Comments on: Force-Directed Graph Layout With Proce55ing

By: Russell Jurney

Russell Jurney — Wed, 02 Dec 2009 09:10:59 +0000

I was having trouble getting my own graph to layout properly, so I’ve just ported yours to Java, and am using it. Thanks!

By: stefanK

stefanK — Sun, 08 Nov 2009 21:31:08 +0000

Hi,

i am in about the same situation as dennis. I am interested in graphs and i am learning processing and having your data files would be very helpful to understand how to define the graph.
thx,

By: Dennis

Dennis — Mon, 26 Oct 2009 23:41:03 +0000

Hey… I am interesting in learning how to do a force directed graph with Processing and your example is one of the better ones. Is there anyway you can email the data files. That way I have a compilable foundation?

By: KattyS

KattyS — Thu, 02 Oct 2008 04:31:07 +0000

Hello. My name is Kate. I wish to become Striptease. I constantly train.
I have removed video and I wish you to ask to look it and to tell to me about my errors, or to prompt, as I can improve the technics.
[url=]http://cananecatah.blogspot.com[/url]
Thank you.

By: banksean

banksean — Fri, 20 Jul 2007 00:17:10 +0000

Oh dear, you’ve actually looked at the source. [runs away]

By: Justin Washtell

Justin Washtell — Fri, 20 Jul 2007 00:10:27 +0000

Sorry, I’m writing this as I’m discovering it. Your app is kinda in 3D, but the movement is restricted to the plane, am I correct? Anyway, you can very quickly and massively optimise your gravitation routine by making the inner loop “for(j=i+1…” rather than “for(j=0…” that will half the number of comparisons as you suggest… and also getting rid of the SQRT in there, which you only multiply back out again afterwards because you are interested in the square of the distance (unless you were to try the experiment I mentioned above).

By: Justin Washtell

Justin Washtell — Thu, 19 Jul 2007 23:53:29 +0000

Just a thought. Your gravity formula: it should really only be r^2 when simulating three-dimensional space. In a hypothetical 2D world, as per your simulation, gravity - like light intensity - should fall off linearly with distance (the emanating photons/gravitons are distributed around the perimiter of an increasingly large notional circle surrounding each massive body, as opposed to the surface of a notional sphere). I appreciate this is all a bit hypothetical given that you’re not trying to simulate the real world (and especially not a bizarre 2D version of it which doesn’t actually exist) but I thought it might be interesting to see what difference it makes to the behaviour of your simulation?

By: banksean

banksean — Thu, 05 Jul 2007 14:28:33 +0000

I’ve got a more up-to-date version of this lib/applet here: http://www.cricketschirping.com/weblog/?p=966

It uses a more performant set of calculations (runge-kutta) that should handle more nodes than this version. The basic interface to the Graph/Node classes is very similar though.

By: carlo

carlo — Wed, 04 Jul 2007 22:32:44 +0000

Hi, and thanks for this applet, it’s really nice to play with by simply changing the function

F = stiffness * (currentLength - naturalLength)

with other sin/cos/tan etc. code, it produces bizarre and interesting movements.

I was wondering if there’s a maximum of nodes it can handle, because if I create more than, say, around 200 nodes it disappears just after having been drawn. If I push the space bar, it is redrawn and it disappear again. Any ideas?

Thanks again and go on!

By: tom

tom — Sat, 16 Dec 2006 22:55:12 +0000

Thanks so much! Been looking for a force-directed graph for Processing (there was one as part of a physics lib, but it didn’t support interaction).

This is perfect for my (super secret operation) / next little project.