Prédateur 0.1.3b2

[JuLieN]

I just released the last "fully working" version of Prédateur. As I yet stated, this is still quite old (about two months), as I've started to rewrite Pred from scratch using bitboards.

Have fun!

Download it here.

[Tony Thomas]

Thanks Julien, hopefully it doesnt have the time allocation problem. If it uses the same time algorithm, I think you could improve the strength in blitz greatly by optimizing it. I prefer fast games, (1min+1 sec).. I will let you know in a couple of days.

[JuLieN]

Hey Tony!

You're right, my time allocation alorithm is very basic. It devides the remaining time by 8, adds the time increment if there's one, and removes 200 microseconds.

I only test Pred in 3"+2s blitz cadence and yes, it often lose games it should win or draw because it lacks of time in the end. There's even a nasty conseuqence : when it is in serious lack of time, Arena complains that it sometime plays an illegal move. As these versions of Prédateur were not meant to be released, and because I've started a Pred "2" version from scratch, I've not investigate this bug anyway.

[gerold]

I just released the last "fully working" version of Prédateur. As I yet stated, this is still quite old (about two months), as I've started to rewrite Pred from scratch using bitboards.

Have fun!

Download it here.

Thanks interesting program. I like your web site. Nice.
Do you plan on making opening book for your program.

Best to you,

Gerold.

[swami]

Thanks Julien, I could add it to Division 10 tournaments edition 9 at CCB.

[JuLieN]

Thanks interesting program. I like your web site. Nice.
Do you plan on making opening book for your program.

Best to you,

Gerold.

Hey Gerold! Thx!

Honestly, I don't like opening books, as they don't show the engines' real strength. When I create engine tournaments, I always disable their opening book and make them play each others two times (each will have white once and black once). I only use an opening book when I validate a new version of my engine, to make sure that, on 200 games, it wont crash nor play illegal move (and I don't want it to play 200 times the same game. )

Anyway, it can soon be very boring for a human, if the engine always play the same openings. A minimal opening book with only 4-5 plies in depth might be ideal. I might make one one day.

Meanwhile, if you use Pred in Arena, you can use a standard opening book. Just set it to a 4-5 plies depth.

[JuLieN]

Thanks Julien, I could add it to Division 10 tournaments edition 9 at CCB.

OMG, now I'll be ashamed, hehe.

All right, stay tuned for Pred 0.2.x wich will avenge its poor 0.1.x brothers.

[hgm]

You're right, my time allocation alorithm is very basic. It devides the remaining time by 8, adds the time increment if there's one, and removes 200 microseconds.

Why by 8? You expect your engine only to survive 8 moves?

[Roman Hartmann]

Hey Tony!

You're right, my time allocation alorithm is very basic. It devides the remaining time by 8, adds the time increment if there's one, and removes 200 microseconds.

...

Hi,
many engines have a simple time allocation algorithm similar to yours. Usually the remaining time is divided by 30 or 40 if fisher time is used. So you might improve the play of your engine quite a bit with only changing the time allocating code.

best regards
Roman

[Matthias Gemuh]

You're right, my time allocation alorithm is very basic. It devides the remaining time by 8, adds the time increment if there's one, and removes 200 microseconds.

Why by 8? You expect your engine only to survive 8 moves?

by 25 would solve his timing problems quite well.


Matthias.