Arimaa Forum - Print Page


    
      
        Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> General Discussion >> Puzzles
        
(Message started by: Bildstein on Nov 23^rd, 2008, 7:21pm)

Title: Puzzles
Post by Bildstein on Nov 23^rd, 2008, 7:21pm

In puzzle 16, I think there is another solution, and to my mind it is more straightforward: move the silver elephant behind the silver rabbit. In fact if you do this, it doesn't matter which way your camel pushes the gold cat, because the silver rabbit can't be frozen.

I think the idea in the suggested solution is quite neat, using the elephant to stop the gold rabbit being able to make way for the gold camel to freeze the silver rabbit, but I think my solution is much more intuitive in this case.

Actually now I think about it, could the problem be altered to put another gold piece in the way, so the silver elephant can't come to the rescue of the silver rabbit? Then the suggested solution might be the only correct one.

Title: Re: Puzzles
Post by tize on Dec 18^th, 2008, 10:49am

I was testing Marwin on some of the puzzles, and he confirms that puzzle 16 (http://arimaa.com/arimaa/puzzles/show.cgi?p=p16) has 2 solutions.

He also found three solutions to puzzle 41 (http://arimaa.com/arimaa/puzzles/show.cgi?p=p41)

ed5w ec5w rf7w da5n
rb8s ed5w ec5w rf7w
Hd6e ed5n da5n ed6n

I think that these puzzles need to be changed.
To show all solutions in the answer, alted to only have one solution, or removed.

Title: Re: Puzzles
Post by Fritzlein on Dec 18^th, 2008, 11:46am

Well-spotted, Bildstein.

Tize, I agree that any alternate solutions should be listed. It is deflating to think you have solved a puzzle and then be told some other answer is correct, which is fine if you are wrong but unfair if you are right. Unfortunately, I believe the puzzle mechanism assumes that there will only be one correct answer. Also most puzzlists assume that there is only one solution. Therefore, as much as I dislike taking anything out of our tiny collection, I guess puzzles 16 and 41 should come down. :(

Puzzle 16 is probably not too hard to replace, because goal in two is a fairly common theme, but I quite like the unusual theme of finding the only move that loses in two instead of one, so puzzle 41 is a greater loss.

By the way, tize, did marwin verify that all the other puzzles have a unique answer? I'm rather impressed if so. It's hard for a human puzzle-composer to verify the uniqueness of their solution in a game with such a huge branching factor. Also I'm impressed with marwin for being able to return information about second-best and third-best moves. It seems a lot of programs just verify that their top move is best, and don't guarantee an accurate evaluation of any other move.

Title: Re: Puzzles
Post by tize on Dec 18^th, 2008, 4:22pm

I have only run marwin on a few of the puzzles, most of the puzzles are to deep for marwin to prove anything.

As for marwin giving info about the non-best move, well he don't. But when he has proven that a move is forced loss that move is removed, and when the list of available moves gets less then 10 he shows all the list.

I then checked these three positions for p1 victories. So it's not that fantastic, and not an automatic process.

Title: Re: Puzzles
Post by omar on Dec 19^th, 2008, 4:39pm

Thanks guys for noticing that these two puzzles have multiple solutions. I actually don't mind puzzles that have multiple solutions, but right now the puzzle generator doesn't support that. I've hidden these puzzles for now; maybe we can bring them back after we have a puzzle system that handles multiple solutions.