I’m making large Breadth-first searches with combinatorial explosions, where a cache/memoization using the process dictionary is our only salvation, but only after having gotten it wrong quite a few times. Source at https://gist.github.com/ferd/f23d917e0739df3ca32fc5f34424c755
Ends up taking about 21 and 15 seconds for parts 1 and 2 respectively, but I didn’t feel like making it faster.
