[VoxBo] buggy makeGmatrix utility?

[Daniel Kimberg]

"Brad" from Mad-Town wrote:
> For several GLMs that we've run recently, the shape of the "param"
> displayed in the Voxel Surfer does not correspond to the shape of the
> HRF that was loaded at the "convolve Diagonal set covariate(s)" step
> in the "set up diagonal set options" gui.  And so seemingly the wrong
> filter was used to smooth the covariates of interest in the G matrix.
> Is this a previously documented behavior? (and if 'yes' is there a
> fix; if 'no' what might we do to circumvent it?)

Does this happen consistently with the same GLMs?  Is it just evident in the voxel surfer, or can you also see the error in glm/gdw as you're building it?  Also, can you try looking at the foo.G and the foo.preG files using vecview to see if the covariates look okay there?  Lastly, if the problem seems limited to the voxel surfer, is it the IDL-based voxel surfer or vbview or both?

If it happens consistently and it's immediately apparent (or apparent with vecview), possibly it would be best for me to send you the absolute latest versions of gdw/glm/gds, and for you to send me the condition function and HRF file, so that I can try to replicate the problem.

dan