This directory has various system models for tomographic image reconstruction
and related problems.

@fatrix2
@Fatrix
This object provides a general "container" for linear operations A*x and A'*y.
Fatrix is the obsolete version; fatrix2 is the current version.

@Gblock
block system models (based on one of Gtomo2_dsc, Gtomo2_sparse, Gtomo2_wtfmex)
for ordered-subsets (aka block iterative) algorithms.  this is largely obsolete;
the newer @Fatrix object includes this functionality

@Gtomo2_dsc
on-the-fly forward or backprojection using a .dsc file (see ASPIRE user manual)
the system matrix is not precomputed so this uses very little memory.

@Gtomo2_wtfmex
uses .wtf created by ASPIRE - loaded in mex memory space rather than matlab
to save memory (singles instead of doubles).  for large problem sizes even
this may use too much memory in which case try Gtomo2_dsc instead.


Gblur
linear shift invariant blur (image restoration)

Gdsft
discrete-space Fourier transform (nonuniform k-space samples).
a slow but exact version of Gnufft.

Glinear
tomographic system matrix based on pixel-driven linear interpolation.
not recommended.

Gmri
For MRI image reconstruction, possibly including nonuniform k-space samples,
off-resonance effects, and relaxation effects.

Gnearest
tomographic system matrix based on pixel-driven nearest-neighbor interpolation.
not recommended.

Gnufft
Nonuniform FFT (NUFFT) for applications like MRI that record Fourier samples.

Gsparse
an object made from matlab sparse matrix, essentially for testing.
also useful as a base for Gblock objects

Gtomo2_strip
strip integral tomographic system model.  useful but uses lots of memory.

Gtomo2_table
table-based tomographic system model.  fast yet small memory.  recommended.

Gtomo3
3D system models, some precomputed, some on-the-fly, based on f3d_mex.

Gtomo_blob
work in progress

Gtomo_nufft
NUFFT-based forward/backprojection for 2D tomography

Final option:
	!wt gen tomo.dsc
	(creates tomo.wtf using ASPIRE)
	A = wtfmex('load', 'file.wtf');