioInformatics/Bmax_neuro/data/imaging_data/biomax/Workflow_000000000000822/2_MotionFiltered/despiked_sprl_shft_al_brain_disdaq_mcf_mregress_REML -verb

++ N.B.: 3dREMLfit command above written to file 3dDeconvolve.REML_cmd
++ (b) Visualization/analysis of the matrix via ExamineXmat.R
++ (c) Synthesis of sub-model datasets using 3dSynthesize
++ ==========================================================
** ERROR: symeigval_double error code = 1
e[7m*+ WARNING:e[0m ----- !! Signal+Baseline matrix condition [X] (1195x27): nan ** BEWARE **
++ Signal+Baseline matrix singular values:
nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan
*** MCW_malloc post-corruption! serial=689278 size=216 source=matrix.c line#1664
← calculate_results ← 3dDeconvolve main
** ERROR: symeigval_double error code = 1
e[7m*+ WARNING:e[0m ----- !! Baseline-only matrix condition [X] (1195x27): nan ** BEWARE **
++ Baseline-only matrix singular values:
nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
-nan -nan
*** MCW_malloc post-corruption! serial=689291 size=216 source=matrix.c line#1664
← calculate_results ← 3dDeconvolve main
** ERROR: symeigval_double error code = 1
e[7m*+ WARNING:e[0m ----- !! stim_base-only matrix condition [X] (1195x10): nan ** BEWARE **
++ stim_base-only matrix singular values:
nan -nan -nan -nan -nan
-nan -nan -nan -nan -nan
*** MCW_malloc post-corruption! serial=689303 size=80 source=matrix.c line#1664
← calculate_results ← 3dDeconvolve main
++ ----- polort-only matrix condition [X] (1195x17): 1.03361 ++ VERY GOOD ++[/left]

Thanks
Dev

2

Hi Dev,

The text output shows no change in memory usage after reading the data file. How big is that file?

Also, the warnings about ‘nan’ are concerning. We do not allow that in the binary files. Note that ‘float_scan’ can detect and possibly overwrite such values.

  • rick
3

Dear Sir,

The file size is 2GB,I am not sure what is wrong with my code and why it is unable to load the file properly.

Thanks
Vasudev

4

Hi Vasudev,

Would you mail me the covar_design.1D file? That seems to be the source of the problem. It should not use NAN either.

Thanks,

  • rick
5

I am annexing covar_design file for your review. please check it

https://www.dropbox.com/s/5pjtwo1cvfb4q36/covar_design.1D?dl=0
Thanks
Vasudev

6

Hi Vasudev,

Sorry for dropping this a bit. Yes, that covar_design.1D file has a column of nan’s. Maybe some machines will handle it, maybe not. But in general, we do not handle those.

But I must expect that particularly since the entire column is nan, it is corrupting the regression computations. That should be equivalent to a column of zeros, which the program can handle (though it might still need -GOFORIT). Can you use zeros, instead?

  • rick
7

If treating the NaN’s in a text file as zeros is okay, then consider setting AFNI_1D_ZERO_TEXT = YES either in your .afnirc file or on the command line or script.

cat > test.1D
NAN
nan
NaN
^C

1dcat test.1D

** Error: Failed parsing data row 0 of 1D file ‘test.1D’
Check for illegal non-numeric characters in:
‘NAN’
You can have text entries set to 0 with -ok_1D_text or by
setting environment variable AFNI_1D_ZERO_TEXT to YES.
** FATAL ERROR: Can’t read input file ‘test.1D’
** Program compile date = Dec 12 2019

1dcat -DAFNI_1D_ZERO_TEXT=YES test.1D
0
0
0