Dear Ed, dear Yoshiki,
thanks for your replies… I am now able to reproduce your results in
Ruby-Gsl,
(eliminating some typos),
and I’ve learnt that there are some conflicting definitions of singular
value
decompositions, which lead to part of my difficulties:
Ruby-Gsl defines these as M= U S V^tr,
where U can have uneven dimensions, but many authors require U to
be square (and softwares, eg. Mathematica) - then it has the dimensions
of the image space of the linear mapping M defines … this definition
makes
it easier to calculate pseudo-inverses, as it doesn’t lead to dimension
problems
in multiplication…
I still don’t know why I don’t get singular values to higher precision
…
does one have to convert to Float or something ?
Thank you for your time,
Best reagrds,
Axel