Hi,
This is an issue which is quite annoying with TCL/TK but at the moment
we are stuck with it. The best solution I can offer is to set up a
smaller
version with the GUI, save the design and edit the design.fsf file in
order
to create a larger model. This avoids having to use the GUI for the
larger
model. The tricky thing is knowing exactly what to change in the .fsf
file.
The easiest way is probably to set up two different ones with the GUI,
each with a slightly different size and then look at the differences
between
the files (e.g. using the "diff" command in the terminal or, better,
in emacs).
Hopefully that will fix things for you.
All the best,
Mark
On 21 May 2009, at 21:17, Greg Burgess wrote:
> Unfortunately, the feat_gui becomes unusable when we switch to Real
> EVs and try to set
> up the model by hand. It seems that there are too many Real EVs (9
> FIRs x 9 EVs) for the
> GUI to handle. Even using a paste window, it just sits and hangs for
> nearly an hour.
>
> We've tried this on a Quad-core Mac with 8GB of RAM and an 8-core
> Mac with 16GB of
> RAM, so I don't think that resources are the issue. I'd appreciate
> any suggestions about
> how to make it easier to use different basis sets.
>
> Thanks,
> --Greg
>
>
> On Thu, 21 May 2009 09:12:29 +0100, Steve Smith
> <[log in to unmask]> wrote:
>
>> Hi - if you want a 'mixed' design in this way I presume that the best
>> solution is just to change the setup contrasts option from "Original
>> EVs" to "Real EVs" and hand-setup the contrasts for every single
>> basis-
>> set EV?
>>
>> Cheers.
>>
>>
>> On 20 May 2009, at 19:33, Greg Burgess wrote:
>>
>>> Hello FSL listers,
>>>
>>> I'm trying to run a lower-level state-item analysis. In these
>>> analyses, you include both
>>> blocked and event-related regressors simultaneously in your model.
>>> One constraint is that
>>> your event-related regressors need to be modeled with an FIR basis
>>> set and your blocked
>>> regressors need to be modeled with an assumed response shape such as
>>> the double-
>>> gamma (c.f. Visscher et al. 2003 for the rationale behind these
>>> constraints). The
>>> contrasts will compare the event-related EVs to other event-related
>>> EVs, and the blocked
>>> EVs to other blocked EVs.
>>>
>>> However, when I try to set up the EVs in the full model setup, I get
>>> the following error: In
>>> order to setup contrasts in "Original EVs" mode whilst using basis
>>> functions or sinusoidal
>>> harmonics, all the original EVs must be the same basic shape and
>>> generate the same
>>> number of real EVs. Please change the EV setup.
>>>
>>> If I attempt to create the model using Real EVs, the GUI becomes
>>> quickly overwhelmed
>>> (i.e., unusable / freezes) trying to create a matrix of (9 event-
>>> related EVs x 9 FIRs) +
>>> three blocked EVs. Even if it didn't freeze, it's just a pain to set
>>> up that many contrasts
>>> rather than just using Original EVs.
>>>
>>> Is there a way to force feat_gui to allow me to use different basis
>>> functions for the
>>> different EVs, as long as I contrast only EVs that have the same
>>> basis functions?
>>> Otherwise, is there a relatively easy way to combine two separate
>>> design.fsf's into one
>>> design.fsf that will solve for the blocked and event-related Evs
>>> simultaneously?
>>>
>>> Thanks,
>>> --Greg
>>>
>>> ____________________________________________________________________
>>> Greg Burgess, Ph.D.
>>> Research Associate, Institute of Cognitive Science
>>> University of Colorado - Boulder
>>> Email: [log in to unmask]
>>>
>>
>>
>> ---------------------------------------------------------------------------
>> Stephen M. Smith, Professor of Biomedical Engineering
>> Associate Director, Oxford University FMRIB Centre
>>
>> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>> +44 (0) 1865 222726 (fax 222717)
>> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>> ---------------------------------------------------------------------------
>
|