In this particular case the problem can probably be resolved by being a bit stricter with the check in interior_contains:
Instead of checking return self.polyhedron()._is_positive( self.eval(Vobj) ) one should have return self.polyhedron()._is_positive( self.eval(Vobj) ) and not self.polyhedron()._is_zero( self.eval(Vobj) ) Am Dienstag, 13. Oktober 2015 09:20:10 UTC+2 schrieb jplab: > > Hi everyone, > > Currently, I am working on the ticket: > > http://trac.sagemath.org/ticket/18442 > > which implements a barycentric subdivision of a polytope. For this I use > the containment check. In the tests, it works fine for a regular pentagon > over AA but not over RDF. The problem can be seen with the following code: > > sage: P = polytopes.regular_polygon(5) > sage: a_vertex = P.vertices()[0] > sage: for facet in P.Hrepresentation(): print facet.contains(a_vertex), > facet.interior_contains(a_vertex) > > True False > True True > True False > True True > True True > sage: P = polytopes.regular_polygon(5, base_ring=RDF) > sage: a_vertex = P.vertices()[0] > sage: for facet in P.Hrepresentation(): print facet.contains(a_vertex), > facet.interior_contains(a_vertex) > True True > True True > True True > True True > True True > > (The first output is as expected: the vertex is not contained in the > interior of exactly 2 facets. This is not the case in RDF) > > Right now, I have the following question in mind: Does it make sense to > have polyhedron over RDF? By which I mean: it can not even deal with > containment of faces properly. Reasonably, the user knows that and will use > a different ring... But what if? > > I know that it is practical to have polyhedron defined over RDF for > different reasons, but there is a fundamental change in the results (and > pretty much any method doing computations) if the ring changes. I'm NOT > asking to remove RDF as a possible ring. The reason I'm asking is that I > have to use such containment check to implement the method barycentric > subdivision for polyhedron, but then it breaks for polyhedron defined on > RDF because of that. So there are methods that simply do not work when the > ring is RDF. > > Maybe this analogy could explain my thoughts: > > If we compute the eigenvalues of a matrix defined over RDF, sage warns us > that the results may go bad. It doesn't if the matrix is rational... Could > such a warning be issued when working with RDF? And then put NotImplemented > whenever it is known that RDF could cause trouble (and then fix it for that > specific ring?). > > I could simply set the barycentric subdivision for polyhedron over RDF not > to be implemented, but I feel this would not answer the problem raised > anyway. > > Thanks! > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
