New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add more advanced pawn shelter/storm code #10
Conversation
|
Well, I'm not sure what git did for the last 3 commits, I will flatten the commits again if we go with this version. I forgot to mention, but I tested against the version with the undefended rook bonus as well. |
|
Wow, this is a huge patch ! I will add to my test queue. Thanks Gary. |
|
Yes, I was trying to think of how to break it down, but all the parts sort of play together. I'm attempting to tune it now, we'll see how it goes :). |
|
Ran 16k games at 40/2 seconds, result was 4203 - 4025 - 7772 (+4), after the recent updates. It seems to be performing much worse than before, although I'm really not sure why, as there weren't any changes I saw that could affect this. This was against master with the undefended rook change in place. I'll try removing that, and re-run overnight. |
|
On Mon, Mar 26, 2012 at 01:42, Gary Linscott
If you prepare a patch rebased against current master so that I can |
|
New pull on the way. |
Idea from Ron Britvich Code reworked by Marco Costalba and Joona Kiiski Bench: 8095369 Resolves mcostalba#3 Resolves mcostalba#10
The idea is based off of the Robbolito shelter/storm code. The king is penalized for the pawns in front of him moving away, and the enemy pawns advancing against him. If the possibility of castling exists, the minimum score of the castled king safety is used.
It certainly needs some tuning, and probably re-tuning of the king safety in general as well.
However, in the current state, it scores quite well at 8k games at 40/2 seconds on my core i7, 2153 - 1952 - 3895 (+9).