Agents are essentially actors that takes actions based on some heuristic, either randomly in the case of RandomAgent or based on some trained information, in the case of MCTSAgent.

class agents.RandomAgent[source]

Implementation of a random agent, which simply selects a random action from the current action space each turn

class agents.HumanAgent[source]

Human agent, which waits for human input to determine what action to take. Note that they should input an integer corresponding to the index of the action they want to select

class agents.MinimaxAgent(**kwargs)[source]

Implementation of minimax which allows you to specify a cutoff horizon for the search.


Minimax – Algorithm that runs the minimax search

class agents.MCTSAgent[source]

Agent that uses Monte Carlo Tree Search (MCTS)


MTCS – The mcts search class