01-29-2021, 01:14 AM
Hi,
I am very exited to discover this project!
I haven't yet read much of the documentation, so please only if possible answer my most salient questions which haven't been already documented unless the answers are obscure.
The Ludii language looks to be extremely expressive/complex! It should therefore be possible to more elegantly state properties of games, which should speed up theorem proving, which I believe can be shown to dominate mere search in some contexts. A quick search of this thread for "theorem" had no results. Are there efforts to prove positional properties / trajectory invariants? (e.g. for instance with chess - opposite colored bishops can not capture, this pawn is overprotected, etc). I am working on a project to translate textbooks on Chess and Go into Prolog predicates ( https://github.com/aindilis/chap2/blob/m...nalysis.pl ) ( https://github.com/aindilis/nlu-mf ) for use in positional theorem proving. My goal there is to 1) extract libraries of strategies, 2) ground Natural Language Understanding in a concrete domain for use in speeding up Rapid Knowledge Formation (RKF), 3) extract strategies that can be transferred for use with my Free Life Planner project ( https://github.com/aindilis/free-life-planner ).
I haven't read the grammar definition document yet, but it looks to be based on s-expressions. Is it Knowledge Interchange Format, First Order Logic, or something similar? Are the Ludii language constants defined in terms of a smaller set of primitives somewhere, preferably in a logic? Or is it defined programmatically in Java? Are there translations to or from Game Description Language? How does it compare to leading General Game Playing solvers? How does it compare as a framework to other GGP systems like Zillions of Games? Are there efforts to solve games, such as with Gamer: https://core.ac.uk/download/pdf/297278922.pdf
I have a stub project called Ender (named after the character from the novel Ender's Game who supposedly "wins all the games."). I intend to now let Ludii/Polygames do most of the heavy lifting. Would theorems (of positional properties) make useful features for Polygames?
I lost my first draft of the post, so I'm just going to post this now before I lose it again. More later.
Thank you,
Andrew
I am very exited to discover this project!
I haven't yet read much of the documentation, so please only if possible answer my most salient questions which haven't been already documented unless the answers are obscure.
The Ludii language looks to be extremely expressive/complex! It should therefore be possible to more elegantly state properties of games, which should speed up theorem proving, which I believe can be shown to dominate mere search in some contexts. A quick search of this thread for "theorem" had no results. Are there efforts to prove positional properties / trajectory invariants? (e.g. for instance with chess - opposite colored bishops can not capture, this pawn is overprotected, etc). I am working on a project to translate textbooks on Chess and Go into Prolog predicates ( https://github.com/aindilis/chap2/blob/m...nalysis.pl ) ( https://github.com/aindilis/nlu-mf ) for use in positional theorem proving. My goal there is to 1) extract libraries of strategies, 2) ground Natural Language Understanding in a concrete domain for use in speeding up Rapid Knowledge Formation (RKF), 3) extract strategies that can be transferred for use with my Free Life Planner project ( https://github.com/aindilis/free-life-planner ).
I haven't read the grammar definition document yet, but it looks to be based on s-expressions. Is it Knowledge Interchange Format, First Order Logic, or something similar? Are the Ludii language constants defined in terms of a smaller set of primitives somewhere, preferably in a logic? Or is it defined programmatically in Java? Are there translations to or from Game Description Language? How does it compare to leading General Game Playing solvers? How does it compare as a framework to other GGP systems like Zillions of Games? Are there efforts to solve games, such as with Gamer: https://core.ac.uk/download/pdf/297278922.pdf
I have a stub project called Ender (named after the character from the novel Ender's Game who supposedly "wins all the games."). I intend to now let Ludii/Polygames do most of the heavy lifting. Would theorems (of positional properties) make useful features for Polygames?
I lost my first draft of the post, so I'm just going to post this now before I lose it again. More later.
Thank you,
Andrew