Endgame study solving competition
harold van der heijden
(my previous message on the same subject was incomplete)
ARVES, the Endgame Circle of The Netherlands and Flanders (named after
Alexander Rueb), recently organized an endgame study solving
competition.
17 competitors had to solve 6 endgame studies in 90 minutes. Points were
awarded for 'key-move', 'full solution' and accidental busts.
Two computer programs competed; DIEP (operated by Vincent Diepeveen on a
200 MHz Pentium) and Chess Genius 3.0 (on an 'old' 386SX/25).
1. Rob Bertholee (NL) 21
2. Marcel Van Herck (BEL) 16
3. Harold van der Heijden (NL) 14
4. DIEP 10
5. Koen Versmissen 8
6. Uitenbroek/Wissmann 7
8. team/Benak 6
10. Chess Genius 5
11.17
The positions were:
1) Kc7, Sc6, pb7; Kc3, Bg1, a7, f7, g5. White to play and win.
2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
Any solutions (especially by computer programs) will be highly
appreciated. If you want to try it, restrict yourself to the 90 minutes,
and give a full solution (not just a key-move).
If you use a computer, please state program/CPU/MHz etc. Next week I
will present the results, if any are forwarded to me.
Good luck!
Harold van der Heijden
chessman
>
>1) Kc7, Sc6, pb7; Kc3, Bg1, a7, f7, g5. White to play and win.
>2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
>3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
>4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
>5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
>6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
>
are the S=Knights ?? assuming K are Kings, R are rooks B are bishops
and Q are queens
chessman
>(my previous message on the same subject was incomplete)
>
>The positions were:
>
>1) Kc7, Sc6, pb7; Kc3, Bg1, a7, f7, g5. White to play and win.
>2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
>3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
>4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
>5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
>6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
>
also am i correct in assuming the piece listed first are white??
brucemo
chessman wrote:
>
> On Mon, 16 Dec 1996 21:15:45 +0100, harold van der heijden
> <har...@pi.net> wrote:
>
> >
> >1) Kc7, Sc6, pb7; Kc3, Bg1, a7, f7, g5. White to play and win.
> >2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
> >3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
> >4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
> >5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
> >6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
> >
>
> and Q are queens
S = "springer" = knight.
You know the rest.
Here it is in the archaic "CI" format, which you can convert to EPD and re-post
if you wish.
bruce
-----
echo position 1
svfe 8/pPK2p2/2N5/6p1/8/2k5/8/6b1 w - - 0 1
srch
echo position 2
svfe 1N2B3/b4p2/2K5/k7/8/4p3/1P6/8 w - - 0 1
srch
echo position 3
svfe K7/8/8/8/7k/2N3p1/1N5p/R7 w - - 0 1
srch
echo position 4
svfe 8/8/2pP2R1/4r3/5p2/8/2K5/k7 w - - 0 1
srch
echo position 5
svfe 6b1/8/5B1P/5KP1/2n5/4P3/p7/4b2k w - - 0 1
srch
echo position 6
svfe 6R1/2Q5/1p6/8/4q3/p7/k7/rN1K4 w - - 0 1
srch
harold van der heijden
chessman wrote:
>
> On Mon, 16 Dec 1996 21:15:45 +0100, harold van der heijden
> <har...@pi.net> wrote:
>
> >
>
> >The positions were:
> >
> >2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
> >3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
> >4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
> >5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
> >6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
> >
>
yes
brucemo
chessman wrote:
>
> On Mon, 16 Dec 1996 21:15:45 +0100, harold van der heijden
> <har...@pi.net> wrote:
> >1) Kc7, Sc6, pb7; Kc3, Bg1, a7, f7, g5. White to play and win.
> >2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
> >3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
> >4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
> >5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
> >6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
> >
One more thing:
What are the rules?
I'd like to do this test and see how my program would have done.
bruce
Vincent Diepeveen
>chessman wrote:
>>
>> On Mon, 16 Dec 1996 21:15:45 +0100, harold van der heijden
>> <har...@pi.net> wrote:
>
>> >1) Kc7, Sc6, pb7; Kc3, Bg1, a7, f7, g5. White to play and win.
>> >2) Kc6, Be8, Sb8, b2; Ka5, Ba7, e3, f7. White to play and draw.
>> >3) Ka8, Ra1, Sb2, Sc3; Kh4, g3, h2. White to play and win.
>> >4) Kc2, Rg6, d6; Ka1, Re5, c6, f4. White to play and win.
>> >5) Kf5, Bf6, e3, g5, h6; Kh1, Be1, Bg8, Sc4, a2. White to play and draw.
>> >6) Kd1, Qc7, Rg8, Sb1; Ka2, Qe4, Ra1, a3, b6. White to play and draw.
>> >
>
>One more thing:
>
>What are the rules?
First: try to find the keymove.
Position 4 (Rg6-h6!) is the only difficult beginning move.
It is quite hard for programs. Diep needs about 12 minutes for this problem
to see that Rh6 is far better than the rest.
The rest of the beginning moves your program will play without doubt
within a second. For position 2 it is interesting to see how many seconds
your program needs to see that it is an absolute draw so score 0.00.
Moves are less
interesting in my viewpoint. The score the program gives is the thing
what is interesting.
Problem 3 you may skip, except if it doesn't know how to capture pawns... :)
This problem is hard for certain humans, but only when they forget to see the
first move. It is a typical 2 ply position for a program, which searching
all first moves will definitely not miss it.
Rest of the positions are hard, but programs simply play the best move.
You only need to enter sometimes a stupid move. Position 5 for example is
hard, but it appears that programs play simply the best move, not because
they see a draw, but because they see that all other moves loose.
There is after few moves however a critical position where you need
to let the program search for several time (which i didn't do).
if i remember well (write down moves by head, sorry if i mistake):
after g6, Nxe3+ Kg5 Nd5,h7 bh4+,Kh4 Nxf6,h8Q a1Q black must play the critical
move Kg5! instead of Qh6.
>I'd like to do this test and see how my program would have done.
>bruce
Vincent
--
+----------------------------------------------------+
| Vincent Diepeveen email: vdie...@cs.ruu.nl |
| http://www.students.cs.ruu.nl/~vdiepeve/ |
+----------------------------------------------------+
harold van der heijden
I will comment on the solutions and your remarks next week, when other
program(mers) had a change to send me solutions.
(Please solutions to my E-mail address, otherwise it's less fun for
other potential competitors...).
Anyway, one general remark I do want to make now is: You are very wrong
in many of your assumptions. It was a endgame-study-solving-competition,
not a competition to find to the quickest win (mate) in a position. Of
course a good endgame study involves 'play' by black, and also a trick,
a nice surprise. If a computer program cannot select these moves, even
if it could beat the world champion in over-the-board play, it still
cannot solve (many) endgame studies.
Also, consider the fact that sub-optimal moves by black (that is: if
Black e.g. would not sacrifice a piece making only a surprising white
move the right solution) are readily recognized or even found in a
solving competition by (human) endgame study experts. So there is
nothing mysterious about it, only current chess playing programs are not
designed to solve an endgame study properly.
The same goes for solving direct mate problems. Any computer program
finds a mate-in-two within a few milli seconds. But few (if any) are
able to find the thematic solution. The problem world makes good use of
programs to check problems for correctness (only one solution possible?
problem busted by better black defence?) but no expert would claim that
programs can solve a problem.
For endgame studies, we even don't have good software to check studies
for correctness, although 'game-playing' software is becoming much
better. But with special programming tricks, I think that much better
results could be achieved.
More on this next week!
Harold van der Heijden
chessman
snip
>
>If you use a computer, please state program/CPU/MHz etc. Next week I
>will present the results, if any are forwarded to me.
>
>Good luck!
>
>Harold van der Heijden
week is up...:) , please solution and results