burlap.behavior.stochasticgames.madynamicprogramming

Interface SGBackupOperator

All Known Implementing Classes:

CoCoQ, CorrelatedQ, MaxQ, MinMaxQ

public interface SGBackupOperator

A stochastic games backup operator to be used in multi-agent Q-learning or value function planning. This operator is meant to be applied to a next state, takes the set of Q-values for that state for all agents and returns the backup operator. The classic Bellman MDP approach would be o use a max operator, but in stochastic games, different solution concepts require different operators.

Author:

James MacGlashan

Method Summary

Modifier and Type	Method and Description
double	performBackup(State s, int forAgent, java.util.List<SGAgentType> agentDefinitions, AgentQSourceMap qSourceMap)

Method Detail

performBackup

double performBackup(State s,
                     int forAgent,
                     java.util.List<SGAgentType> agentDefinitions,
                     AgentQSourceMap qSourceMap)