Changes between Version 10 and Version 52 of Ticket #27571


Ignore:
Timestamp:
06/10/19 15:27:54 (4 months ago)
Author:
dcoudert
Comment:

I have updated the ticket description as this ticket fix all doctests in families.py.

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #27571

    • Property Reviewers changed from to Dima Pasechnik
    • Property Cc ​stumpc5 added
    • Property Commit changed from 8fcd5842211b9a469d3a631cb7f28cdc1a944be1 to 50a69e8454a4b77beaffc726c6babc472488641a
  • Ticket #27571 – Description

    v10 v52  
    1 This ticket fix this py3 failing doctest in `graphs.MathonPseudocyclicStronglyRegularGraph`
     1This ticket fix py3 failing doctest in `graphs.MathonPseudocyclicStronglyRegularGraph`:
    22{{{
    33Failed example:
     
    2424    [-3 -2  1 -1  3 -4  2  4  0]
    2525}}}
    26 This is an ordering issue related to method `automorphism_group`. The way the automorphism group is computed depends on a ordering of the vertices. Currently:
    27 - for the `'sage'` algorithm, the default order is `list(G)`, but setting `partition=[G.vertices()]` will change that order
     26and
    2827{{{
    29 sage: G = graphs.PaleyGraph(9)
    30 sage: G.automorphism_group(algorithm='sage')
    31 Permutation Group with generators [(1,2)(a,a + 2)(2*a,2*a + 1), (1,2*a + 2)(2,a + 1)(a + 2,2*a + 1), (0,1)(a + 1,a + 2)(2*a,2*a + 2)]
    32 sage: G.automorphism_group(algorithm='sage', partition=[list(G)])
    33 Permutation Group with generators [(1,2)(a,a + 2)(2*a,2*a + 1), (1,2*a + 2)(2,a + 1)(a + 2,2*a + 1), (0,1)(a + 1,a + 2)(2*a,2*a + 2)]
    34 sage: G.automorphism_group(algorithm='sage', partition=[list(G)[::-1]])
    35 Permutation Group with generators [(2*a + 1,2*a)(a + 2,a)(2,1), (2*a + 1,0)(2*a,a + 1)(a,1), (2*a + 2,2*a + 1)(a + 1,a)(2,0)]
     28Failed example:
     29    G9.automorphism_group(algorithm="bliss").order() # optional - bliss
     30Expected:
     31    9
     32Got:
     33    3
    3634}}}
    37 - for `'bliss'`, the order is always `list(G)`
    38 and the order `list(G)` is not always the same in py2 and py3...
    3935
    40 **So, in this ticket, we do**
     36**in this ticket, we do**
    4137- ensure that the ordering with `'sage'` and `'bliss'` is the same in `automorphism_group` and `canonical_label`/`canonical_form`
    4238- clean bliss to avoid for instance indexing a dictionary with a possibly not hashable key. The new solution is slower but safe.
    4339- specify the partition in the doctest of `MathonPseudocyclicStronglyRegularGraph` to make the doctest stable with py2 and py3
    44 
    45 
    46 
    47 Other doctests are still failing with py3 and bliss and could be fixed in another ticket.
     40- pass correct domain to `PermutationGroup` in `bliss.pyx`