Center Control

[Henk]

Strange Skipper computes SEE for each square in center and takes the sum ( > 0: 1 < 0: -1) * weight. But if I leave this term out it's score on the center control test positions "STS(v12.0) Center Control." is almost equal (a very bad score) and there is no overlap with another term.

Or is computing SEE too expensive.

[Sven]

Strange Skipper computes SEE for each square in center and takes the sum ( > 0: 1 < 0: -1) * weight. But if I leave this term out it's score on the center control test positions "STS(v12.0) Center Control." is almost equal (a very bad score) and there is no overlap with another term.

Or is computing SEE too expensive.

I am not sure whether SEE of central squares is a correct measurement for center control when considering positions with hanging pieces. Take a position with a white pawn on c3, a black queen on d4 and a black pawn on c5. Both sides have equal control over the central square d4 but SEE(d4, white) will return a positive score due to the hanging queen and SEE(d4, black) is either meaningless (black cannot capture own queen) or 0.

[Henk]

By the way Skipper also adds Field.Value.

Code: Select all

        public int SEE(IField field)
        {
            MoveBase move =  GetSmallestAttackingMove(field);
            if (move == null) return 0;
            var capture = move.GetCapture(this);
            return field.Value + Math.Max(0, (capture == null) ? SEENonCapture(move) : SEECapture(move, capture));                   
        }

[Sven]

By the way Skipper also adds Field.Value.

Code: Select all

        public int SEE(IField field)
        {
            MoveBase move =  GetSmallestAttackingMove(field);
            if (move == null) return 0;
            var capture = move.GetCapture(this);
            return field.Value + Math.Max(0, (capture == null) ? SEENonCapture(move) : SEECapture(move, capture));                   
        }

How does that use the color for which you want to calculate a square's SEE?

[Henk]

r7/4kp2/3Rp2b/2p1P3/bp3P1P/rB2N3/P1P5/1K1R4 w - - bm Bd5; c0 "Bd5=10, Nc4=2"; id "STS(v12.0) Center Control.100";

Even Stockfish can't find Bd5. Are these test positions ok ?

[d] r7/4kp2/3Rp2b/2p1P3/bp3P1P/rB2N3/P1P5/1K1R4 w - - bm Bd5

[Sven]

r7/4kp2/3Rp2b/2p1P3/bp3P1P/rB2N3/P1P5/1K1R4 w - - bm Bd5; c0 "Bd5=10, Nc4=2"; id "STS(v12.0) Center Control.100";

Even Stockfish can't find Bd5. Are these test positions ok ?

Can't follow, on my PC Stockfish finds 1.Bd5 very quickly.

[Henk]

On my machine Stockfish thinks Nc4 is best move. It also considers Nd5. Don't see Bd5.

[Sven]

On my machine Stockfish thinks Nc4 is best move. It also considers Nd5. Don't see Bd5.

Which SF version, what kind of machine, how long did you wait? Can you post the analysis output?

[Henk]

Stockfish6-64 and I saw depth 30