The output is a lie...

Further work on approximations this week/month. Output from the grid approximation looks promising but appears to be affected by the grid in unintended ways as mentioned in my previous approach. The Barnes Hut Quadtree being used by the FDP algorithm needs much more work, though the structure and tree itself is fine.

Monthly meeting yesterday, minutes hidden from you all (mwahaha), but are found in the usual place. The meeting itself was not as comfortable as normal, I had very little work to show (as most of my developments have been in the coding domain and have very little observable outputs at the moment). We did however mention my intentions and the work I have done so it was not completely in vain.

Continuation of the approximation refinement is planned for the rest of this work and tomorrow, with some of my time being spent on some mandatory postgraduate courses  which happen to take up the majority of my time as I am forced to print every answer one by one (here's an idea, wait until the end and then show the user a page of their answers, or even better, don't use flash).

Sarcasm aside, I am still trying to stop my grid approximation outputs clumping together and will update here as soon as I have results. My last few attempts, using all "boxes" containing vertices as opposed to the immediately surrounding boxes, changing the weightings, changing the strength of forces, changing the way vertices move between boxes, have not produced any significant changes. However, I feel I am getting close to a fix so bare with me!

Not much else to say I'm afraid, so here is a picture showing vertices being stuck to the border of their grid, drawn by creating a new grid every FDP iteration (purely to see what would happen, I don't intend to do this as that's madness and very inefficient).