| Zobrist Hashing |
Website Links For Hashing |
Information AboutZobrist Hashing |
| CATEGORIES ABOUT ZOBRIST HASHING | |
| computer chess | |
| computer go | |
| game artificial intelligence | |
| hashing | |
|
This method has several advantages: it is reasonably collision resistant, the code is easy to write and the values are easy to compute, and where necessary it can be updated incrementally for speed -- if only one point changes, it is just necessary to do two XOR operations: original hash XOR old point value XOR new point value = new hash. If one of the states is the empty state and the corresponding value is 0, then only one operation is needed. In Computer Go , this technique is used with Transposition Table s and Superko detection. Normally the values used range in size from 32 bits to 96 bits (longer hashes provide more collision resistance). Frequently an exact hash collision is taken to mean the positions are the same, without checking the actual positions. Often the original positions are not even stored in the transposition table in order to save time and memory. SEE ALSO |
|
|