True. For some people, alpha-beta is a four letter word.
Vas
Vas
analysing complex rookendgames with a computer can be a bit annoying :-).
rookendgames are definitely in my opinion one of the weakest parts of engines.
this is of course quite understandable. often enough a relative long series of precise moves leads to R vs. connected pawns or so not covered by TBs.
how should the engine ever solve this by search?
on the other hand: analysing completely without an engine one often misses some tactics, but rook endings are great :-)
in my opinion it's impossible to rely on general principles only in these endings. rybka must be tought more specific chess knowledge in these cases.
I guess this is planned for a rook-endgame modul in the future?
rookendgames are definitely in my opinion one of the weakest parts of engines.
this is of course quite understandable. often enough a relative long series of precise moves leads to R vs. connected pawns or so not covered by TBs.
how should the engine ever solve this by search?
on the other hand: analysing completely without an engine one often misses some tactics, but rook endings are great :-)
in my opinion it's impossible to rely on general principles only in these endings. rybka must be tought more specific chess knowledge in these cases.
I guess this is planned for a rook-endgame modul in the future?
R vs pawns can probably be handled well with some special heuristics. This shouldn't be too far away.
A special rook endgame module also makes a lot of sense and it will help and we'll probably do it fairly soon, but you should not expect miracles from this. A lot of rook endgames are conceptual and would require really massive searches to understand in an alpha-beta way.
Vas
A special rook endgame module also makes a lot of sense and it will help and we'll probably do it fairly soon, but you should not expect miracles from this. A lot of rook endgames are conceptual and would require really massive searches to understand in an alpha-beta way.
Vas
The following position is still winning.
Are you sure? I examined those lines:
1. Ke1 Kf7 2. Kd1 Kg7 3. Kc1 (3. Ra7+? Kf8 4. Kc1=) 3... Kh7!! (3... Kf7 4. Kb1 Ra5 5. Kb2 Rb5+ (5... Rxg5 6. a7 Ra5 7. Rh8+-) 6. Kc3
Ra5 7. Kb4 Ra1 8. Kb5 Rb1+ 9. Kc5 Rc1+ 10. Kb6 Rb1+ 11. Ka7+-) 4. Kb1 Ra5 5. Kc2
Rxg5!= *
R7/6k1/P5p1/6P1/8/8/r7/6K1 w - - 0 1
Are you sure? I examined those lines:
1. Ke1 Kf7 2. Kd1 Kg7 3. Kc1 (3. Ra7+? Kf8 4. Kc1=) 3... Kh7!! (3... Kf7 4. Kb1 Ra5 5. Kb2 Rb5+ (5... Rxg5 6. a7 Ra5 7. Rh8+-) 6. Kc3
Ra5 7. Kb4 Ra1 8. Kb5 Rb1+ 9. Kc5 Rc1+ 10. Kb6 Rb1+ 11. Ka7+-) 4. Kb1 Ra5 5. Kc2
Rxg5!= *
After 4 Kb1 Ra5 White has 5 Ra7+
Ouch i missed that! Too bad Kh7 was beautyfull :-)
Ra7+ is the point of putting the pawn on g5. You have to give this check while the black king can't go to the sixth rank. It doesn't matter if the black king is on f7, g7 or h7.
My analysis is BTW from our dinner, and not computer checked, but I'm pretty sure it's sound.
Vas
My analysis is BTW from our dinner, and not computer checked, but I'm pretty sure it's sound.
Vas
aha, the Vancura position. if the white king is on f4 the winning idea is Rc1 Rb8 Ra1 Rb6 Ra5 Ke4 Kf7 Kd4 Ke7 and so on. 1...Ra5 also leads to the same theme, other moves lose afaik. In other cases putting the rook on c6 or f6 is enough, e.g. if the white king is on f3 then simply Rf1+ and Rf6.
Btw, I like the following position:
It's good for youth training, first analyse White to move and then ask how black can defend. Such miniatures where you have to defend against a thread are used rarely, only "black/white to move and win" is trained, which is not good I think...
Btw, I like the following position:
R7/P4k2/8/8/8/1K6/8/r7 b - -
It's good for youth training, first analyse White to move and then ask how black can defend. Such miniatures where you have to defend against a thread are used rarely, only "black/white to move and win" is trained, which is not good I think...
fen
Here Rybka has lost very important tempo Re3?! In rook endings it often costs half a point: 36 Kf2 (or Rc4) c2 (36 - g6 37 Rc4 Nxd5 38 Nxd5 Rxd5 39 Rxc3 Kg7 40 a5 §R->a4 c4 41 Rxc4 Rxa5 42 a4 Kf6 43 Re4 §g4,K->b4) 37 Rc4 Nxd5 38 Nxd5 Rxd5 39 Rxc2 Rd3 40 Rxc5 h6 (40 - g6 41 g4! §g5 Rxa2 42 a5 h6 43 h4! Ra4 44 g5!) 41 h4! Rxa2 42 a5 Kh7 43 h5! §K->b5 g6 44 g4 Ra4 45 Kf3 Ra3+ 46 Kf4 Ra4+ 47 Ke5
Mário
3r3k/2n1N1pp/8/2pP4/P3R3/P1p5/6PP/6K1 w - -
Here Rybka has lost very important tempo Re3?! In rook endings it often costs half a point: 36 Kf2 (or Rc4) c2 (36 - g6 37 Rc4 Nxd5 38 Nxd5 Rxd5 39 Rxc3 Kg7 40 a5 §R->a4 c4 41 Rxc4 Rxa5 42 a4 Kf6 43 Re4 §g4,K->b4) 37 Rc4 Nxd5 38 Nxd5 Rxd5 39 Rxc2 Rd3 40 Rxc5 h6 (40 - g6 41 g4! §g5 Rxa2 42 a5 h6 43 h4! Ra4 44 g5!) 41 h4! Rxa2 42 a5 Kh7 43 h5! §K->b5 g6 44 g4 Ra4 45 Kf3 Ra3+ 46 Kf4 Ra4+ 47 Ke5
Mário
Can anybody check this analysis with Rybka? Is there better defence for Black? Character § marks the plan or threat.
i can't wait to see Rybka play the Sicilian.... :-D
Why is that ?
that's my favourite opening.. that and the english.
If Kasparov had have Rybka in time, (I'm sure) he would demolish Kramnik's Berlin Wall!
(he would have demolish Kramnik's Berlin Wall)
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill
