ModelBasedAgent

Description:

A Model based approach where a supplied model is used to select moves. Note: because this method relies solely on the learnt Model it is unlikely to perform well on it's own compared to MCTS methods.

• Create a class which inherits from AI State.
• Train a model and construct it (see Model for more infomation).
• Construct this using one of the supplied constructors.
• When you are ready to calculate the next move pass the current state into run()
• The AI will run on a sperate thread so keep checking back to see if the done flag is true. (The AI runs on a separate thread to decouple it from your code.)
• Once the AI is done and you have retrieved the next state from next call reset() to reset the AI.

Constructors:

Name	Description
ModelBasedAgent(Model _model)	Constructs an AI using the supplied model

Methods:

Return Type	Name	Description
void	run(AIState initalState)	Starts the AI with a supplied starting state.
void	reset()	Used once the next move have been retrieved to reset the AI ready for the next iteration

Fields:

Type	Name	Description
bool	done	True if the AI has finished thinking, false otherwise
bool	started	True if this AI iteration has started
AIState	next	The next move, represented in an AI state. Null if the AI is thinking