Hello,
I'm trying to run selxavg3-sess for an analysis with TER < TR, and it errors with "design is ill-conditioned". I searched the mailing list archive, and the only past problem of this sort that I found was associated with a faulty tpef file, which I am not using.
The details: There is only one run. TR is 1.2s, and stimuli were presented at intervals of 1/3 TR (0.4s) (the desired TER). The paradigm file reflects this, it is of the form:
0 0 0.4 1 0.8 0 1.2 2 1.6 1 2 0
I loaded it into matlab and verified that the time points all have a difference of 0.4s.
The analysis is, mkanalysis-sess -analysis analysis_name -TR 1.2 -TER 0.4 -paradigm 03TR.par -designtype event-related -nconditions 2 -timewindow 19.2 -prestim 2.4 -mcextreg -funcstem fmcstcsm6 -polyfit 2 -runlistfile runlist_03TR -noautostimdur -force
I define two contrasts,
mkcontrast_sess -analysis analysis_name -rmprestim -contrast AV-REST -a 1 -c 0 mkcontrast_sess -analysis analysis_name -rmprestim -contrast AVT-REST -a 2 -c 0
And then selxavg3-sess,
selxavg3-sess -analysis analysis_name -sf subjectname -df sessdir -overwrite
And this errors. Like I said, the only thing I found in the mailing list was a problem with tpef, and so I left it out altogether (as above), and the problem persists. If I change TER to equal TR (1.2), then it works. So the problem has something to do with TER being less than TR. Perhaps something in the par file then? I checked it in matlab, and diff(par(:,1)) = 0.4 for all of the events.
Any thoughts would be very appreciated! Ben.
Did you optimize the stimulus schedule with the idea that you would do sub-TR FIR estimation? Eg, you would have had to have run optseq2 with this in mind. If not, then you probably cannot do an FIR analysis. Instead, assume a shape (eg, -gammafit 2.25 1.25)
doug
Ben Letham wrote:
Hello,
I'm trying to run selxavg3-sess for an analysis with TER < TR, and it errors with "design is ill-conditioned". I searched the mailing list archive, and the only past problem of this sort that I found was associated with a faulty tpef file, which I am not using.
The details: There is only one run. TR is 1.2s, and stimuli were presented at intervals of 1/3 TR (0.4s) (the desired TER). The paradigm file reflects this, it is of the form:
0 0 0.4 1 0.8 0 1.2 2 1.6 1 2 0
I loaded it into matlab and verified that the time points all have a difference of 0.4s.
The analysis is, mkanalysis-sess -analysis analysis_name -TR 1.2 -TER 0.4 -paradigm 03TR.par -designtype event-related -nconditions 2 -timewindow 19.2 -prestim 2.4 -mcextreg -funcstem fmcstcsm6 -polyfit 2 -runlistfile runlist_03TR -noautostimdur -force
I define two contrasts,
mkcontrast_sess -analysis analysis_name -rmprestim -contrast AV-REST -a 1 -c 0 mkcontrast_sess -analysis analysis_name -rmprestim -contrast AVT-REST -a 2 -c 0
And then selxavg3-sess,
selxavg3-sess -analysis analysis_name -sf subjectname -df sessdir -overwrite
And this errors. Like I said, the only thing I found in the mailing list was a problem with tpef, and so I left it out altogether (as above), and the problem persists. If I change TER to equal TR (1.2), then it works. So the problem has something to do with TER being less than TR. Perhaps something in the par file then? I checked it in matlab, and diff(par(:,1)) = 0.4 for all of the events.
Any thoughts would be very appreciated! Ben. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Yes, the paradigm came from optseq2 with a PSD window 0 19.2 0.4.
Looking more into the details, it seems that the design is pretty ill-conditioned, cond(Xn'*Xn) is 6.2e5, and the cut-off in fast_selxavg3.m is 1e4.
Like I said, the design came from optseq2 and the cost/efficiency is the same as other paradigms that have worked well (albeit with TER=TR). Perhaps this one was pushing it just a little too much?.
Thanks! Ben.
2010/4/15 Douglas N Greve greve@nmr.mgh.harvard.edu:
Did you optimize the stimulus schedule with the idea that you would do sub-TR FIR estimation? Eg, you would have had to have run optseq2 with this in mind. If not, then you probably cannot do an FIR analysis. Instead, assume a shape (eg, -gammafit 2.25 1.25)
doug
Ben Letham wrote:
Hello,
I'm trying to run selxavg3-sess for an analysis with TER < TR, and it errors with "design is ill-conditioned". I searched the mailing list archive, and the only past problem of this sort that I found was associated with a faulty tpef file, which I am not using.
The details: There is only one run. TR is 1.2s, and stimuli were presented at intervals of 1/3 TR (0.4s) (the desired TER). The paradigm file reflects this, it is of the form:
0 0 0.4 1 0.8 0 1.2 2 1.6 1 2 0
I loaded it into matlab and verified that the time points all have a difference of 0.4s.
The analysis is, mkanalysis-sess -analysis analysis_name -TR 1.2 -TER 0.4 -paradigm 03TR.par -designtype event-related -nconditions 2 -timewindow 19.2 -prestim 2.4 -mcextreg -funcstem fmcstcsm6 -polyfit 2 -runlistfile runlist_03TR -noautostimdur -force
I define two contrasts,
mkcontrast_sess -analysis analysis_name -rmprestim -contrast AV-REST -a 1 -c 0 mkcontrast_sess -analysis analysis_name -rmprestim -contrast AVT-REST -a 2 -c 0
And then selxavg3-sess,
selxavg3-sess -analysis analysis_name -sf subjectname -df sessdir -overwrite
And this errors. Like I said, the only thing I found in the mailing list was a problem with tpef, and so I left it out altogether (as above), and the problem persists. If I change TER to equal TR (1.2), then it works. So the problem has something to do with TER being less than TR. Perhaps something in the par file then? I checked it in matlab, and diff(par(:,1)) = 0.4 for all of the events.
Any thoughts would be very appreciated! Ben. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: www.nmr.mgh.harvard.edu/facility/filedrop/index.html
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
try it with a prestim of 0
Ben Letham wrote:
Yes, the paradigm came from optseq2 with a PSD window 0 19.2 0.4.
Looking more into the details, it seems that the design is pretty ill-conditioned, cond(Xn'*Xn) is 6.2e5, and the cut-off in fast_selxavg3.m is 1e4.
Like I said, the design came from optseq2 and the cost/efficiency is the same as other paradigms that have worked well (albeit with TER=TR). Perhaps this one was pushing it just a little too much?.
Thanks! Ben.
2010/4/15 Douglas N Greve greve@nmr.mgh.harvard.edu:
Did you optimize the stimulus schedule with the idea that you would do sub-TR FIR estimation? Eg, you would have had to have run optseq2 with this in mind. If not, then you probably cannot do an FIR analysis. Instead, assume a shape (eg, -gammafit 2.25 1.25)
doug
Ben Letham wrote:
Hello,
I'm trying to run selxavg3-sess for an analysis with TER < TR, and it errors with "design is ill-conditioned". I searched the mailing list archive, and the only past problem of this sort that I found was associated with a faulty tpef file, which I am not using.
The details: There is only one run. TR is 1.2s, and stimuli were presented at intervals of 1/3 TR (0.4s) (the desired TER). The paradigm file reflects this, it is of the form:
0 0 0.4 1 0.8 0 1.2 2 1.6 1 2 0
I loaded it into matlab and verified that the time points all have a difference of 0.4s.
The analysis is, mkanalysis-sess -analysis analysis_name -TR 1.2 -TER 0.4 -paradigm 03TR.par -designtype event-related -nconditions 2 -timewindow 19.2 -prestim 2.4 -mcextreg -funcstem fmcstcsm6 -polyfit 2 -runlistfile runlist_03TR -noautostimdur -force
I define two contrasts,
mkcontrast_sess -analysis analysis_name -rmprestim -contrast AV-REST -a 1 -c 0 mkcontrast_sess -analysis analysis_name -rmprestim -contrast AVT-REST -a 2 -c 0
And then selxavg3-sess,
selxavg3-sess -analysis analysis_name -sf subjectname -df sessdir -overwrite
And this errors. Like I said, the only thing I found in the mailing list was a problem with tpef, and so I left it out altogether (as above), and the problem persists. If I change TER to equal TR (1.2), then it works. So the problem has something to do with TER being less than TR. Perhaps something in the par file then? I checked it in matlab, and diff(par(:,1)) = 0.4 for all of the events.
Any thoughts would be very appreciated! Ben. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: www.nmr.mgh.harvard.edu/facility/filedrop/index.html
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
freesurfer@nmr.mgh.harvard.edu
-
Ben Letham -
Douglas N Greve