PS - A mugshot of the culprit:

http://www.rlmiller.org/bug.jpg

On Nov 23, 10:16 am, Robert Miller <[EMAIL PROTECTED]> wrote:
> The complexity predicting the running time of this algorithm is
> difficult to analyze, and depends very deeply on the structure of the
> graph itself. With G and Pi as you have them, I'm looking at
> sage: g = Graph(G)
> sage: g.show(partition=Pi)
> and it seems to be moderately symmetric. Usually when this is the
> case, the algorithm performs best- it is a balance between enough
> information to narrow the search, but not so much information that
> that becomes a bottleneck.
>
> I have a few questions.
>
> 1) What is the size of the automorphism group the function finally
> outputs, and do you suspect that it is incorrect? Most likely if the
> output is incorrect, you will have some subgroup of the automorphism
> group. If so, what size do you expect the automorphism group to be?
>
> > > I ran across a graph and partition for which it takes the search_tree
> > > function 6 hours on my machine to complete, and I was wondering if
> > > this is something to be expected, or a possible bug.  I've been
> > > working with graphs of similar sizes to this one with no real
> > > problems, and in looking at the graph with the visualization
> > > functions, there didn't seem to be anything especially odd about it.
>
> 2) Can you give an example of a similar graph that goes quickly?
>
> > > I also profiled the search_tree function running on this graph; it
> > > wasn't all that illuminating, but I can post the results of that here
> > > too, if it would help.
>
> 3) That would probably mean more to me than to you, could you post it?
>
> -- Robert Miller
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-support@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at 
http://groups.google.com/group/sage-support
URLs: http://sage.math.washington.edu/sage/ and http://sage.scipy.org/sage/
-~----------~----~----~----~------~----~------~--~---

Reply via email to