Dear list,
I was exploring the SPM m-files and I noticed that in least squares
estimation the pseudoinverse is used. The thing I was wondering is why the
code does not use the Matlab function pinv for that but calculates the
pseudoinverse with singular value decomposition. Isn't that what the pinv
does also?
So, why make own function for that? How does that (spm_sp('x-',X)) differs
from the function pinv and why?
|