Bug in order_ideal_complement_generators: 'down'

[nthiery]

The down option is broken in order_ideal_complement_generators due to
a glitch:

    sage: P = Poset( ( [1,2,3], [ [1,3], [2,3] ] ) )
    sage: sage: P.order_ideal_complement_generators([1,2], direction='up')
    set([3])
    sage: P.order_ideal_complement_generators([1,2], direction='down')
    set([3])

The result should be [] in the later case.

Upcoming patch on the Sage-Combinat queue

Apply:

attachment: trac_12848-posets-order_ideal_complement_generators_fix-nt-v2.patch
attachment: trac_12848-posets-modifications.patch

CC: @sagetrac-sage-combinat

Component: combinatorics

Keywords: posets, days49

Author: Nicolas M. Thiéry, Frédéric Chapoton

Reviewer: Darij Grinberg, Anne Schilling

Merged: sage-5.11.beta3

Issue created by migration from https://trac.sagemath.org/ticket/12848

[anneschilling]

comment:1

I am looking at the patch on the sage-combinat queue. The method

•    def upper_set(self, elements, direction='up'):
  

•        if direction=='up':
  

•            return self.order_filter(elements)
  

•        else:
  

•            return self.order_ideal(elements)
  

is missing a doctest.

[nthiery]

comment:2

Replying to @anneschilling:

I am looking at the patch on the sage-combinat queue. The method

•    def upper_set(self, elements, direction='up'):
  

•        if direction=='up':
  

•            return self.order_filter(elements)
  

•        else:
  

•            return self.order_ideal(elements)
  

is missing a doctest.

Ah right. And we need to discuss the name for this method, and use it in order_ideal_complement_generators

[fchapoton]

comment:3

could you please upload the patch here ?

[fchapoton]

comment:4

Attachment: trac_12848-posets-order_ideal_complement_generators_fix-nt.patch.gz

Here is a modified patch, that should pass all tests and is fully documented.

There is an annoying issue with the names of the methods "upper_ideal", "lower_ideal", "order_ideal", "order_filter"

[anneschilling]

comment:5

There is an annoying issue with the names of the methods "upper_ideal", "lower_ideal", "order_ideal", "order_filter"

What is the precise issue with the names?

Anne

[fchapoton]

comment:6

Well, the issue is that of "order_ideal" is a synonym for "lower ideal" and "order_filter" is a synonym for "upper_ideal". Before this patch, everywhere one uses "order_ideal" and "order_filter". In my opinion, these are bad names, and "lower ideal" and "upper ideal" are much better.

And, well, I think that "upper set" and "lower set" are very bad too.

In any case, there is still some cleanup to do.

[fchapoton]

comment:7

ok, here is a new patch, 100% doctest, should pass all tests.

In principle, it should not break anything anywhere.

[fchapoton]

comment:9

The bot has turned green, and the patch is ready for review !

[fchapoton]

comment:10

anybody out there ? maybe it is ready to go ?