It wasn't until my last set of modifications to my solver that I truely
appreciated 8x4 FreeCell as a game. Why? Because the win rate is only
tied to a few move types to win almost all of the solvable deals in the
first 1,000,000 deals.
I recently upgraded my solver to replace some convoluted pruning logic
with simpler, yet more comprehensive, pruning logic. At the same time, I
decided to add additional constraints to my Pass_1 logic in order to
filter out move types that I felt were of little value in finding a
solution. I was mildly surprised to discover that there were only two
solvable deals that couldn't be solved with my latest Pass_1 logic.
Bottom Line: I was hoping to start a discussion about partitioning deals
into difficulty classes based on the move types needed to solve them,
but that now appears to be a fruitless venture. Oh Well.
Received on Fri Apr 05 2013 - 11:14:30 IDT