backend_polymake: Work around polymake bug with zero inequalities over quadratic extensions

[mkoeppe]

Following up on #22683:

Easy workaround for a segfault, as discussed and suggested at
https://forum.polymake.org/viewtopic.php?f=8&t=547

See also: #22710: Meta-ticket: polymake

Depends on #22683

CC: @tscrim

Component: geometry

Keywords: polymake

Author: Matthias Koeppe

Branch/Commit: 93af258

Reviewer: Travis Scrimshaw

Issue created by migration from https://trac.sagemath.org/ticket/22723

[mkoeppe]

Branch: u/mkoeppe/backend_polymake__work_around_polymake_bug_with_zero_inequalities_over_quadratic_extensions

[mkoeppe]

Commit: 4070c66

[mkoeppe]

Dependencies: #22683

[mkoeppe]

Last 10 new commits:

ef23af4	Polyhedron_base._polymake_init_: Add test for RDF polyhedra
205879f	Matrix, MatrixSpace: Add coercion to polymake interface
e49765e	Polymake: Fix tests whose error messages changed because we do not use files
51af468	PolymakeElement.__iter__: New
2f0e059	PolymakeElement._sage_: Implement for *Vector, *Matrix, QuadraticExtension
218daca	Add Polyhedron_polymake
5ab94a2	Merge tag '7.6' into t/22683/backend_polymake_for_polyhedron
8fd81c7	Polyhedron_polymake._polymake_: Add doctest
952b860	PolymakeElement.__iter__: Doc fixes
4070c66	Polyhedron_polymake: Work around polymake bug with zero inequalities over quadratic extensions

[mkoeppe]

Description changed:

--- 
+++ 
@@ -1,4 +1,6 @@
+Following up on #22683:
+
 Easy workaround for a segfault, as discussed and suggested at
 https://forum.polymake.org/viewtopic.php?f=8&t=547
 
-
+See also: #22710: Meta-ticket: polymake

[mkoeppe]

Author: Matthias Koeppe

[tscrim]

comment:6

Your check doesn't result in any speedup when the list passes as it has to check every element and has to double check when it fails. I think you're better off just running the filter directly:

ieqs = [ v for v in list if not all(self._is_zero(x) for x in v) ]

The other option would be a more hybrid approach like this:

for i,v in enumerate(ieqs):
    if all(self._is_zero(x) for x in v):
        ieqs = ieqs[:i] + [v for v in ieqs[i+1:] if all(self._is_zero(x) for x in v))
        break

Although I think always filtering is best because it is the simplest code-wise.

[sagetrac-git]

Changed commit from 4070c66 to 93af258

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. Last 10 new commits:

8812e9c	Merge tag '7.6' into t/22658/polyhedron_methods_using_the_polymake_pexpect_interface
b47f6f9	Polyhedron_base._polymake_init_: Add work around for empty polytopes with Polymake 3.0
3b49a54	Polymake: Fix another test whose error message depends on whether we use files
07a1757	Polyhedron_base._polymake_init_: Fix comment
09942e7	coercion tutorial: Update introspection results
19d05f1	Add more 'optional - polymake
cd583e2	Remove one unneccessary 'optional - polymake
7bcc04c	Merge tag '8.0.beta0' into t/22658/polyhedron_methods_using_the_polymake_pexpect_interface
d7d4234	Add '# optional - polymake'
93af258	Merge branch 't/22658/polyhedron_methods_using_the_polymake_pexpect_interface' into t/22723/backend_polymake__work_around_polymake_bug_with_zero_inequalities_over_quadratic_extensions

[mkoeppe]

comment:8

Thanks. I agree. I've simplified the code.
Also merged the updated #22658.

[tscrim]

comment:9

LGTM. Thanks.

[tscrim]

Reviewer: Travis Scrimshaw

[vbraun]

Changed branch from u/mkoeppe/backend_polymake__work_around_polymake_bug_with_zero_inequalities_over_quadratic_extensions to 93af258