|(1) Posted by Bob Baker [Sunday, Jul 28, 2019 09:15]|
Renewed Construction Challenge
Some time ago I posted a challenge to construct a legal drawn position with no promoted pieces in which a large number of half-moves are '!' moves for both sides under the Nunn convention. The idea was to see if an 18-ply position I had constructed could be improved. Rather quickly, Geir Østmoe produced a 32-ply position, which has not been surpassed.
A variation of the challenge was suggested in which none of the moves are captures, based on my original position which ended with 11 non-capture plies. By mining the Syzygy tablebases, Arpad Rusz found a 16-ply sequence to be the longest for 5-man or smaller positions. I would have bet a large sum that by this time, Geir would have posted a position surpassing Arpad's record position, but it seems I would have lost that bet. I want to state the challenge again to see if Geir or anyone else is ready to take a shot at it.
Challenge: Construct a legal drawn position with no promoted pieces in which the first 17 or more half-moves include no captures and are '!' moves for both sides under the Nunn convention.
Considering Arpad's result, it will have to be larger than a 5-man position.
|(2) Posted by Arpad Rusz [Monday, Jul 29, 2019 21:51]|
Two positions with 6 pieces:
19 unique half-moves (but there are also some captures):
This is like a study (16 unique half-moves):
|(3) Posted by Geir Sune Tallaksen Østmoe [Monday, Jul 29, 2019 23:35]|
I can confirm that you would have lost the bet. I tried, but gave up. 16 ply without captures is very difficult.
|(4) Posted by Rewan Demontay (Real Name: J. Malcom) [Tuesday, Jul 30, 2019 01:37]|
Hey Bob, just for fun, why not allow for promoted pieces as a separate challenge? Of course the challenge now is to go beyond 32-ply and have a longer captureless sequence.
The following diagram of mine does not fall under the Nunn Convention, but it's a starting point to inspire others, I suppose. I created this myself. Stockfish can easily find one of the many possible solutions
(= 13+12 )
|(5) Posted by Bob Baker [Wednesday, Jul 31, 2019 16:02]|
Rewan, seeing your diagram helped me to understand why I specified no promoted pieces. (I'm kidding. I know it's just for fun.)
|(6) Posted by Bob Baker [Wednesday, Jul 31, 2019 16:29]|
Arpad, your second position sets a new kind of record: Longest sequence in which the game must end without either side having had a non-unique drawing move.
|(7) Posted by Rewan Demontay (Real Name: J. Malcom) [Wednesday, Jul 31, 2019 17:07]|
Hey Arpad, please could you mention who the authors are for each of those two studies that you showed?
|(8) Posted by Bob Baker [Thursday, Aug 1, 2019 16:30]|
In fact, who should be credited as the author of a position found by a computer program mining a set of tablebases?
|(9) Posted by Rewan Demontay (Real Name: J. Malcom) [Thursday, Aug 1, 2019 17:51]|
Oh, they were computer generated?!
|(10) Posted by Arpad Rusz [Friday, Aug 2, 2019 00:31]|
Those two examples were mined by me from the Syzygy tablebases with the help of SEE.
|(11) Posted by Rewan Demontay (Real Name: J. Malcom) [Friday, Aug 2, 2019 00:39]|
What is this "SEE" that you speak of?
|(12) Posted by Hauke Reddmann [Friday, Aug 2, 2019 10:39]|
@Bob: Good point, the writer of the program definitely should
get SOME of the credit. (Cf. precedence case by H. Mertes
and his KS/KS helpmates published as "Mertes & Co.")
|(13) Posted by Bob Baker [Friday, Aug 2, 2019 12:42]|
The new kind of record I mentioned for Arpad's second position can easily be increased to 20 plies by including a preamble with a position like "1Kr5/4p3/Q1N1q1p1/P2k4/6P1/8/8/8 w - - 0 1". I'm pretty sure Geir can come up with a longer preamble.
|(14) Posted by Arpad Rusz [Saturday, Aug 3, 2019 22:49]|
'SEE' is Syzygy Endgame Explorer, a mining tool written by me. (https://ruszchessstudies.blogspot.com/search/label/SEE)
|(15) Posted by Bob Baker [Sunday, Aug 4, 2019 16:33]|
@Hauke: I suppose some credit is due to each of the entities that made finding the position possible. That would include the creator of the tablebases, the provider of the tablebases for public use, the author of the mining tool, the person who used the mining tool to locate the position, and the computer on which the mining tool executed.
Some of those have a second level. Who is due the credit for being the creator of the tablebases? That would include the author of the program which found and stored the tablebase positions and the computer on which that program executed. In the case of the Lomonosov tablebases, that would be the Lomonosov 33,072-processor supercomputer named after Mikhail Lomonosov, the 18th century scientist who founded Moscow State University. For the Syzygy tablebases, Ronald de Man generated the 6-man tables and Bojun Guo generated the 18 terabytes of the 7-man tables, but I don't know what supercomputers they used.
No more posts
MatPlus.Net Forum Misc Renewed Construction Challenge