On Mon, Sep 5, 2011 at 8:44 AM, Dima Pasechnik <dimp...@gmail.com> wrote:
>
>
> On Monday, 5 September 2011 22:41:50 UTC+8, rjf wrote:
>>
>> is 4^(-2) (use various kinds of integers) integer rational float?
>> ditto for
>> 5^(-2) ?
>>
>> Seems to me that the presence of python integers is an inconsistency
>> waiting to appear,
>
> as brilliantly uncovered by our 2nd year undergrads :)
> With the patch I propose one has
> sage: type(int(5)^-2)
> <type 'sage.rings.rational.Rational'>
> sage: type(int(5)^int(-2))
> <type 'float'>
> with the idea that if you want to stay Pythonic, you can (the 2nd example),
> but otherwise you get promoted to the better situation (the 1st example),
> when everything is exact.
> Presently, int()^whatever stays Pythonic, no matter it causes a loss of
> precision, or not.
> One gets bizzare things like:
> sage: def f(x): return x^-3
> ....:
> sage: f(int(3))
> 0.037037037037037035
> sage: var('x')
> x
> sage: p(x)=x^-3
> sage: p(int(3))
> 1/27
> i.e. symbolic functions and the procedures behave drastically differently,
> etc...
This is what Sage *should* do, IMHO.
> Certain purists argue that I am wrong, and that the loss of precision must
> be preferred...
These "purists" are usually actually "applied-ists", and for numerical
computation they are often "right". When I'm doing lots of numerical
computation (which I do actually do sometimes), I tend to agree, and
set the relevant modes in Sage.
That said, for consistency with the rest of Sage, e.g., that int(1) +
Integer(2) is an Integer, we should have int(3)^Integer(2) be an
Integer.
>
>>
>> and
>> the only proper use of python ints is as a sage integer which happens
>> to be, at the moment,
>> small in magnitude.
>> Lisp does this right, with fixnums automatically promoted to bignums
>> if they get too big.
>> Coercion from rational to integer happens if the rational has a
>> denominator of 1.
>> There are lots more of these, e.g complex with imag part zero etc.
>> Do you want
>> these all to be equal....
>> 1+0*i, 1/1, 1, int(1)? 1.0 1.0d0? maybe.
>> RJF
>>
>> On Sep 5, 6:52 am, Dima Pasechnik <dim...@gmail.com> wrote:
>> > this is now #11779
>
> --
> 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
>
--
William Stein
Professor of Mathematics
University of Washington
http://wstein.org
--
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
