RFC on FD.reified.sumCN

[Christian Schulte]

Well, I think Tobias fix was quite accurate in the sense that propagation
has become correct. Furthermore, implemnting true nonlinear reified
equations is quite painful while it usefulness is really limited. For me it
is not so clear that doing anything better than that is worth the effort.
Another option is to go for expansion in the first place and save all the
trouble.

Regards
Christian

--
Christian Schulte, http://www.imit.kth.se/~schulte/ 

-----Original Message-----
From: mozart-hackers-bounces at ps.uni-sb.de
[mailto:mozart-hackers-bounces at ps.uni-sb.de] On Behalf Of Raphael Collet
Sent: Tuesday, February 08, 2005 9:00 AM
To: hackers at mozart-oz.org
Subject: RFC on FD.reified.sumCN


Hello guys,

I need some explanation on the builtin that posts the propagator for
FD.reified.sumCN.

I recently observed a problem with FD.reified.sumCN (the propagator for
reified polynomial equations).  See bug report #1509 for the description of
the problem.  I have taken a look at the implementation, and found something
strange the builtin fdp_sumCNR:

OZ_BI_define(fdp_sumCNR, 5, 0)
{
  OZ_EXPECTED_TYPE(OZ_EM_VECT OZ_EM_INT "," OZ_EM_VECT OZ_EM_VECT OZ_EM_FD 
                   "," OZ_EM_LIT "," OZ_EM_FD "," OZ_EM_FDBOOL);

  PropagatorExpect pe;

  OZ_EXPECT(pe, 2, expectLiteral);
  sum_ops op = getSumOps(OZ_in(2));

  // wait for linearity
  OZ_EXPECT(pe, 1, expectVectorLinearVector);

  ...
}

This "wait for linearity" looks absurd.  The builtin actually blocks until
the equation becomes linear.  This makes the propagator almost useless in
practice!  This line has been added by Tobias Mueller in revision 1.23 of
card.cc, which happened on 2000/03/27.  The log file
says: "fixed bug in nonlin sum".  Doesn't help.

I'd like to fix this.  Can someone tell me the reason for this statement?
Does it avoid a bug (in which case the bug is *not* fixed)?


Thanks in advance,

raph
____________________________________________________________________________
_____
mozart-hackers mailing list
mozart-hackers at ps.uni-sb.de      
http://www.mozart-oz.org/mailman/listinfo/mozart-hackers