carldaman wrote:PK wrote:OK, there is something that actually helps Rodent a bit. Chains of three pawns pointing towards king side, as specified by Lyudmil, get a bonus of 20 centipawns provided that the enemy king is there. Furthermore, that bonus climbs to 28 cp if there is an nemy pawn on the "congestion point" as defined by Carl. I have made another adjustement that makes the code a bit messy, namely the chains whose spearhead is on the 4th rank need two opposing pawns to obtain the second bonus.
To make this work, I needed to reduce speed overhead by hashing pawn chains eval. Rodent already had king and pawn hashtable that became a bit bloated as a result. Right now it keeps passer eval (midgame and endgame), pawn structure eval (ditto), king shield (midgame only) and separately chain eval (midgame only).
There are many things to tweak (increasing pawn storm bonus comes to mind, as in the KID positions with the chain in place Rodent sometimes attacks with the "g" pawn, sometimes not), but I'm glad that the basic formula finally brought some benefit.
Very interesting and encouraging developments, Pawel. I've always thought you'd be among the likeliest candidates to not only take on this challenge, but to also succeed.
My recommendation is to give an extra bonus for storming with the g and h pawns when the extended chain conditions are already in place.
Regards,
CL
You are right, Pawel is my hero, especially if he provides some further info.
From now on, I will be playing only with Rodent.
Do you see the Stockfish guys, Carl?
When it comes about implementing simple and mostly useless patches, they are the first to push a test, but when there is a real challenge, like the good knowledge of implementing a pointed chain eval, that would benefit positional chess analysis very much, and users would definitely like it, they all disappear into thin air.
Where are you, Ralph, Joerg, Stefan?
Rodent is already leaps and bounds in front of SF in terms of positional knowledge.
I think Rodent showed the way: it is evident now, that you can implement sound knowledge like pointed chains/triades, even when the engine search is not optimal for such eval factors, and still get some marginal value, in case you find a way to minimise the overhead, as Pawel says. In case some factors like triades are more rare to appear on the board, you might even suffer in terms of elo from such implementation, but, if you do a wise thing like hashing pawn chains, then a similar patch might also succeed as computational costs would be lower.
I really do not imagine, if SF for the time being does not even have pointed chain and longer chain eval, when they will start tackling the much more difficult concepts like closed positions (as pointed chains and closed positions are not synonymous), square control, positional compensation?
I hope Pawel says a bit more about this, but, I think no one can dispute, that if he managed to gain close to 1 elo from triades, then the pointed chain/longer chains concept promises some 49 elo more, until you get to 50, as this term is objectively worth some 50 elo points.
