Barton Willis wrote:
...
> Not so good -- this is with 5.14.0. Is there another option variable
> that controls the accuracy of float solutions when algexact : false?
The function $algsys rounds the results by calling 'roundroots1 which in
turn calls 'maxima-rationalize. The variable '$ratepsilon is being bound
to 1.0e-7 by $algsys so can not be controlled. Perhaps this rounding is
intended to help detect equivalent solutions in the presence of numerical
inaccuracy, and if so then perhaps a value of '2 * flonum-epsilon' would
be more appropriate, or perhaps '$algdelta?
Regards
Douglas Crosher