Arotau

Last few days

Since the last PhD entry:

• Met briefly with Alan on Thursday.
• In the AT-HBR completeness proof, need to check out the case when HBR produces the nucleus clause.
• Ditched the old NR chapter, and started again with a completely new outline. Copied and pasted various bits into the new chapter.
• Spent the weekend working on AT-HBR experiments. Finally managed to fix the last remaining bugs, but it all just runs too slowly. Figured out a faster method of eliminating trivial clauses, and some ideas for incremental updating of the AT.
• Realised on Monday that I needed to get back to the NR chapter if I was to get it finished by Tuesday. While working on the results sections, I decided that I needed more experiments, especially after finding that some of the earlier experiments had run out of memory on Alan’s machine (which only has 256MB of RAM). Now I need to rerun them on my machine (which has 1GB), but they will take on the order of 30 hours to run. This also means that I can’t run any more AT-HBR experiments until the NR ones have finished, as my machine is the only one with enough memory to mean that rerunning the experiments won’t be necessary.
• Still need to code up the code to fix the branch order, so as to replicate the heuristic interaction work from the HBR chapter.