Tetris Attack AI - Java-Games

nathan@visi.com (Nathan Mates) wrote in
news:41f15836$0$95816$a1866201@visi.com:

> In article
> <1106321284.065599.223240@z14g2000cwz.googlegroups.com>,
> <swanofnever@gmail.com> wrote:
>>the main thing is: at each iteration there are (at most) 60
>>possible switches.
With how many positions?

> I don't think you fully read my first response to you. You
> *really
> need to look up MinMax trees*. I repeat that: go to
> http://www.google.com/ and look that up. They're directly
> related to this exact problem. I mean it.

I hope you looked at "alpha beta".
It's not useable for my current game bacause it has branching
factor over 200, but it might work better for you.

Nathan Mates wrote:
> I don't think you fully read my first response to you. You *really
> need to look up MinMax trees*. I repeat that: go to
> http://www.google.com/ and look that up. They're directly related to
> this exact problem. I mean it.
>
> Chess has far more possibilities, yet it's been done on computers
> simpler than a SNES. Minmax makes it possible.

Um...the difference between Chess and Tetris Attack being that chess
moves are made alternately by opposing players and in Tetris Attack you
make all the moves...?

The whole concept of MinMax relies on opposition maximizing their
advantage for every move.

How do you apply that to Tetris Attack where you don't have opposing
moves and the mini-max principle does not apply very clearly?


If I understand Tetris Attack correctly it looks like you could get a
pretty good AI by just grouping tokens equal to the token dropping
around the impact area. Your target should be the 7 squares around the
impact point. The 8th square is right above it and obviously it is
empty.

Now make a list of all the tokens equal to the one dropping. Calculate
how many "swaps"/"moves" it would take to get each of them into the
target 7 squares. Keep moving the tokens to the 7 sqaures until they
hit a target square or an equal token until you have impact. Reapeat
for new token.

T
PS: I might be confused about how Tetris Attack works, never played it.

r@mail.com wrote:
> nathan@visi.com (Nathan Mates) wrote in
> news:41f15836$0$95816$a1866201@visi.com:
> > In article
> > <1106321284.065599.223240@z14g2000cwz.googlegroups.com>,
> > <swanofnever@gmail.com> wrote:
> >>the main thing is: at each iteration there are (at most) 60
> >>possible switches.
> With how many positions?
> > I don't think you fully read my first response to you. You
> > *really
> > need to look up MinMax trees*. I repeat that: go to
> > http://www.google.com/ and look that up. They're directly
> > related to this exact problem. I mean it.
> I hope you looked at "alpha beta".
> It's not useable for my current game bacause it has branching
> factor over 200, but it might work better for you.

Thanks!

waylon_w@users.cidermail.com