2012 in review (part 1 of 2)

So it turns out I have a lot to say about how my 2012 went. Part 2 to follow!

In January, I began a researchassistant position with my advisor, Ryan, marking the first time in grad school that I've had no major responsibilities other than research (that is, no classes and no teaching). In December, Ryan and I had submitted an NSF grant proposal about lattice-based deterministic parallel computation, so now we began fleshing out the ideas we'd described in the proposal, putting together a semantics for a small language with lattice-based data structures that we called LVars.

Alex and I started training for the 2012 Toronto Marathon, and I opened aFitocracy account on January 1 with ambitious plans to log 1,000 miles by year's end.

I spent some time helping organize InWIC, a conference for women in computing in Indiana; the main thing I did was work together with my friends Will and Ian at Studio Cypher to design a cooperative puzzle game for the InWIC attendees to play.

At the end of the month, I went on a delightful road trip to POPL in Philadelphia with a whole pile of folks from IU and CMU. Highlights included playing board games in the hotel lobby and being too scared to say hello to Simon Peyton Jones in the elevator (later I got over it and we had a lovely chat).

In February, Ryan and I continued working on our project. As our LVar-based language began to solidify, I started working on a determinism proof for it, taking the determinism proof for the Featherweight CnC language in this paper as a starting point and frequently soliciting Amr's advice. It had actually been the Featherweight CnC proof that had gotten me started thinking about the relationship between monotonicity and determinism in the first place, so it was interesting to go back to that proof and try to adapt it to our setting.

It began to sink in that I was headed toward doing a Ph.D. with Ryan on deterministic parallelism, which was pretty exciting. Deterministic parallelism was still rather unfamiliar territory for me, but now that I had a big proof to work on, I was in my element. Amr also agreed to join my nascent thesis committee.

As March began, I was getting close -- or so I thought! -- to finishing the proof, and we considered submitting something to ICFP, but with the proof still not done, we decided we'd better wait and shoot for POPL in July instead.

During IU's spring break in mid-March, Alex and I went to Boston for a few days, where I spent some time working with Amal on the still-unfinished multi-language parametricity project I was doing with her. While there, I told Amal about what I was working on with Ryan, and when I showed her one of the properties I'd had to prove to make our determinism proof go through, she pointed out that it looked sort of like a frame rule. At the time, I didn't really understand what that meant, but the idea stuck with me.

While I was in Boston, Amr found a bug in my proof that had to do with the names of memory locations and the circumstances under which they could be renamed. It wasn't a showstopper, but it would continue to plague us, in one guise or another, for several months.

When I got back to Bloomington, after a couple of meetings in which Ryan, Amr, and I all found ourselves evaluating expressions by hand on Amr's whiteboard, we realized that we really needed a runnable model of our language, so I started hacking a Redex model together. I had a lot of fun bringing the paper semantics to life. On March 30th, which was my 30th birthday, I celebrated by giving a whiteboard talk at PL-wonks about what Ryan, Amr, and I had been doing, followed by a Redex demo.

The next day, Alex and I