tictactoe_policy package

Submodules

tictactoe_policy.cli module

tictactoe_policy.dataset module

tictactoe_policy.evaluation module

tictactoe_policy.game module

Board configuration and rules for configurable Tic-Tac-Toe.

class tictactoe_policy.game.GameConfig(board_size: int = 3, win_length: int | None = None)

Bases: object

Immutable game dimensions.

Parameters:

• board_size – Board width and height, from 3 through 8.

• win_length – Consecutive marks required to win. Defaults to the board size when omitted.

Raises:

ValueError – If either dimension is outside its valid range.

Variables:

• board_size – Configured board width and height.

• win_length – Configured number of consecutive marks needed to win.

board_size: int = 3

Board width and height.

property cell_count: int

Return the total number of board cells.

Returns:

board_size squared.

win_length: int | None = None

Consecutive marks required to win.

tictactoe_policy.game.MAX_BOARD_SIZE = 8

Largest supported board width and height.

tictactoe_policy.game.MIN_BOARD_SIZE = 3

Smallest supported board width and height.

class tictactoe_policy.game.TicTacToeGame(config: GameConfig | None = None)

Bases: object

Validate boards and implement configurable Tic-Tac-Toe rules.

Parameters:

config – Game dimensions. Standard 3x3 rules are used when omitted.

Variables:

• config – Dimensions used by this game.

• lines – Every winning line expressed as row-major cell indices.

static apply_move(board: Sequence[int], move: int, player: int) → tuple[int, ...]

Apply one legal move without modifying the input board.

Parameters:

• board – Board cells in row-major order.

• move – Index of the empty cell to occupy.

• player – Player mark, either 1 or -1.

Returns:

A new immutable board containing the move.

Raises:

ValueError – If the player or move is invalid.

infer_player(board: Sequence[int]) → int

Infer whose turn follows an X-first reachable board.

Parameters:

board – Non-terminal board cells in row-major order.

Returns:

1 when X moves next or -1 when O moves next.

Raises:

ValueError – If the board is invalid or terminal.

is_terminal(board: Sequence[int]) → bool

Determine whether a board is won or full.

Parameters:

board – Board cells in row-major order.

Returns:

True when play cannot continue.

legal_moves(board: Sequence[int]) → list[int]

List the unoccupied cells on a board.

Parameters:

board – Board cells in row-major order.

Returns:

Legal move indices in ascending order.

validate_board(board: Sequence[int], check_reachable: bool = True) → tuple[int, ...]

Validate a board and return an immutable representation.

Parameters:

• board – Board cells in row-major order, using 1 for X, -1 for O, and 0 for an empty cell.

• check_reachable – Validate turn counts and winner consistency.

Returns:

The validated board as a tuple.

Raises:

ValueError – If the dimensions, cells, turn counts, or winners are invalid.

winner(board: Sequence[int]) → int

Return the sole winner of a board.

Parameters:

board – Board cells in row-major order.

Returns:

1 for X, -1 for O, or 0 when there is no sole winner.

winners(board: Sequence[int]) → set[int]

Find every player with a completed winning line.

Parameters:

board – Board cells in row-major order.

Returns:

Set containing 1 and/or -1 for winning players.

tictactoe_policy.logger module

tictactoe_policy.minimax module

tictactoe_policy.model module

tictactoe_policy.policy module

tictactoe_policy.search module

tictactoe_policy.symmetry module

tictactoe_policy.training module

Module contents