Hi all,
after a lot of work on the infrastructure (you can see the Freecell
Solver trunk about that: (
http://svn.berlios.de/viewvc/fc-solve/fc-solve/trunk/ ) I have ran
Freecell Solver on the first 10 million deals of Baker's Game based on
that of the MS-Freecell / Freecell-Pro layouts. The preset chosen was
such that first ran the default heuristic (which may have some false
negatives) and then ran an atomic moves-based solution that guarantees
an accurate verdict.
Here are the results. “gen_states” are the generated states. “iters”
are the iterations' count (they are only different for the solved
boards). sol_lens are the solutions' lengths.
Count solved/gen_states == 7,431,962
Max solved/gen_states == 893,848
Min solved/gen_states == 66
Mean solved/gen_states == 774.802232
StdDev solved/gen_states == 2,043.686579
Median solved/gen_states == 261
Count solved/iters == 7,431,962
Max solved/iters == 893,777
Min solved/iters == 53
Mean solved/iters == 737.625714
StdDev solved/iters == 2,041.805056
Median solved/iters == 223
Count solved/sol_lens == 7,431,962
Max solved/sol_lens == 246
Min solved/sol_lens == 63
Mean solved/sol_lens == 107.457259
StdDev solved/sol_lens == 10.863272
Median solved/sol_lens == 107
Count unsolved/gen_states == 2,568,038
Max unsolved/gen_states == 1,411,608
Min unsolved/gen_states == 38
Mean unsolved/gen_states == 6,668.607073
StdDev unsolved/gen_states == 8,535.782652
Median unsolved/gen_states == 4,160
Count unsolved/iters == 2,568,038
Max unsolved/iters == 1,411,608
Min unsolved/iters == 38
Mean unsolved/iters == 6,668.607073
StdDev unsolved/iters == 8,535.782652
Median unsolved/iters == 4,160
A few observations:
1. The FreeCell FAQ (
http://www.solitairelaboratory.com/fcfaq.html )
claim that Baker's Game has a win rate of 75%) was confirmed as 74.3%
of the deals are solvable.
2. The maximal iterations count for unsolved boards is 1.4 million,
which indicates that unsolvable Baker's Game deals do not usually span
as many positions as their Freecell counterparts.
3. The solution lengths for unoptimised solutions are pretty long at
107 moves mean and median.
You can find the data that was used to calculate the summary here:
http://prdownload.berlios.de/fc-solve/Bakers-Game-First-10Million-Deals-Statistics-4-Freecells-v0.2.0.perl-storable.xz
Regards,
-- Shlomi Fish
--
------------------------------------------
Shlomi Fish http://www.shlomifish.org/
Electrical Engineering studies. In the Technion. Been there. Done
that. Forgot a lot. Remember too much.
Received on Sun Jun 12 2011 - 14:47:26 IDT