Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Looking for Fine endgame test suite.

Author: Robert Hyatt

Date: 10:54:09 03/27/03

Go up one level in this thread


On March 27, 2003 at 09:08:04, David Dahlem wrote:

>On March 27, 2003 at 06:31:04, Tim Foden wrote:
>
>>Hi All,
>>
>>As I have the following Fine 70 position:
>>
>>[D]8/k7/3p4/p2P1p2/P2P1P2/8/8/K7 w - - bm Kb1; id "Fine70";
>>
>>I assume that there are 1..69, and maybe 71..x also available.  :)
>>
>>Does anyone have these, preferably as EPD?
>>
>>Cheers, Tim.

A couple of notes below.  However, with the fixes I give below, where Fine's
book
is wrong in several places as endgame tables (and sometimes raw computer search
analysis) shows,  crafty gets 34 of 35 right at 10 secs/move on my dual 2.8.  At
5
secs/move, it gets 33 right, and at 1 sec/move it gets 33 also.  I get the same
number
right whether I use 1 or 2 cpus on this test, btw.

total positions searched..........          35
number right......................          34
number wrong......................           1
percentage right..................          97
percentage wrong..................           2
total nodes searched..............    33395031
average search depth..............         4.1
nodes per second..................     1918152






>
>3k4/4pp2/7K/3pPp2/3P1P2/2P5/8/8 w - - bm Kg7; id "Fine 90b";
>4k3/5ppp/8/8/8/8/PPP5/3K4 w - - bm Ke2; id "Fine 68";

This one needs checking.  a4 is steadily climbing.  It starts off big, drops to
just below
zero, then starts climbing at depth 17/18/etc.  I have now let it run thru
depth=22 and the
score for a4 is only up to +.55, so I am not sure since I don't know whether
this is a
white to move and win or white to move and draw position.  If the latter, then
a4 might
well be a second solution.  If it is to move and win, a4 might not be good
enough.


>4K3/pk6/2p5/2P5/1pP4P/1P6/8/8 w - - bm Kd7; id "Fine 78";
>6k1/7p/5P1K/8/8/8/7P/8 b - - bm Kf7; id "Fine 25";
>8/1k3ppp/8/3K4/7P/5PP1/8/8 b - - bm Kc7; id "Fine 66";
>8/3p4/3k4/8/3K4/8/2P1P3/8 w - - bm e4 c4; id "Fine 43";
>8/4p3/4k3/8/4K3/8/3PP3/8 w - - bm Kd4 d4; id "Fine 36";
>8/7p/3p4/8/6pP/8/4PK2/7k w - - bm Kg3; id "Fine 59c";
>8/7p/5K2/3k4/6P1/8/6P1/8 b - - bm Ke4; id "Fine 45";
>8/8/2p5/1kp5/8/2K5/1P1P4/8 b - - bm c4; id "Fine 53fb";
>8/8/2p5/1kp5/8/2K5/1P1P4/8 w - - bm b3; id "Fine 53fw";
>8/8/2p5/1p3k2/1P5P/5K2/8/8 w - - bm h5; id "Fine 50";
>8/8/2pp3k/8/1P1P3K/8/8/8 b - - bm d5; id "Fine 51";
>8/8/3k1p2/5P2/3K4/8/3P4/8 w - - bm Ke4 d3; id "Fine 22";

This one is wrong.  Or at least there is a better move.  Kc4 leads to a mate in
20 (endgame tables) and there is no shorter (or equal) mate there.



>8/8/3k1ppp/8/P2K2PP/8/8/8 w - - bm Kc4; id "Fine 62";

Ke4 also wins easily, score > +11.0  a5 might win the same way but I didn't
check.



>8/8/3kp3/8/3PP3/3K4/8/8 w - - bm Kc3 Ke3; id "Fine 37";

Another one that is wrong.  Kc4 is a mate in 24, and that is the shortest mate
possible here.


>8/8/4k3/8/4K3/8/4P3/8 w - - bm e3; id "Fine 11";
>8/8/4pk2/8/3PPK2/8/8/8 w - - bm Ke3; id "Fine 36a";


Kg4 is the shortest mate (25) here.  All other moves are worse.


>8/8/5P2/8/6K1/7p/6k1/8 w - - bm f7; id "Fine 14";
>8/8/6kp/p7/1p4P1/1P3K2/P1P5/8 w - - bm c4; id "Fine 73";
>8/8/7k/8/4p1K1/8/5P2/8 b - - bm e3; id "Fine 16";
>8/8/8/1k6/3pP3/1K1P4/8/8 w - - bm Kc2; id "Fine 31";
>8/8/8/3P1k2/7P/3K1p2/8/8 b - - bm Kg4 Kf4; id "Fine 20";
>8/8/8/4k3/5pP1/5P2/4K3/8 b - - bm Kd4; id "Fine 31a";

Ke6 is a forced draw.  There is no forced win here.  I didn't check to see if
the
suggested best move Kd4 is also a forced draw or not.


>8/8/8/4p1p1/8/5P2/6K1/4k3 w - - bm Kh1; id "Fine 43c";

Kg1 is a forced draw (EGTB).  Kh1 might be also, I didn't try.  But the best
white can do here is draw.



>8/8/8/5p1p/4k2P/6P1/5P2/5K2 b - - bm Kf3; id "Fine 56";
>8/8/8/8/4k1p1/8/4KPP1/8 w - - bm f3; id "Fine 38";

g3 is a mate in 26 and it is the shortest mate possible here.  f3 may be a mate,
I don't
know, but g3 is the shortest mate.


>8/8/8/8/8/P4p2/k6K/8 w - - bm a4; id "Fine 13";
>8/8/k7/2p5/K1P5/8/P7/8 w - - bm Ka3; id "Fine 21a";
>8/8/p3k3/1pP4p/1P6/P3K3/8/8 w - - bm Ke4; id "Fine 64";
>8/p7/2k5/8/2K5/8/PP6/8 w - - bm Kb4; id "Fine 39";

Kb4 and a4 are both mate in 29.  a4 should be an alternate solution.



>8/p7/2k5/8/2P5/2K5/P7/8 w - - bm Kd4; id "Fine 24";
>8/pp2k3/8/3PPK2/8/8/8/8 w - - bm d6; id "Fine 47";
>8/pp3k2/2p3p1/4p3/8/2P5/P1P3PP/6K1 w - - bm h4; id "Fine 81";
>8/pp6/8/PPP2k1p/8/5K2/8/8 w - - bm a6 b6 c6; id "Fine 60";
>
>Regards
>Dave



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.