Adam Shelly
12/13/2005 6:09:00 AM
Cool.
I'm suprised RemoveRight did better than DeepScore.
I was looking more at HistoryPlayer, (which should do better than
Pessimistic, since it uses the same choice for any unknown situations)
and I realized that when scoring a move, it is giving too much weight
to the subsequent turns. So it can choose the absolute best move on
turn 2, for instance, then make a bad move 3 turns later, and end up
ranking the turn 2 choice as the worst possibility. So for now, the
history information it keeps is mostly useless, except for a speedup.
My history algorithm needs some tuning (if it can be salvaged at all
:)
I'd be curious to see what happens if you add the other submitted
players to the tournament. Can you post the tourney framework?
-Adam
On 12/12/05, David Balmain <dbalmain.ml@gmail.com> wrote:
> Hi Adam,
> Just ran a tournament with your players. In case you or anyone else is
> interested;
>
> APessimisticKalahPlayer scored 527 points in 2265.782466 seconds
> AHistoryKalahPlayer scored 478 points in 98.2062879999999 seconds
> RemoveRightKalahPlayer scored 469 points in 0.004744 seconds
> DeepScoreKalahPlayer scored 460 points in 0.055886 seconds
> ScoreKalahPlayer scored 450 points in 0.020305 seconds
> RemoveRandomLowKalahPlayer scored 312 points in 0.012781 seconds
> RemoveRandomHighKalahPlayer scored 309 points in 0.013103 seconds
> RemoveHighKalahPlayer scored 264 points in 0.005501 seconds
> RemoveLowKalahPlayer scored 187 points in 0.007318 seconds
>
> Cheers,
> Dave
>
> On 12/13/05, Adam Shelly <adam.shelly@gmail.com> wrote:
> > Of course I introduced an error while cleaning up my code:
> > That will teach me to skip unit tests...
> >
> > in AdamsPlayers.rb, line 111
> > i = best_move(b) if taketurn
> > should be
> > m = best_move(b) if taketurn
> >
> > -Adam.
> >
> >
> > On 12/12/05, Adam Shelly <adam.shelly@gmail.com> wrote:
> > > Here are my Kalah Players.
> >
> >
> >
>
>