Hi all.

I just implemented a new method that performs the "strong" minimization of the systems of generators and constraints of an NNC_Polyhedron. This is based on a couple of rather simple observations that Elisa and I pointed out some time ago.

The method is, for the moment, public and is named NNC_minimize(). I also added a few tests (named NNCminimize1-2-3) that seem to confirm the correctness of the approach.

Surprisingly, my prototype implementation requires NO scalar product besides those computed in the standard minimization process (obviously, it does require some further adjacency tests, which are based on the already computed sat_c matrix).

If the approach passes some more involved test, then we could start exploiting it where needed. In this sense, I still have to investigate whether or not the approach is applicable when we do NOT have some of the matrices minimized (e.g., what happens if con_sys is up-to-date but NOT minimized ?)

ciao, Enea.