Index

## 1. Computer chess

• chess-playing programs essentially explore huge numbers of potential future moves by both players and apply a relatively simple evaluation function to the positions that result
• in the early days of computer chess, 1950s it was thought that by studying how a machine could play chess we would understand how the human brain works
• alas this proved to be untrue
• Claude Shannon wrote the first paper on computer chess in 1950
• in his paper he mentions two techniques which could be used to solve this problem
• type A strategy - brute force looking ahead say 6 moves and applying the minimax algorithm (AlphaBeta was not discovered until a decade later)
• type B strategy - which examines only the interesting moves at each ply
• he concluded that type B was the better technique
• time has shown the opposite!
• easier to compute all moves (type A), than write a program to work out the interesting moves - and faster..
• Shannon estimated that with approximately thirty moves possible in a typical real-life position, he expected that searching the approximately (over 700,000,000) positions involved in looking three moves ahead for both side (six plies) would take about sixteen minutes, even in the "very optimistic" case that the program evaluated a million positions every second
• it took about forty years to achieve this speed

## 2. Modern chess programs

• type B chess programs were abandonded in 1973 in favour of type A style programs
• modern chess programs evaluate a full ply width of moves up until a certain level
• then they selectively reduce the width
• try and eliminate null moves (circular board positions etc)

## 3. End game tablebases

• one of the severe weaknesses of computer chess was in the end game
• much work was done by Ken Thompson and others (1977 onwards) to explore check mate and the move before this
• and the move before this etc..
• they found some supprising facts
• technically king and rook can draw against king and queen
• before it was generally accepted that king and queen could win
• read the wikipedia entry for more details

## 4. Chess evaluation function within a game tree

• in these notes the board will be labelled a-h along the x-axis and 1-8 along the y-axis

## 5. Evaluation function

• in Othello we have seen that a simple material evaluation function delivers a reasonably strong computer opponent
• in Chess, just using material in an evaluation function requires an enormous amount of computing speed to deliver good results
• thus even the most powerful of chess computers use a more sophisticated evaluation function than just material advantage

## 6. Chess evaluation function

• calculates material

∞ king
900 Queen
500 Rook
300 Knight
300 Bishop
100 Pawn

• maybe Knight are worth more than Bishops...
• maybe a Rook is worth two Knights...
• possibly these values should change over the stages of the game

## 7. Positional evaluation

• mobility
• development and castling
• centre control
• king attack
• king safety
• pawn structure
• piece attack and piece defense

## Index

1. Computer chess
2. Modern chess programs
3. End game tablebases
4. Chess evaluation function within a game tree
5. Evaluation function
6. Chess evaluation function
7. Positional evaluation
8. Chess Pin
9. Chess Fork
10. Pawn Structure
Index

Thisdocumentwasproducedusinggroff-1.22.