>> Question: Should charpoly automatically set ratmx to true?
> Sometimes with ratmx=true, charpoly is slower:
> ... I think it is better to leave the decision to the user.
Can we characterize (that seems like the right word here!) the cases
where ratmx is likely to be better or worse for a given matrix
calculation without actually carrying out the calculation? If so, the
default value of ratmx should be auto, and the matrix package should
choose on its own.... Presumably there will be cases where auto will
not choose optimally, so true and false should remain as options.
-s