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

        Arimaa >> General Discussion >> Endgame Puzzle 11
        
(Message started by: jdb on Apr 18th, 2011, 8:43am)
Title: Endgame Puzzle 11
Post by jdb on Apr 18th, 2011, 8:43am
Silver to play and win

Note: The gold move is part of the puzzle.

1w Df2 Ra4 Ca5
1b rb4 cb5
2w Df2w De2w Dd2w Dc2w


http://109.74.204.218/arimaa/plan_js/plan_js.html
Title: Re: Endgame Puzzle 11
Post by Hippo on Apr 18th, 2011, 9:52am
I understand what you are trying to write here, but I am afraid it does not work.

Gold made a missclick (see 2w) now it's silver who is winning.
1w De2 Ra4 Ca5 Rc7
1b rb4 cb5  
2w Rc7s Rc6x De2w Dd2w Dc2w
would worked well.

Or the puzzle could be

silver made a missclick in this strange ending (see 2s) but it's silver who is still winning.
1w Dh1 Ra4 Ca6
1b rc5 cb6 rf7
2w Dh1n Dh2w Dg2w Ca6s
2b rf7s rf6x cb6s rc5s rc4w
3w Df2w De2w Dd2w Dc2w

That could be harder to solve.
Title: Re: Endgame Puzzle 11
Post by Fritzlein on Apr 18th, 2011, 10:49am

on 04/18/11 at 09:52:58, Hippo wrote:
I understand what you are trying to write here, but I am afraid it does not work.

Clearly this is meant as a stepping stone to solve Puzzle 10, but I agree with Hippo that there is some subtle difficulty in presenting the puzzle.

Since the solution requires Silver to win by repetition, how do we know that Silver's first move 2s rb4eeee isn't itself a repetition?  I assume that is what Hippo is trying to fix by including a capture in the move list, so that the position can't be a repetition.

Puzzles that can't be solved without knowing who wins a repetition fight are always going to be problematic.  If you include a move list of any sort along with the position, it will be a dead giveaway as to how it should be solved.  Also if you say anything along the lines of, "assume every position in the solution is occurring for the first time," it gives away the theme just as surely.  But if you don't give a longer move list or an explicit disclaimer, then you don't have enough information from the position alone to solve the puzzle.

I don't see any way around this conundrum.  
Title: Re: Endgame Puzzle 11
Post by robinz on Apr 18th, 2011, 10:58am
I'm once again intrigued as to what these posts say, but am trying to resist peeking  :)

I must say I never came across this position in trying to solve puzzle 10. But then, I can't solve this one either - the only moves I can find that cause Gold any problems lead to losing a repetition fight (which seems to be the purpose behind jdb including the gold move just before the puzzle). I have looked for other ways for silver to win, but haven't found one that works - what am I missing?  ???
Title: Re: Endgame Puzzle 11
Post by jdb on Apr 18th, 2011, 12:26pm

Quote:
If you include a move list of any sort along with the position, it will be a dead giveaway as to how it should be solved.  Also if you say anything along the lines of, "assume every position in the solution is occurring for the first time," it gives away the theme just as surely.


To clarify:

In all of these puzzles, there is no move history before the setup moves. Assume the puzzle starts with an empty board. On 1g gold places his pieces, and on 1s silver places her pieces. After that the moves are played as in a normal game.
Title: Re: Endgame Puzzle 11
Post by Hippo on Apr 18th, 2011, 2:15pm

on 04/18/11 at 12:26:53, jdb wrote:
To clarify:

In all of these puzzles, there is no move history before the setup moves. Assume the puzzle starts with an empty board. On 1g gold places his pieces, and on 1s silver places her pieces. After that the moves are played as in a normal game.


With that convention I would prefere formulation without 2w ... gold to move but siver to win.
Title: Re: Endgame Puzzle 11
Post by Eltripas on Apr 18th, 2011, 11:37pm

on 04/18/11 at 14:15:22, Hippo wrote:
2w ... gold to move but siver to win.


Then the puzzle would have a lot of answers  :P
Title: Re: Endgame Puzzle 11
Post by Hippo on Apr 19th, 2011, 1:20am

on 04/18/11 at 23:37:38, Eltripas wrote:
Then the puzzle would have a lot of answers  :P


OK:
So what about ...
1w Ef3 Df2 Ra4 Ca5  
1b rb4 cb5

Expect the perfect play of both players where loser gains points for prolongation of the game, who is winning? In which turn the game would be over?
Title: Re: Endgame Puzzle 11
Post by jdb on Apr 19th, 2011, 8:12am

on 04/19/11 at 01:20:00, Hippo wrote:
OK:
So what about ...
1w Ef3 Df2 Ra4 Ca5  
1b rb4 cb5

Expect the perfect play of both players where loser gains points for prolongation of the game, who is winning? In which turn the game would be over?


I think there is a misunderstanding. The reason for adding the 2w move to the puzzle, is to force silver to win without using repetition.
Title: Re: Endgame Puzzle 11
Post by Fritzlein on Apr 19th, 2011, 9:02am

on 04/19/11 at 08:12:06, jdb wrote:
I think there is a misunderstanding. The reason for adding the 2w move to the puzzle, is to force silver to win without using repetition.

Ooh,  I totally misunderstood that.  Thanks for clarifying!  How about
2s rb4e rc4e rd4e re4e
3g Db2e Dc2e Dd2e De2e
3s cb5s cb4n Ra4e rf4e
4g Df2e Ca5s Rb4e Rc4e
4s cb5e cc5e rg4w rf4w
5g Dg2w Df2w De2w Dd2n
5s cd5n Rd4n cd6n Rd5n
or
5g Dg2w Df2w Ca4e Cb4e
5s cd5n Rd4n cd6n Rd5n
This incidentally answers Hippo's question about how you resolved repetition fights in your endgame tablebase: you didn't!  :)
Title: Re: Endgame Puzzle 11
Post by Hippo on Apr 19th, 2011, 11:07am

on 04/19/11 at 08:12:06, jdb wrote:
I think there is a misunderstanding. The reason for adding the 2w move to the puzzle, is to force silver to win without using repetition.


In that case the second ... harder formulation takes place :). OK I have to think about it :).

Oh why I didn't noticed ... the original Fritzlein solution to puzzle 10 actualy loses on repetition (7s is illegal).

And now I have solved the puzzle myself ... leading to the same answer as Fritzlein, except there is more options at main variant 5s.

Adding variant 4g Ca5s Ca4s Ca3e Df2e
4s Rb4e cb5s rg4w rf4w
Title: Re: Endgame Puzzle 11
Post by Fritzlein on Apr 19th, 2011, 5:17pm

on 04/19/11 at 11:07:48, Hippo wrote:
Oh why I didn't noticed ... the original Fritzlein solution to puzzle 10 actualy loses on repetition (7s is illegal).

Really?  I looked at my solution to Puzzle 10 again and I still think it wins, at least according to jdb's convention that no position in the solution has occurred earlier in the game.  Of course, the solution to Puzzle 11 gives us an unconditional way to solve Puzzle 10, one that doesn't rely on assumptions about how the Puzzle 10 position was reached, but at worst my original solution to Puzzle 10 seems to be "unknown outcome", not a repetition loss.
Title: Re: Endgame Puzzle 11
Post by Hippo on Apr 19th, 2011, 5:58pm

on 04/19/11 at 17:17:51, Fritzlein wrote:
Really?  I looked at my solution to Puzzle 10 again and I still think it wins, at least according to jdb's convention that no position in the solution has occurred earlier in the game.  Of course, the solution to Puzzle 11 gives us an unconditional way to solve Puzzle 10, one that doesn't rely on assumptions about how the Puzzle 10 position was reached, but at worst my original solution to Puzzle 10 seems to be "unknown outcome", not a repetition loss.

Oh sorry, I didn't remember it well :( gold dog went from e3 not f2 so gold generated first position in the cycle. So your solution was correct, just not the one the endgame tables provided.
Title: Re: Endgame Puzzle 11
Post by Tuks on Jun 6th, 2011, 7:09am
ok, thanks hippo, your right, this is one solution, it was easy to find the other once this one was found


Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.