removed bug and new methods in Poset

[stumpc5]

Poset.dual contained a bug:

P = Poset(([1,2],[[1,2]]))
Q = P.dual()
Q.show()
... AssertionError: `self` should be an acyclic graph

The patch contains the solution for this bug as well as new methods
is_lattice
is_isomorphic
is_selfdual
panyushev_complement
panyushev_orbits

CC: @saliola @fchapoton

Component: combinatorics

Keywords: poset, days30

Reviewer: Franco Saliola

Issue created by migration from https://trac.sagemath.org/ticket/10065

[fchapoton]

comment:2

The procedure is_isomorphic has two problems :

• one doctest is failing (see the report of the build bot). This is related to a deprecated use of size instead of cardinality, somewhere. Maybe we should correct this deprecated use first.

• it fails on Lattices

P=Posets.PentagonPoset()
P.is_isomorphic(P)
ValueError: The input is not a finite poset.

Maybe this can be corrected by having a method is_isomorphic for Lattices which is inherited from Posets.

[fchapoton]

comment:3

One more thing to be corrected :

The doc of !panyushev_orbits! is badly formatted. It does not display correctly in the notebook.

[stumpc5]

comment:4

Replying to @fchapoton:

• one doctest is failing (see the report of the build bot). This is related to a deprecated use of size instead of cardinality, somewhere. Maybe we should correct this deprecated use first.

I only updated the doctest - the size/cardinality is in generic_graph.py. I am in the process to make changes to the is_isomorphic method in there anyway. I will make this little change in that other patch.

• it fails on Lattices

P=Posets.PentagonPoset()
P.is_isomorphic(P)
ValueError: The input is not a finite poset.

Maybe this can be corrected by having a method is_isomorphic for Lattices which is inherited from Posets.

This was just a minor mistake in the is_isomorphic method for Posets. I updated it.

The doc of panyushev_orbits is badly formatted. It does not display correctly in the notebook.

Done...

[stumpc5]

comment:5

I uploaded a changed version of the patch containing a faster way to compute the Panyushev complement. Beside this, I didn't do any changes.

[fchapoton]

comment:6

This last change makes some tests fail. See the shortlog by the buildbot.

[stumpc5]

comment:7

Sorry Fred, I often forget to rebuild before making the test... if I didn't do more mistakes, it should work now.

@Franco: I also had to edit the patches

trac_6641-poset_antichains_backtracker.patch
addon_poset_antichains_backtracker-cs.patch

(and uploaded them to the combinat queue) for ticket #6641. Could you probably fold them and add them to your ticket? Thanks!

[stumpc5]

Attachment: trac_10065_smaller_poset_improvements-cs.patch.gz

[saliola]

comment:8

These issues are addressed in #10998.

[saliola]

Changed keywords from poset to poset, days30

[sagetrac-mariah]

comment:10

If this is a sage-duplicate, then I believe this ticket can be closed.