TalkChess.com

I have read that some computer chess engines will use the distance to king as a variable in the Evaluation Function.

Does anyone know an efficient way of calculating this distance?

Look it up in a table indexed by (square - kingSquare).

I am not sure that works for me. I have the board indexed like so:

0 1 2 3 4 5 6 7
8 9 10 11 12 13 14 15
16 17 18 19 20 21 22 23
24 25 26 27 28 29 30 31
32 33 34 35 36 37 38 39
40 41 42 43 44 45 46 47
48 49 50 51 52 53 54 55
56 57 58 59 60 61 62 63

So by simply subtracting the values 24 woud be right next to 23 which is not true.

For whatever reason, HG was assuming a 0x88 representation where the square difference index works. For more information, see

http://chessprogramming.wikispaces.com/0x88

This is why it is better to use 0x88 numbering:
You can convert one into the other by nr0x88 = nrPlain + (nrPlain & 070); or of course use another table lookup for that, nr0x88 = table[nrPlain].

Well, it depends on what distance you want as well . The dumb method is still pretty fast for king walk distance:

distance = abs(file(king) - file(piece)) + abs(rank(king) - rank(piece));

Calculating file/rank is trivial:

file = index & 7;
rank = index >> 3;

Your King cannot walk diagonally?

True, but you can precalculate the table at engine startup using the long way round (shown below) and reference the table when eval is called.

Richard

hgm wrote:Your King cannot walk diagonally?

LOL !
I'd say : Max(Abs(FileKingX-FileTargetX) , Abs(FileKingY-FileTargetY)) .

So it's the maximum distance in vertical/horizontal

hgm wrote:Look it up in a table indexed by (square - kingSquare).

or

distance = Max( abs(Rank(king) - rank(square)), abs(file(king) - file(square)));

That computes the so-called "taxicab distance" You take the largest of the vertical and horizontal distances.