Author: Dieter Buerssner
Date: 11:57:04 03/24/04
Go up one level in this thread
On March 23, 2004 at 19:27:13, Jaime Benito de Valle Ruiz wrote:
>On March 23, 2004 at 18:25:54, Anson T J wrote:
>
>>A game begins with 1.e4 and ends in the fifth move with knight takes rook mate.
>>Reconstruct the game.
>>
>>[d]rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq e3 0 1
>>
>>Note: This puzzle has apparently stumped many smart people, including Garry
>>Kasparov, Anatoly Karpov, Mikhail Botvinnik, and Ken Thompson (a la Unix and C).
>>Major kudos if you solve it yourself!
>
>Although this is a deep perft 11, the code of any engine can be modified to just
>search for this condition:
I did exactly this. It needs quite some time. BTW, it is more of a depth 9 perft
(from the question it could be depth 8, even). And at the last depth, only at
few moves must be looked (knight moves that capture rook). In those few cases
another depth can be considered added, to check for the checkmate condition. In
the question, the first ply (e4) was already given.
>...g8f6 2. f2f3 f6e4 3. d1e2 e4g3 4. e2e7 d8e7 5. e1f2 g3h1
1 2 3 4 5 6 7 8 9
the "perft depth".
I first tried a perft without hash, and found out, that there is no solution at
depth 8 (so black must mate). This already took almost an hour (P4, 2.53 GHz).
So, I was too lazy, to do depth 9 this way. I used a hash for perft ( I already
have perft functions with and without hash, so the modifications to tailor for
the question took only little time), and after 20 minutes your solution was
found (whole run about 1 hour). I stored the whole positions in the hash, so
there is not the theoretical chance, that hash collisions can change the result.
Result: there is no other mate position, than in the line you gave. Because of
the hash, the program can not decide, whether there are any transpositions. I
tried a perft 8 after e4 Nf6 without hash, and only one transposition is
possible:
(e2e4 g8f6) d1e2 f6e4 f2f3 e4g3 e2e7 d8e7 e1f2 g3h1
Regards,
Dieter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.