Nathann,
Using the following instead fixes the problem:
g.add_edges( (Mod(i,n),Mod(i+j,n)) for i in range(n) for j in range(1,k+1) )
This is more consistent, since we are actually using the same vertex
objects. However, that should just work, right? Why doesn't it?
This is coming from the code around line 1000 of c_graph, which goes
from vertex labels to ints and back. The IntegerMod case was not in
mind when this code was written. The real problem is that when the
Python int 0 gets passed to get_vertex, it does not add the entry to
the translation dictionary. The idea being that it would be more
efficient for the majority of graphs, which only use [0, 1, ..., n-1]
as labels, going through dictionaries a bunch would be a waste.
However, when the IntegerMod 0 gets passed in, it does not find an
equal object in the dict, and since it is not one of {int, long,
Integer}, it assigns a new int in the translation dictionary for it.
So that's what's happening, what do people think about what we should
do about this? Technically the input is a bit fuzzy, but this raises
the question of how many other objects are there which will pass the
test int(0) == IntegerMod(0, 20)... I am very reluctant to support
adding all the labels to the dicts, unless someone can show that there
really isn't any overhead there...
--
Robert L. Miller
http://www.rlmiller.org/
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
