Here's a tic tac toe board
CHESS IS NOTHING BUT A LARGER VERSION OF TIC-TAC-TOE!!!!!!
Sort:
WindowsEnthusiast wrote:
teju17 wrote:
SO, what if WE make a chess engine that is NOT A.I, and JUST AN INTERACTIVE MEMORY IN WHICH WE INSERT THE MEMORY LIKE: IF E4, PUT E6. AND IT NEVER LOSES EVEN TO ALPHA ZERO,........................
should i start planning on how to make one?
Sure, you can start planning on how to make one. A good place to start is to get a degree in computer science so you understand the challenges inherent to this. I don't know if you have any idea of how difficult this is; if it were this easy, chess would've been solved long ago.
no computer degree needed. stackoverflow is enough
blitz2009 wrote:
Are you guys that stupid? Do you really think computers will solve chess. You guys don’t even understand how many possible positions there are. It’s almost impossible to even say how much positions are possible so how will computers solve chess. I hear this crap almost everyday now. If you think it’s possible go spend the rest of your dumb life making some computer that can solve chess. You don’t think stockfish developers have already tried this? It’s simply not possible if you think it’s possible your just a arrogant idiot so just shut up.
clearly you know nothing about A.I.
@blitz2009, it is possible, but it will take too much time, and the result wouldn't even be that fruitful.
kk
blitz2009 wrote:
Clearly ik nothing about AI? Can you just shut up? Your in all of my forum posts talking negatively about my posts. I have already blocked you. Second of all let’s all just shut up. If you want to go spend the rest of your life about this than go do it
how am I talking negatively about your posts. literally that's all you do in YOUR posts. "Are you guys that stupid?." "Your just a arrogant idiot so just shut up." you should honestly be reported, you've been doing this in countless threads for way too long
Aug 21, 2021
0
#68
WindowsEnthusiast wrote:
Not so. While the game tree out to four moves is trivial to evaluate for a modern chess engine, generalized Tic-Tac-Toe is PSPACE-complete, while generalized chess is EXPTIME-complete.
No one cares about NxN chess, we only care about 8x8 chess.
blitz2009 wrote:
You shut up idiot go police some other forums you loser
bro you're just an 11/12 year old from what your username says, you're prob gonna need some mental help when you grow up
What about after 5 moves?
But if you limit the search to only the results that are tried and true, then what??
blitz2009 wrote:
Are you guys that stupid? Do you really think computers will solve chess. You guys don’t even understand how many possible positions there are. It’s almost impossible to even say how much positions are possible so how will computers solve chess. I hear this crap almost everyday now. If you think it’s possible go spend the rest of your dumb life making some computer that can solve chess. You don’t think stockfish developers have already tried this? It’s simply not possible if you think it’s possible your just a arrogant idiot so just shut up.
I'm a career computer scientist. You have no idea what you're talking about.
llama47 wrote:
WindowsEnthusiast wrote:
Not so. While the game tree out to four moves is trivial to evaluate for a modern chess engine, generalized Tic-Tac-Toe is PSPACE-complete, while generalized chess is EXPTIME-complete.
No one cares about NxN chess, we only care about 8x8 chess.
These computational complexity results for generalized chess are still a useful guide for 8-by-8 chess algorithm development.
SameerAchhab1 wrote:
WindowsEnthusiast wrote:
teju17 wrote:
SO, what if WE make a chess engine that is NOT A.I, and JUST AN INTERACTIVE MEMORY IN WHICH WE INSERT THE MEMORY LIKE: IF E4, PUT E6. AND IT NEVER LOSES EVEN TO ALPHA ZERO,........................
should i start planning on how to make one?
Sure, you can start planning on how to make one. A good place to start is to get a degree in computer science so you understand the challenges inherent to this. I don't know if you have any idea of how difficult this is; if it were this easy, chess would've been solved long ago.
no computer degree needed. stackoverflow is enough
Stack Overflow won't teach you how to make a quantum computer of sufficient size to solve chess.
tygxc wrote:
#59
Quantum computers are commercially available and hold the potential to be much much faster than conventional computers as reported in scholarly papers. They are capable of boolean operations and those are what is necessary to solve chess.
Short answer: there are 10^40 sensible and legal positions, the 2 prong method needs to visit about the square root of those hence 10^20.
If you want it scholarly:
There is a scholarly thread here by John Tromp who calculated the exact number of possible positions with no excess promotions. He also estimated the fraction of legal positions by sampling.
There is also the scholarly paper by Schaeffer of how he solved checkers which showed he needed only visit a small fraction of total possible positions.
The 50 moves rule and the 3 fold repetition rule are essential for solving chess. Without either 50 moves rule or 3 fold repetition rule chess is no finite game and cannot be solved. With 3 fold repetition rule alone chess is finite and thus solvable, but the complication is enormous.
Forward search is how checkers was solved. A checkers program called Chinook calculated forward towards an endgame table base. For chess this would mean to look at the top 4 possibilities by white and then prove there is 1 move by black that ultimately reaches a draw in say an 8 men endgame table base.
You sidestepped my point about how quantum computers most likely cannot provide an exponential speedup over classical computers. This isn't to say quantum computers aren't, in practical terms, potentially much faster than classical computers; it's just to say that quantum computers most likely can provide at best some subexponential (but in some cases still superpolynomial, cf. Shor's algorithm for integer factorization) speedup.
Also, a nitpick: quantum computers can do boolean operations, but that is not doing them justice. Operations on qubits do not always have classical analogues and that is one source of quantum supremacy as would be needed to solve chess. Quantum computers also provide only a probabilistic result.
The 3-fold repetition rule can be kept, yes (as any forced endless cycle can and should be evaluated as a draw), but the 50-move rule has already been ignored for endgame tablebases as an artificial limitation.
The trouble with a forward search is evaluating which positions are drawn in the first place without exhaustive search, a task far less trivial in chess than in checkers. A 14-piece tablebase, sufficient for most late middlegames, would already blow up the number of positions needed to roughly 10^20 most likely. You can't assume the number of chess positions will be reduced to roughly its square root like it was for checkers. I would be skeptical of any such claim. For one, triangulation is not possible in checkers as far as I can tell, so zugzwang can be identified much more easily and earlier.
#79
Present computers are not yet up to the task right now. So technology has to advance to make it feasible. People point out that Moore's Law is bound to hit physical boundaries like the size of the atom and the speed of light. Quantum computers are a probable replacement for present technology.
"any forced endless cycle can and should be evaluated as a draw"
++ that poses a huge practical problem. Players can shuffle pieces in a 32 men position almost endlessly. Calculation does not halt. We need to keep the 50 move rule, but we can do without it for the table base. That is also how ICCF handles it: 50 moves rule is in force, but you can claim a table base win that exceeds 50 moves.
"The trouble with a forward search is evaluating which positions are drawn"
++ In principle it is easy: calculate forward until a table base of drawn positions is reached and then look up if the position is either in it, then it is a draw, or is not in it, then it is no draw.
"far less trivial in chess than in checkers."
++ true: if checkers is a larger tic-tac-toe, then chess is a larger checkers.
"A 14-piece tablebase, sufficient for most late middlegames, would already blow up the number of positions needed"
++ It is not yet clear how many men tablebase are needed to solve chess. The present 7 men seems too small, but maybe 8 or 9 or 10 will do. In theory 7 would suffice if the forward search is extremely powerful. Present table bases contain illegal positions and non sensible positions like with 3 light square bishops. Those can be pruned. Present table bases also contain DTZ and DTM information unnneeded to solve chess. Last but not least for the purpose of solving chess the table bases only needs to hold the draw positions. So the table base for the purpose of solving chess should just be an array of known draws with exactly x men. It does not need to contain wins and it does not need to contain positions with x-1 men.
"You can't assume the number of chess positions will be reduced to roughly its square root like it was for checkers."
++ Maybe more, maybe less, it will become clear after it is done. For example if you are calculating if 1 e4 e5 2 Nf3 Nc6 3 Bb5 Nf6 is a draw or not, all positions with either a white pawn on e2 or a black pawn on e7 do not need to be assessed. With each pawn move or exchange large numbers of positions will never need access.
Already now in TCEC the engines running a short time on modest hardware hit their 7 men table base around move 10, and much more around move 20. So it is plausible that an engine running a long time on powerful hardware will hit an 8 men tabe base much more frequently. Maybe enough, maybe not yet.
Solving chess will probably happen variation by variation e.g. start with proof that the Berlin is a draw. Also for checkers only a fraction of possible openings needed analysis for the proof.
Interesting opinion by the late GM Sveshnikov:
"Chess is an exact mathematical problem. The solution comes from two sides: the opening and the endgame. The middlegame does not exist. The middlegame is a well-studied opening. An opening should result in an endgame.... Give me five years, good assistants and modern computers, and I will trace all variations from the opening towards tablebases and 'close' chess. I feel that power."
tygxc wrote:
Interesting opinion by the late GM Sveshnikov:
"Chess is an exact mathematical problem. The solution comes from two sides: the opening and the endgame. The middlegame does not exist. The middlegame is a well-studied opening. An opening should result in an endgame.... Give me five years, good assistants and modern computers, and I will trace all variations from the opening towards tablebases and 'close' chess. I feel that power."
I actually did some more research as of late and it appears that quantum computers' speedup likely won't apply to chess. See for example this: https://quantumcomputing.stackexchange.com/questions/5823/will-quantum-computers-be-able-to-solve-the-game-of-chess
Aug 24, 2021
0
#79
tygxc skrifaði:
Interesting opinion by the late GM Sveshnikov:
".... Give me five years...
#86
It is interesting that the late GM Sveshnikov was of the opinion that it would be possible to solve chess in 5 years with present computers and present i.e. 7 men table bases.
Maybe the Lomonosov University could start such a 5 year program, supplying the computers and the good assistants.
Solving chess is a formidable task. Maybe it is easier to slice the salami. Start with just one of the 500 ECO codes, for example the Berlin C67. This not only reduces the effort by a factor of 500, but the calculation towards the 7-men table base can start from a 26 men starting position. 26 is not only closer to 7, research by John Tromp also shows that 28 men has most possible chess positions, so it is an advantage to start below 28 men.
#61
There are exactly 19201527561695835455154058755564594798074 positions possible with one set of chess men. 5.2% of those are estimated to be legal.
The question is not if computers will solve chess, but when.
You may not like it, but computers will probably solve chess before the end of this century.