Hard Test Suite 2008 (Vincent Lejeune) CEGT results

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Vinvin
Posts: 5223
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: Rybka 3 alternatives

Post by Vinvin »

Andrew wrote:
Ovyron wrote:It seems the Persistent Hash implementation is bugged and it ignores the "Play Depth" parameter, so make it 4 more bugs.
Has this been mentioned on the Rybka forum anywhere? I tried a 60'+0
game from a set position between two of the personalities, each with
its own persistant hash file and default parameters for this.

After playing I restarted a new game from the same position and it
played out most of the same game very quickly. I've also noticed
it sometimes instant plays moves for several moves in a row
(this is with pers. hash off) which can't be good.
Post your msg here : http://rybkaforum.net/cgi-bin/rybkaforu ... w.pl?bid=2
Directly to the producer ! ;-)
Vinvin
Posts: 5223
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: HTS-2008 (Vincent Lejeune) Verses Rybka 3 and the winner

Post by Vinvin »

Hi Dan, thanks for testing !
75% of the responses of your question about alteratives move are here http://home.scarlet.be/vincentlejeune/h ... -beta3.pgn , follow the main line and you should see a better score than the Rybka3 move, for the other 25% , I investigate.

This take time ... but you already know that ;)
User avatar
smirobth
Posts: 2307
Joined: Wed Mar 08, 2006 8:41 pm
Location: Brownsville Texas USA

Re: Hard Test Suite 2008 (Vincent Lejeune) CEGT results

Post by smirobth »

Very interesting test suite, I love hard problems and a bunch of these are really cool. I have looked at several of the positions (from beta 1, since I downloaded it about 2 weeks ago and have not had internet access again until recently.). Here are three of them that I think might be flawed:

[D]8/5k2/p7/2pP1Pp1/2P3P1/2N3K1/P4N2/5q2 b - - 0 1
1... Kf8 (1... Qxc4 ? 2.Nfe4 {Draw} [R.S.: I don't think it is a draw, for example Black can now play 2... Kf8 3.Kf2 Qd4+ 4.Kf3 c4 and the black queen will eventually work its way over to f4, pick off the g4 pawn, with an eventual win)]

[D]6k1/4pp1p/3p2p1/7n/p3P3/r4P2/P1PK2P1/1R5R w - - 0 1
1.Rxh5 gxh5 2.Rb2
[R.S.:I think there are other ways for White to also trap and win the a3 rook, for example 1.Rb8+ Kg7 2.Ra1 or simply 1.Rb2 both should also eventually win the rook with an easy win for White.]

[D]8/8/7p/3KNN1k/2p4p/8/3P2p1/8 w - - 0 1
1.Kc6! (1.Ng7+ Kg5 2.Nf3+ Kg4 3.Nf5 [R.S.: I think white can probably also draw with 3.Ke4 h3 4.Nf5 g1=Q 5.Nxg1 h2 6.Nxh6+ Kh5 7.Nf3 h1=Q 8.Nf5 for example: 8...Kg4 9.Ne3+ Kg3 10.Nf5+ Kf2 11.N5d4 and Black cannot make progress.]
- Robin Smith
User avatar
Ovyron
Posts: 4556
Joined: Tue Jul 03, 2007 4:30 am

Re: Rybka 3 alternatives

Post by Ovyron »

Andrew wrote:Has this been mentioned on the Rybka forum anywhere?
Yes, with Play Depth = 64 Rybka should not be playing instantly at all. When Vas is online he's very responsive and replied to all of my post, perhaps he's looking in the issue (Rybka 3.1 anyone?)
Vinvin
Posts: 5223
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: Rybka 3 alternatives

Post by Vinvin »

Ovyron wrote: (Rybka 3.1 anyone?)
Here's the poll for a bugs fixed version : http://rybkaforum.net/cgi-bin/rybkaforu ... l?tid=5971