Danny A. Jones suggested I send a copy of this message to the Yahoo
group fc-solve-discuss, so here it is:
During the past month or so, Ryan A. Miller ran the first 100 million
deals of
FreeCell Pro, using Holroyd's Patsolve solver (F9). The runs
produced a list of
1272 impossibles and 136 intractables.
All of these 1408 deals were run by me through the regular (F7)
solver in
FCPro at 5 freecells. All but one of the deals, 14720822, are
solvable with
5 freecells (14720822 was originally found a couple of years ago by
Dr.
Tom Holroyd using his Patsolve solver).
Next the 136 intractables were run through the F7 solver at 4
freecells.
This initially produced 101 wins, 5 impossibles, 26 intractables, and
4 deals
which crashed the program. A second run of these 26 intractables at
longer
time limits and using solver mode W found 19 more wins, 3 more
impossibles,
and only 4 intractables. The four deals which crashed the program
were then
put into Gary Campbell's solver, and all four were found to be
winnable. His
solver also found that one of the four remaining intractables was
impossible,
and found a win for a second; the F7 solver in FCPro (with a still
longer time
limit and mode X) found a solution to another (Danny A. Jones' solver
also
found one for the same deal), and Danny A. Jones' solver found that
the last
one was impossible. Adding two impossibles to the eight already
found by FCPro makes ten in all, and a total of 1282 for the first
100 million
deals. A file including summaries of all of the runs is available
from me by
request.
Michael Keller
Received on Wed Oct 12 2005 - 14:47:28 IST