next up previous
Next: iNFFT - Plan Up: Library Previous: iNFFT - General procedure

iNFFT - Parameter

The inverse nfft basically wraps an already initialised direct nfft (direct_plan), i.e. one only specifies the type of iteration by setting one of
LANDWEBER, STEEPEST_DESCENT, CGNR_E, CGNE_R.
in infft_flags. The additional flag NORMS_FOR_LANDWEBER, only applicaple if LANDWEBER is set, causes the library to compute the residuals $ \Vert$$ {r}$$ _l\Vert _{\mbox{\boldmath\scriptsize {${W}$}}}^2$ and $ \Vert$$ {\hat z}$$ _l\Vert _{\mbox{\boldmath\scriptsize {${\hat W}$}}}^2$; ITERATE_2nd, only applicaple if CGNR_E or CGNR_E is set, computes the cgne/cgnr-iterate out of the cgnr/cgne-algorithm, see [14] for details. Weights and damping factors are used if the flags
PRECOMPUTE_WEIGHT, PRECOMPUTE_DAMP
are specified, respectively. One has to initialise the members my_iplan.w, my_iplan.w_hat in these cases. The utility library, see lib/utils.c, provides a set of functions for computing voronoi weights and certain damping factors, see Subsection 4.3. The default for infft_flags is CGNR_E.



Stefan Kunis 2004-09-03