I really appreciate the commitment to bench-marking in this one. The memoization speedup for number processing was particularly surprising.