Weekly Journal: Week 4 (June 11 to June 18)

Victoria Manfredi
vmanfred@cs.smith.edu


I have been debugging two algorithms and I think that they are almost debugged. When I compare them to "standard" output one of the algorithms is still doing something unexpected every once in a while. But they are almost debugged, and once they're correct, then I can return to the Q-learning algorithm which I am excited about.

Well, I think the myoptimal and derivative following algorithms are finally correct. I have also reimplemented the Q-learning algorithm and I think it may be correct as well :-)

I have been making graphs of the outputs of the different algorithms when they compete against each other (using the debugged algorithms!) and the results are interesting and appear to match the results from a paper by Amy (the results are not identical to those found in Amy's paper because the constraints on each algorithm are different: in the paper, the pricebots have different ranges for what the buyers will buy at, while in the program I am working on the pricebots all have the same range of prices for what the buyers will buy at).


Smith CS Page / Brown CS Page / Vicky's Homepage / Amy's Homepage / Weekly Journal