Feature request: Mathematica's SpringElectricalEmbedding

[oameye]

Mathematica's SpringElectricalEmbedding
gives very nice layouts for graphs with giving very symmetrical results. It is also the default in Mathematica

Looking at the docs it seemed that SFDP [Spring-Electric model suggested by Hu (2005, The Mathematica Journal, pdf)] should give the same result, but I can't replicate Mathematica's result. Any thought, how the get the same as in Mathematica?

[oameye]

Okay, I found that Align(Spring()) gives a very similar result

[hexaeder]

Hm looks like the underlying force model is similar (all nodes repel each other, edges pull nodes together). Our Spring code essentially does something like Euler integration of such a system which, ideally, converges towards a low energy solution. However for larger graphs it routinely gets stuck in some non-optimal configuration, I guess Mathematica will produce better results, as they can actually use an optimizer rather than time evolving the system?
We could probably do better with default parameters of the forces, masses and damping. Also, often Spring stops becaus it reaches the maximum number of interations rather than converging, so it can be usefull to increas the iterations keyword argument.

If performance allows, I'd suggest you also play around with Align(Stress()). In my experience, the Stress() layout is much better in actually converting towards a global minima in a sane manner (i.e. compare how chaotic the spring iterations looks compared to the beautiful unfolding using Stress.