Just for comparison, I ran FCELL over the first 1000 games
(this is the released version that has been available on my
website for over two years at
http://numin8r.us/programs).
Not counting automoves, it got:
a minimum solution of 39 moves (game # 251, for example),
a maximum solution of 63 moves (many, such as #305, etc.),
and a median length of 58 (e.g. game #500, ironically).
I'm currently working on a redesigned FCELL that should be
able to get solution lengths approaching a minimum. The
released version (v6.1) was designed simply to get a solution
that could be followed by a human to aid in understanding it.
I believe that solutions over 100 moves are only good for
proving that a game HAS a solution.
-Gary Campbell
----- Original Message -----
From: "Shlomi Fish" <shlomif_at_iglu.org.il>
To: <fc-solve-discuss_at_yahoogroups.com>
Sent: Friday, April 17, 2009 6:56 AM
Subject: Statistical Analysis of the Lengths of Freecell Solver's Solutions
> Hi all!
>
> Inspired by the recent thread about the solution, I set to actually count and
> perform a solution analysis of the lengths of Freecell Solver's solutions. I
> wanted to try both the native Freecell Solver moves and the Freecell
> Pro/standard notation (with automoves) solution. For this, I resurrected the
> files under the following directory:
>
> http://svn.berlios.de/viewcvs/fc-solve/trunk/ext-ifaces/FC-Pro/
>
> I found a few bugs there, and wrote a routine to verify the integriry of the
> intermediate positions, and tested that the solution was OK for the first
> 1,000 boards using the board-loop.rb script. Eventually, that was the case.
>
> Then I integrated this code into the main Freecell code here:
>
> http://svn.berlios.de/viewcvs/fc-solve/trunk/fc-solve/source/
>
> I created the fc_pro_iface.c and fc_pro_range_solver.c and created a new
> uninstalled executable. This one was able to count the number of moves in both
> the native FCS moves and the FC-Pro moves (which don't count auto-moves and
> are shorter).
>
> Here are the results. Note that I forgot to use the -opt flag, and as I tried
> it now it makes the range solver crash due to a bug (which I intend to solve).
>
> {{{{{{{{{{{{{{{{{
> FCS Native Moves - no auto-moves
> --------------------------------
> Min: 74
> Max: 301
> Average: 122.708303384481
> StdDev: 16.9779929322848
> Median: 120
>
> FC-Pro moves - with auto-moves
> ------------------------------
> Min: 38
> Max: 253
> Average: 91.5467670864709
> StdDev: 17.5190178761448
> Median: 89
> }}}}}}}}}}}}}}}}}
>
> At first I thought I spent too much time working on this, but now I see that
> the fact I detected the bug with the -opt and the range solver was worth it.
>
> Regards,
>
> Shlomi Fish
>
> --
> -----------------------------------------------------------------
> Shlomi Fish http://www.shlomifish.org/
> Funny Anti-Terrorism Story - http://xrl.us/bjn7t
>
> God gave us two eyes and ten fingers so we will type five times as much as we
> read.
>
>
Received on Fri Apr 17 2009 - 07:41:11 IDT