I have fixed this bug.  Prior to the next build,
there are a couple of workarounds:

   q: */ 12323 12211 12433x   NB. wrong
12323 151819363

   qco=: /:~ @ (1&p: (# , ;@($:&.>"_)@(-.@[#])) ]) @ q:
   qco */ 12323 12211 12433x 
12211 12323 12433

   qco1=: /:~@;@(q:&.>"_)^:_
   qco1 */12323 12211 12433x 
12211 12323 12433



----- Original Message -----
From: Roger Hui <[email protected]>
Date: Friday, June 24, 2011 14:18
Subject: Re: [Jprogramming] possible bug in q:
To: Programming forum <[email protected]>

> Yes, now that you've found it, I see that there's been
> a bug in q: since day 0.  The problem is in J's faulty
> implementation of Cliff Reiter's factoring code.
> http://webbox.lafayette.edu/~reiterc/j/vector/factor_ecj.html
> 
> The specific problem is that when an ancillary 
> factoring routine finds a factor, the implementation
> incorrectly assumes that that factor can not be
> further factored.  I try to fix this shortly.
> 
> Thank you for finding and reporting this error.
> 
> 
> 
> ----- Original Message -----
> From: Andrew Nikitin <[email protected]>
> Date: Friday, June 24, 2011 12:11
> Subject: [Jprogramming] possible bug in q:
> To: J programming <[email protected]>
> 
> > 
> > 
> >    q:12323x*12211x*12433x
> > 12323 151819363
> > 
> >    9!:14 ''
> > j602/2008-03-03/16:45

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

Reply via email to