Closed diagiraldo closed 3 years ago
Hi Diana,
I've sent a request to obtain access to the Google Drive data.
There's a chance that #2269 will intrinsically resolve your issue, as the deeper structural changes happening there are very relevant to the case where both t-tests and F-tests are performed. But then again, given it's the all_stats()
function that's failing, maybe not, the changes there are mostly around the GLM classes... I've not yet fully resolved and completed those changes, but you could just try compiling and running it and see what happens. #2260 should not be relevant here as you don't have non-finite values in your data.
Rob
Thanks for the data Diana.
My dismissal of non-finite values was slightly premature: while there are no non-finite values in your input data inside your mask (and hence no warning regarding such was generated at the command-line), the fixel data outside of this mask does contain non-finite values. It is the combination of this fact with the presence of an F-test that exposed the error (using -fonly
actually still has the same error, it just happened to not be detected in release mode when you tried it).
Rob
Describe the bug I am trying to run the
fixelcfestats
to perform hypothesis testing for multiple contrasts + an f-test, but the process terminates abruptly while calculating the basic properties of the default permutation, even when I use the-notest
flag:It should be mentioned I am including two different measures in the testing so I have 2 exchangeability blocks and variance blocks while the design matrix also has 2 blocks, however It works without problem when:
-fonly
flag -> Only F-test-ftests
option -> Only t-testsTo Reproduce
Data I used to reproduce the error: https://drive.google.com/file/d/1U3rIOQ9T8DvVXH3wCNycKuohx4pM_VLD/view?usp=sharing
Platform/Environment/Version
OS: Ubuntu 18.04.4 LTS
MRtrix3 version: 3.0.2-246-g861c004f
Advanced debugging information
This is what happens when debugging: