Re: Problems in mathn, rational, complex, matrix

On Sun, 1 Feb 2004, Gavin Sinclair wrote:

> On Sunday, February 1, 2004, 7:02:31 PM, Mathieu wrote:
> > You have to do Matrix[[7.0,6.0],[3.0,9.0]], or the equivalent thing with
> > Rationals. For the determinant to be meaningful, the Matrix must not be
> > filled with Integers.

A number like 1.0 is mathematically an Integer too. However, in a language
like Ruby, the class of an Object is an information that distinguishes 1.0
from 1 (although 1.0==1) and that plays a role in determining the
operation to use. To instruct the Matrix class to use Float#/, you have to
replace your integers of class Integer (eg: 7) by integers of class
Float (eg: 1.0). Likewise if you want it to use Rational#/, you have to
replace them by integers of class Rational (that is, with denominator 1,
Rational(7,1)).

> I'd say that Ruby should do the same thing, however it is implemented.
> It is not Ruby's custom to expose computers' limitations on
> mathematics (e.g. overflow errors).

huh?

Ruby pretty much exposes overflow errors of Float in the same way as C
would do (IEEE standard). It doesn't expose overflow errors on Integers,
_because_ its Integers are arbitrary-precision, but you still can get
problems like out-of-memory errors and insanely long computations.

________________________________________________________________
Mathieu Bouchard                       http://artengine.ca/matju