Website founded by
MatPlus.Net Forum General Reciprocal Stipulations
You can only view this page!
|Page: [Previous] [Next] 1 2 |
|(21) Posted by Hauke Reddmann [Thursday, Mar 9, 2023 22:13]|
Kevin said: "Here is why Thomas Maeder's Boolean logic proposal is insufficient:..."
That's because Boolean logic has no notion of time. What already might suffice
is some sort of temporal, process or Petri net logic (I had this stuff by the ton
in my CS studium). https://en.wikipedia.org/wiki/Temporal_logic
Note that chess time progresses in halfmoves and temporal logic already has a
"at next time" operator, so "#2" (including the fact there is only one key)
should be expressable as a standard temporal logic formula, as soon as #
is defined as a condition.
|(22) Posted by Kevin Begley [Friday, Mar 10, 2023 04:22]|
I received an email from Vlaicu detailing an idea which greatly simplifies (and extends) this entire methodology (in some surprising ways).
hint: it eliminates the need for "self-" as an independent stipulation. It may also eliminate the need for "==" and "##" as TARGETs.
It seems like a major breakthrough to me, yet it's based upon a simple realization.
Though I could not immediately find any holes in it (and, trust me, I did try), it will require more thought (to identify any special cases).
I'm left wondering: is there anything this will not cover?
I have encouraged him to either post here, or link his article here (if that's the direction he wants to go).
Until then, I'll be busy wondering how I didn't think of this myself.
Symbolic, Axiomatic, Temporal Logic?!
Way to wake the entire neighborhood!
At the moment I was nearly halfway through your linked article, I suddenly realized: time is passing.
Can we actually make use of this (in some specific way)?
|(23) Posted by Hauke Reddmann [Friday, Mar 10, 2023 10:02]|
@Kevin: Making use...that mostly depends on whether you want
an AI composing chess problems (especially fairy). :-)
Disclaimer: Such a temporal logic formula would merely check the
correctness of a problem (not the esthetics, which would
open another can of worms of the size of Shai-Hulud).
|(24) Posted by Vlaicu Crisan [Friday, Mar 10, 2023 12:37]|
I would suggest just a slight alteration in the definition of the goal elements:
- SIDE TO MOVE: White, Black
- DIRECTION: Forward, Backward
- PLAY TYPE: Antagonistic, Collaborative
- TARGET: Opponent, Own
- AIM: Mate, Stalemate, Capture etc.
- LENGTH (number of moves)
The combination of side to move, direction, play type, target, aim and length defines the goal.
The stipulation is the combination of one or several nested goals.
For a proper classification, the side to move, aim and length can be removed.
We have the following primary categories (orthodox and heterodox):
- Forward + Antagonistic + Opponent = DIRECT
- Forward + Collaborative + Opponent = HELP
- Forward + Antagonistic + Own = SELF
Based on the primary categories, we can define the derived categories (e.g. HELPSELF or SEMI-REFLEX).
Both series and parry-series are actually fairy conditions:
- Series: Black / White doesn't move
- Parry-Series: Black / White moves only when in check
Did I miss anything essential here?
|(25) Posted by Kevin Begley [Friday, Mar 10, 2023 14:21]|
I would add that the TARGET COLOR (what you call the TARGET) can be greatly expanded:
1) OWN color (e.g., s#n reduces to antagonistic play to checkmate the King of OWN color).
2) Opponent's color (e.g., directmate is antagonistic play to checkmate the King of the Opponent's color).
3) Both colors (e.g., h==n is reduced to collaborative play to stalemate the King of both colors simultaneously).
4) Either color (e.g., if you stipulate a direct-#n where you checkmate of your own King is not a defense, you can stipulate this merely using TARGET COLOR = EITHER).
5) Neither color (e.g., you could stipulate a kind of NOT CHECK problem, where the goal is to get both Kings out of check simulatenously).
6) Either color but NOT Both colors (e.g., stalemate white or stalemate black but do not stalemate both).
7) Neither color unless it applies to Both colors (you get the idea... note that this is different from Both colors, in that it limits the path to get to both).
8) other possibilities?
I think TARGET is a better word for AIM, and clearly Target COLOR is what you mean by TARGET.
I suspect explicitly stipulating the TARGET COLOR, rather than doing this with independent aims (==, ##, etc) or independent stipulations (self-) is a revolutionary idea.
It collapses the elements necessary to stipulate a problem.
If a super amazing AI composer is created, I will invent a secret fairy condition, and tell nobody who might tell the AI.
After a few years, I will reveal my secret to the AI, and see how I've done. Then I will invent a new secret.
This will not be an issue. It will make us better composers (ironically, it will force us to revert to composing without assistance, the way the first composers did it).
Only we will know how good we are, of course (and how much the AI helped us). I am well prepared to live with that. We must be the ultimate judge of our own work.
I suspect we would learn a great deal from studying the choices a super-AI-Composer makes (the same way we learn from super talented human composers).
When a super AI fairy element inventor comes along, and teams up with the super-AI-composer, I'll just ignore what they produce, and compare later (ask it where I could have improved).
I don't foresee any need for a Butlerian Jihad.
...can of worms of the size of Shai-Hulud
The Cans! The Worms! Is there a relationship?
|(26) Posted by shankar ram [Wednesday, Mar 22, 2023 08:35]|
In the latest issue of Problem Paradise (http://flowpaper.com/flipbook/PP1019L54/?wp-hosted=1#page=1), Torsten's original 1389 (page 39) has been published:
Problem Paradise 101, 2023
(= 3+3 )
7.5h[ws] -> [s#1 & hs#1.5] 2 Solutions
(Cooperative sequence in 7.5 moves, W starting, to reach a position with Black to move, with s#1 by Black and hs#1.5, 2 Solutions)
The term "Reci-HS#" has been used for this. However, that was earlier used for a different type with the notation being m.nh[w|s w] -> #1 & s#1 (Cooperative sequence in m.n moves, to reach a position with White to move, with #1 and s#1).
Apparently, the term "Reci-HS#" is more appropriate in Torsten's new problem, as both W and B get a chance to s#1!
Vlaicu has kindly written: "Please note this sequence doesn't match Shankar Ram's definition from his article on Reciprocal Stipulations, hence the proper explanations are provided here."
Actually, there is a type with an illegal "out-of-turn" move by White: m.n[w|s s] -> [s#1 & s#1 by B], meaning a cooperative sequence leading to a position with Black to move, and with s#1 (by W) and s#1 by Black. Here, obviously, the s#1 by White is an illegal "out-of-turn" move. This can be converted to a legal type by adding a move by Black to the 1st goal and converting it from s#1 to hs#1.5: m.n[w|s s] -> [hs#1.5 and s#1 by B], which is what has happened in Torsten's problem!
More grist to my upcoming article!
|(27) Posted by shankar ram [Sunday, Mar 26, 2023 15:05]|
But, I know many of you are adamant about never reading more than 50 words (maybe we're not all so smart).
There's hope for ordinary mortals yet! I fed the opening post of this thread to chatGPT and asked it to summarize it. This is what it produced:
The post is a forum discussion about chess problems, specifically on the topic of "Reciprocal Stipulations." The author expresses their opinion that there is a lack of understanding of the difference between a goal and a stipulation among problemists.
They also argue that there should be more formalized standards for AIMs, goals, stipulations, and fairy conditions. The author critiques Shankar Ram's article on reciprocal stipulations and argues that he lacks an appreciation of formalized terms/nomenclature. The author also argues that the "FAIRIES" category in the FIDE Album is a categorical mistake and proposes calling it "MISCELLANEOUS" or "POTPOURRI."
Finally, the author provides a detailed explanation of what a formal chess problem stipulation is and what its elements are, including the concept of a GOAL.
No more posts
|Page: [Previous] [Next] 1 2 |
MatPlus.Net Forum General Reciprocal Stipulations