[Chicago] Graph traversal in Python: ideas

Matthew Kemp mattkemp at gmail.com
Sun Sep 11 18:28:18 CEST 2011


Instead of coming up with a new specification for graphs why don't we start
with an existing implementation. They probably figured out most of the
corner cases and gotchas. In the past I've used NetworkX (
http://networkx.lanl.gov/). It supports most common graph functionality:
directed and undirected, weighted and unwieghted, standard graph traversal
algorithms. Also, arbitrary data can be attached to nodes and edges. Not
sure about hypergraph support.

Just my $.02


On Sat, Sep 10, 2011 at 9:01 PM, Tal Liron <tal.liron at threecrickets.com>wrote:

>  On 09/10/2011 08:54 PM, Tal Liron wrote:
>
> So, I've taken the challenge more seriously:
>
> I should also add that it's possible to turn this proposal into support for
> hypergraphs.
>
> All that would need to be done is to have "to" and "from" be arbitrary
> objects instead of keywords, so that edges can connect to as many vertices
> as desired, rather than just having directionality. And then something like
> the "fromto" keyword could just be a tuple of these.
>
> I don't think this would make any serious difference to the implementation.
> However, hypergraphs are known to cause migraines: I've tried to keep the
> proposal as painless as possible at this point!
>
> Another thought is that "to" and "from", rather than being keywords, can be
> special values, like True and False. So, you could then use To and From for
> relationships, and ToFrom can equate to a tuple: (To, From). Voila:
> hypergraphs with Tylenol.
>
> -Tal
>
>
>
> _______________________________________________
> Chicago mailing list
> Chicago at python.org
> http://mail.python.org/mailman/listinfo/chicago
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/chicago/attachments/20110911/6372d242/attachment.html>


More information about the Chicago mailing list