[SOLVED] Suitable layout and speed for large dataset

[Shine]

Hi,
I have a rather large dataset in a bipartite form, one set has around 3000 nodes and the other 8000 nodes. There are also around 100,000 edges between the nodes. I chose Fruchterman Reingold with the speed of 50 and 100. The results are different in two cases. I also tried other layouts but I do not know which output is right. Any idea?
Also none of the edges are displayed in the graph even though I marked that edges to be shown. Why is it the case?

[rotten]

Layouts are different algorithms for placing your nodes on the page. None of them are 'right or wrong'. They are different ways of looking at the same data.

The parameters for the layout algorithms are useful for tweaking the view. Keeping notes as to which ones you used might be helpful if you are comparing two charts side-by-side or trying to get back to the same picture you had before...

Because you usually start out with the nodes deployed randomly on the page, it is possible that some of these algorithms will result in rotated, or even different appearing charts (depending on the algorithm - others should always end up looking the same).

The layout could implement the named algorithm incorrectly, I suppose, that would make them 'wrong' in a sense. You'd still be getting a unique view of your data though. Do you feel that one of the layout algorithms is not doing what you thought it should do?

[mbastian]

For large datasets, I would advise 'Yifan Hu'. It is something close from what a Fruchterman Reingold will do, but much faster.

The algorithm is set to stop itself when it evaluates to be finished. If this happens to soon, set the 'Step Ratio'to 0.99.