On 2013-02-22, Henry Baker <hbaker1 at pipeline.com> wrote:
> Speaking of least squares, would it be possible to utilize
> a least squares approach within "linsolve" and "solve" to
> handle overdetermined systems with _too many equations_.
> I suppose I'm only talking about systems with _numeric_
> constant coeffients rather than variable coefficients.
IIRC there is an SVD algorithm in the lapack package, named dgesvd or
something like that.
best
Robert Dodier