On 19 mai, 11:52, Volker Braun <vbraun.n...@gmail.com> wrote: > On Sunday, May 19, 2013 10:36:18 AM UTC+1, vdelecroix wrote: > > If we want to round down the > > expression "- expr" then we should round up "expr" and then take the > > opposite. > > If you care about errors then you should be evaluating the symbolic > expressions with interval arithmetic. Your proposal is just to do interval > arithmetic and then take the upper/lower bound as desired. That would be > correct, but there should also be a way to evaluate symbolic expressions > *quickly* without doing interval arithmetic.
I care about - the specification and the behavior of numerical_approx - how should be implemented conversion from an exact real constant to a RealField or a RealIntervalField In numerical approximation, it is written "Return a numerical approximation with at least the requested number of bits or digits of precision". It is False and even critical with the example a = (106*pi - 333) / (355 - 113*pi) as even the third digit is False with the option digits=8. On the other hand, I definitely think that a conversion from an exact real number to a RealField or a RealIntervalField must be the perfect answer (both are unsatisfactory) ! @Volker: Do you have examples of a conversion of a symbolic expression to RR with critical speed ? I found that interval arithmetic prevent from getting False answer while it does not affect too much the speed (from a user point of vue). -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
