This week I have been implementing and making the necessary ammendments to my own contraction approximation algorithm, and this brings us the latest output from my implementions:
The graphs are drawn remarkably well for a first attempt (no refinement) and show that the contraction approximation works well enough to give a layout showing some of the features of the graph. Still, there is some work to be done to give more attractive results as these results are somewhat squashed on the z-axis (again...).
Removing the z-axis from the FDP algorithm, squashes the graphs again so they become thin stick-like graphs at a slant. This is perculiar as this behaviour suggests that every dimension has moved over (so 3rd is now 2nd, and that if I were to include a 4th, it would draw the 3rd, which is of course madness).
Some investigation and experimentation should resolve this. In an attempt to allow for clumping as opposed to edge contraction, I will be looking at changing the algorithm so a user can select how many vertices can be clumped, and in turn, allow the algorithm to decide whether edge contraction is sufficient or if grouping vertices would be more efficient.
I will continue with everything and update next Monday or Tuesday (only one update this week I'm afraid... I can hear your sign of relief by the way...). Have a good Friday/weekend!
No comments:
Post a Comment