On Friday 15 May 2009 01:44:00 Gary Campbell wrote:
> > A meta-move is a move that consists of several single or multiple card
> > moves done at once. For example, my solver may opt to move a card that's
> > covered by several cards to the foundations, by first moving the cards on
> > top of it to free freecells, or to free stacks. I have a move function
> > that attempts to do this, and as far as the solver is concerned, it moves
> > from the original state to the derived state with the card already placed
> > in the foundations in one step.
> >
> > I should note that FCS still keeps track of the individual moves that are
> > conducted by this move, so it can later replay the solution step-by-step.
>
> If I understand it, this appears to be very
> similar to what I call Macro move sequences.
> I note a "theme" for a sequence of moves
> and prune all moves that don't contribute to
> it while it is going on. This might be a good
> topic for further discussion.
I used to call meta-moves "multi-moves". According to the report here, written
by two students of "Graduate-Level A.I.", these should be called meta-moves:
http://kevin.atkinson.dhs.org/freecell/
http://www.google.com/search?q=%22meta%20moves%22 - this search, however,
returns FCS-related results at the top.
>
> > Well, on game No. 24,795,893 my solver (with its "-l gi" configuration)
> > has done 11,414,822 iterations and consumed over 50% of my memory (2.5 GB
> > RAM + 1 GB swap) before I gave up and killed it.
> >
> > With game No. 53,687,601 my solver (with its "-l gi" configuration) has
> > done 13,593,338 iterations and consumed over 50% of my memory, before I
> > gave up and killed it.
> >
> > With the Cookie game #2:
> >
> > I could not solve this game.
> > Total number of states checked is 1158165.
> > This scan generated 642361 states.
> > }}}
> >
> > How does your solver fare on those?
>
> I think V6.1 runs over 5 minutes without
> a resolution on all three, but versions
> of the solver I'm working on now have
> produced "impossible" fairly quickly.
> but I'm not satisfied that the same versions
> haven't produced some false negatives as
> well.
Do you mean that you're not satisfied that they did produce some false
negatives? Does making these deals produce "impossible" quickly involved
applying some heuristics that resulted in more possible false negatives?
> Suffice it to say, these are among
> my test cases, and I won't consider my
> solver to be "powerful" unless it handles
> them adequately.
>
> As for the rest of the posting, I have no
> response. I shall continue to study it.
OK.
Regards,
Shlomi Fish
>
>
>
> ------------------------------------
>
> Yahoo! Groups Links
>
>
>
--
-----------------------------------------------------------------
Shlomi Fish http://www.shlomifish.org/
Freecell Solver - http://fc-solve.berlios.de/
God gave us two eyes and ten fingers so we will type five times as much as we
read.
Received on Fri May 15 2009 - 02:13:24 IDT