Connectivity of complete graphs

[Nathann Cohen]

Fixes the problem reported at :

​http://ask.sagemath.org/question/793/vertex-connecitivity-incomplete-pun-intended?answer=1306#1306

The connectvity of a complete graph is n-1. Currently, an exception is returned

Requires : #11367

Apply : trac_11910.patch​

[Jason Grout]

When testing on sage-4.8.alpha3, I get this (presumably from #11367):

sage -t  "devel/sage-main/sage/graphs/generic_graph.py"     
Exception MemoryError: MemoryError() in 'sage.graphs.distances_all_pairs.c_eccentricity' ignored
Exception MemoryError: MemoryError() in 'sage.graphs.distances_all_pairs.c_eccentricity' ignored
**********************************************************************
File "/home/grout/sage-trees/sage-4.8.alpha3/devel/sage-main/sage/graphs/generic_graph.py", line 6805:
    sage: g.vertex_connectivity()
Expected:
    9.0
Got:
    9
**********************************************************************
1 items had failures:
   1 of  21 in __main__.example_84
***Test Failed*** 1 failures.
For whitespace errors, see the file /home/grout/.sage//tmp/generic_graph_22844.py
	 [63.6 s]
 
----------------------------------------------------------------------
The following tests failed:


	sage -t  "devel/sage-main/sage/graphs/generic_graph.py"
Total time for all tests: 63.6 seconds

Other than that, the code looks good. The MemoryErrors? look interesting, though. I don't think it's from this patch, but do you have any idea what it could be from?

[Jason Grout]

Interestingly, when I pop the patch and rerun tests, I don't get those memory errors anymore, so maybe it is this patch causing the memory errors. Do you have any clue about what is going on?

[Nathann Cohen]

Hmmmm... O_o

The memory error most probably comes from patch #12052.

Nathann

[Nathann Cohen]

Hmmm... Which kind of version are you using ? I applied the patch against alpha4 and there is nothing wrong there ! O_o

Nathann

[David Coudert]

I have installed the patch with sage 5.0-beta4.

It is working correctly, but I also have a problem with the tests. This is a rounding problem. You should ensure that unweighted connectivity returns an integer.

sage -t  "devel/sage-myclone/sage/graphs/generic_graph.py"
IBM ILOG License Manager: "IBM ILOG Optimization Suite for Academic Initiative" is accessing CPLEX 12 with option(s): "e m b q ".
**********************************************************************
File "/path-to-sage/sage-5.0.beta4/devel/sage-myclone/sage/graphs/generic_graph.py", line 7020:
    sage: g.vertex_connectivity()
Expected:
    9.0
Got:
    9
**********************************************************************
1 items had failures:
   1 of  21 in __main__.example_85
***Test Failed*** 1 failures.
For whitespace errors, see the file /path-to-home/.sage//tmp/generic_graph_2957.py
         [53.6 s]
 
----------------------------------------------------------------------
The following tests failed:


        sage -t  "devel/sage-myclone/sage/graphs/generic_graph.py"
Total time for all tests: 53.6 seconds

[Nathann Cohen]

Ahaahh. Actually for once it is the opposite : the function returns an integer, and a float was expected in the doctest... Of course the value returned is always an integer, as it is g.order()-1. Stupid me :-p

Thank you for the review ! Patch updated ! :-)

Nathann

[David Coudert]

The patch now pass the tests successfully !

Install ok, docbuild ok, test ok, functionality ok => positive review.

D.

[Nathann Cohen]

Install ok, docbuild ok, test ok, functionality ok => positive review.

Wouhouuuuuuu ! :-)

And it looks like ISGCI (#11880) should be included in Sage in a not so distant future :-)

Thanks !

Nathann