Re: Study about accuracy of statistical software, incl. Octave

[Jason Riedy]

And Marco Caliari writes:
> The other deficiencies are much harder to fix. I will give a look.

std() could be fixed relatively easily by calling the BLAS's
routines (SNRM2, DNRM2, SCNRM2, DZNRM2) rather than relying on
sqrt (sumsq (...)).  The half drop in precision is a typical
failure mode for implementations that don't scale.  I suspect
that fixing std() may fix corrcoef and help anova.

Or, if you feel like implementing an even more accurate method
(scaling only by powers of the radix), see the pseudocode for
Rtsmsq in
  http://www.eecs.berkeley.edu/~wkahan/ieee754status/why-ieee.pdf
You can detect overflow without the overflow flag; look for Infs.

Note the use of "... comment" in the pseudocode. ;)

Jason