Dear MNEusers,
Since having installed MNE 2.7 on a x86-64bit machine running kernel 2.6.18-164.10.1.el5 of CentOS, and NVIDIA driver v. 190.53 I can no longer remotely access the mne programs. It appears that the problem is only encountered using a remote session either VNC or an SSH session with X11 forwarding. In these cases both mne_analyze and mne_opengl_test result in errors. mne_analzye returns a Desired visual not found. Segmentation fault error, and the opengl test returns:
mne_opengl_test version 1.1 compiled at Dec 21 2009 19:47:38
nvert = 140970
ntri = 281936
GLUT_WINDOW_BUFFER_SIZE 32
GLUT_WINDOW_RED_SIZE 8
GLUT_WINDOW_GREEN_SIZE 8
GLUT_WINDOW_BLUE_SIZE 8
GLUT_WINDOW_DEPTH_SIZE 24
GLUT_WINDOW_ALPHA_SIZE 8
500 x 500
X Error of failed request: BadRequest (invalid request code or no such
operation)
Major opcode of failed request: 144 (GLX)
Minor opcode of failed request: 181 ()
Serial number of failed request: 38
Current serial number in output stream: 45
Segmentation fault
I wonder if anyone has encountered this problem? Is it possible that the newest Nvidia driver may be causing an issue with MNE? Perhaps rolling back a version or two? Can anyone recommend nvidia drivers they're using successfully?
Kambiz Tavabi PhD
Biomedical Imaging Laboratory
The Children's Hospital of Philadelphia
34th Street and Civic Center Boulevard
Philadelphia, Pa. 19104
Tel: 267.426.0302
email: tavabik at email.chop.edu
I used mne_forward_solution per the messages written which are listed below my signature. When I look in the output file, TestFwd.fif using mne_show_fiff I find no results. The output from that is shown at the bottom of this message. I truncated a couple of things but left what was before and after them to show that they do not include the forward solution results.
Any thoughts or suggestion would be most appreciated. Thanks.
Don
Don Krieger, Ph.D., D.ABNM
Department of Neurological Surgery
University of Pittsburgh
(412)648-9654
Well, the first thing I noticed, is there are only two points in your
source space. Usually, this is quite odd. Is this intentional? The
problem may be further upstream.
Thanks for responding. Yes, the two points are intentional.
Here's a dump of where the dipoles are made using mne_list_source_space:
# CoordinateSystem "Head"
# begin end X (mm) Y (mm) Z (mm) Q(nAm) Qx(nAm) Qy(nAm) Qz(nAm) g/%
1 1 30.0 50.0 80.0 10.0 2.67 5.35 8.02 100.0
2 2 40.0 40.0 70.0 10.0 5.96 2.98 7.45 100.0
## Name "ALL dipoles" Style "Dipoles2"
If there's a problem upstream, is there something in the output from mne_forward_solution which indicates where it is? The only thing I see which might be that is:
" Setting up compensation data...
No compensation set. Nothing more to do."
Well, I don't really understand what the problem is (but I imagine
someone will have the real answer). My two thoughts are that either
there is a problem with the tags you included in mne_forward_solution
(mne_do_forward_solution is designed to avoid the complexities of
these commands), or whatever procedure you used to generate this
source space may be not interpretable by mne_forward_solution. An
alternative which would work, would be to generate a "grid source
space" then to do the inverse solution by removing the uninteresting
dipoles from the forward model and calculate the inverse solution in
Matlab.
Thank you again for responding. I very much appreciate it.
I generated the source space using mne_volumne_source_space. That part appears to be OK since mne_list_source_space applied to the results yields back the right answer.
I haven't been able to get mne_do_forward_solution to work. It gives the error message: "Could not find the MRI data directory ..." But it appears that "do_forward_" is a wrapper around "forward_" which does some automatic file naming.
I'm not trying to get to an inverse solution at this point. I just want the forward solution values at the MEG sensors, given specific dipole locations and orientations.
Well, you may want to try to do this with an actual MRI instead of
just using the surfaces from fsaverage. There are some dependencies
upon having an actual MRI (which there isn't for fsaverage). I would
try on a real MRI.
I think the problem is that numbers are actually there but mne_show_fif doesn't show them. I think I have to convert the forward solution file to a matlab file and then access the forward solution matrix using matlab. Perhaps you or someone else on the list can confirm this.
Don
Don Krieger, Ph.D., D.ABNM
Department of Neurological Surgery
University of Pittsburgh
(412)648-9654
The mne_show_fiff listing is absolutely correct. It does not output
large data like matrices in full. You do not need to do any
conversion, just read the file with the mne_read_forward_solution
function provided in the MNE matlab toolbox.