I was told by a long-time user whose opinions I 
respect that it would be more useful to be able 
compute to some specified precision than to have symbolic manipulations.  So, 
some day, Joey will be
able to say in J:

   y=: 40 x: 2.5
   y
2.5d40
   ^ y
12.18249396070347343807017595116796618318
   1 o. y
0.5984721441039564940518547021861622717035
   ^. y
0.9162907318741550651835272117680110714501

And there will be no degradation in the speed of
computations on ordinary 64 bit IEEE floating 
point numbers.



----- Original Message -----
From: John Randall <[EMAIL PROTECTED]>
Date: Wednesday, December 6, 2006 12:16 pm
Subject: Re: [Jgeneral] Bug? Sin of pi.

> Joey K Tuttle wrote:
> > At 07:18  -0500 2006/12/06, John Randall wrote:
> >>Joey Tuttle writes:
> >>>In any case, I'm happy to sit back and wait for arbitrary
> >>>precision/accuracy facilities to be introduced in J.
> >>
> >>Symbolic languages such as Maple and Mathematica have arbitrary
> >>precision, but you hit a switch to do native floating-point
> >>calculation when you want speed.
> >>
> >
> > Of course, as the case with 123x currently, I presume that a
> > switch to (surely slower) arbitrary precision would continue
> > to be required. That is what I'm happy to wait for...
> 
> Fair enough, but you still pay the price for the slower
> simplification/evaluation model anyway.  That is why Maple and 
> Mathematicaare not used for serious numerical calculation, except 
> as front ends to
> standard libraries like LAPACK.
> 
> Even with arbitrary precision, pi is not representable exactly, 
> since you
> are still limited to rational numbers.  If argument reduction is 
> used in
> calculating sin, it is quite likely that sin(pi) is not 0: you need
> symbolic representation (or luck) for that.


----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to