Good God that looks horrifying...

Small Update today, the approximations are beginning to take shape, or not if you take a look at the output. Below are some images of some well known graphs in the field of graph drawing, showing their distortion.

 The image shows, from left to right; 3elt, a 10x10x10 grid, data and a 55x55grid. All are drawn in 2D. As can be seen, the output resembles its better drawing (see previous posts), but the global untanglement is not adequete.

This is most likely due to the calculation of forces between grids (if you see, the tangles generally occur over the grid lines which suggests the cause), which will need to be experimented with to get the most pleasing layouts.

The drawing times are far more impressive:

Graph	Normal (ms)	Grid (ms)
55grid	67850	9186
3elt	168175	19330
data	62814	12513
10grid3d	8311	2112

The runtime is far smaller, proving the strength of the grid approximation. The Barnes Hut implementation will be posted as soon as a reasonable output is given. For now, that's all folks.

<edit> Note to self; the grid variant can either use the average position of vertices or go through the vertices in an adjacent square to the current square. This is the cause for the incredibly fast results currently. Also, if iterating over the vertices instead of the average, the number of squares should be reduced so less vertices per square </edit>