Page 1 of 1

HilbertBasis of a Cone<Rational>

Posted: 13 Dec 2011, 09:23
by sebasguts
Hello,

everytime I want to compute a HILBERT_BASIS of a Cone<Rational>, polymake says "WARNING: available properties insufficient to compute 'HILBERT_BASIS'". Which properties are needed to compute a HILBERT_BASIS or is there a way to force the computation and all needed properties.

Greetings,
Sebastian

Re: HilbertBasis of a Cone<Rational>

Posted: 13 Dec 2011, 10:10
by paffenholz
polymake should compute all needed properties by itself, provided there is a way to do so. For HILBERT_BASIS it should be sufficient to provide RAYS/INPUT_RAYS or FACETS/INEQUALITIES. The cone must be pointed.

Can you please provide an example of a cone where the computation fails?

Re: HilbertBasis of a Cone<Rational>

Posted: 13 Dec 2011, 10:18
by sherrmann
I think you need to install either 4ti2 or normaliz to compute a hilbert basis. See http://www.polymake.org/doku.php/external_software.

Re: HilbertBasis of a Cone<Rational>

Posted: 13 Dec 2011, 11:27
by sebasguts
Works pretty fine now. Maybe you want to change your Error message, so one can see that he has to install software. I'm going to install All The Programms now.

Thank you very much for your help.
Sebas

Re: HilbertBasis of a Cone<Rational>

Posted: 13 Dec 2011, 18:01
by gawrilow
Works pretty fine now. Maybe you want to change your Error message, so one can see that he has to install software.
Unfortunately, it's not trivial. polymake works with a set of production rules; some of, being dependent on external software, are automatically disabled if the required software is not installed on your system. Then, for the internal reasoning machinery, it does not make a big difference whether the rules are disabled or your input is not sufficient to apply them. The only thing it can state is that there is no feasible chain of rules which would lead from the given input to the required property. If polymake would report every rule it discards during the reasoning, you would be drowned in spam.

The only hint polymake currently gives to you about this issue is a warning at the very beginning of the session that some external dependences are missing, inviting you to inspect them by a shell command show_unconfigured. If anybody has an idea how to improve the diagnostics, I'll be glad to discuss and, eventually, implement this.