OK, well I have plenty of projects so I won't work too hard on that right now and see what you come up with. I need something that works for a talk I am giving in about a month, but if necessary I can use the "bandaid" code I already have, which works about 98% of the time for the stuff I care about, and I know how to make it a little better. I am happy to review code in this area or discuss it further.
Cheers, Marshall Hampton On Sep 20, 11:35 am, "Arnaud Bergeron" <[EMAIL PROTECTED]> wrote: > 2008/9/20 mhampton <[EMAIL PROTECTED]>: > > > > > > > Hi everybody, > > > This is mainly about making the lrs optional package standard to > > improve some polytope-related calculations. > > > My original motivation for lrs (linear reverse search) is that is a > > very different algorithm/implementation for computing exact convex > > hulls in arbitrary dimensions. Currently the default for this in Sage > > is cddlib, which uses the "double description" method. There are > > classes of polytopes for which lrs is much faster than cddlib, > > although in my experience cddlib does better most of the time. lrs is > > also nice in that it doesn't use much memory. Polymake also uses > > these two methods, plus a third, the "beneath and beyond" method. I > > still need to modify the Polyhedron class to make use of lrs if it is > > present. > > > I am now thinking more about nice renderings of polytopes, and for > > that I need better triangulation code. I have a really bad algorithm > > in polyhedra.py right now that needs to be improved or replaced. I > > see at least three options for this: > > > 1) I could just improve the code I already have. I've actually > > already done this, and I guess I will submit a ticket soon. But I am > > not an expert in this area and I don't think even my improved versions > > would be good enough for serious users. > > > 2) The option I like best in the short term is to make lrs standard > > and use it for computing triangulations. The lrs algorithm computes a > > triangulation anyway, and its probably one of the faster methods > > available for doing so. lrs is small (spkg is 120kb) and compiles > > quickly. Its mature code, its been maintained and improved for about > > 10 years. > > > 3) Eventually it would be good to add TOPCOM. Apparently it is what > > people use who do research that involves triangulations. But this is > > a bigger task than I can take on right now. Its a much larger piece > > of code than lrs, but it would add more functionality. > > > So I can do option (1) very soon, and option (2) in the next month or > > so unless people have objections. > > I am beginning to work on the triangularization code just now, since > it was on the list of requested items when I asked for thing to do in > graphics. It should be ready in about a week if all goes well. > > Arnaud > > > To help, install the optional package and test it by downloading: > >http://www.d.umn.edu/~mhampton/lrs_test.ext > > and run $SAGE_ROOT/local/bin/lrs your_path_to_lrs_test/lrs_test.ext > > where of course you replace "your_path_to_lrs_test" with the path to > > the file. I've done this on a couple of intel macs (10.4 and 10.5), I > > don't expect any problems on linux and I have no idea about Solaris. > > > Cheers, > > Marshall Hampton > > -- > La brigade SnW veut vous recruter -http://brigade.snw.googlepages.com --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
