[Networkit] Bachelor-Project about generic semiring operators

Christian Staudt christian.staudt at kit.edu
Tue Jun 28 17:02:05 CEST 2016


Hi Thomas,
sounds good. At this point my recommendation is that you go ahead with implementation and tests, then show us code and test results. And I’m not just saying that because I don’t have much time (last week at KIT…). I think your design ideas are basically sound. Have some more trust in them and try them out. Then iterative improvements are possible.

Chris


> On 28 Jun 2016, at 16:25, Thomas Schmidt <thomas.3.schmidt at uni-konstanz.de> wrote:
> 
> Hi Chris,
> 
> > Have you thought about how this new functionality will be available on
> > the Python layer? It should, but it doesn’t have to be a 1:1 mapping
> > of the underlying C++. We can help with the mapping.
> 
> I think the best way would be for the Python layer to, just like on the
> C++ layer, add the relax function as a parameter. I am not really sure
> how to map a template class, though. But since, in python, addition and
> multiplication can be defined for arbitrary classes as well, I think it
> would be best, if one could just define such a class, assign objects of
> it as the edgeweights, and then just call Dijkstra on it.
> 
> Regarding that, something just came to my mind: How should the
> edgeweights be added? Should there be an extra function for it? So far,
> I just created a vector and handed it over to the Dijkstra constructor.
> 
> But anyways, if Python's duck typing could be utilized this way, it
> would be great. Do you know if this is possible?
> 
> Thomas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ira.uni-karlsruhe.de/pipermail/networkit/attachments/20160628/f029de04/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.ira.uni-karlsruhe.de/pipermail/networkit/attachments/20160628/f029de04/attachment.sig>


More information about the NetworKit mailing list