Hi Remi,
> In SPM2, for the coregistration algorithm, why is spm_powell trying to
> find a minimum of the evaluated function, while according to the theory
> coregistration is performed when mutual information is at maximum?
algorithms for function "optimisation" tend to search for the minimum,
though this is really nothing more than a convention. When in reality
one wants to maximise the function one simply minimises the negation of
the function (i.e. -MI).
Good luck Jesper
|