Re: [igraph] fixed length of edges in plot

[Stijn van Dongen]

I have a suspicion it would in general, for N nodes, require plotting in an
N-dimensional space, where each node is plotted on a unit vector.  In the past
I have plotted some protein homology graphs, where very close clusters of
homologs where spread out in plot space, simply because 2D-space is too
crowded.  Another experiment is to plot a cube in higher dimensions. One
example is this 5-dimensional hypercube:
   http://www.ebi.ac.uk/~stijn/cube5-bl.png

The paired colour coding indicates nodes that are maximally far away from each
other. The green nodes, although at distance 5 in the graph, still end up very
close in the (BioLayout-produced) layout.  This is not because the lay-out is
bad: it is simply very hard to cram high-dimensional objects into two
dimensions.

The message from all this is that although graph visualisations are a great
tool, they can also be very misleading.

best,
Stijn



On Fri, Sep 17, 2010 at 09:43:02AM +0200, Gábor Csárdi wrote:
> Hi Marco,
> 
> I think for many graphs this is impossible. Even just deciding whether
> this is possible of not, for a given graph, seems to be a difficult
> problem. But I might be wrong.
> 
> Best,
> Gabor
> 
> On Fri, Sep 17, 2010 at 12:41 AM, Marco <address@hidden> wrote:
> > Hi all,
> >
> > I was wondering: is there a way, when plotting a graph, to force all
> > edges to be of a fixed length? This would render my graph plots somewhat
> > more readable, I think.

-- 
Stijn van Dongen         >8<        -o)   O<  forename pronunciation: [Stan]
EMBL-EBI                            /\\   Tel: +44-(0)1223-492675
Hinxton, Cambridge, CB10 1SD, UK   _\_/   http://micans.org/stijn