(mdp) One-Player Stochastic Games (MDP Games)

ASWinReach

class ggsolver.mdp.ASWinReach(graph, final=None, player=1, **kwargs)

graph()

Returns the input game graph.

is_solved()

Returns if the game is solved or not.

reset()

Resets the solver.

solution()

Returns the solved game graph. The graph contains two special properties:

• node_winner (node property): Maps each node to the id of player (1/2) who wins from that node.

• edge_winner (edge property): Maps each edge to the id of player (1/2) who wins using that edge.

solve()

Alg. 45 from Principles of Model Checking. Using the same variable names as Alg. 45.

state2node(state)

Helper function to get the node id associated with given state.

win_acts(state)

Retuns the list of winning actions from the given state.

win_region(player)

Returns the winning region for the player.

winner(state)

Returns the player who wins from the given state.

PWinReach

class ggsolver.mdp.PWinReach(graph, final=None, player=1, **kwargs)

graph()

Returns the input game graph.

is_solved()

Returns if the game is solved or not.

reset()

Resets the solver.

solution()

Returns the solved game graph. The graph contains two special properties:

• node_winner (node property): Maps each node to the id of player (1/2) who wins from that node.

• edge_winner (edge property): Maps each edge to the id of player (1/2) who wins using that edge.

solve()

Abstract method.

state2node(state)

Helper function to get the node id associated with given state.

win_acts(state)

Retuns the list of winning actions from the given state.

win_region(player)

Returns the winning region for the player.

winner(state)

Returns the player who wins from the given state.