| Safe Haskell | None |
|---|---|
| Language | Haskell2010 |
BishBosh.Search.AlphaBeta
Description
AUTHOR- Dr. Alistair Ward
DESCRIPTION
- Performs an https://www.chessprogramming.org/Alpha-Beta search, implemented using https://www.chessprogramming.org/Negamax.
- Moves are dynamically re-ordering using the killer-heuristic.
- https://www.chessprogramming.org/Repetitions & https://www.chessprogramming.org/Transpositions are detected.
Synopsis
- extractSelectedTurns :: NPlies -> Result x y positionHash criterionValue weightedMean -> (DynamicMoveData x y positionHash, [Turn x y], NPlies)
- negaMax :: (Enum x, Enum y, Eq criterionValue, Ord positionHash, Ord x, Ord y, Real weightedMean, Show x, Show y) => SearchDepth -> SearchState x y positionHash criterionValue weightedMean -> Reader (Result x y positionHash criterionValue weightedMean)
Types
Type-synonyms
Data-types
Functions
extractSelectedTurns :: NPlies -> Result x y positionHash criterionValue weightedMean -> (DynamicMoveData x y positionHash, [Turn x y], NPlies) Source #
- Drop the specified number of turns; typically those made before starting the search.
- CAVEAT: abandons the fitness component of the quantified game.
Arguments
| :: (Enum x, Enum y, Eq criterionValue, Ord positionHash, Ord x, Ord y, Real weightedMean, Show x, Show y) | |
| => SearchDepth | The depth to which the tree should be searched; i.e. the number of plies to look-ahead. |
| -> SearchState x y positionHash criterionValue weightedMean | |
| -> Reader (Result x y positionHash criterionValue weightedMean) |
- Implements a depth-first search (implemented as nega-max), with alpha-beta pruning.
- alpha is the minimum fitness of which the maximising player is assured.
- beta is the maximum fitness of which the minimising player is assured.