Endgame: Is this a draw?

[Trahald]

Can Shredder access 6 piece TB in it's search? Or can any other engine use the 6 piece Nalimov TB in search?

I ask because it would be very interesting to see if any engines could hold this position with Black against perfect play. Rybka 3 Human for one does not come close to lasting 50 moves against the 6 piece TB after giving up the pawn.

[Event "Blitz 4m+2s"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "Murray, Russell"]
[Black "Rybka 3 Human"]
[Result "1-0"]
[SetUp "1"]
[FEN "5n2/1R5p/8/5K2/3k4/2b5/6B1/8 b - - 0 1"]
[PlyCount "80"]
[TimeControl "240+2"]

{512MB, Fritz12.ctg, Cahill} 1... Ng6 {7} 2. Rxh7 {34} Ne5 {6} 3. Bf1 {16} Kd5
{5} 4. Rc7 {29} Kd4 {5} 5. Rc8 {0} Ba1 {0} 6. Ra8 {0} Bc3 {5} 7. Rd8+ {26} Ke3
{0} 8. Ba6 {19} Bb2 {8} 9. Rd1 {20} Bc3 {5} 10. Bb5 {19} Bd4 {3} 11. Re1+ {28}
Kd2 {0} 12. Rf1 {22} Nf7 {8} 13. Ba6 {13} Be3 {0} 14. Rb1 {16} Bd4 {0} 15. Ke4
{17} Bg7 {4} 16. Rf1 {22} Nd6+ {0} 17. Kd5 {23} Ne8 {0} 18. Rf7 {18} Bh8 {22}
19. Rf2+ {17} Kc3 {0} 20. Rf3+ {20} Kb4 {0} 21. Kc6 {17} Bd4 {0} 22. Bc8 {18}
Ng7 {0} 23. Rf4 {14} Kc3 {6} 24. Kd5 {21} Be3 {0} 25. Rf3 {24} Kd2 {0} 26. Bd7
{24} Bb6 {36} 27. Kc4 {22} Bc7 {5} 28. Rh3 {15} Ke1 {8} 29. Kd4 {21} Kd2 {0}
30. Rh7 {21} Bd8 {2} 31. Rh2+ {29} Kc1 {0} 32. Bg4 {21} Bc7 {0} 33. Rg2 {30}
Bb6+ {10} 34. Kc4 {19} Bd8 {1} 35. Kb3 {19} Be7 {0} 36. Rc2+ {22} Kb1 {0} 37.
Rd2 {20} Bg5 {0} 38. Rd1+ {17} Bc1 {0} 39. Re1 {20} Ne8 {0} 40. Bf5+ {19} Ka1 {
0} 41. Rxc1# {8} 1-0

[rlsuth]

Can Shredder access 6 piece TB in it's search? Or can any other engine use the 6 piece Nalimov TB in search?

I ask because it would be very interesting to see if any engines could hold this position with Black against perfect play. Rybka 3 Human for one does not come close to lasting 50 moves against the 6 piece TB after giving up the pawn.

[Event "Blitz 4m+2s"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "Murray, Russell"]
[Black "Rybka 3 Human"]
[Result "1-0"]
[SetUp "1"]
[FEN "5n2/1R5p/8/5K2/3k4/2b5/6B1/8 b - - 0 1"]
[PlyCount "80"]
[TimeControl "240+2"]

{512MB, Fritz12.ctg, Cahill} 1... Ng6 {7} 2. Rxh7

Any program that uses tablebases can win instantly after 1... Ng6 {7} 2. Rxh7 as can be seen if you enter the position here: http://www.shredderchess.com/online-che ... abase.html

In other words, Ng6 is the worst move that black can play.

.

[govert]

Can Shredder access 6 piece TB in it's search? Or can any other engine use the 6 piece Nalimov TB in search?

I ask because it would be very interesting to see if any engines could hold this position with Black against perfect play. Rybka 3 Human for one does not come close to lasting 50 moves against the 6 piece TB after giving up the pawn.

[Event "Blitz 4m+2s"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "Murray, Russell"]
[Black "Rybka 3 Human"]
[Result "1-0"]
[SetUp "1"]
[FEN "5n2/1R5p/8/5K2/3k4/2b5/6B1/8 b - - 0 1"]
[PlyCount "80"]
[TimeControl "240+2"]

{512MB, Fritz12.ctg, Cahill} 1... Ng6 {7} 2. Rxh7

Any program that uses tablebases can win instantly after 1... Ng6 {7} 2. Rxh7 as can be seen if you enter the position here: http://www.shredderchess.com/online-che ... abase.html

In other words, Ng6 is the worst move that black can play.

.

Actually, it's a mate in 72 after Rxh7, so the question is if there is a capture within the first 50 of these 72 moves.

Following "best play" according tho the DTM tablebase we get a capture after 52 moves so that is not good enough.

[Trahald]

Any program that uses tablebases can win instantly after 1... Ng6

AFAIK most engines which access the Nalimov TB are limited to 5 pieces max. On the Naum website it says that using 6 piece TB's in search would actually hurt performance because the size is just too great.

Hence the reason why Rybka 3 Human on a fast quad core with full 5 piece TB plays the move 1... Ng6, the fact that it leads to mate in 72 is way beyond it's capabilities.

My point is that since Rybka with 5 piece TBs can play directly into a losing position with Black, then it is almost certain that other engines would do the same, therefore the position should not be adjudicated as a draw.

So again I wonder if there is an engine that can use 6 piece tablebases in search, or if there is an interface that can take over when 6 pieces on the board are left.

Regarding the initial position before Ng6 it is my guess that White could force a win if there were no such thing as the 50 move draw, but with perfect defense Black can hold for this reason. That is only my guess however.

[Eelco de Groot]

Can Shredder access 6 piece TB in it's search? Or can any other engine use the 6 piece Nalimov TB in search?

I'm not sure Russell. Paulo Soares would know more about Shredder. I don't think the online 6 men tablebases could be used by Shredder in search, but the Shredderbases can if they are on your harddisk, Nalimov tablebases too, I don't know if there are 6 men Shredderbases yet, only 3,4,5 pieces! 6 pieces would be too big to hold in memory though even the 6men bitbases are way too big.

Also not possible: I thought I remembered Vytron on Rybkaforum once using the online tablebases to automatically adjudicate 6 men positions from tournaments, I thought that must be possible in the Shredder GUI, it seemed feasible to program this at least. But now I can't find the option in my GUI or Shredder manual so I think I was mistaken about this being possible.

There are many big hardware guys who have downloaded the full 6 piece tablebases and are using it with Rybka so that is possible but as the case with Naum, it probably does not give a measurable boost, compared to for instance only using 5 men tablebases. And Naum is reportedly better at making use of tablebases in its search than Rybka 3.

Regards, Eelco

[Jim Walker]

I'm not sure but I would bet that Crafty can use 6 piece TBs in search. I haven't tried Crafty in several years though.
Jim