Search found 282 matches

Go to advanced search

by gawrilow
11 Dec 2019, 10:48
Forum: Helpdesk
Topic: From V-representation to H-representation
Replies: 1
Views: 31

Re: From V-representation to H-representation

INEQUALITIES is a pure input property, it's what you can pass to the Polytope or Cone object constructor. It can have duplicate and redundant inequalities, and hidden equations. FACETS is what you get out of the convex hull solver - this is the "canonical" H-representation, free of redundancies and ...
by gawrilow
10 Dec 2019, 21:23
Forum: Helpdesk
Topic: Error while using nmz
Replies: 3
Views: 44

Re: Error while using nmz

'nmz' was just an abbreviation used in the paper, probably because 'normaliz' would have been too broad on some pictures. The label to be used in the 'prefer' and 'prefer_now' commands is spelled 'libnormaliz'. You can figure this out e.g. by tab completion. Probably we should also offer a command l...
by gawrilow
10 Dec 2019, 17:10
Forum: Helpdesk
Topic: Error while using nmz
Replies: 3
Views: 44

Re: Error while using nmz

It's not an error, just a warning. Somewhere in your file ~/.polymake/prefer.pl must be a statement looking like 'prefer nmz;' This probably comes from some extension you have been using before. Currently, there is no such label in polymake core rulebase. Therefore the warning. If you don't know (an...
by gawrilow
28 Nov 2019, 16:25
Forum: Helpdesk
Topic: Error in using lp2poly
Replies: 21
Views: 378

Re: Error in using lp2poly

TOSimplex can return all feasible solutions for the integer variables of a MILP, but this feature is missing in the interface to polymake
You are cordially invited to extend the interface 8-)
by gawrilow
27 Nov 2019, 13:52
Forum: Helpdesk
Topic: Error in using lp2poly
Replies: 21
Views: 378

Re: Error in using lp2poly

For every eliminated coordinate a Fourier-Motzkin elimination step is executed, which can potentially double the number of inequalities. You are trying to eliminate 90-2=88 coordinates... Conversion to Float is not an option, since it only can make your polyhedron infeasible, as you have experienced...
by gawrilow
27 Nov 2019, 12:03
Forum: Helpdesk
Topic: Error in using lp2poly
Replies: 21
Views: 378

Re: Error in using lp2poly

If you enter the lines one by one in the interactive shell, you'll see at which step it hangs.

You don't need to enter the first two lines, starting with "use".
by gawrilow
27 Nov 2019, 11:03
Forum: Helpdesk
Topic: Error in using lp2poly
Replies: 21
Views: 378

Re: Error in using lp2poly

poly2lp produces a Polytope with an LP subobject, which is by definition a linear program without any integrality constraints. The list of INTEGER_VARIABLES is merely attached to it as a kind of comment. Since recently, polymake can also solve MILPs, but they have to be created as MILP subobjects. p...
by gawrilow
26 Nov 2019, 21:51
Forum: Helpdesk
Topic: Computing the coordinates of the Vertices_in_facets ?
Replies: 2
Views: 57

Re: Computing the coordinates of the Vertices_in_facets ?

Numbers in VERTICES_IN_FACETS refer to the rows of VERTICES matrix.
You can do something like this:

Code: Select all

# $p is the Polytope object for my $i (0 .. $p->N_FACETS-1) { print "facet $i contains vertices:\n", $p->VERTICES->minor($p->VERTICES_IN_FACETS->row($i), All), "\n"; }
by gawrilow
26 Nov 2019, 20:57
Forum: Helpdesk
Topic: Error in using lp2poly
Replies: 21
Views: 378

Re: Error in using lp2poly

If you leave the polyhedron with floating-point coordinates: `$p=lp2poly<Float>("input file");` it is reported as feasible by the cdd solver, and its linear span vectors contain some very adventurous coordinates like -1.110223025e-16. The solution area of your problem seems to be defined by the intr...
by gawrilow
26 Nov 2019, 20:48
Forum: Helpdesk
Topic: Error in using lp2poly
Replies: 21
Views: 378

Re: Error in using lp2poly

Well, the error message deserves an improvement, but the fact is that your polyhedron seems to become infeasible after a conversion to rational coordinates. There are two suspicious equations in your initial LP with floating-point fractional RHS; it could be that CPLEX still finds some "almost feasi...

Go to advanced search

cron