Zurichess Nidwalden

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

Moderators: hgm, Rebel, chrisw

brtzsnr
Posts: 433
Joined: Fri Jan 16, 2015 4:02 pm

Re: Zurichess Nidwalden

Post by brtzsnr »

I remember that I posted some different binaries here: http://talkchess.com/forum3/viewtopic.p ... ss#p740729

I was hoping to get some input from the community. It's a different search, built from ground zero using evolutionary algorithms. I only guided the search a bit (by building the atom operations), but the whole logic is learned evolutionary. I think I didn't get on the parity of manual tuned search, was about 10-30 Elo weaker.
tpoppins
Posts: 919
Joined: Tue Nov 24, 2015 9:11 pm
Location: upstate

Re: Zurichess Nidwalden

Post by tpoppins »

brtzsnr wrote: Mon Oct 22, 2018 11:32 pm That is not true. Overall the new binary searches 10% faster. Otherwise, the logic is similar.
10% faster is good. Elo-wise it might translate into a +10 Elo improvement, which would need at least 4500 games to corroborate.
Tirsa Poppins
CCRL