DJ13 Welcome Test "Reloaded"

[FriedmannC]

Hi Timo! I'm positive the next matches will also be very exciting! There is still room for DEEP JUNIOR to be improved! Maybe an update would bring it closer to its stronger opponents and hopefully I expect the next versions to become more powerfull! So far it' s doing well against FRITZ 13, NAUM 4.2 AND CHIRON 1.1. It's still pretty weak comparing to the top engines.
Thank you for taking into consideration my suggestion regarding Zappa Mexico II 64! Also THINKER 5.4D provides exciting chess and why not - Protector is an agressive opponent.
I really hope Vitruvius 1 will prove to be an engine with THE DYNAMIC STYLE, DIFFERENT FROM IVANHOE'S, AS ITS PROGRAMMER MENTIONED!
What's your opinion about a match DEEP RYBKA 4.1- HOUDINI 2.c and STOCKFISH 2.2.2- HOUDINI 2.c on big hardware and LTC? IMHO, I still believe Rybka can manage to win the match (ofcourse, it will be very tight) and Stockfish could pull up a surprise, as it seems to be high-tuned to play against Houdini!
I noticed that private engines like CRYPTIC and THINKER are very strong; in the engine room there are games where HOUDINI has difficult problems against CRYPTIC at least! We may expect Houdini will lose the number 1 spot in favour of an already known program or it could be surpassed by an engine that hasn't been revealed by now. Time will prove if I'm right.
All the best!
Catalin

[noctiferus]

Maybe a Zappa with Dissident Aggressor settings?

[FriedmannC]

Good point!

[Robert Flesher]

Maybe a Zappa with Dissident Aggressor settings?

I created the Dissident Aggressor settings to play humans, and thus it will be crushed against any of the top tier engines. Although, it can win the occasional brilliant game vs the top engines, the aggressive insanity is just too much. It is better versus weaker engines, or playing humans.

[Uri Blass]

If we look at 4 top matches (6 CPU) according to IPON Junior should score about 21,5%. Currently it's score is 31% and I think it will be in the end about 30% quite sure. But it's difficult to say if Junior benefits from longer time control, because draw % increases with longer time control of course. In CEGT: draw ratio 40/4 31% 40/20 36% and 40/120 44%. With infinite level all games are draw

I think that we can look at results against weaker opponents

Junior has 50% against Naum after 20 games(4 wins for Junior 4 wins for Naum)

Not counting draws
Junior is losing 5-4 against Chiron after 23 games and Junior has 5-5 against Fritz13 after 23 games

total result against weak engines is 32.5 out of 66

I think that these engines score clearly better against Junior at blitz.

results against top programs are now
4.5/19 against Houdini
8/20 against Critter
7/21 against Stockfish and
6/21 against Rybka
5/20 against Komodo
total result is 30.5/101 that is slightly above 30%

[TimoK]

Draw rate rises dramatically with more time. Ok, far too few games in my test, yet.

IPON (DJ13 vs. Top 18):


DJ13-Test (DJ13 vs. Top 8):


Maybe I will setup a second test after this one where Junior has to face the rest of IPON Top 18. Then we'd have a better comparision.

Best regards
Timo

[FriedmannC]

Hi Timo, please test Deep Junior 13 X 64 against Zappa Mexico II X 64, not against Zappa Dissident Agressor. I wasn't aware that this setting of Zappa is weaker. Thank you very much!
All the best,
Catalin

[Jouni]

Junior current score is 32% against 4 top engines (90 games). I predict it will not change more than +-2%. Interestingly Houdini has already 3 losses, but Rybka is unbeaten! So is it the best engine for long analysis? Where is comment by Ansari

[TimoK]

Hi Catalin,

since I usually use default settings for my tests (as an exception Critter uses "Everywhere" in the "Tablebase Usage" option) I will test Zappa with default settings, too. I am curious if Zappa Mexico II is a match to DJ13. At least they will be playing under the conditions people say they both like most (long TCs, large Hash, multi-threaded).

Best regards
Timo

[TimoK]

Yeah, Rybka stays unbeaten until now, but also Komodo hasn't lost a single game, yet. Maybe these two engines like the long TCs most?!