Lomonosov tablebases

Technical questions regarding Openings, Middlegames, Endings etc.
User avatar
Jon Tait
Posts: 149
Joined: Sat Jul 25, 2009 8:31 am
Contact:

Post by Jon Tait » Mon Feb 27, 2017 6:32 am

Christopher Kreuzer wrote:Does anyone here actually have access to the Lomonosov tablebases?
Yes, I do. :)
Christopher Kreuzer wrote:I wasn't aware that these were available. Has anyone here made any practical use of them during analysis?
Yes, I use them regularly to check non-trivial seven piece endings. Here's one such case from an online game of mine with K+2N vs. K+3P:

http://200opengames.blogspot.co.uk/2016 ... -line.html
Francis Fields wrote:Does anyone here think that tablebases help understanding?
Yes and no. Yes, because as with all the endgame databases they often throw up surprises. No, because they don't actually explain anything; you still have to try and figure out why what they're telling you is correct – as in the above game, where it took me quite a while to understand why I was losing. And Nunn wrote a whole series of books about this, and that was just five piece endgames.
blog inspired by Bronstein's book, but using my own games: http://200opengames.blogspot.co.uk/

Roger de Coverly
Posts: 21297
Joined: Tue Apr 15, 2008 2:51 pm

Re:

Post by Roger de Coverly » Mon Feb 27, 2017 9:45 am

Jon Tait wrote: Yes, I use them regularly to check non-trivial seven piece endings. Here's one such case from an online game of mine with K+2N vs. K+3P:
In that game, although the shortest win was more than 50 moves, captures and pawn moves periodically reset the counter. Is it known whether the tablebase actually checks this before announcing win/draw?

User avatar
Jon Tait
Posts: 149
Joined: Sat Jul 25, 2009 8:31 am
Contact:

Re: Re:

Post by Jon Tait » Mon Feb 27, 2017 12:19 pm

Roger de Coverly wrote:In that game, although the shortest win was more than 50 moves, captures and pawn moves periodically reset the counter. Is it known whether the tablebase actually checks this before announcing win/draw?
dunno, sorry
blog inspired by Bronstein's book, but using my own games: http://200opengames.blogspot.co.uk/

Nick Burrows
Posts: 1705
Joined: Sat Aug 14, 2010 12:15 pm

Re: Re:

Post by Nick Burrows » Mon Feb 27, 2017 1:42 pm

Roger de Coverly wrote:
In that game, although the shortest win was more than 50 moves, captures and pawn moves periodically reset the counter. Is it known whether the tablebase actually checks this before announcing win/draw?

I think engines that use tablebases are configured to allow for the 50 move rule, but tablebases themselves are not.

Matt Fletcher
Posts: 271
Joined: Wed Mar 09, 2011 9:42 pm

Re: Lomonosov tablebases

Post by Matt Fletcher » Mon Feb 27, 2017 11:54 pm

There was an interesting finish involving tablebases (actually 5-piece not 7-piece) in the TCEC Superfinal between Stockfish 7 and Houdini 5 in November last year.

This is the final position, with Black to move:



Both computers had been evaluating the position as a draw for some moves, then as soon as it got to a tablebase position, it was adjudicated a win for White. I assume it's because it was mate in 72 (and first piece drops in 62) so their 50-move-rule sense kicked in. Who knew Queen vs 2 bishops was a win?!

The whole game is worth looking at - check out Black's 29th in particular, as well as White's King walking to a8:



From the final position, the following is apparently best play for anyone interested (I find it quite pretty!):



Nick Burrows
Posts: 1705
Joined: Sat Aug 14, 2010 12:15 pm

Re: Lomonosov tablebases

Post by Nick Burrows » Tue Feb 28, 2017 10:52 am

Why wasn't it a 50 move draw?

Kevin Thurlow
Posts: 5817
Joined: Wed Apr 30, 2008 12:28 pm

Re: Lomonosov tablebases

Post by Kevin Thurlow » Tue Feb 28, 2017 12:03 pm

"Who knew Queen vs 2 bishops was a win?!"

I did, but only because I read Lev Polugayesky's book - I think it was "Grandmaster Performance", where he analysed (at extreme length) Q vs BB. Coincidentally, I was talking to Neil McDonald last night and we discussed the position. I said I would find my copy of the book and use a tablebase to check the analysis.

Joshua Gibbs

Re: Lomonosov tablebases

Post by Joshua Gibbs » Tue Feb 28, 2017 2:31 pm

Nick Burrows wrote:Why wasn't it a 50 move draw?
QFT

Joshua Gibbs

Re: Lomonosov tablebases

Post by Joshua Gibbs » Tue Feb 28, 2017 2:32 pm

[
quote="Nick Burrows"]Why wasn't it a 50 move draw?
QFT shouldn't it be a draw move 50?

Angus French
Posts: 2151
Joined: Thu May 15, 2008 1:37 am
Contact:

Re: Lomonosov tablebases

Post by Angus French » Tue Feb 28, 2017 3:11 pm

Kevin Thurlow wrote:"Who knew Queen vs 2 bishops was a win?!"

I did, but only because I read Lev Polugayesky's book - I think it was "Grandmaster Performance", where he analysed (at extreme length) Q vs BB. Coincidentally, I was talking to Neil McDonald last night and we discussed the position. I said I would find my copy of the book and use a tablebase to check the analysis.
There's some discussion but no conclusion in the notes to Polugayevski - Geller, Skopje 1968 (Grandmaster Performance, p122, second column)... Perhaps it's elsewhere.

Matt Fletcher
Posts: 271
Joined: Wed Mar 09, 2011 9:42 pm

Re: Lomonosov tablebases

Post by Matt Fletcher » Tue Feb 28, 2017 7:19 pm

Nick Burrows wrote:Why wasn't it a 50 move draw?
I don't know for sure, but I suspect it's somewhere in the rules of the tournament - ie as soon as you get to a tablebase position, the result is whatever the outcome would be with best play, regardless of move limits.

Tim Harding
Posts: 2318
Joined: Sat Oct 23, 2010 8:46 pm
Location: Dublin, Ireland
Contact:

Re: Re:

Post by Tim Harding » Wed Mar 01, 2017 7:40 pm

Roger de Coverly wrote:In that game, although the shortest win was more than 50 moves, captures and pawn moves periodically reset the counter. Is it known whether the tablebase actually checks this before announcing win/draw?
No they do not take account of the 50 move rule. They just say how many moves to mate with optimal play (if there is a mate - otherwise they say draw).

And in the case of ICCF correspondence tournaments, if the tablebase says there is a forced mate, it is adjudicated a win even if it would take more than 50 moves without pawn move or capture.
Tim Harding
Historian and FIDE Arbiter

Author of 'Steinitz in London,' British Chess Literature to 1914', 'Joseph Henry Blackburne: A Chess Biography', and 'Eminent Victorian Chess Players'
http://www.chessmail.com

Kevin Thurlow
Posts: 5817
Joined: Wed Apr 30, 2008 12:28 pm

Re: Lomonosov tablebases

Post by Kevin Thurlow » Wed Mar 01, 2017 10:54 pm

"And in the case of ICCF correspondence tournaments, if the tablebase says there is a forced mate, it is adjudicated a win even if it would take more than 50 moves without pawn move or capture."

Just to clarify, in ICCF events, if you reach a position with tablebase win or draw, you can immediately claim it - you don't wait until the end of the playing session, even if it's months away.

Roger de Coverly
Posts: 21297
Joined: Tue Apr 15, 2008 2:51 pm

Re: Lomonosov tablebases

Post by Roger de Coverly » Wed Mar 01, 2017 11:04 pm

Kevin Thurlow wrote: Just to clarify, in ICCF events, if you reach a position with tablebase win or draw, you can immediately claim it - you don't wait until the end of the playing session, even if it's months away.
Presumably then the ICCF are setting aside the 50 move rule?

Matt Fletcher
Posts: 271
Joined: Wed Mar 09, 2011 9:42 pm

Re: Re:

Post by Matt Fletcher » Thu Mar 02, 2017 12:37 am

Tim Harding wrote: And in the case of ICCF correspondence tournaments, if the tablebase says there is a forced mate, it is adjudicated a win even if it would take more than 50 moves without pawn move or capture.
That's interesting - seems like TCEC are using the same rules.

Post Reply