This page is meant to show an example of a relatively simple evaluation function. It is designed specifically for those who have good programming skills but want to get some ideas about chess knowledge required to attain a reasonable playing strength.

Features

Proposed evaluation function should deal with the following issues

material (together with some scaling and additional clauses)

Throughout the draft it will be assumed that the program has the following capabilities

While making and unmakingmoves, it updates the material score for each side, subdivided to pieces and pawns

It updates the value derived from the piece-square tables for all pieces except the king in the same manner

When asked, it can return the position of either king

When asked, it can return the number of white/black pawns, bishops, knights or rooks

It posesses a function int isPiece(int cl, int sq, int pc), returning 1 if a piece pc of color cl stands on the square sq and 0 otherwise.

Initialization

Distance

When the program is turned on, it must fill the tables with bonuses for distance between any two squares that will be used for evaluating king tropism. The following formula is borrowed from the masters thesis by Adam Kujawski^{[1]} and modified in such a way that all the scores become positive. It assumes that both rows and columns are numbered from 0 to 7 and that we have a function "abs" returning the value without the sign.

/* initializes the table of distances between squares */void setDist(){int i,j;for(i =0; i <64;++i){for(j =0; j <64;++j){
dist_bonus[i][j]=14-(abs( COL(i)- COL(j))+abs( ROW(i)- ROW(j)));}}}

Taking this function as a basis, tropism tables for various piece types will be derived within its body, just after determining the value of dist_bonus[i][j].

At the beginning we can assume that bk_dist is equal to rk_dist, but there is a better idea, taking into account the relevance of a diagonal. To apply it, we need two tables numbering the diagonals.

At the initialization stage, we also count the value of INITIAL_PIECE_MATERIAL. It will be needed in scaling the king safety values against the remaining enemy material. We want it to be calculated rather than to become a constant because different piece values might be used by the program.

Our simple King safety routine works as follows: evaluating each piece we update two variables: tropismToWhiteKing and tropismToBlackKing, feeding them with the values from the distance tables (see initialization for their details). Then the pawn shield value is computed (I'd like it to be 0 for a complete pawn shield, growing negative as defects accumulate). Finally, the following formula is applied:

## Table of Contents

Home * Evaluation * Evaluation function draftThis page is meant to show an example of a relatively simple evaluation function. It is designed specifically for those who have good programming skills but want to get some ideas about chess knowledge required to attain a reasonable playing strength.

## Features

Proposed evaluation function should deal with the following issues## Prequisites

Throughout the draft it will be assumed that the program has the following capabilities## Initialization

## Distance

When the program is turned on, it must fill the tables with bonuses for distance between any two squares that will be used for evaluating king tropism. The following formula is borrowed from the masters thesis by Adam Kujawski^{[1]}and modified in such a way that all the scores become positive. It assumes that both rows and columns are numbered from 0 to 7 and that we have a function "abs" returning the value without the sign.Taking this function as a basis, tropism tables for various piece types will be derived within its body, just after determining the value of dist_bonus[i][j].

At the beginning we can assume that bk_dist is equal to rk_dist, but there is a better idea, taking into account the relevance of a diagonal. To apply it, we need two tables numbering the diagonals.

Now we have to incorporate the following code

## Others

At the initialization stage, we also count the value of INITIAL_PIECE_MATERIAL. It will be needed in scaling the king safety values against the remaining enemy material. We want it to be calculated rather than to become a constant because different piece values might be used by the program.Here You can view the complete proposal of the initialization routine.

## King Safety

Our simple King safety routine works as follows: evaluating each piece we update two variables: tropismToWhiteKing and tropismToBlackKing, feeding them with the values from the distance tables (see initialization for their details). Then the pawn shield value is computed (I'd like it to be 0 for a complete pawn shield, growing negative as defects accumulate). Finally, the following formula is applied:## References

1994).Programowanie gry w szachy. Masters thesis, University of Warsaw, pdf (Polish)## What links here?

Up one Level