Hi all,

totally new in Povray, here to ask about possibilities:

I was reading a post for displaying a network as a connected graph.
http://www.gossamer-threads.com/lists/python/python/792816

Is there any package to render and be able to import a connected graph of
millions nodes, 100+M links ?
It would be so cool if it would be possible to make a short video, I'd love to
create a presentation on a network analysis we've done with a new technique.

Thank you!

Post a reply to this message

Le 25/09/2013 15:32, gg4u nous fit lire :
> Hi all,
> 
> totally new in Povray, here to ask about possibilities:
> 
> I was reading a post for displaying a network as a connected graph.
> http://www.gossamer-threads.com/lists/python/python/792816
> 
> Is there any package to render and be able to import a connected graph of
> millions nodes, 100+M links ?
> It would be so cool if it would be possible to make a short video, I'd love to
> create a presentation on a network analysis we've done with a new technique.
> 
> Thank you!

Greetings,

Beware, when they talk about graph libraries, usually, it's not for
graphical display (indeed, that's a totally different hard stuff on its
own).

For instance, the boost graph library is about applying algorithms to a
graph (as in Graph Theory) such as finding the shortest path(s) between
two nodes or other funny things for mathematicians (such as testing an
heuristic to the salesman problem, or the Chinese postman problem,...)

Absolutely no visualisation is ever done in such libraries.

One tool/library of interest for visualisation (with many engines for
placement) is graphviz : describes a graph as text, get a picture... or
another text file with placement of nodes and links.

About rendering millions of nodes, with 100+M links: a HD screen is
1920x1080, that's only 2 millions pixels. there is no way to represent
the millions of nodes and being able to distinguish the links at once.

Even on a 4K screen, it will be hard to see.

about visual navigation in such graph, the best approaches usually drop
the traditional Cartesian space (either 2D or 3D) to use hierarchical
clusters or hyperbolic view.


> http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.60.3531&rep=rep1&type=pdf

> http://www.cs.ubc.ca/~tmm/courses/old533/readings/hermann.pdf

One hope to finish: if you intend to have a graph of geographically
located nodes, VRML might be of some help for the visualisation.
(but the syntax is a bit hostile, despite being plain text, if you want
a world of 2 millions nodes and 100+M links (which, btw, is a very high
connectivity for a computer network): you will need a resilient VRML
engine to handle so much data (unless you have the data in a database
and only serve the nearby nodes))
(seems VRML is dead, replaced by X3D)

The hardest part in graphical graph is the placement of the elements.

Post a reply to this message

>gg4u  on date 25/09/2013 15.32 wrote:
> Hi all,
>
> totally new in Povray, here to ask about possibilities:
>
> I was reading a post for displaying a network as a connected graph.
> http://www.gossamer-threads.com/lists/python/python/792816
>
> Is there any package to render and be able to import a connected graph of
> millions nodes, 100+M links ?
> It would be so cool if it would be possible to make a short video, I'd love to
> create a presentation on a network analysis we've done with a new technique.
>
> Thank you!
>
>
>
It seems more a Graphviz affair.
Someone asked for a similar problem:
http://www.graphviz.org/content/how-create-very-large-graph-more-1-million-edges-or-nodes

Paolo

Post a reply to this message