March 11th, 2007

tzb

Worklog: Finally submitted

Well, both the papers have been submitted. With any luck one or both of them will be accepted. I'm not too happy about how the SOCP paper turned out -- mostly because I did so little and nothing is really novel -- but at least IÂ got something written, and after all, it is a workshop, so works-in-progress should have a place there.

Probably the most valuable part of the experience was learning GNU R, which I used to analyze (in a fairly weak sense) the performance results and make the graphs. R is pretty cool and makes it really easy to do things like divide a column ("vector") of results by another column to do things like compare the speedup of two test cases against a baseline. I put some notes up on the wiki about it. The language is a very quirky, but it gets the job done. I also spent some time solving all the niggling little details when using LaTeX and wrote my solutions down. Also, Referencer is very cool. This would be a tool worth porting to Windows.

Now I'm off to write a commissioned article for Developerworks...