Hi Marco, I knew I would have failed, I always fail, unless SF is my direct opponent.mcostalba wrote:Thanks Lyudmil,Lyudmil Tsvetkov wrote:Hi Marco.
Very nice test!
If I were you, I would collect positions top engines can solve, and not vice-versa, cause I think former ones are considerably less.
For fun sake, try this one, from a simulated game with Louis' SF we played back in November 2014 here on the forum.
[d]1r3r2/4bpkp/1qb1p1p1/3pP1P1/p1pP1Q2/PpP2N1R/1Pn1B2P/3RB2K w - - 0 6
best move Qf6, with mate in 25.
here is the result of the DBT test (120+120sec 1 thread 512MB hash, 1.2Mnps):
Code: Select all
+---+---+---+---+---+---+---+---+ | | r | | | | r | | | +---+---+---+---+---+---+---+---+ | | | | | b | p | k | p | +---+---+---+---+---+---+---+---+ | | q | b | | p | | p | | +---+---+---+---+---+---+---+---+ | | | | p | P | | P | | +---+---+---+---+---+---+---+---+ | p | | p | P | | Q | | | +---+---+---+---+---+---+---+---+ | P | p | P | | | N | | R | +---+---+---+---+---+---+---+---+ | | P | n | | B | | | P | +---+---+---+---+---+---+---+---+ | | | | R | B | | | K | +---+---+---+---+---+---+---+---+ Fen: 1r3r2/4bpkp/1qb1p1p1/3pP1P1/p1pP1Q2/PpP2N1R/1Pn1B2P/3RB2K w - - 0 6 Key: 56062F5662BB9942 Checkers: Warm-up best move: Rh6, score: cp -184 Forcing best move: Qf6+ After forcing best move, score: cp -89
So I am sorry but test failed, your position if strong (engine failed to find bm in normal search), but not so strong.
Still, I think this is by far the hardest position for engines, I bet top engines will solve it satisfactorily much later after they have solved other test positions, for example the Qe5 sacrifice from Gusev-Averbakh, as it involves deeper search and more intricate eval.
Also, while SF expects black win, the position is a white win, while in most of the remaining test positions above the transition is just from draw to win, loss to draw, etc.
We will check this position and your test suite in 10 years' time, and see who was rigth.
Keep up the good work.