[gecode-users] propagation consistency of count (single value) constraint

[Christian Schulte]

Hi,

It is better than value consistency and bounds consistency but for some
special cases not domain consistent.

Thanks for the error report, I fixed them.

Cheers
Christian

--
Christian Schulte, www.gecode.org/~schulte
Professor of Computer Science, KTH, cschulte at kth.se
Expert Researcher, SICS, cschulte at sics.se


-----Original Message-----
From: users-bounces at gecode.org [mailto:users-bounces at gecode.org] On Behalf
Of Kish Shen
Sent: Wednesday, August 26, 2015 10:21 PM
To: users at gecode.org
Subject: [gecode-users] propagation consistency of count (single value)
constraint

I noticed an inconsistency in my (ECLiPSe) documentation for the count (for
single values) constraint in the Gecode interface, and the current Gecode
reference documentation for these constraints:

count(home, x, y, INT_EQ, z)

this is for IntVarArg x, while y and z can be int or IntVar

My documentation states that the constraint can be posted with ICL_DOM.
but this seems to be inconstent with the current reference documentation.

Looking at Gecode's reference documentation, it seems that depending on if y
and z are int or IntVar, either the constraint is propagated at ICL_DEF only
(not achieving domain consistency), or at two different consistency levels:

ICL_DOM, which does stronger propagation (but not domain consistent), and
less propagation with all other IntConLev.

So if I understand the reference documentation correctly, ICL_DOM is used
only to distinguish it from an alternative that perform less propagation
(where the alternative exist, otherwise there is only one level of
propagation). How does the propagation achived compared to bound and value
consistencies?

I also noticed a minor typo in the reference documentation: in at least two
places, domain is spelt as comain, as in 'propagation is not comain
consistent'.

Cheers,

Kish





_______________________________________________
Gecode users mailing list
users at gecode.org
https://www.gecode.org/mailman/listinfo/gecode-users