4

Possible Duplicate:
What's the fastest force-directed network graph engine for large data sets?

We currently have a dynamically updated network graph with around 1,500 nodes and 2,000 edges. It's ever-growing. Our current layout engine uses Prefuse - the force directed layout in particular - and it takes about 10 minutes with a hefty server to get a nice, stable layout.

I've looked a little GraphViz's sfpd algorithm, but haven't tested it yet...

Are there faster alternatives I should look at?

  • I don't care about the visual appearance of the nodes and edges - we process that separately - just putting x, y on the nodes.
  • We do need to be able to tinker with the layout properties for specific parts of the graph, for instance, applying special tighter or looser springs for certain nodes.
  • Bonus credit for stats/benchmarks!

Thanks in advance, and please comment if you need more specific information to answer!

Community
  • 1
  • 1
peteorpeter
  • 4,037
  • 2
  • 29
  • 47
  • 1
    @peteorpeter I'd suggest looking at [Cytoscape](http://www.cytoscape.org/) (with its [scripting plugins](http://cytoscape.wodaklab.org/wiki/ScriptingPlugins) facilities), Python [NetworkX](http://networkx.lanl.gov/), or [aiSee](http://www.aisee.com/) to get other ideas. – chl Mar 31 '11 at 10:35
  • @chl Thanks - look like some excellent possibilities! – peteorpeter Mar 31 '11 at 13:55
  • @peteorpeter Cross-posting is not encouraged. (Your +100 bounty would have been ok here too.) I'm afraid we'll have to close your question now. – chl Apr 06 '11 at 17:34
  • @chl Apologies for the confusion. SO and the Stack Exchanges are in an interesting spot ATM where the differentiation isn't complete... I put in a close vote myself. Question: why migrate it to SO? Is that required to close it? Thanks again for the response. – peteorpeter Apr 06 '11 at 21:27
  • @peteorpeter I agree that it is difficult to choose between SO and stats.SE for questions related to R or software issues. (We have had several discussions on our Meta about this, e.g. [Vote on policy for R questions](http://bit.ly/fcIw9i).) However, it is not a good idea to duplicate question as one purpose of SE is to create a lasting record of great solutions to specific questions, that can be accessed from e.g. googling or referring sites. Disseminating responses to duplicated questions over SE sounds rather inefficient in this respect. Note that your question is still visible on stats.SE. – chl Apr 07 '11 at 05:56
  • @peteorpeter (...) Migrating the question from stats.SE is the only solution for your two questions to be merged into a single one, here on SO. It allows to keep old comments and answers attached to the master question. While moving your question, I've flagged your two questions here for mods attention, but maybe they missed it. – chl Apr 07 '11 at 06:00
  • @chl Ahh, I see. Good points all. Consider me educated! – peteorpeter Apr 07 '11 at 13:55

0 Answers0