On Mon, Feb 18, 2013 at 06:20:44AM -0800, Volker Braun wrote: > You are talking about lattice polytopes, right?
Actually no; I build a bunch of plain Polyhedrons to compute some intersections and plot them. Running to catch my bus! > There are three different representations: > * The LatticePolytope class, which is mostly geared towards reflexive > polytopes in small dimensions. It uses a pexpect interface with PALP. It > also has some compile-time limits with number of points and dimension that > make it unsuitable for arbitrary lattice polytopes. > * The PPL-based polytopes over ZZ: Polyhedron([[1,0],[0,1],[0,0]], > base_ring=ZZ). I think this should be eventually the default lattice > polytope class, but it lacks some functionality specific to lattice > polytopes. And its not backwards compatible with LatticePolytope since the > latter interface isn't particularly nice in some cases. > * The LatticePolytope_PPL class that is not public and currently only > in http://trac.sagemath.org/12553. Its rather bare-bones but designed to > be fast enough to be able to iterate through all 4d reflexive polytopes > (~500 million). I've been using it for internal computations for a while > now. > So my suggestion would be to base your stuff on Polyhedron() whether or > not your have lattice polytopes. If you need any particular functionality > thats not available then let me know. > On Monday, February 18, 2013 1:58:53 PM UTC, Nicolas M. Thiery wrote: > > Hi Volker, > > I am building pictures (for representing root systems) having > typically 1000 small polyhedrons of small dimension (2). It's very > nice to be able to express my problem at this level! However, a lot of > time seems to be spent in communications: > > sage: %prun RootSystem(["G",2,1]).coambient_space().plot(bounding_box=3) > 11585493 function calls (11505945 primitive calls) in 95.937 > seconds > > Ordered by: internal time > > ncalls tottime percall cumtime percall filename:lineno(function) > 9756 36.522 0.004 36.522 0.004 {posix.read} > 3252 21.061 0.006 21.061 0.006 {posix.fork} > 53198 7.324 0.000 13.506 0.000 weakref.py:55(__getitem__) > 3887440 6.768 0.000 7.237 0.000 face.py:235(__cmp__) > 13433 2.890 0.000 2.890 0.000 {built-in method poll} > 8657 1.886 0.000 3.290 0.000 weakref.py:79(__setitem__) > > Is this to be expected (i.e. should I try to use less polyhedrons)? > Would you have recommendations? Things are roughly 3 times faster with > the ppl backend, but at this point I can't always have data in QQ. > > I can push my code to the sage-combinat queue tomorrow in case you > would like to experiment with it. > > Thanks! > > Cheers, > Nicolas > -- > Nicolas M. ThiA-A?A 1/2ry "Isil" <nth...@users.sf.net> > http://Nicolas.Thiery.name/ Nicolas -- Nicolas M. Thiéry "Isil" <nthi...@users.sf.net> http://Nicolas.Thiery.name/ -- You received this message because you are subscribed to the Google Groups "sage-combinat-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-combinat-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-combinat-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-combinat-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
