Package teamwork :: Package policy :: Module JIPolicy
[hide private]
[frames] | no frames]

Module JIPolicy

source code

Classes [hide private]
  JIPolicy
generic class for a communication policy, where there is a single possible message and it refers to the achievement of some joint goal
  STEAMPolicy
subclass of joint intentions policies, but following the STEAM algorithm for decision-theoretic selectivity
  SilentPolicy
generic policy class that never specifies communication
  BestAvailablePolicy
a generic communication policy class that searches a space of candidate policies and finds the best one for a given COM-MTDP.
  GloballyOptimalJIPolicy
subclass of joint intentions policies that performs a brute force search to determine the best policy of communication, in reference to a single JPG
  SingleMsgPolicy
subclass of joint intentions policies that generates all possible communication behaviors for an agent with a single joint commitment (this class is no longer very useful)
  LocallyOptimalJIPolicy
subclass of joint intentions policies that generates locally optimal decisions with respect to a single JPG
  RepeatMsgs
policy subclass, useful as a helper to the LocallyOptimalJIPolicy class
Functions [hide private]
 
generatePolicies(policySpace, actions, observations, horizon, debug=None) source code
 
generateActionCombos(table, actions, result) source code
Variables [hide private]
  StringTypes = (<type 'str'>, <type 'unicode'>)