double BayesianGameForDecPOMDPStage::
ComputeImmediateReward(Index jtI, Index jaI) const
{
//Index jaohI = _m_jaohReps.at(jtI);
JointBeliefInterface* jb = _m_JBs.at(jtI);
BeliefIteratorGeneric bit = jb->GetIterator();
double r = 0.0;
do{
Index s = bit.GetStateIndex();
double p = bit.GetProbability();
r += p * _m_pu->GetReward(s, jaI);
}while (bit.Next() );
return r;
}
/**Calculates the jaohI corresponding to jaI_arr and joI_arr and also
* returnes the P(jaohI) and the expected obtained reward for previous
* time steps GIVEN this joint action history.
*
* \todo TODO add following to .h
*
* basically this function is a form of
* PlanningUnitMADPDiscrete::GetJAOHProbs(Recursively)
* that also computes the reward.
*
* */
void BayesianGameForDecPOMDPStage::ProbRewardForjoahI(
//input args
Index ts, Index jtI, Index* jaI_arr,Index* joI_arr,
//output args
Index& jaohI, double& PjaohI, double& ExpR_0_prevTS_thisJAOH )
{
//first we get the initial jaoh
JointActionObservationHistoryTree * jaoht =
_m_pu->GetJointActionObservationHistoryTree(0);
double CPjaohI = 1.0;
PjaohI = CPjaohI; // == 1.0
// get the initaal belief
JointBeliefInterface* jb = _m_pu->GetNewJointBeliefFromISD();
Index tI = 0;
while(tI < ts)
{
//calculate the R for tI
double ExpR_0_prevTS_thisJAOH_thisT = 0.0;
#if USE_BeliefIteratorGeneric
BeliefIteratorGeneric it=jb->GetIterator();
do
{
double R_si_ja = _m_pu->GetReward(it.GetStateIndex(), jaI_arr[tI]);
if(DEBUG_BG4DECPOMDP4)
cout << "R(s="<<it.GetStateIndex()<<",ja="<<jaI_arr[tI]<<")="<< R_si_ja << "\n";
ExpR_0_prevTS_thisJAOH_thisT += it.GetProbability() * R_si_ja;
} while(it.Next());
#else
for(Index sI=0; sI < _m_pu->GetNrStates(); sI++)
{
double R_si_ja = _m_pu->GetReward(sI, jaI_arr[tI]);
if(DEBUG_BG4DECPOMDP4)
cout << "R(s="<<sI<<",ja="<<jaI_arr[tI]<<")="<< R_si_ja << "\n";
ExpR_0_prevTS_thisJAOH_thisT += jb->Get(sI) * R_si_ja;
}
#endif
ExpR_0_prevTS_thisJAOH += ExpR_0_prevTS_thisJAOH_thisT;
if(DEBUG_BG4DECPOMDP4)
{
cout << "calculating expected reward R(oaHist,a) for tI="<<tI
<<"oaHist:"; jaoht->GetJointActionObservationHistory()->Print();
cout << endl; cout << "R(b,a) (exp reward for jtI=" << jtI <<
", tI="<<tI<<") is "<< ExpR_0_prevTS_thisJAOH_thisT <<endl;
}
jaoht = jaoht->GetSuccessor( jaI_arr[tI], joI_arr[tI] );
jaohI = jaoht->GetIndex();
CPjaohI = jb->Update( *_m_pu->GetReferred(), jaI_arr[tI], joI_arr[tI] );
PjaohI = PjaohI * CPjaohI;
tI++;
}
delete jb; //free the belief allocated with 'new'
if(DEBUG_BG4DECPOMDP4)
{
cout << "expected previous reward (up to ts-1) for (jtI=" << jtI <<
") ";
jaoht->GetJointActionObservationHistory()->Print();
cout << " is "<< ExpR_0_prevTS_thisJAOH <<endl << endl;
}
}
