This week I will be exploring the wonder that is the Barnes Hut approximation.
When I say exploring, I mean implementing, and when I say wonder, I mean hellish firestorm from which nothing can survive (oh ok, its not that bad). I will be implementing this, the FR grid variant implemented by Walshaw in his multilevel graph drawing algorithm, and also my own adaptation using the coarsening scheme as a basing structure, in order to compare the effectiveness and speed of each approach (also, if time allows, I may look into other approximation methods if anything is quicker).
As always, it comes down to speed, which in turn comes down to how quickly a graph converges, which again is speed, measured by how quickly a graph converges, known as spee.. STOP (hammertime).
The approximations have been proven to speed up the n-body problem, but other qualities such as how the approximation affects the layout is scarcely covered. In order to prove my own adaptation is quicker, I will need the BH tree to compare against.
Other than this, there is no other developments to declare in this post, so all I can say is so long, and thanks for all the fish.
No comments:
Post a Comment