Ð81Ð
¥ See  http://afni.nimh.nih.gov/doc/misc/3dDeconvolveSummer2004/
¥ Equation solver: Gaussian elimination to compute R matrix pseudo-inverse was replaced by SVD (like principal components)
H Advantage: smaller sensitivity to computational errors
H ÒCondition numberÓ and Òinverse errorÓ values are printed at program startup, as measures of accuracy of pseudo-inverse
H Condition number < 1000 is good
H Inverse error < 1.0e-10 is good
¥ 3dDeconvolve_f program can be used to compute in single precision (7 decimal places) rather than double precision (16)
H For better speed, but with lower numerical accuracy
H Best to do at least one run both ways to check if results differ significantly  (SVD solver should be safe)
Upgrades Since July 2004