On Wed, Jun 13, 2012 at 9:43 PM, rjf <fate...@gmail.com> wrote: > > > oh, just a note on precision in Maxima with bigfloats. > > If you have 2 numbers of precision N and M, and you add them together, the > resulting number will be of precision K where K is the global value of the > specified precision. this is fpprec [in decimal, approximately] or > ?fpprec > in binary. > > Thus the inherent precision of some number as input is maybe less important > than > the global setting.
This last statement is *precisely* the problem. It doesn't "preserve the precision" of either input (e.g. by taking the min of the input precisions). Perhaps we could set ffprec to be the min of the input precisions (plus a default, plus anything still in scope?) before doing any operations. This is still an inferior solution (e.g. when summing a series whose terms are computed to the same absolute precision but have wildly varying relative precisions which does come up in computing L-functions or even generalizations of the hypergeometric functions). - Robert -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
