[PPL-devel] Opposing inequality to equality conversion.

[Roberto Bagnara]

Hello Axel.

Axel Simon wrote:
> I'm trying to verify a claim in Halbwachs' paper on factoring polyhedra 
> ("Some ways to reduce the space dimension in polyhedra computations" is 
> the journal version). Specifically, he claims that omitting a common 
> equality relation from the convex hull computation yields not 
> significant speedup. I've tried to run a convex hull calculation, once 
> with the common equalities added and once with the common equalities 
> omitted and found quite a difference in performance.

The difference is in favor of which technique?  (This is probably implicit
in what you write afterward, but I would like to be sure.)

> However, I added 
> the equality as two opposing inequalities. My question: are the two 
> opposing inequalities not immediately turned into one equality by the 
> Constraint class? If not, then this might explain the speed difference 
> I'm seeing.

If I understand correctly, you are asking whether the Constraint_System
class immediately turns two opposing inequalities into the corresponding
equality.  The answer is negative.  This kind of simplification only
happens (lazily) as part of the conversion/minimization process.
Please do not hesitate to come back to us in case I misunderstood
the question.
All the best,

     Roberto

-- 
Prof. Roberto Bagnara
Computer Science Group
Department of Mathematics, University of Parma, Italy
http://www.cs.unipr.it/~bagnara/
mailto:bagnara at cs.unipr.it