RE: TIB: Chess AI
[Prev][Next][Index][Thread]
RE: TIB: Chess AI
On Wed, 08 Sep 1999 23:32:54 PDT, Josh Cunningham wrote:
>Ah, but here is the problem, their are well more then a hundred moves at
>one
>time! You can't just look at the more 'important' pieces on the board, or
>just moving a man to take an opponent. You have to be able to 'look ahead'
>at least one round, else your AI will be pointless. I have though about a
>chess AI engine for a long time ~2 years. You would need to know not only
>the best move at the time but the possibility of a good move at least one
>round from now. That is where the problem lies. Now I can do this, but
>the
>actual implementation isn't fast. What the calcs has to do is figure out
>the best move it can make, with keeping in mind one move from then; and
>then
>it needs to figure out the best move you can make, again keeping in mind at
>least one move ahead. All that is needed is a algorithm to figure out the
>best move, then have the calc 'switch' sides, and figure out your best
>move,
>then move accordingly. But all of this isn't efficient at all. On a side
>note, I have an old apple IIC that has a great chess program. The AI for
>it
>is quite good, especially considering the IIC is only a 2MHz machine! I
>know that the 6305 processor and the Z-80 are quite similar, maybe someone
>could understand the Asm code of it enough to port it to the Calc, which is
>a much faster machine.
>If you need any help, or just questions, please don't hesitate to ask.
>
>Josh Cunningham
>
>>From: "Jeff Tyrrill" <jeff_tyrrill@msn.com>
>>Reply-To: ti-basic@lists.ticalc.org
>>To: <ti-basic@lists.ticalc.org>
>>Subject: RE: TIB: Chess AI
>>Date: Wed, 8 Sep 1999 17:21:42 -0700
>>
>>
>>The program doesn't need to keep a list of all possible moves, except
>>perhaps on one turn. There are no more than probably a hundred, at the
>>very
>>most, moves at any one turn, and usually much less than that. Also, the
>>program could store them a piece at a time, so there would only be around
>>20-25 at the most moves to keep track of at a time, usually much less.
>>When
>>the program is testing each possible move to find the best, it only has to
>>remember the best one(s), not all of them.
>>
>>Somebody already made a TI-Basic chess computer for the TI-85. It's on
>>ticalc.org. It works, but is extremely slow, too slow to be tolerable IMO,
>>but it demonstrates that a calculator chess computer is possible, even in
>>TI-Basic. The program was ported to 85 Basic from a GW Basic program, also
>>included in the ZIP file. On the easiest difficulty level, moves take 1-2
>>minutes to calculate, but they're as good as random moves. On the second
>>difficulty level, moves take several minutes to calculate. I didn't play
>>long enough to see whether the moves were any good, because it was taking
>>so
>>long! :)
>>
>>-----Original Message-----
>>From: owner-ti-basic@lists.ticalc.org
>>[mailto:owner-ti-basic@lists.ticalc.org]On Behalf Of The Incredible
>>Bulk, AKA Prime
>>Sent: Tuesday, September 07, 1999 8:09 AM
>>To: ti-basic@lists.ticalc.org
>>Subject: Re: TIB: Chess AI
>>
>>
>>
>>Suffice to say, the list of possible moves would cause
>>a memory overload and kill the program. You'd have to
>>do it on something w/more space, aka a TI-92+ or 89
>>(maybe even a 86, if you think about it.). Also, If
>>you can get your hans on one, you could try it as an
>>APP (residing in the flash memory area) to make it a
>>little easier to work with, but impossible to tamper.
>>
>>--- Adam Davis <adavis@baladyne.com> wrote:
>> >
>> > This is a very complex problem. Chess has never
>> > been 'solved'.
>> >
>> > So they use prediction for most chess programs. The
>> > program determines a number
>> > of moves it can make, then using it's own rules,
>> > determines a number of moves
>> > it's opponent would then make. Then it finds
>> > (through some quantifying process)
>> > which positions end up with giving itself the best
>> > chance to win. then it sees
>> > which of it's moves would have the best 'score', and
>> > makes that move. Really
>> > fast computers go and look ahead at 5-50 moves after
>> > its current position.
>> >
>> > -Adam
>> >
>> > LtERT01@aol.com wrote:
>> > >
>> > > I'm developing a chess program for the 83 (you may
>> > have heard of it,
>> > > Chessmaster 83, listed on ticalc.org and
>> > ti-files.org) and I want to try to
>> > > write an AI for it. If anybody out there has any
>> > suggestions on how to do
>> > > that, I'd be appreciative. So far, all I can
>> > figure out to do is to have it
>> > > go through all the possible moves, and assign them
>> > values based on piece
>> > > count and position, but that would take a long
>> > time and it would be pretty
>> > > hard to get the calculator to quanitify a board
>> > position. So... I need a
>> > > little help.
>> > >
>> > > Eric Tollefson
If someone makes a good Chess AI program for the 83, I can attempt to port
it to ASM. That would get rid of the speed problems.
James Vernon
jimbob_v@hotmail.com
______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com