CPAIOR is an interesting environment for gathering researchers from diverse but close areas, ranging from mathematical programming to artificial intelligence. That reflects the 4 organized master classes, all of which related to search: the first about mixed-integer programming (MIP, that I missed for being late), then constraint programming (CP) with Giles Pesant (I arrived at the middle of it), satisfiability (SAT) with Marijn Heule and now A* with Nathan Sturtevant. The two later speakers were invited for the sake of bringing something different to the conference.
Despite being possible to describe SAT as a proper subset of CP, research on that topic has a very different focus for being concerned with only one type of constraint (predicates) and a bi-valued domain (true and false). With a more strict focus, the SAT community has been presenting outstanding results in the last years. In fact, there are people that envy how fast SAT solvers has been progressing compared to the CP ones. However, that’s the cost of being generalist.
Nathan Sturtevant presented an interesting animated example that does explain why Dijkstra’s shortest path algorithm can be very slow for certain cases, endorsing A* search.
PS: Lunch and coffee breaks are very tasty. I can’t wait for the barbecue and the gala dinner.
Thiago Serra 