Is this Right? Force Layout 2 does not take Dynamic Edge Value into Consideration!

Once it's running
Post Reply
kameothrills
Posts: 3
Joined: 26 Oct 2017 20:14

Is this Right? Force Layout 2 does not take Dynamic Edge Value into Consideration!

Post by kameothrills » 26 Oct 2017 20:27

Hello everyone! :)

do you guys know how to make Force Layout 2 work with *dynamic edge weights*? As you can see on the video here: https://streamable.com/cr0ct , the weight of one edge grows over time but the node in the center does not move.

Or is this not possible with Gephi? I'd be open to other alternatives but at this time I'm very much interested in observing *weight value change over time*

Thank you very much!!
Last edited by kameothrills on 28 Nov 2017 15:09, edited 1 time in total.

kameothrills
Posts: 3
Joined: 26 Oct 2017 20:14

Re: Is this Right? Force Layout 2 does not take Dynamic Edge Value into Consideration!

Post by kameothrills » 26 Oct 2017 20:47

Here's another slightly more complex example where two edges grow over time but still no change in layout. https://streamable.com/x9174

User avatar
eduramiba
Gephi Code Manager
Posts: 975
Joined: 22 Mar 2010 15:30
Location: Madrid, Spain

Re: Is this Right? Force Layout 2 does not take Dynamic Edge Value into Consideration!

Post by eduramiba » 26 Oct 2017 21:09

Uhm, that's strange. According to the code (https://github.com/gephi/gephi/blob/mas ... .java#L214) it should use the dynamic weight in the current interval at each iteration.

It's possible that weight is not affecting results for such simple graphs with so few nodes and edges. Can you try with graphs with a t least 20 or 30 nodes?

kameothrills
Posts: 3
Joined: 26 Oct 2017 20:14

Re: Is this Right? Force Layout 2 does not take Dynamic Edge Value into Consideration!

Post by kameothrills » 26 Oct 2017 22:34

So, I did some tests, bear with me :P


It seems to work well with timestamps!!

50 Nodes, Timestamps, Force Layout 2 and most edge weights constants, some grow some decay.

I define the timestamps like this:

For nodes: <[2000.0, 2001.0, 2002.0, 2003.0, 2004.0, 2005.0, 2006.0, 2007.0, 2008.0, 2009.0, 2010.0, 2011.0, 2012.0]>

For edges: <[2000.0, 1.0]; [2001.0, 1.0]; [2002.0, 1.0]; [2003.0, 1.0]; [2004.0, 1.0]; [2005.0, 1.0]; [2006.0, 1.0]; [2007.0, 1.0]; [2008.0, 1.0]; [2009.0, 1.0]; [2010.0, 1.0]; [2011.0, 1.0]; [2012.0, 1.0]>

For varying edges similar to above but with different values.

https://streamable.com/4bzij

That works like a charm as seen on the video.


For intervals however, that doesn't seem to be the case.

Perhaps it's the way I'm defining the intervals?

Again, 50 Nodes, Intervals, Force Layout 1, 2 and FruRein, most edge weights constant, one grows.

FruRein: https://streamable.com/zzgdm

ForceAtlas2: https://streamable.com/osvku

ForceAtlas: https://streamable.com/6mg62


It boggles the mind, what are your thoughts?


Thank you! :)
Last edited by kameothrills on 28 Nov 2017 15:09, edited 1 time in total.

User avatar
eduramiba
Gephi Code Manager
Posts: 975
Joined: 22 Mar 2010 15:30
Location: Madrid, Spain

Re: Is this Right? Force Layout 2 does not take Dynamic Edge Value into Consideration!

Post by eduramiba » 27 Oct 2017 10:32

The visualization (where edge weight is correcly displayed with intervals) gets the weight differently, so it might be a bug https://github.com/gephi/gephi/blob/mas ... .java#L246

Can you please create an issue in github?

Post Reply