On Tuesday, 6 September 2011 11:08:02 UTC+8, leif wrote: > > On 5 Sep., 17:44, Dima Pasechnik <dim...@gmail.com> wrote: > > Presently, int()^whatever stays Pythonic, no matter it causes a loss of > > precision, or not. > > Not true: > > sage: 3r^ZZ(-3) > 0.037037037037037035 > sage: 3r^QQ(-3) > 1/27 >
what is "Not true"? Surely conversion into float is a loss of precision. No, really, the current behaviour of int()^-k cannot be defended. If one thinks of int()^-k as 1/int()^k, and this is, mathematically, the usual definition of a negative power, it must be an Integer, even if you hold that int()^k is int; this is the way Sage division works. (unless you want to stay Pythonic, and get int(1)/int()^k, i.e. 0 in Python 2, which would be even more stupid than the current conversion into float) Now, as int()^-k should to be Integer, so should be int()^k, too. > > I'm strongly against making the result of exponentiation (of e.g. > Python ints) depend on the exponent's *type*; it should only depend on > its *value*, in this case yielding either Python ints or floats in > turn. > > Exponentiation isn't commutative (nor associative), in contrast to > multiplication and addition, where a factor / summand of type Integer > makes the product / sum be of Sage's Integer type. > > > -leif > -- 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
