Arotau

Meeting

Brief meeting with Alan again:

• Described my experimental woes over the weekend. He was as surprised as I was to discover that his machine has no swap.
• Alan suggested an even more improved NR idea: every decision node where the number of undef vars is > 90%, say. Much better than the very crude approximation for the top of the search space I currently have, which completely ignores the effect of unit propagation.
• Decided to set myself some serious deadlines: HBR chapter by tomorrow morning, and the NR chapter by Thursday morning.