Re: [igraph] Need a graph layout that uses edge weights

[Gabor Csardi]

Rick,

The Fruchterman-Reingold layout now has a parameter to provide edge
weights. Weights bigger than 1 result more "attraction" along the
edges, weights less than 1 do the opposite. (Maybe you can even do
negative weights which result repulsion. I've no idea whether this is
good for anything.) It seems to work well:

> g <- graph.ring(10)
> E(g)$weight <- 1:2
> E(g)$weight 
 [1] 1 2 1 2 1 2 1 2 1 2
> l <- layout.fruchterman.reingold(g, weights=E(g)$weight)
> plot(g, layout=l)

You can download a 0.5 C library or R source package from
http://cneurocvs.rmki.kfki.hu/igraph/download

G.


On Tue, Jun 05, 2007 at 02:29:31PM -0400, Rick Borovoy wrote:
>    Hi,
> 
> 
> 
>    Can anyone help me find / make a layout algorithm that factors in edge
>    weights?   For example, in the case of a spring-type algorithm, the spring
>    constants would be greater on larger-weighted edges.
> 
> 
> 
>    The real problem:  I want to draw a network diagram where the nodes
>    correspond to groups of people, and the link weights correspond to how
>    many connections there are between two groups.   It should get laid out so
>    that groups that have the most connections between them are close to each
>    other in the graph.
> 
> 
> 
>    As a work-around, I created a graph with n edges between a  given pair of
>    nodes, where n = the number of connections between the two groups.  It did
>    not produce the desired effect, however, when I laid it out using Spring,
>    Fructerman-Rheingold, Kamanda.Kawai, etc.
> 
> 
> 
>    Any guidance would be much appreciated.
> 
> 
> 
>    Thanks,
> 
>    Rick Borovoy
> 
> 

> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help


-- 
Csardi Gabor <address@hidden>    MTA RMKI, ELTE TTK