Page 1 of 1

False results when computing bases/ranks on minors

Posted: 01 Feb 2012, 15:39
by hampe
I sometimes get false results when I compute linear algebra stuff like basis_cols, rank, etc on matrix minors. This does not happen for all possible input, but I managed to find a small example that reliably produces a wrong result on my machine (polymake 2.10) as well as on the web version of polymake (which says it's 2.11):

Code: Select all

$m = new Matrix<Integer>([[0,0,1,0],[1,1,1,1],[0,0,0,1]]); $i = new IncidenceMatrix([[0,2]]); print basis_cols($m->minor($i->row(0),All)); # ->produces {2} instead of {2,3}
Similarly, if I write rank($m->minor...), I get 1 instead of 2. This doesn't change if I replace Integer by Rational and it happens on the perl side as well as on the c++ side (using similar code). If I do basis_cols(new Matrix(...)), the result is correct.

Re: False results when computing bases/ranks on minors

Posted: 01 Feb 2012, 17:15
by gawrilow
I'll look into it.

Generally, you'd better post such bug reports directly as Trac tickets. The Trac system shares the user credentials with the SVN repository.

Re: False results when computing bases/ranks on minors

Posted: 01 Feb 2012, 17:36
by hampe
No problem - how do I do that? :D I have never used trac before. I can probably look up some general tutorials on trac on the internet, but everything I found so far involved elaborate setups of databases etc... is there a simpler way to use this? (Maybe even a GUI frontend? *hope*) :)
If not, maybe I can find someone at the workshop who will show me how to use it... :-)

Re: False results when computing bases/ranks on minors

Posted: 01 Feb 2012, 18:00
by gawrilow
Nothing simpler than that: you just log in under http://www.polymake.org/trac/polymake and click on "New Ticket".

Re: False results when computing bases/ranks on minors

Posted: 01 Feb 2012, 18:17
by hampe
oh wow ... that IS easy :D Thank you!