In reference to the question I recently sent about getting a NaN result
for a corrected cluster statistic:
> I am searching a very small area of about 43 voxels, and the corrected
>cluster statistic is reported as NaN, with a divide by 0 error. I first
>thought that this was because I was looking at a sub-resel size, but other
>analyses I've done have reported 0 resels and still given me a corrected
>cluster statistic.
I thought that it might be helpful for folks to see the error code. I'm
trying to get a grant together, and a rapid response would be really
helpful if anyone has an idea!! It was:
Warning: Divide by zero.
> In C:\Program Files\spm99\spm_P.m at line 85
In C:\Program Files\spm99\spm_list.m at line 401
Warning: Returning NaN for out of range arguments
> In C:\Program Files\spm99\spm_Pcdf.m at line 84
In C:\Program Files\spm99\spm_P.m at line 102
In C:\Program Files\spm99\spm_list.m at line 401
Line 85 is the one containing the [beta = statement]
elseif STAT == 'T'
beta = (gamma(D/2 + 1)/En)^(2/D);
p = exp(-beta*(k^(2/D)));
|