Ronald de Man aka syzygy.
Again unfortunately i have a (last) necessity to respond now because of self defense after some unjust erroneous
statements but i can't and will not continue with this ad infinitum most of all it has become a silly yes/not
discussion instead of a proper academic debate (but then this forum also is not suitable for such debates
Your suggestion that i haven't read any definitions of weakly and strong solutions
is beyond preposterous (*). Rather than continuing your condescending remarks i again suggest you
contact vd Herik or else (maybe better) Jos Uiterwijk (see below) to get their opinion about some
nomenclature and definitions, in particular your obsessive desires to have rigorous proofs for almost
everything in game theory, whereby this nowadays is not required anymore because since a few
decades with have other disciplines such as complexity theory and graph/network theory
whereby most of the findings in that area (especially emerging stuff in complexity theory
don't require rigorous mathematical proofs because they are too complex for that.
Nash and von Neumann/Morgenstern also already drove game theory into the
field of complexity, and certain elements in modern game theory (as the Nash equilibrium
in many situations have evolved beyond the stage that you always need a rigorous proof;
so rather then talking about math purists (in the sense of people erroneously adhering
to theoretical math formalism in the Hilbert style which doesnt existas Goede proved), i
should better talk about 'old fashioned math purists' (go back to 1880 i suggest).
Anyway, your earlier comment about the Riemann function already was insulting, as if i
would not be able to comprehend such stuff, while i already in earlier postings talked
about 'conjecture' (that chess is a draw) in an attempt to compromise with you;
a futile attempt apparently, because you seem try to continue to throw mud against me
now almost in similar style as this mr Kobra. So i'll have to report these remarks
(first Riemann) and then your last insinuation that i haven't read the definitions
of weak and strongly solving to the moderator (who imo should close this thread).
Amen, and as i said, i have other (and far better) things to do.
(*) except maybe when i'm disputing *your* interpretation of 'ultraweak solving'
and again, only definition we have so far is 'determining the outcome'
Determining can be done without the need for a rigorous mathemetical proof,
but like i said, i delay my judgment on this issue depending on what vd Herik
(upon your request) is going to say about this issue, ie. confirming your opinion
on the need for a rigorous proof (like for Hex) or not.
(**) https://project.dke.maastrichtuniversit ... er_jos.htm
Is there any project coming to solve chess?
Moderator: Ras
-
- Posts: 1032
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: Is there any project coming to solve chess?
Last edited by jefk on Sun Nov 26, 2023 7:02 pm, edited 1 time in total.
-
- Posts: 10895
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: Is there any project coming to solve chess?
I do not know if there is a name for a strong strategy that also has the following properties:
1)If the position is a win then you play the move that leads to the shortest win.
2)if the position is a draw then you play the move that leads to the longest draw assuming the opponent prefers a shorter draw.
3)If the position is a loss then you play the move that leads to the longest loss.
-
- Posts: 5738
- Joined: Tue Feb 28, 2012 11:56 pm
Re: Is there any project coming to solve chess?
If you did not read them, then you have shown that you did not understand them. I did not wish to imply that you did not understand something if there was still the potential explanation that you had chosen not to read the definitions.
Bye bye.
-
- Posts: 5738
- Joined: Tue Feb 28, 2012 11:56 pm
Re: Is there any project coming to solve chess?
I understand what you mean, but a "weak solution" (or a "strong solution") does not require the strategy to be the quickest possible or otherwise "optimal".Uri Blass wrote: ↑Sun Nov 26, 2023 6:43 pmI do not know if there is a name for a strong strategy that also has the following properties:
1)If the position is a win then you play the move that leads to the shortest win.
2)if the position is a draw then you play the move that leads to the longest draw assuming the opponent prefers a shorter draw.
3)If the position is a loss then you play the move that leads to the longest loss.
In any event, if 1.e4 f6 2.d4 g5 3.Qh5# is a perfect game, then it might be (part of) an optimal winning strategy for white. A (winning) weak solution for white requires white to have a winning strategy against any counterplay by black.
However, it is of course difficult to be sure that, if white has a mate in N in the initial position, then 1.e4 leads to a loss in N-1 for black.
We can also not be sure that, if white has a mate in N' after 1.e4 f6, then the position after 1.e4 f6 2.d4 is a loss in N'-1 for black. There might be a "better" move for white than 2.d4 in that position.
But requiring a DTM-optimal strategy seems gross overkill to me. Also, it raises the question what to do with the 50-move rule. I guess you would need a DTM50-optimal strategy, where the best move in a position may vary with the value of the 50-move counter.
Since a game of chess only has 3 possible outcome values: win (+1), draw (+1/2), loss (0), it seems better to just require a strategy that achieves the optimal value.
-
- Posts: 354
- Joined: Thu Jul 21, 2022 12:30 am
- Full name: Chesskobra
Re: Is there any project coming to solve chess?
It is these claims that are not helping your cause here. Nowadays no rigorous proofs are required in these areas? Do you follow any literature in complexity theory or game theory or graph theory or any other area of mathematics? Anyway, you have probably already left this thread (for the 2nd time?) after multiple clarifications just before leaving. I won't be debating with you, although I ain't leavin the thread.jefk wrote: ↑Sun Nov 26, 2023 6:35 pm Ronald de Man aka syzygy.
nomenclature and definitions, in particular your obsessive desires to have rigorous proofs for almost
everything in game theory, whereby this nowadays is not required anymore because since a few
decades with have other disciplines such as complexity theory and graph/network theory
whereby most of the findings in that area (especially emerging stuff in complexity theory
don't require rigorous mathematical proofs because they are too complex for that.
Nash and von Neumann/Morgenstern also already drove game theory into the
field of complexity, and certain elements in modern game theory (as the Nash equilibrium
in many situations have evolved beyond the stage that you always need a rigorous proof;
so rather then talking about math purists (in the sense of people erroneously adhering
to theoretical math formalism in the Hilbert style which doesnt existas Goede proved), i
should better talk about 'old fashioned math purists' (go back to 1880 i suggest).
-
- Posts: 1032
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: Is there any project coming to solve chess?
- syzygy
so i don't understand the definition of weakly solving , yeah sure, that must be it mr syzygy, Lol (hint, irony,sarcasm).
It's actually the other way around, You Do Not understand the
definition of ultraweakly solving, because to determine is not same as to prove to determine is to determine (beyond reasonable doubt);
which i've done. And to prove is to give a mathematical proof; which i not have done.
Y ou tried to find a way out by stating that to determine is to establish and that to establish is to prove but
that again is Not logical reasoning, it's cognitive bias.
Also by assuming that vd Herik 'means' proving instead of determining when he states to determine is only an assumption
and quite frankly i don't care what vd H thought when he used the word to determine, if he wanted to state the
need for a proof, then he should have said so.
syzygy you suggested to find winning lines agains a 'deterministic' engine, apparently you never
heard about the concept of book learning which by now is known for some four decades or so i estimate
(it was mentioned eg by Hyatt as a simple method to avoid repetition of lines; some engines
haven an experience.bin file for such purposes, you know)
PS indeed ChessKobra, in complexity and network theory you don't always need a rigorous proof;
read the book 'a new Kind of Science' by Stephen Wolfram. In some (many) situations eg.
for cellular automata (or the game of life by Conway) you ony can determine the outcome by actual
simulation| you cannot calculate the result on beforehand, nor can you Prove anything.
You only can Determine the outcomes (almost same lecture to you as i given to syzygy)
Did you study math ? Where chesskobra, which endmarks (A/B/C)? what was your Msc thesis ? and what is your ICCF rating ?
goodbye
(yes i'm out of this thread and imo the thread should be closed anyway discussion with me closed because both of you don't
come any further besides repetitive yes/no arguments and some ad hominem claims or doubting my knowledge
(eg where i have studied complexity theory etc. (well that's none of your business mr Kobra, in fact i have two academic
master titles and that's sufficient to also being able to do some more self study rather then making insinuating
and condescending derogatory comments and engaging in meaningless discussions in -almost- meaningless (talkchess) forum like you do
(where the moderators apparently are on holiday or so).
so i don't understand the definition of weakly solving , yeah sure, that must be it mr syzygy, Lol (hint, irony,sarcasm).
It's actually the other way around, You Do Not understand the
definition of ultraweakly solving, because to determine is not same as to prove to determine is to determine (beyond reasonable doubt);
which i've done. And to prove is to give a mathematical proof; which i not have done.
Y ou tried to find a way out by stating that to determine is to establish and that to establish is to prove but
that again is Not logical reasoning, it's cognitive bias.
Also by assuming that vd Herik 'means' proving instead of determining when he states to determine is only an assumption
and quite frankly i don't care what vd H thought when he used the word to determine, if he wanted to state the
need for a proof, then he should have said so.
syzygy you suggested to find winning lines agains a 'deterministic' engine, apparently you never
heard about the concept of book learning which by now is known for some four decades or so i estimate
(it was mentioned eg by Hyatt as a simple method to avoid repetition of lines; some engines
haven an experience.bin file for such purposes, you know)
PS indeed ChessKobra, in complexity and network theory you don't always need a rigorous proof;
read the book 'a new Kind of Science' by Stephen Wolfram. In some (many) situations eg.
for cellular automata (or the game of life by Conway) you ony can determine the outcome by actual
simulation| you cannot calculate the result on beforehand, nor can you Prove anything.
You only can Determine the outcomes (almost same lecture to you as i given to syzygy)
Did you study math ? Where chesskobra, which endmarks (A/B/C)? what was your Msc thesis ? and what is your ICCF rating ?
goodbye
(yes i'm out of this thread and imo the thread should be closed anyway discussion with me closed because both of you don't
come any further besides repetitive yes/no arguments and some ad hominem claims or doubting my knowledge
(eg where i have studied complexity theory etc. (well that's none of your business mr Kobra, in fact i have two academic
master titles and that's sufficient to also being able to do some more self study rather then making insinuating
and condescending derogatory comments and engaging in meaningless discussions in -almost- meaningless (talkchess) forum like you do
(where the moderators apparently are on holiday or so).
-
- Posts: 10895
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: Is there any project coming to solve chess?
If I understand correctly the disagreement is about what is the meaning of ultraweakly solving a game and if you need mathematical proof that the game is a draw or you only need to be sure above reasonable doubt.jefk wrote: ↑Mon Nov 27, 2023 4:19 am - syzygy
so i don't understand the definition of weakly solving , yeah sure, that must be it mr syzygy, Lol (hint, irony,sarcasm).
It's actually the other way around, You Do Not understand the
definition of ultraweakly solving, because to determine is not same as to prove to determine is to determine (beyond reasonable doubt);
which i've done. And to prove is to give a mathematical proof; which i not have done.
Y ou tried to find a way out by stating that to determine is to establish and that to establish is to prove but
that again is Not logical reasoning, it's cognitive bias.
Also by assuming that vd Herik 'means' proving instead of determining when he states to determine is only an assumption
and quite frankly i don't care what vd H thought when he used the word to determine, if he wanted to state the
need for a proof, then he should have said so.
syzygy you suggested to find winning lines agains a 'deterministic' engine, apparently you never
heard about the concept of book learning which by now is known for some four decades or so i estimate
(it was mentioned eg by Hyatt as a simple method to avoid repetition of lines; some engines
haven an experience.bin file for such purposes, you know)
PS indeed ChessKobra, in complexity and network theory you don't always need a rigorous proof;
read the book 'a new Kind of Science' by Stephen Wolfram. In some (many) situations eg.
for cellular automata (or the game of life by Conway) you ony can determine the outcome by actual
simulation| you cannot calculate the result on beforehand, nor can you Prove anything.
You only can Determine the outcomes (almost same lecture to you as i given to syzygy)
Did you study math ? Where chesskobra, which endmarks (A/B/C)? what was your Msc thesis ? and what is your ICCF rating ?
goodbye
(yes i'm out of this thread and imo the thread should be closed anyway discussion with me closed because both of you don't
come any further besides repetitive yes/no arguments and some ad hominem claims or doubting my knowledge
(eg where i have studied complexity theory etc. (well that's none of your business mr Kobra, in fact i have two academic
master titles and that's sufficient to also being able to do some more self study rather then making insinuating
and condescending derogatory comments and engaging in meaningless discussions in -almost- meaningless (talkchess) forum like you do
(where the moderators apparently are on holiday or so).
I do not think that this discussion is important and I can learn nothing about the education of the participants only because of their opinion about the subject.
-
- Posts: 1032
- Joined: Sun Jul 25, 2010 10:07 pm
- Location: the Netherlands
- Full name: Jef Kaan
Re: Is there any project coming to solve chess?
Uri Blass
well spoken and I agree with you, which is why i asked the moderators to close this thread
And while the nitpicking probably will continue if the thread isn't closed,
an oldfashioned math purist might say, ok determining doesn't mean proving
but the word solution in ultraweak solution means proving.
Then i say, well ok, but not rigorous proving; in fact i have proven that chess is a draw, via Mcts and Zermelo.
Coz i claim:
Lemma:
1) *if* there is a winning strategy then Mcts will find it (like it will
find a possible winning strategy for Black against 1.g4 (attention towforce(!)
Emperical finding:
2) we cannot find such a winning strategy (neither can the Chin database
ergo (proof by contradiction)
3) There is No such winning strategy
Ergo (conclusion)
4) Zermelo (chess is a draw
QED
Bbecause of point 1 and 2 have not been established (proven) rigorously well i todisagree, it's
just a matter of time (increasing search) second, believe it or not, oldfashioned math purists,
not all math proofs have to be solid, there is a difference between ordinary proofs and
100 pct rigorous(*) solid proofs. Have a look a these two AI answers for example:
https://poe.com/s/ImNaHFYZymonlcRAh8pe
https://poe.com/s/xD0WGaEH0uiStTBM4IPE
(what it wrote about Hex might not be entirely correct, but anyway you may get the main idea)
So goodbye again, i'm intending to draft a paper (with no hurry), for proofreading by someone as eg. Jos
Uiterwijk (for me vd H is out of this field(**)) and in the meantime the one or two oldfashioned math
purists here without Phd are not qualified to judge on this forum my prelimenary work and
condemning it in improper hostile and negative ways before i even have done a proper writeup
Amen, byebye
(*) syzygy came up (twice) with the example of a Riemann function in a nasty way suggesting i
don't know the difference between a conjecture and a proof; Such an example isn't fair because
the Riemann function is infinite and chess isn't because of the 50 move draw rule; the max
game length in chess is only something below 200 or 300 or so (look it yup). Also the Riemann
function is not a game (2 person perfect infop blabla) like for which the Zermelo theorem applies.
The Zermelo theorem is not a conjecture, it has been proven (in set theory)
https://arxiv.org/abs/1111.6991
although initially some oldfashioned math purists disagreed (later it was a certain Kalmar
who made additional comments and the theorem became well accepted
https://en.wikipedia.org/wiki/L%C3%A1sz ... alm%C3%A1r
(**) dunno if he still is ICGA chairman but i'll look it up
well spoken and I agree with you, which is why i asked the moderators to close this thread
And while the nitpicking probably will continue if the thread isn't closed,
an oldfashioned math purist might say, ok determining doesn't mean proving
but the word solution in ultraweak solution means proving.
Then i say, well ok, but not rigorous proving; in fact i have proven that chess is a draw, via Mcts and Zermelo.
Coz i claim:
Lemma:
1) *if* there is a winning strategy then Mcts will find it (like it will
find a possible winning strategy for Black against 1.g4 (attention towforce(!)
Emperical finding:
2) we cannot find such a winning strategy (neither can the Chin database
ergo (proof by contradiction)
3) There is No such winning strategy
Ergo (conclusion)
4) Zermelo (chess is a draw
QED
Bbecause of point 1 and 2 have not been established (proven) rigorously well i todisagree, it's
just a matter of time (increasing search) second, believe it or not, oldfashioned math purists,
not all math proofs have to be solid, there is a difference between ordinary proofs and
100 pct rigorous(*) solid proofs. Have a look a these two AI answers for example:
https://poe.com/s/ImNaHFYZymonlcRAh8pe
https://poe.com/s/xD0WGaEH0uiStTBM4IPE
(what it wrote about Hex might not be entirely correct, but anyway you may get the main idea)
So goodbye again, i'm intending to draft a paper (with no hurry), for proofreading by someone as eg. Jos
Uiterwijk (for me vd H is out of this field(**)) and in the meantime the one or two oldfashioned math
purists here without Phd are not qualified to judge on this forum my prelimenary work and
condemning it in improper hostile and negative ways before i even have done a proper writeup
Amen, byebye
(*) syzygy came up (twice) with the example of a Riemann function in a nasty way suggesting i
don't know the difference between a conjecture and a proof; Such an example isn't fair because
the Riemann function is infinite and chess isn't because of the 50 move draw rule; the max
game length in chess is only something below 200 or 300 or so (look it yup). Also the Riemann
function is not a game (2 person perfect infop blabla) like for which the Zermelo theorem applies.
The Zermelo theorem is not a conjecture, it has been proven (in set theory)
https://arxiv.org/abs/1111.6991
although initially some oldfashioned math purists disagreed (later it was a certain Kalmar
who made additional comments and the theorem became well accepted
https://en.wikipedia.org/wiki/L%C3%A1sz ... alm%C3%A1r
(**) dunno if he still is ICGA chairman but i'll look it up
-
- Posts: 354
- Joined: Thu Jul 21, 2022 12:30 am
- Full name: Chesskobra
Re: Is there any project coming to solve chess?
The lemma is wrong. MCTS does not converge min-max for arbitrary games. It is known to converge for certain board filling games like Go and Hex. See the wiki link I posted before and this reference: Browne et al. A survey of Monte Carlo tree search methods IEEE TRANSACTIONS ON COMPUTATIONAL INTELLIGENCE AND AI IN GAMES, VOL. 4, NO. 1, MARCH 2012.
Also, even if it converged for some games, for large complex games it would require enormous number of simulations (may be of the order of the number of games). Lc0 uses a combination of MCTs and NN, so that is another problem using any results about MCTS.
-
- Posts: 5738
- Joined: Tue Feb 28, 2012 11:56 pm
Re: Is there any project coming to solve chess?
I thought you were gone. This is getting boring now.