I would like to measure kangaroo DP for puzzle #120
I always wondered what the relation was between DPsize and runtime. Maybe somebody should draw a few graphs. They will be parabolic because setting the size too low or high exponentially increases runtime, and if we could even make a 3D graph lining up the DP curves from all range sizes (powers in this case).
The idea would be to find the smallest points on the graph to see if there's a way to improve the automatically calculated dpsize value. (I once experimented with setting an arbitrary dpmask, but that had very bad effects for runtime.)