Getting ready for Thermopylae

[hgm]

I have tested all engines we have now to my satisfaction, so I wanted to start the official Spartan Chess tournament (Thermopylae 2011) soon. A suitable tournament form seems the following:

Participants:

Code: Select all

1. Nebiyu
2. Spartacus
3. Oberon
4. ChessV
5. Fairy-Max
6. Sjaak
7. Catalyst

Start with a round-robin with 4 games per pairing, (played in 2 cycles), where Nebiyu and Spartacus (which are a bit stronger than the rest) play with a time-odds handicap of 8 or 10. This will keep their games more interesting, as most other participants will then actually stand some chance against them. This also has the advantage that it is not only their mutual games that decide who is going to win the tourney, but also who does better against the other opponents.

After that, the first 4 will go into a play-offs round-robin with 2 games per pairing, no handicaps, where the score from the qualification group keeps counting.

@Daniel: which version of Nebiyu should I take? The Nebiyu_Chess 1.0 from thepackage that I downloaded to test the alien stuff?

@Gregory: No hope for a Spartan Quadrox, I suppose?

[Evert]

Do I have time to send you an updated copy of Sjaak with a number of bugs fixed (including the annoying time-loss issue that I thought I'd fixed before)?

[George Tsavdaris]

I have tested all engines we have now to my satisfaction, so I wanted to start the official Spartan Chess tournament (Thermopylae 2011) soon. A suitable tournament form seems the following:

Participants:

Code: Select all

1. Nebiyu
2. Spartacus
3. Oberon
4. ChessV
5. Fairy-Max
6. Sjaak
7. Catalyst

Start with a round-robin with 4 games per pairing, (played in 2 cycles), where Nebiyu and Spartacus (which are a bit stronger than the rest) play with a time-odds handicap of 8 or 10. This will keep their games more interesting, as most other participants will then actually stand some chance against them. This also has the advantage that it is not only their mutual games that decide who is going to win the tourney, but also who does better against the other opponents.

I don't really agree with this setting, but i have to ask first:
If no time handicap is given then Nebiyu and Spartacus will win all their games against the other engines? 50 out of times for example? Or 10 out of 10? What is the percentage approximately?

After that, the first 4 will go into a play-offs round-robin with 2 games per pairing, no handicaps, where the score from the qualification group keeps counting.

Great let it start!

[Daniel Shawul]

Mostly I was working on my Go engine but there are some side improvements on the spartan as well. Fixed serious bug in reductions and
also a separate piece square table for pawns and hoplites.
Download https://sites.google.com/site/dshawul/N ... ects=0&d=1 .

The handicaps is a good idea at least until we get some more engines and start different divisions. For George I am joking

[Greg Strong]

@Gregory: No hope for a Spartan Quadrox, I suppose?

Not anytime real soon, I'm afraid

[Evert]

Ok, I just sent you an updated version of the program. At very short time controls (40 moves/10s) this seems to be considerably stronger when tested against earlier versions, at both normal chess and Spartan. BayesElo gives it +138 with an error bar of 28 against the last version you tested, which is actually pretty sad considering the fairly poor chess that it plays...

[hgm]

Mostly I was working on my Go engine but there are some side improvements on the spartan as well. Fixed serious bug in reductions and
also a separate piece square table for pawns and hoplites.
Download https://sites.google.com/site/dshawul/N ... ects=0&d=1 .

How sure are you that these improvements actually make it play stronger? I am playing it against Spartacus now, but it is only at 4.5 out of 20 (which is even exaggerated because Spartacus threw a game by crashing in totally won position), while Spartactus actually had a very hard time to keep up with the old version (16:12:16, I believe).

[Daniel Shawul]

It definately looks like I have screwed up. For one I was using the reduction and pruning I was using for GO. I also do not think the pawn pushes are helping from the games I saw. So I am undoing the changes and will update you after two hours or so.

[Daniel Shawul]

Ok I have fixed it now. The version number for the spartan is incremented to 1.1 to avoid confusion. It probably will perform good,
but if it doesn't perform use the old one. Unfortunately I don't have that version with me. Download at the previous link.

[PK]

hi,

I've just uploaded version 0.04 at the link:

www.koziol.home.pl/oberon.zip

It has slightly better move ordering and some minor time management enhancements. There are two executables: one for Spartan, one for normal chess.