If the title was an unsuccessful hint, this week I have been working on the multimatching algorithm for replacing the currently used edge contraction method. This has been started countless times before and progress has been made - however, due to the amount of time between my work on it, I've had to spend the majority of a day catching up on my own work.
Additionally, I realised that some of my work is not longer applicable - the paper I have written for the IV2012 conference uses the edge contraction for comparisons, but the newer multimatching uses hashmaps for its storage (and other tasks), resulting in an unfair comparison. I now get to choose whether to continue with the previous approach and provide compareable results or continue using hashmaps for this report and provide new results. <EDIT> In a sudden moment of clarity (and remembering the obvious) - the multimatching method should work for edge contraction in the same way, albeit a little slower. Therefore I will be using the multimatching method for all result. </EDIT>
Much of the report itself has been put together (as mentioned in my previous blog entry) but no significant work has gone into refining it. Some may call this laziness, I call it preparation and excitement for my return to coding. In the mean time, I am going to return to the code and hopefully finish that before the end of the week. I will update again when any progress is made. Sayonara!
No comments:
Post a Comment