Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Another King example!

45 views
Skip to first unread message

Har...@t-online.de

unread,
Jul 13, 1998, 3:00:00 AM7/13/98
to
Hi all,

The following position appeared when the King had to move with black.
I checked this a bit on K6-200. The King finds 1...f5! between 10 and 20
sec (depending if normal or active setting). Junior4.6 needs 3min, CSTal
5min.
What about the other programs?


3rr1k1/1q3ppp/p7/2Qnn3/1P2N3/B6P/5PP1/R2R2K1 b - -


1...f5! 2.Nd6 Rxd6 3.Qxd6 Nf4 4.f3 Nxf3+ 5.gxf3 Qxf3 -+
IE 6.Ra2 h5 7.b5 Qg3+ 8.Kh1 Qxh3+ 9.Kg1 Qg4+

Kevin Begley

unread,
Jul 13, 1998, 3:00:00 AM7/13/98
to

Hmmm, try this puzzle which I composed (with White to move):

8/nrkP1pK1/1b1q1P2/1P2p3/3n1R1p/4QP1P/7B/R7 w

See how long it takes it to find that 1.Qxe5!! wins.
Shouldn't be too long.

Kevin.

Dann Corbit

unread,
Jul 13, 1998, 3:00:00 AM7/13/98
to
Kevin Begley wrote in message <35AA8A...@earthlink.net>...
7th ply, 5 seconds for Crafty to choose that move. Eval climbs steadily
from there. Crafty had a cow trying to find the previous one. I stopped
the process after 40 minutes (in the 13th ply), still without solution. I
suspect that each searching technique varyation will produce different
results. I suspect that an 8 ply brute force approach might occasionally be
faster to solve some problems than the cleverest search yet devised. I'd
like to see if any program can find Marshall's remarkable mate. I forget
the "common name" of the game, but Marshall said it was the best move he
ever made. It was an incredible combination, with Marshall throwing away
all of his power and ending up with a stunning win. I suspect after loss of
several valuable pieces with no results yet, most programs would just stop
looking.
--
Hypertext C-FAQ: http://www.eskimo.com/~scs/C-faq/top.html
C-FAQ ftp: ftp://rtfm.mit.edu, C-FAQ Book: ISBN 0-201-84519-9
Try "C Programming: A Modern Approach" ISBN 0-393-96945-2
Want Software? Algorithms? Pubs? http://www.infoseek.com

Robert Hyatt

unread,
Jul 14, 1998, 3:00:00 AM7/14/98
to
Kevin Begley <kraw...@earthlink.net> wrote:

: Har...@t-online.de wrote:
:>
:> Hi all,
:>
:> The following position appeared when the King had to move with black.
:> I checked this a bit on K6-200. The King finds 1...f5! between 10 and 20
:> sec (depending if normal or active setting). Junior4.6 needs 3min, CSTal
:> 5min.
:> What about the other programs?
:>
:> 3rr1k1/1q3ppp/p7/2Qnn3/1P2N3/B6P/5PP1/R2R2K1 b - -
:>
:> 1...f5! 2.Nd6 Rxd6 3.Qxd6 Nf4 4.f3 Nxf3+ 5.gxf3 Qxf3 -+
:> IE 6.Ra2 h5 7.b5 Qg3+ 8.Kh1 Qxh3+ 9.Kg1 Qg4+

: Hmmm, try this puzzle which I composed (with White to move):

: 8/nrkP1pK1/1b1q1P2/1P2p3/3n1R1p/4QP1P/7B/R7 w

: See how long it takes it to find that 1.Qxe5!! wins.
: Shouldn't be too long.

: Kevin.

Actually pretty easy here.

7 4.15 ++ 1. Qxe5!!
7 4.27 -0.75 1. Qxe5 Qxe5 2. Rxd4 Bxd4 3. Bxe5+
Bxe5 4. b6+ Rxb6 5. Rxa7+ Kd8 6. Kxf7
Rxf6+
7-> 4.48 -0.75 1. Qxe5 Qxe5 2. Rxd4 Bxd4 3. Bxe5+
Bxe5 4. b6+ Rxb6 5. Rxa7+ Kd8 6. Kxf7
Rxf6+
8 4.88 -0.75 1. Qxe5 Qxe5 2. Rxd4 Bxd4 3. Bxe5+
Bxe5 4. b6+ Rxb6 5. Rxa7+ Kd8 6. Kxf7
Rxf6+
8-> 8.89 -0.75 1. Qxe5 Qxe5 2. Rxd4 Bxd4 3. Bxe5+
Bxe5 4. b6+ Rxb6 5. Rxa7+ Kd8 6. Kxf7
Rxf6+
9 10.52 ++ 1. Qxe5!!
9 16.40 0.99 1. Qxe5 Ne6+ 2. Kxf7 Qxe5 3. Rc4+ Kxd7
4. Bxe5 Ng5+ 5. Kg6 Nxf3 6. Bf4 Ke6
9-> 22.88 0.99 1. Qxe5 Ne6+ 2. Kxf7 Qxe5 3. Rc4+ Kxd7
4. Bxe5 Ng5+ 5. Kg6 Nxf3 6. Bf4 Ke6
10 28.65 1.22 1. Qxe5 Ne6+ 2. Kxf7 Qxe5 3. Rc4+ Kxd7
4. Bxe5 Ng5+ 5. Kg6 Nxf3 6. Re4 Ke6
7. Bd4+ Kd5


--
Robert Hyatt Computer and Information Sciences
hy...@cis.uab.edu University of Alabama at Birmingham
(205) 934-2213 115A Campbell Hall, UAB Station
(205) 934-5473 FAX Birmingham, AL 35294-1170

Kevin Begley

unread,
Jul 14, 1998, 3:00:00 AM7/14/98
to

9 plies... that's interesting, but I think crafty missed a few things...
After 1.Qxe5 Ne6+? 2.Qxe6!! is a major improvement.
What does crafty say?

And, I assume it rejected 1...Qxe5 in light of 2.Rxd4 ...Bxd4 3.Bxe5+
...Bxe5 the stunning improvement, 4.Rxa7!!! but I think 1...Qxe5 is
objectively better. Maybe I'm biased, though, since it was the real
point of the puzzle.

But, clearly it's the most interesting line, due to the amazing counter
sacrifice: 4...Bxf6+!! 5.Kxf6 ...Kd8!!,

and there's still a few fireworks left 6.Ra5! (6.b6?? ...Rxb6+ 7.Kxf7?
...Rf6+! =) ...Kxd7 (6...Ra7?! 7.Ra6! ...Rb7 8.b6! +-)

If you let Crafty run a little longer, does it ever agree with this?

And, if you remove the f3-pawn, does it realize that the position is
a draw?


Kevin.

Dusan Dobes

unread,
Jul 14, 1998, 3:00:00 AM7/14/98
to
: Har...@t-online.de wrote:
: > I checked this a bit on K6-200. The King finds 1...f5! between 10 and 20

: > sec (depending if normal or active setting). Junior4.6 needs 3min, CSTal
: > 5min.
: > What about the other programs?
: >
: > 3rr1k1/1q3ppp/p7/2Qnn3/1P2N3/B6P/5PP1/R2R2K1 b - -
: >
: > 1...f5! 2.Nd6 Rxd6 3.Qxd6 Nf4 4.f3 Nxf3+ 5.gxf3 Qxf3 -+
: > IE 6.Ra2 h5 7.b5 Qg3+ 8.Kh1 Qxh3+ 9.Kg1 Qg4+

Kevin Begley (kraw...@earthlink.net) wrote:
: Hmmm, try this puzzle which I composed (with White to move):
: 8/nrkP1pK1/1b1q1P2/1P2p3/3n1R1p/4QP1P/7B/R7 w
: See how long it takes it to find that 1.Qxe5!! wins.

Phalanx XVI on a pentium/150, 6M hash:
f5! 2min 52s
Qxe5! 4min 55s


6 -> 0:25.80 672789
7 -39 4247 1082793 Re8-e6 Ba3-b2 Rd8-d7 Bb2xe5 Re6xe5 Ne4-d6
Qb7-a8 Rd1-e1 Re5xe1 Ra1xe1
7 -34 4495 1148248 Ne5-d3 !
7 2 4879 1267877 Ne5-d3 Rd1xd3 Re8xe4 Qc5-a5 Rd8-d7 Ra1-d1
Re4-e5
7 3 6480 1673923 Pf7-f5 Ne4-d6 Rd8xd6 Qc5xd6
7 4 9879 2539895 Rd8-d7 Ne4-d6 Rd7xd6 Qc5xd6
7 -> 1:40.49 2583013
8 -17 15851 4150364 Rd8-d7 Ne4-d6 Rd7xd6 Qc5xd6 Nd5-f4 Pf2-f3
Ne5xf3 Kg1-h1 Nf4xg2 Qd6-d7 Qb7xd7 Rd1xd7
8 -12 17224 4561781 Pf7-f5 !
8 15 19513 5209862 Pf7-f5 Ne4-d6 Rd8xd6 Qc5xd6 Nd5-f4 Pf2-f3
Ne5xf3 Kg1-h1 Nf3-e1 Rd1-d2 Ne1xg2
8 -> 4:00.75 6354349
9 19 30581 8140595 Pf7-f5 Ne4-d6 Rd8xd6 Qc5xd6 Nd5-f4 Pf2-f3
Ne5xf3 Kg1-h1 Nf4xg2 Qd6-d7 Qb7xd7 Rd1xd7
Re8-e2
9 -> 9:23.45 14844239
10 0 72372 19206787 Pf7-f5 Ne4-d6 Rd8xd6 Qc5xd6 Nd5-f4 Pf2-f3
Nf4xg2 Rd1-d5 Ng2-h4 Pf3-f4 Ne5-c4
10 -> 30:52.31 47306044
11 32 240627 61518972 Pf7-f5 Ne4-d6 Rd8xd6 Qc5xd6 Nd5-f4 Pf2-f3
Ne5xf3 Kg1-h1 Nf3-h4 Qd6-d2 Nf4xg2 Qd2-a2
Kg8-h8 Qa2-d5 Qb7xd5 Rd1xd5
interrupted
search aborted
Depth=11, Value=32, Time=4143.18, Last turn=172.24, Nodes=104104490, N/s=25126
PV = Pf7-f5 Ne4-d6 Rd8xd6 Qc5xd6 Nd5-f4 Pf2-f3 Ne5xf3 Kg1-h1
Nf3-h4 Qd6-d2 Nf4xg2 Qd2-a2 Kg8-h8 Qa2-d5 Qb7xd5 Rd1xd5


6 -> 0:36.15 649107
7 -48 4487 799954 Kg7-h7 Nd4xb5 Rf4-c4 Kc7xd7 Qe3-e4 Rb7-c7
Rc4xc7 Bb6xc7 Qe4xh4
7 -> 1:31.73 1664419
8 -43 16842 2994329 Kg7-h7 Nd4xb5 Qe3-e4 Na7-c6 Rf4xh4 Bb6-d4
Ra1-a8 Qd6xd7
8 -38 29492 5292100 Qe3xe5 !
8 17 29835 5360975 Qe3xe5 !!
8 91 30854 5554651 Qe3xe5 Qd6xe5 Rf4xd4 Bb6xd4 Bh2xe5 Kc7xd7
Be5xd4 Na7xb5 Bd4-e5
8 -> 5:32.31 5990008
9 119 36617 6594207 Qe3xe5 Qd6xe5 Rf4xd4 Bb6xd4 Bh2xe5 Kc7xd7
Be5xd4 Na7xb5 Bd4-f2 Nb5-d6 Ra1-d1
9 -> 9:21.47 10304839
10 154 66938 12322389 Qe3xe5 Qd6xe5 Rf4xd4 Bb6xd4 Bh2xe5 Kc7xd7
Be5xd4 Na7xb5 Bd4-c5 Kd7-e6 Bc5-f2
interrupted
search aborted
Depth=10, Value=154, Time=814.94, Last turn=294.92, Nodes=15146263, N/s=18585
PV = Qe3xe5 Qd6xe5 Rf4xd4 Bb6xd4 Bh2xe5 Kc7xd7 Be5xd4 Na7xb5
Bd4-c5 Kd7-e6 Bc5-f2


Robert Hyatt

unread,
Jul 14, 1998, 3:00:00 AM7/14/98
to
Kevin Begley <kraw...@earthlink.net> wrote:

: 9 plies... that's interesting, but I think crafty missed a few things...


: After 1.Qxe5 Ne6+? 2.Qxe6!! is a major improvement.
: What does crafty say?

On my notebook, it thinks this loses, due to fxe6... It thinks the original
move Qxe5 is ok... but only "ok" as winning a pawn is not quite a won game
here...

Here's the PV for fxe6 (- means black is juch worse)

10-> 2:38 -2.64 2. ... fxe6 3. f7 Kxd7 4. f8=Q Bd4+
5. Rxd4 Qxd4+ 6. Kf7 e5 7. Qe8+ Kc7
8. Bxe5+ Kb6+ 9. Kf6 Qf2 10. Ra6+ Kc5
11. Bd6+ Kc4

I'll try to answer your other questions later. A GM is playing Crafty
on ICC, which means I can't use the ALR right now to get deeper analysis
than my notebook can provide...

Eduardo Suastegui

unread,
Jul 14, 1998, 3:00:00 AM7/14/98
to
I ran the Fritz 5 analsys module under CB Light, and after 3 minutes it
settles on ...Re6 and ...Qc6, with the latter marked as the best move.
Perhaps this is not the best way to test this. However, I also analyzed the
position using the Fritz 5 program, and got the same result.

--
Eduardo Suastegui
http://home.earthlink.net/~esuastegui/eschess
"Once upon a time, chess players would gather at coffee houses.
Now, they ponder chess over a cup of Java."
Remove the 101 when replying via email.

Dann Corbit

unread,
Jul 14, 1998, 3:00:00 AM7/14/98
to
Description:
ftp://sunsite.unc.edu/pub/Linux/games/strategy/phalanx-16.lsm
Tar file:
ftp://sunsite.unc.edu/pub/Linux/games/strategy/phalanx-16.tar.gz

--
Hypertext C-FAQ: http://www.eskimo.com/~scs/C-faq/top.html
C-FAQ ftp: ftp://rtfm.mit.edu, C-FAQ Book: ISBN 0-201-84519-9
Try "C Programming: A Modern Approach" ISBN 0-393-96945-2
Want Software? Algorithms? Pubs? http://www.infoseek.com

mclane wrote in message <6oh06f$138$1...@steve.prima.ruhr.de>...


>do...@bart.math.muni.cz (Dusan Dobes) wrote:
>>Phalanx XVI on a pentium/150, 6M hash:
>>f5! 2min 52s
>>Qxe5! 4min 55s
>

>Aha. But what is Phalanx XVI ?!?
>
>I have never heard about this program. Can you tell us some more about
>it ? That would be nice, thanks in forward.
>best wishes
>
>mclane
>


mclane

unread,
Jul 15, 1998, 3:00:00 AM7/15/98
to

Dusan Dobes

unread,
Jul 16, 1998, 3:00:00 AM7/16/98
to
mclane (mcl...@prima.ruhr.de) wrote:

: Aha. But what is Phalanx XVI ?!?

: I have never heard about this program. Can you tell us some more about
: it ? That would be nice, thanks in forward.

Phalanx is my hobby project. It started in March 1997. It's developed
under Linux and GCC, I also prepared binary distribution for Win32 with
latter versions. Licensing policy is GPL (free and in sources).
Interface is xboard/winboard/RoboFICS compatible. It plays on FICS
as 'pikozrout', it's current blitz rating is 2380, standard 2210.

sources:
ftp://sunsite.unc.edu/pub/Linux/games/strategy/phalanx-16.tar.gz
(sunsite has many mirrors)

ftp://ftp.math.muni.cz/pub/math/people/Dobes/phalanx-16.tar.gz

binaries for Win32:
ftp://ftp.math.muni.cz/pub/math/people/Dobes/winphx16.zip

Dusan Dobes

0 new messages