[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[igraph] Is there a need to transform an adjacency matrix: clustering
From: |
Dylan Young |
Subject: |
[igraph] Is there a need to transform an adjacency matrix: clustering |
Date: |
Fri, 03 Oct 2014 11:20:03 +0100 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.1.1 |
Hello,
I have a weighted digraph that I would like to analyse for clusters
using the spinglass.community algorithm (the digraph includes negative
weights). Direction in the matrix is from i to j.
I have created an adjacency matrix in igraph (using R) using:
mat.adj <- graph.adjacency(my.matrix, weighted = TRUE, mode =
"directed", diag = TRUE)
when I plot the graph the edge direction between vertices is correct.
However, I wondered if the matrix (my.matrix) needs to be transformed
before creating the adjacency matrix for implementation of the
spinglass.community algorithm so that the edge direction is from j to i
as noted in Newman 2010 pp. 114 (Networks, An Introduction). I presumed
not because the graph plots correctly without transforming.
I use the following code for the spinglass algorithm:
mat.communities <- spinglass.community(mat.adj, spins=15,
implementation="neg", gamma=1.0, gamma.minus=1.0, weights=NULL,
update.rule="config")
Could you advise?
Thanks,
Dylan
- [igraph] Is there a need to transform an adjacency matrix: clustering,
Dylan Young <=