Work this week has been exceptionally slow.
Tuesday was spent looking through previous implementations to find if there is anything I can use (and also to reinforce my understanding of what my code actually does). Turns out my understanding was correct in the first place and I was just having a mental breakdown...
Wednesday and today (Thursday) has been spent attempting to get output from my Veldhuizen implementation. Although there is an output, it is not correct due to a small bug which has yet to be debugged.
Attempt: The plan is to have the FDP algorithm run over the graphs for a small number of times to get the premature coordinates of the uncoarsened graph, a "key frame" similar to that in Veldhuizen's description. Each iteration of this step would keep changing and outputting the latest coordinates until a pre-set finishing point.
Problem: The way in which the algorithm works, the coordinates are based through the different graphs, but are not saved (so the coordinates are always starting at zero).
Fix: save the coordinates, simple as. The issue with this fix is that its not so simple as that. The method of storing graphs makes retrieving the levels difficult.
As an alternate route, to try and speed up my work, I may just choose to hold graphs in their own Graph type Component, which will hold the vertices and edges for that graph. That would make this step easier but less efficient, which for some reason I am obsessed with - even though I don't need to be yet!
In other new, I am mistaken when I refer to Java as comparing memory locations in my latest implementation (as referenced in my previous post). Java still compares Object or Component (whatever I am using at the time), but I only see the memory location because I am printing the Object/Component.
Also this week, I marked my first batch of coursework.
Anyway, back to work... one week left before my monthly meeting with Chris Walshaw and Alan Soper. I am intending to have something physical to show other than thousands of lines of useless code (such as graphs and pretty lights to distract them while I run away). The aim is still to have my Veldhuizen work completed (to my understanding of what it does, as recently I've been avoiding the scary maths equations he uses).
That's for all for now, my next may very well be from a padded cell, so see you soon? Heh, cya!
No comments:
Post a Comment