Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Nunn's K & P suite

Author: Robert Hyatt

Date: 05:45:54 02/04/00

Go up one level in this thread


On February 04, 2000 at 02:13:20, Jeremiah Penery wrote:

>On February 04, 2000 at 00:02:09, Robert Hyatt wrote:
>
>>On February 03, 2000 at 17:08:37, Amir Ban wrote:
>>
>>>
>>>There's an article by Nunn & Friedel in the latest ICCAJ called "Brains of the
>>>Earth" on a contest to solve 6 very hard king & pawn endings. The article says
>>>solving this suite in 24 hours was an exceptional achievement, and Kasparov made
>>>an amazing achievement of solving it in 45 minutes. It's also possible to
>>>understand from the article that Hiarcs could not solve the set.
>>>
>>>So I set up the EPD, intending to let Junior think about it all night, but I was
>>>surprised to find that Junior (no tablebases) can find all 6 key moves in less
>>>than 2 minutes each.
>>>
>>>Here is the EPD:
>>>
>>>6k1/6p1/8/4K1P1/8/7P/8/8 w - - ; bm Kf4;
>>>k7/4p3/4p3/8/8/3P1P2/5P2/K7 w - - ; bm Kb2;
>>>8/4p3/2kp4/4p3/6K1/5P2/3P4/8 w - - ; bm Kg5;
>>>k7/8/1p6/p1p5/2P4K/8/PP6/8 w - - ; bm a4;
>>>8/8/p7/8/1P6/7p/P4k1P/3K4 w - - ; bm a3;
>>>8/5p1p/8/6k1/8/6P1/5PP1/7K w - - ; bm Kh2;
>>>
>>>I don't quite understand it.
>>>
>>>Amir
>>
>>position 2:
>>
>>I think this must be obvious.  Kb2 is the _only_ move I get.  My first output
>>is at ply=14, with Kb2 as the move.  I stopped it at depth=28, 30 seconds,
>>move was still Kb2, eval still +1.5
>
>This is strange... Here's what I'm getting:

remember that (a) my disks are _extreme_.  10K/80mbytes/sec drives.  And
I run with a big hash, which helps here (hash=384M).  And I use 4 cpus of
course. :)




>
>18->  46.18   1.94   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kd4 e6
>                     8. Ke3 Kg6 9. Ke4 Kf7
>19    48.41     ++   1. Kb2!!
>19     2:16   2.98   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kd4 Kg4
>                     8. Ke4 Kh3 9. f5 Kg4 10. f4 <HT>
>19->   2:17   2.98   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kd4 Kg4
>                     8. Ke4 Kh3 9. f5 Kg4 10. f4 <HT>
>20     3:44   2.98   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kd4 Kg4
>                     8. Ke4 Kh3 9. f5 Kg4 10. f4 Kg3
>20->   3:50   2.98   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kd4 Kg4
>                     8. Ke4 Kh3 9. f5 Kg4 10. f4 Kg3
>21     3:58     ++   1. Kb2!!
>21     9:16   3.54   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kc5 Kg6
>                     8. Kc6 Kg7 9. f5 Kh6 10. f6 e6 <HT>
>21->   9:19   3.54   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                     Kd5 5. f4 Ke4 6. Kc4 Kf5 7. Kc5 Kg6
>                     8. Kc6 Kg7 9. f5 Kh6 10. f6 e6 <HT>
>22     9:25     ++   1. Kb2!!
>
>I'm not sure quite why it's taking so long (My HD is going crazy with TB probes,
>and it probably is a fairly slow drive, but I don't see why it's making this
>take _this_ long...).  I'm also not sure why I'm getting such a high score, when
>everyone else is getting much lower scores.


No idea there.



>
>>position 3:
>>
>>Kg5 is the only move I get from the first printed output thru 25 plies.  Score
>>is -1.0, but no other move ever shows up as better...
>
>I get similar results here. :(
>
>>position 4:
>>
>>a4 is also the only move I get, from the first output at 14 plies thru 27 plies
>>where I stopped it.  eval=.36, so it isn't crushing...  but it does agree with
>>the move.
>
>Same here...
>
>>position 5
>>
>>a3 is found instantly.  And it is crushing.  By ply 24, 30 seconds, score is +7
>>and climbing every iteration.
>
>Yuck.  It takes my computer about 3 minutes to find a3 on ply 16.  After that,
>the plies go fairly quickly and the score rises a bunch.
>
>>position 6.
>>
>>The first "hard" one (one that takes over a few milliseconds to get the
>>suggested best move).  Crafty likes Kg1 until depth 22, 20 seconds.  At
>>that point it switches to Kh2 and sticks with it.
>
>I'm getting Kh2 from ply 13 on (10 seconds), but it is not finding the best
>continuation of 2. Kh1 in reasonable time (a couple minutes :).
>
>All my times/scores are from a Modified Crafty 17.7 on a PII-300, using (at
>least) the KPPKP TB.  I'll also run the harder ones (2,3,4,6) without TBs, and
>see what I get...
>
>2) 24->  57.58   1.23   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                        Kd5 5. f4 Ke4 6. Kc4 Kxf4 7. Kd5 Kf5
>                        8. f3 Kg5 9. Ke4 e6 10. f4+ Kg4 11.
>                        Ke3 Kf5 12. Kf3 Kg6 13. Ke4
>   25     2:01     ++   1. Kb2!!
>   25->  13:17   1.62   1. Kb2 Kb7 2. Kc3 e5 3. d4 Kc6 4. dxe5
>                        Kd5 5. f4 Ke4 6. Kc4 Kxf4 7. Kd5 Kf5
>                        8. f3 Kg5 9. Ke4 e6 10. f4+ Kg4 11.
>                        Ke3 Kf5 12. Kf3 Kg6 13. Ke4
>Hardly convincing, but the score may continue to rise...I'll have to let these
>run longer to see.
>
>3) 22->   4:08  -0.32   1. Kg5 d5 2. Kf5 Kd6 3. d3 d4 4. Kg6
>                        Kc5 5. Kf5 Kd5 6. Kg6 Kd6 7. Kf7 Kd7
>                        8. Kg6 Kc6 9. Kf7 e4 10. fxe4 Kd6 11.
>                        Kg6 e5
>I think this is going to end up in a draw score, like Dann was getting.
>
>4) 21     1:01   0.00   1. a4 Kb8 2. Kg5 Kc7 3. Kf5 Kd6 4.
>                        Kf4 Ke6 5. Ke4 Kd6 6. Kf3 Kc7 7. Kf4
>                        Kd6 8. Kf3
>
>6) 21     1:37   0.82   1. Kh2 f6 2. Kg1 h5 3. Kf1 Kf5 4. f3
>                        Ke6 5. g4 hxg4 6. fxg4 Ke5 7. g3 Kd4
>                        8. Kf2 Ke4 9. Ke2 <HT>
>
>
>I'm really going to have to let these things search overnight, maybe with TBs,
>to see if I can get anything interesting...



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.