Well, my adviser wasn't quite as enamored with my algorithm as I had hoped. He want's me to apply Metropolis directly first. I'm not wild about that because a) I know Metropolis works and b) so does everybody else. It's just not particularly efficient.
I'm trying to look at the bright side. It will be a good baseline. If my algorithm can beat Metropolis in a straight-up fight, I'm onto something. If not, I'll need to look for improvements. Meanwhile, much of grad school is sucking it up and formalizing results you already know on the off chance you may be wrong so, coding Metropolis it is.
I've decided to write it in C# simply because it will take me half the time of writing it in Java. I'll regret that if we port the data layer to Hadoop but, we're six weeks into the semester; I need some tangible results real soon. I should have it working over the weekend.
No comments:
Post a Comment