Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Another insane Deep fritz 10.1 piece sacrifice!
- - By rocket (***) [se] Date 2012-04-23 10:00 Edited 2012-04-24 13:48
I bet you will not find any other strong engine play this piece sacrifice because it's losing:smile:

Yet Deep fritz  10 likes it even at slower time controlls. I know the previous game I posted many engines sacrificed a piece as well but not this game...

White: Houdini 2.0

Black Deep fritz 10.1

1. D4 D5 2. Nf3 Nf6 3. c4 dxc4 4 Nc3 Nc6 5 Qa4 e6 6. Qxc4 Nb4 7 Qb3 c5 8 dxc5 Bxc5 9 a3 Qb6??(sacrifices a piece for a pawn and a check on the king). 10 axb4 Bxf2 11 Kd1

White simply developed the pieces and won easily. Once again I don't see how this qb6 line is concidered optimal for so long since an engine evaluates a position after a number of moves(ahead of the current position we are looking at) and not simply after the check on the king... It must calculate some seriously flawed lines:grin:
Parent - By Peter Grayson (****) [gb] Date 2012-04-23 16:44
... and you would not find DF10.1 playing this line if you used the supplied DFritz10.ctg opening book. 5..e6? would not have been played. No point complaining about the moves an engine makes when you direct it into a line that is weak and the engine author(s) sought to avoid. A good example of why engine specific opening books should be considered the same as tablebases. If the engine author supplies a book then it may make sense to use it.

Give the engine long enough and it will play 5..Nd5.

Regards,
PeterG
Parent - By Permanent Brain (*****) Date 2012-04-23 18:18
Please simply press Ctrl-C in your Fritz and Ctrl-V here when posting, to paste game notations (preferably, you set game: PGN in your clipboard options). Thanks! :lol:
Parent - By dragon49 (****) [us] Date 2012-04-23 19:13
What type of hardware are you using?
- - By rocket (***) [se] Date 2012-04-23 18:40 Edited 2012-04-23 18:45
the position were qb6 is played was approximately equal- it was not a weak line for black.

In an equal middlegame the engine prefers to sacrifice a full piece for a single check on the king which is pretty amazing.

The fritz 11 opening book was used in the game.
Parent - - By Peter Grayson (****) [gb] Date 2012-04-23 22:28
OK, but the book could not have been optimised because 5..Nd5 would have been played from the DF11 book.

PeterG
Parent - - By Pia (****) [ru] Date 2012-04-24 02:58 Edited 2012-04-24 03:00
If it's "normal" book setting, always played 5...e6.

Parent - By Peter Grayson (****) [gb] Date 2012-04-24 08:33 Edited 2012-04-24 08:35

> If it's "normal" book setting, always played 5...e6.


I agree but sub-optimal lines are to extend the book for the user's enjoyment and consequently may not always give best engine performance. Given Deep Fritz 10's SMP wild variability, it is not surprising to find some unusual moves being played, particularly early on. In that respect DF10.1 was no improvement over DF10. So fun yes but doubt if it is substantial enough to be worth posting. Guess the number of replies here confirm the level of interest!?

On reflection though, considering the lack of chess posts as opposed to political posts here now, probably was worth posting if only to try and reactivate chess discussion.

Regards,
PeterG
- - By rocket (***) [se] Date 2012-04-24 14:06
On my intel core duo laptop deep fritz 10 and fritz 10 want to play qb6 up until  about 2mins of ponder when it changes to nd5... so in most time controlls in a middlegame it would have played the move on single dual core

I would be interested to see how fast Deep fritz 10 rejects Qb6 on several cores? please share the result if you have multiprocessor:)
Parent - - By Pia (****) [ru] Date 2012-04-24 14:41 Edited 2012-04-24 14:45
Intel Core2 Q6600 @ ~6600 knps

Analysis by Deep Fritz 10: (10.1)

9...Qd8-b6 10.a3xb4 Bc5xf2+ 11.Ke1-d1 Bc8-d7 12.Kd1-c2 Ra8-c8 13.Kc2-b1 Bd7-c6 14.Nf3-e5 0-0 15.Bc1-g5 Nf6-d7 16.Ne5-c4 Qb6-d4 17.e2-e3 Qd4-g4 18.Qb3-c2
  =/+  (-0.40)   Depth: 9/22   00:00:00  299kN
  +/=  (0.27)   Depth: 19/43   00:02:16  918mN
9...Nb4-d5
  +/=  (0.26)   Depth: 19/46   00:04:16  1719mN
  +/=  (0.26)   Depth: 19/46   00:04:16  1719mN

another try

Analysis by Deep Fritz 10:

9...Nb4-c6 10.Bc1-g5 0-0 11.Ra1-d1 Qd8-e7 12.e2-e3 Rf8-d8 13.Bf1-d3 Bc5-d6
  =  (-0.04)   Depth: 9/19   00:00:00  315kN
9...Qd8-b6 10.a3xb4 Bc5xf2+ 11.Ke1-d1 Bc8-d7 12.Kd1-c2 Ra8-c8 13.Kc2-b1 Bf2-e3 14.Bc1xe3 Qb6xe3 15.Qb3-c2 0-0 16.Qc2-d3 Qe3-b6 17.Qd3-d4 Qb6xd4 18.Nf3xd4 e6-e5 19.Nd4-c2 a7-a6 20.b4-b5 a6xb5
  =  (-0.04)   Depth: 9/19   00:00:00  340kN
  =  (0.24)   Depth: 18/40   00:01:03  414mN
9...Nb4-d5
  =  (0.23)   Depth: 18/43   00:02:32  1002mN
  =  (0.23)   Depth: 18/43   00:02:32  1002mN
Parent - By rocket (***) [se] Date 2012-04-24 14:48
Hah! so it still takes several minutes ... Multiprocessor seems useless compared to dual core with chess engines. There is no difference in this case!
- By rocket (***) [se] Date 2017-12-07 15:05 Edited 2017-12-07 15:09
<No point complaining about the moves an engine makes when you direct it into a line that is weak and the engine author(s) sought to avoid.>

Irrelevant what the book author intended. What matters is the objective nature of the position, and the position is approximately equal after e6. No problem having the engine play it out. I don't know why the author put a question mark on e6.

2017 UPPDATE: On my Asus Notebook 2CPU, Deep Fritz 10.1 still has Qb6 as the principled variation after 21 minutes =0.12:yell:
Up Topic The Rybka Lounge / Computer Chess / Another insane Deep fritz 10.1 piece sacrifice!

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill