[Coin-lpsolver] Querying the feasible bounds

[Spencer Fung]

Lou,

The Arc consistency is only for finite domain problem, isn't it? Actually, I
am also interested on the techniques for pre-solving a LP. Do you know any
paper talk about that? 

BTW, I still want to know the procedure in CLP for getting the bounds after
pre-solve.

Thanks.

Spencer

-----Original Message-----
From: coin-lpsolver-bounces at list.coin-or.org
[mailto:coin-lpsolver-bounces at list.coin-or.org] On Behalf Of Lou Hafer
Sent: Saturday, November 05, 2005 2:47 AM
To: coin-lpsolver at list.coin-or.org
Subject: RE: [Coin-lpsolver] Querying the feasible bounds

Spencer,

	To expand a bit on Matt's answer: Essentially, you're asking the
solver
to infer tighter bounds by looking at the interactions between your
constraints.
This is what a presolve does --- analyses the constraint system and modifies
it (usually, simplifies it) so that the inferences are explicit in the
system.

	Over in AI, various techniques that OR people apply in presolve are
used
(often with more persistence) in constraint propagation.  Arc consistency
(generalised as k-consistency) is one such technique.

	You can think of cutting planes as inferring consequences of
integrality.

	Bottom line, though, is that inference is not free.
	
							Lou

_______________________________________________
Coin-lpsolver mailing list
Coin-lpsolver at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-lpsolver