[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [igraph] Question on igraph_community_fastgreedy

From: Thomas F. Brantle
Subject: RE: [igraph] Question on igraph_community_fastgreedy
Date: Wed, 27 Feb 2008 17:08:06 -0500

Tamas and Gabor:
Please see clarification question below. Thanks.

-----Original Message-----
From: address@hidden
[mailto:address@hidden On Behalf
Of Tamas Nepusz
Sent: Wednesday, February 27, 2008 4:20 PM
To: Help for igraph users
Subject: Re: [igraph] Question on igraph_community_fastgreedy

> const igraph_vector_t *weights
> argument has no description... are these the weights described as  
> the heuristics in the Wakita paper?
No, the heuristics described in the Wakita paper are still waiting to  
be implemented... :( 

*** I assume that this means the current iGraph version still represents the
original CNM algorithm? Also, since Wakita suggests both "programming/data
structures" improvements plus several "Q-decision" heuristic variations, is
it planned to implement both the Q-decision heuristics plus the suggested
data structures or one or the other or both?

These are simply edge weights. In the weighted  
case, the "degrees" of the vertices are defined as the sum of the  
weights of their adjacent edges, and m (in the original formula) is  
replaced by the sum of the weights along all the edges. 

*** Just to clarify, since the input network/graph in CNM (Fast Greedy) is
assumed to be the individual link pairs, then the "weights" are really any
"arbitrary" >0 (possibly >=0) value for each specific link node pair? 

*** So, I'm a little confused by the "sum of the weights of their adjacent
edges" since the input weights are on a link pair by link pair basis. Isn't
their only one weight possible per link pair? Thus the initial Adjacency (or
e-matrix) is a matrix of link pair "weights" (and zeros) as opposed to
"ones" (and zeros). 

*** Sorry if I'm confused on the issue and thanks for the clarifications.

This extension  
was briefly mentioned in the original paper of Newman.

> Also, if I'm not interested in weights, can I use NULL?


igraph-help mailing list

reply via email to

[Prev in Thread] Current Thread [Next in Thread]