Graphing a large graph

I'm using neato to draw a very large graph (around 2500 nodes; around 3 million edges), but the output looks very messy. What's the best way to clean up the output?

Re: Graphing a large graph

Greetings. Unfortunately it's probably not possible to clean up the output with Graphviz.  3 million edges is too many. Even a "4K display" has only about 9 million pixels, so how is it possible to show 3 million edges.  Also the average number of connections per node is over 2000 so most nodes are connected to almost all other nodes.

A matrix representation might be a lot more effective. For example see http://sociograph.blogspot.com/2012/11/visualizing-adjacency-matrices-in-python.html    Please try that and let us know how it works!

There are advanced models like subset (Jacquard distance), simmelian quadrilateral distance, linlog that have been explored in the literature that may help a little when there are nodes of high degree, but your data seems much too large and dense. Also graphistry is a commercial startup and they have been very open to external collaboration. The platform is based on a proprietary GPU layout service, but their current AWS nodes top out around 1-2 million objects. That is still really impressive. They can possibly provision a larger server if you have a budget for the work.

Stephen North

 

 

 

Matrix representation is ok.

Matrix representation is ok. Thank you for your advice though it was not originally desugned for me.
_________________

blog.remind.com/17-tips-for-effective-note-taking

.

Recent comments