You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 

442 lines
35 KiB

#include "storm/modelchecker/multiobjective/pcaa/SparseMaPcaaWeightVectorChecker.h"
#include <cmath>
#include "storm/adapters/CarlAdapter.h"
#include "storm/models/sparse/MarkovAutomaton.h"
#include "storm/models/sparse/StandardRewardModel.h"
#include "storm/utility/macros.h"
#include "storm/utility/vector.h"
#include "storm/exceptions/InvalidOperationException.h"
namespace storm {
namespace modelchecker {
namespace multiobjective {
template <class SparseMaModelType>
SparseMaPcaaWeightVectorChecker<SparseMaModelType>::SparseMaPcaaWeightVectorChecker(SparseMaModelType const& model,
std::vector<PcaaObjective<ValueType>> const& objectives,
storm::storage::BitVector const& actionsWithNegativeReward,
storm::storage::BitVector const& ecActions,
storm::storage::BitVector const& possiblyRecurrentStates) :
SparsePcaaWeightVectorChecker<SparseMaModelType>(model, objectives, actionsWithNegativeReward, ecActions, possiblyRecurrentStates) {
// Set the (discretized) state action rewards.
this->discreteActionRewards.resize(objectives.size());
for(auto objIndex : this->objectivesWithNoUpperTimeBound) {
typename SparseMaModelType::RewardModelType const& rewModel = this->model.getRewardModel(this->objectives[objIndex].rewardModelName);
STORM_LOG_ASSERT(!rewModel.hasTransitionRewards(), "Preprocessed Reward model has transition rewards which is not expected.");
this->discreteActionRewards[objIndex] = rewModel.hasStateActionRewards() ? rewModel.getStateActionRewardVector() : std::vector<ValueType>(this->model.getTransitionMatrix().getRowCount(), storm::utility::zero<ValueType>());
if(rewModel.hasStateRewards()) {
// Note that state rewards are earned over time and thus play no role for probabilistic states
for(auto markovianState : this->model.getMarkovianStates()) {
this->discreteActionRewards[objIndex][this->model.getTransitionMatrix().getRowGroupIndices()[markovianState]] += rewModel.getStateReward(markovianState) / this->model.getExitRate(markovianState);
}
}
}
}
template <class SparseMaModelType>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::boundedPhase(std::vector<ValueType> const& weightVector, std::vector<ValueType>& weightedRewardVector) {
// Split the preprocessed model into transitions from/to probabilistic/Markovian states.
SubModel MS = createSubModel(true, weightedRewardVector);
SubModel PS = createSubModel(false, weightedRewardVector);
// Apply digitization to Markovian transitions
ValueType digitizationConstant = getDigitizationConstant(weightVector);
digitize(MS, digitizationConstant);
// Get for each occurring (digitized) timeBound the indices of the objectives with that bound.
TimeBoundMap lowerTimeBounds;
TimeBoundMap upperTimeBounds;
digitizeTimeBounds(lowerTimeBounds, upperTimeBounds, digitizationConstant);
// Initialize a minMaxSolver to compute an optimal scheduler (w.r.t. PS) for each epoch
// No EC elimination is necessary as we assume non-zenoness
std::unique_ptr<MinMaxSolverData> minMax = initMinMaxSolver(PS);
// create a linear equation solver for the model induced by the optimal choice vector.
// the solver will be updated whenever the optimal choice vector has changed.
std::unique_ptr<LinEqSolverData> linEq = initLinEqSolver(PS);
// Store the optimal choices of PS as computed by the minMax solver.
std::vector<uint_fast64_t> optimalChoicesAtCurrentEpoch(PS.getNumberOfStates(), std::numeric_limits<uint_fast64_t>::max());
// Stores the objectives for which we need to compute values in the current time epoch.
storm::storage::BitVector consideredObjectives = this->objectivesWithNoUpperTimeBound;
auto lowerTimeBoundIt = lowerTimeBounds.begin();
auto upperTimeBoundIt = upperTimeBounds.begin();
uint_fast64_t currentEpoch = std::max(lowerTimeBounds.empty() ? 0 : lowerTimeBoundIt->first, upperTimeBounds.empty() ? 0 : upperTimeBoundIt->first);
this->maxStepBound = std::max(this->maxStepBound, currentEpoch);
while(true) {
// Update the objectives that are considered at the current time epoch as well as the (weighted) reward vectors.
updateDataToCurrentEpoch(MS, PS, *minMax, consideredObjectives, currentEpoch, weightVector, lowerTimeBoundIt, lowerTimeBounds, upperTimeBoundIt, upperTimeBounds);
// Compute the values that can be obtained at probabilistic states in the current time epoch
performPSStep(PS, MS, *minMax, *linEq, optimalChoicesAtCurrentEpoch, consideredObjectives, weightVector);
// Compute values that can be obtained at Markovian states after letting one (digitized) time unit pass.
// Only perform such a step if there is time left.
if(currentEpoch>0) {
performMSStep(MS, PS, consideredObjectives, weightVector);
--currentEpoch;
} else {
break;
}
}
// compose the results from MS and PS
storm::utility::vector::setVectorValues(this->weightedResult, MS.states, MS.weightedSolutionVector);
storm::utility::vector::setVectorValues(this->weightedResult, PS.states, PS.weightedSolutionVector);
for(uint_fast64_t objIndex = 0; objIndex < this->objectives.size(); ++objIndex) {
storm::utility::vector::setVectorValues(this->objectiveResults[objIndex], MS.states, MS.objectiveSolutionVectors[objIndex]);
storm::utility::vector::setVectorValues(this->objectiveResults[objIndex], PS.states, PS.objectiveSolutionVectors[objIndex]);
}
}
template <class SparseMaModelType>
typename SparseMaPcaaWeightVectorChecker<SparseMaModelType>::SubModel SparseMaPcaaWeightVectorChecker<SparseMaModelType>::createSubModel(bool createMS, std::vector<ValueType> const& weightedRewardVector) const {
SubModel result;
storm::storage::BitVector probabilisticStates = ~this->model.getMarkovianStates();
result.states = createMS ? this->model.getMarkovianStates() : probabilisticStates;
result.choices = this->model.getTransitionMatrix().getRowIndicesOfRowGroups(result.states);
STORM_LOG_ASSERT(!createMS || result.states.getNumberOfSetBits() == result.choices.getNumberOfSetBits(), "row groups for Markovian states should consist of exactly one row");
//We need to add diagonal entries for selfloops on Markovian states.
result.toMS = this->model.getTransitionMatrix().getSubmatrix(true, result.states, this->model.getMarkovianStates(), createMS);
result.toPS = this->model.getTransitionMatrix().getSubmatrix(true, result.states, probabilisticStates, false);
STORM_LOG_ASSERT(result.getNumberOfStates() == result.states.getNumberOfSetBits() && result.getNumberOfStates() == result.toMS.getRowGroupCount() && result.getNumberOfStates() == result.toPS.getRowGroupCount(), "Invalid state count for subsystem");
STORM_LOG_ASSERT(result.getNumberOfChoices() == result.choices.getNumberOfSetBits() && result.getNumberOfChoices() == result.toMS.getRowCount() && result.getNumberOfChoices() == result.toPS.getRowCount(), "Invalid state count for subsystem");
result.weightedRewardVector.resize(result.getNumberOfChoices());
storm::utility::vector::selectVectorValues(result.weightedRewardVector, result.choices, weightedRewardVector);
result.objectiveRewardVectors.resize(this->objectives.size());
for(uint_fast64_t objIndex = 0; objIndex < this->objectives.size(); ++objIndex) {
std::vector<ValueType>& objVector = result.objectiveRewardVectors[objIndex];
objVector = std::vector<ValueType>(result.weightedRewardVector.size(), storm::utility::zero<ValueType>());
if(this->objectivesWithNoUpperTimeBound.get(objIndex)) {
storm::utility::vector::selectVectorValues(objVector, result.choices, this->discreteActionRewards[objIndex]);
} else {
typename SparseMaModelType::RewardModelType const& rewModel = this->model.getRewardModel(this->objectives[objIndex].rewardModelName);
STORM_LOG_ASSERT(!rewModel.hasTransitionRewards(), "Preprocessed Reward model has transition rewards which is not expected.");
STORM_LOG_ASSERT(!rewModel.hasStateRewards(), "State rewards for bounded objectives for MAs are not expected (bounded rewards are not supported).");
if(rewModel.hasStateActionRewards()) {
storm::utility::vector::selectVectorValues(objVector, result.choices, rewModel.getStateActionRewardVector());
}
}
}
result.weightedSolutionVector.resize(result.getNumberOfStates());
storm::utility::vector::selectVectorValues(result.weightedSolutionVector, result.states, this->weightedResult);
result.objectiveSolutionVectors.resize(this->objectives.size());
for(uint_fast64_t objIndex = 0; objIndex < this->objectives.size(); ++objIndex) {
result.objectiveSolutionVectors[objIndex].resize(result.weightedSolutionVector.size());
storm::utility::vector::selectVectorValues(result.objectiveSolutionVectors[objIndex], result.states, this->objectiveResults[objIndex]);
}
result.auxChoiceValues.resize(result.getNumberOfChoices());
return result;
}
template <class SparseMaModelType>
template <typename VT, typename std::enable_if<storm::NumberTraits<VT>::SupportsExponential, int>::type>
VT SparseMaPcaaWeightVectorChecker<SparseMaModelType>::getDigitizationConstant(std::vector<ValueType> const& weightVector) const {
STORM_LOG_DEBUG("Retrieving digitization constant");
// We need to find a delta such that for each objective it holds that lowerbound/delta , upperbound/delta are natural numbers and
// sum_{obj_i} (
// If obj_i has a lower and an upper bound:
// weightVector_i * (1 - e^(-maxRate lowerbound) * (1 + maxRate delta) ^ (lowerbound / delta) + 1-e^(-maxRate upperbound) * (1 + maxRate delta) ^ (upperbound / delta) + (1-e^(-maxRate delta)))
// If there is only an upper bound:
// weightVector_i * ( 1-e^(-maxRate upperbound) * (1 + maxRate delta) ^ (upperbound / delta))
// ) <= this->maximumLowerUpperDistance
// Initialize some data for fast and easy access
VT const maxRate = this->model.getMaximalExitRate();
std::vector<std::pair<VT, VT>> eToPowerOfMinusMaxRateTimesBound;
VT smallestNonZeroBound = storm::utility::zero<VT>();
for(auto const& obj : this->objectives) {
eToPowerOfMinusMaxRateTimesBound.emplace_back();
if(obj.lowerTimeBound){
STORM_LOG_ASSERT(!storm::utility::isZero(*obj.lowerTimeBound), "Got zero-valued lower bound."); // should have been handled in preprocessing
STORM_LOG_ASSERT(!obj.upperTimeBound || *obj.lowerTimeBound < *obj.upperTimeBound, "Got point intervall or empty intervall on time bounded property which is not supported"); // should also have been handled in preprocessing
eToPowerOfMinusMaxRateTimesBound.back().first = std::exp(-maxRate * (*obj.lowerTimeBound));
smallestNonZeroBound = storm::utility::isZero(smallestNonZeroBound) ? *obj.lowerTimeBound : std::min(smallestNonZeroBound, *obj.lowerTimeBound);
}
if(obj.upperTimeBound){
STORM_LOG_ASSERT(!storm::utility::isZero(*obj.upperTimeBound), "Got zero-valued upper bound."); // should have been handled in preprocessing
eToPowerOfMinusMaxRateTimesBound.back().second = std::exp(-maxRate * (*obj.upperTimeBound));
smallestNonZeroBound = storm::utility::isZero(smallestNonZeroBound) ? *obj.upperTimeBound : std::min(smallestNonZeroBound, *obj.upperTimeBound);
}
}
if(storm::utility::isZero(smallestNonZeroBound)) {
// There are no time bounds. In this case, one is a valid digitization constant.
return storm::utility::one<VT>();
}
VT goalPrecisionTimesNorm = this->weightedPrecision * storm::utility::sqrt(storm::utility::vector::dotProduct(weightVector, weightVector));
// We brute-force a delta, since a direct computation is apparently not easy.
// Also note that the number of times this loop runs is a lower bound for the number of minMaxSolver invocations.
// Hence, this brute-force approach will most likely not be a bottleneck.
uint_fast64_t smallestStepBound = 1;
VT delta = smallestNonZeroBound / smallestStepBound;
while(true) {
bool deltaValid = true;
for(auto const& obj : this->objectives) {
if((obj.lowerTimeBound && *obj.lowerTimeBound/delta != std::floor(*obj.lowerTimeBound/delta)) ||
(obj.upperTimeBound && *obj.upperTimeBound/delta != std::floor(*obj.upperTimeBound/delta))) {
deltaValid = false;
break;
}
}
if(deltaValid) {
VT weightedPrecisionForCurrentDelta = storm::utility::zero<VT>();
for(uint_fast64_t objIndex = 0; objIndex < this->objectives.size(); ++objIndex) {
auto const& obj = this->objectives[objIndex];
VT precisionOfObj = storm::utility::zero<VT>();
if(obj.lowerTimeBound) {
precisionOfObj += storm::utility::one<VT>() - (eToPowerOfMinusMaxRateTimesBound[objIndex].first * storm::utility::pow(storm::utility::one<VT>() + maxRate * delta, *obj.lowerTimeBound / delta) )
+ storm::utility::one<VT>() - std::exp(-maxRate * delta);
}
if(obj.upperTimeBound) {
precisionOfObj += storm::utility::one<VT>() - (eToPowerOfMinusMaxRateTimesBound[objIndex].second * storm::utility::pow(storm::utility::one<VT>() + maxRate * delta, *obj.upperTimeBound / delta) );
}
weightedPrecisionForCurrentDelta += weightVector[objIndex] * precisionOfObj;
}
deltaValid &= weightedPrecisionForCurrentDelta <= goalPrecisionTimesNorm;
}
if(deltaValid) {
break;
}
++smallestStepBound;
STORM_LOG_ASSERT(delta>smallestNonZeroBound / smallestStepBound, "Digitization constant is expected to become smaller in every iteration");
delta = smallestNonZeroBound / smallestStepBound;
}
STORM_LOG_DEBUG("Found digitization constant: " << delta << ". At least " << smallestStepBound << " digitization steps will be necessarry");
return delta;
}
template <class SparseMaModelType>
template <typename VT, typename std::enable_if<!storm::NumberTraits<VT>::SupportsExponential, int>::type>
VT SparseMaPcaaWeightVectorChecker<SparseMaModelType>::getDigitizationConstant(std::vector<ValueType> const& weightVector) const {
STORM_LOG_THROW(false, storm::exceptions::InvalidOperationException, "Computing bounded probabilities of MAs is unsupported for this value type.");
}
template <class SparseMaModelType>
template <typename VT, typename std::enable_if<storm::NumberTraits<VT>::SupportsExponential, int>::type>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::digitize(SubModel& MS, VT const& digitizationConstant) const {
std::vector<VT> rateVector(MS.getNumberOfChoices());
storm::utility::vector::selectVectorValues(rateVector, MS.states, this->model.getExitRates());
for(uint_fast64_t row = 0; row < rateVector.size(); ++row) {
VT const eToMinusRateTimesDelta = std::exp(-rateVector[row] * digitizationConstant);
for(auto& entry : MS.toMS.getRow(row)) {
entry.setValue((storm::utility::one<VT>() - eToMinusRateTimesDelta) * entry.getValue());
if(entry.getColumn() == row) {
entry.setValue(entry.getValue() + eToMinusRateTimesDelta);
}
}
for(auto& entry : MS.toPS.getRow(row)) {
entry.setValue((storm::utility::one<VT>() - eToMinusRateTimesDelta) * entry.getValue());
}
MS.weightedRewardVector[row] *= storm::utility::one<VT>() - eToMinusRateTimesDelta;
for(auto& objVector : MS.objectiveRewardVectors) {
objVector[row] *= storm::utility::one<VT>() - eToMinusRateTimesDelta;
}
}
}
template <class SparseMaModelType>
template <typename VT, typename std::enable_if<!storm::NumberTraits<VT>::SupportsExponential, int>::type>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::digitize(SubModel& subModel, VT const& digitizationConstant) const {
STORM_LOG_THROW(false, storm::exceptions::InvalidOperationException, "Computing bounded probabilities of MAs is unsupported for this value type.");
}
template <class SparseMaModelType>
template <typename VT, typename std::enable_if<storm::NumberTraits<VT>::SupportsExponential, int>::type>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::digitizeTimeBounds(TimeBoundMap& lowerTimeBounds, TimeBoundMap& upperTimeBounds, VT const& digitizationConstant) {
VT const maxRate = this->model.getMaximalExitRate();
for(uint_fast64_t objIndex = 0; objIndex < this->objectives.size(); ++objIndex) {
auto const& obj = this->objectives[objIndex];
VT errorTowardsZero;
VT errorAwayFromZero;
if(obj.lowerTimeBound) {
uint_fast64_t digitizedBound = storm::utility::convertNumber<uint_fast64_t>((*obj.lowerTimeBound)/digitizationConstant);
auto timeBoundIt = lowerTimeBounds.insert(std::make_pair(digitizedBound, storm::storage::BitVector(this->objectives.size(), false))).first;
timeBoundIt->second.set(objIndex);
VT digitizationError = storm::utility::one<VT>();
digitizationError -= std::exp(-maxRate * (*obj.lowerTimeBound)) * storm::utility::pow(storm::utility::one<VT>() + maxRate * digitizationConstant, digitizedBound);
errorTowardsZero = -digitizationError;
errorAwayFromZero = storm::utility::one<VT>() - std::exp(-maxRate * digitizationConstant);;
} else {
errorTowardsZero = storm::utility::zero<VT>();
errorAwayFromZero = storm::utility::zero<VT>();
}
if(obj.upperTimeBound) {
uint_fast64_t digitizedBound = storm::utility::convertNumber<uint_fast64_t>((*obj.upperTimeBound)/digitizationConstant);
auto timeBoundIt = upperTimeBounds.insert(std::make_pair(digitizedBound, storm::storage::BitVector(this->objectives.size(), false))).first;
timeBoundIt->second.set(objIndex);
VT digitizationError = storm::utility::one<VT>();
digitizationError -= std::exp(-maxRate * (*obj.upperTimeBound)) * storm::utility::pow(storm::utility::one<VT>() + maxRate * digitizationConstant, digitizedBound);
errorAwayFromZero += digitizationError;
}
if (obj.rewardsArePositive) {
this->offsetsToLowerBound[objIndex] = -errorTowardsZero;
this->offsetsToUpperBound[objIndex] = errorAwayFromZero;
} else {
this->offsetsToLowerBound[objIndex] = -errorAwayFromZero;
this->offsetsToUpperBound[objIndex] = errorTowardsZero;
}
}
}
template <class SparseMaModelType>
template <typename VT, typename std::enable_if<!storm::NumberTraits<VT>::SupportsExponential, int>::type>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::digitizeTimeBounds(TimeBoundMap& lowerTimeBounds, TimeBoundMap& upperTimeBounds, VT const& digitizationConstant) {
STORM_LOG_THROW(false, storm::exceptions::InvalidOperationException, "Computing bounded probabilities of MAs is unsupported for this value type.");
}
template <class SparseMaModelType>
std::unique_ptr<typename SparseMaPcaaWeightVectorChecker<SparseMaModelType>::MinMaxSolverData> SparseMaPcaaWeightVectorChecker<SparseMaModelType>::initMinMaxSolver(SubModel const& PS) const {
std::unique_ptr<MinMaxSolverData> result(new MinMaxSolverData());
storm::solver::GeneralMinMaxLinearEquationSolverFactory<ValueType> minMaxSolverFactory;
result->solver = minMaxSolverFactory.create(PS.toPS);
result->solver->setOptimizationDirection(storm::solver::OptimizationDirection::Maximize);
result->solver->setTrackScheduler(true);
result->solver->setCachingEnabled(true);
result->b.resize(PS.getNumberOfChoices());
return result;
}
template <class SparseMaModelType>
std::unique_ptr<typename SparseMaPcaaWeightVectorChecker<SparseMaModelType>::LinEqSolverData> SparseMaPcaaWeightVectorChecker<SparseMaModelType>::initLinEqSolver(SubModel const& PS) const {
std::unique_ptr<LinEqSolverData> result(new LinEqSolverData());
// We choose Jacobi since we call the solver very frequently on 'easy' inputs (note that jacobi without preconditioning has very little overhead).
result->factory.getSettings().setSolutionMethod(storm::solver::GmmxxLinearEquationSolverSettings<ValueType>::SolutionMethod::Jacobi);
result->factory.getSettings().setPreconditioner(storm::solver::GmmxxLinearEquationSolverSettings<ValueType>::Preconditioner::None);
result->b.resize(PS.getNumberOfStates());
return result;
}
template <class SparseMaModelType>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::updateDataToCurrentEpoch(SubModel& MS, SubModel& PS, MinMaxSolverData& minMax, storm::storage::BitVector& consideredObjectives, uint_fast64_t const& currentEpoch, std::vector<ValueType> const& weightVector, TimeBoundMap::iterator& lowerTimeBoundIt, TimeBoundMap const& lowerTimeBounds, TimeBoundMap::iterator& upperTimeBoundIt, TimeBoundMap const& upperTimeBounds) {
//Note that lower time bounds are always strict. Hence, we need to react when the current epoch equals the stored bound.
if(lowerTimeBoundIt != lowerTimeBounds.end() && currentEpoch == lowerTimeBoundIt->first) {
for(auto objIndex : lowerTimeBoundIt->second) {
// No more reward is earned for this objective.
storm::utility::vector::addScaledVector(MS.weightedRewardVector, MS.objectiveRewardVectors[objIndex], -weightVector[objIndex]);
storm::utility::vector::addScaledVector(PS.weightedRewardVector, PS.objectiveRewardVectors[objIndex], -weightVector[objIndex]);
MS.objectiveRewardVectors[objIndex] = std::vector<ValueType>(MS.objectiveRewardVectors[objIndex].size(), storm::utility::zero<ValueType>());
PS.objectiveRewardVectors[objIndex] = std::vector<ValueType>(PS.objectiveRewardVectors[objIndex].size(), storm::utility::zero<ValueType>());
}
++lowerTimeBoundIt;
}
if(upperTimeBoundIt != upperTimeBounds.end() && currentEpoch == upperTimeBoundIt->first) {
consideredObjectives |= upperTimeBoundIt->second;
for(auto objIndex : upperTimeBoundIt->second) {
// This objective now plays a role in the weighted sum
storm::utility::vector::addScaledVector(MS.weightedRewardVector, MS.objectiveRewardVectors[objIndex], weightVector[objIndex]);
storm::utility::vector::addScaledVector(PS.weightedRewardVector, PS.objectiveRewardVectors[objIndex], weightVector[objIndex]);
}
++upperTimeBoundIt;
}
// Update the solver data
PS.toMS.multiplyWithVector(MS.weightedSolutionVector, minMax.b);
storm::utility::vector::addVectors(minMax.b, PS.weightedRewardVector, minMax.b);
}
template <class SparseMaModelType>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::performPSStep(SubModel& PS, SubModel const& MS, MinMaxSolverData& minMax, LinEqSolverData& linEq, std::vector<uint_fast64_t>& optimalChoicesAtCurrentEpoch, storm::storage::BitVector const& consideredObjectives, std::vector<ValueType> const& weightVector) const {
// compute a choice vector for the probabilistic states that is optimal w.r.t. the weighted reward vector
minMax.solver->solveEquations(PS.weightedSolutionVector, minMax.b);
auto newScheduler = minMax.solver->getScheduler();
if(consideredObjectives.getNumberOfSetBits() == 1 && !storm::utility::isZero(weightVector[*consideredObjectives.begin()])) {
// In this case there is no need to perform the computation on the individual objectives
optimalChoicesAtCurrentEpoch = newScheduler->getChoices();
auto objIndex = *consideredObjectives.begin();
PS.objectiveSolutionVectors[objIndex] = PS.weightedSolutionVector;
if(!storm::utility::isOne(weightVector[objIndex])) {
storm::utility::vector::scaleVectorInPlace(PS.objectiveSolutionVectors[objIndex], storm::utility::one<ValueType>()/weightVector[objIndex]);
}
} else {
// check whether the linEqSolver needs to be updated, i.e., whether the scheduler has changed
if(linEq.solver == nullptr || newScheduler->getChoices() != optimalChoicesAtCurrentEpoch) {
optimalChoicesAtCurrentEpoch = newScheduler->getChoices();
linEq.solver = nullptr;
storm::storage::SparseMatrix<ValueType> linEqMatrix = PS.toPS.selectRowsFromRowGroups(optimalChoicesAtCurrentEpoch, true);
linEqMatrix.convertToEquationSystem();
linEq.solver = linEq.factory.create(std::move(linEqMatrix));
linEq.solver->setCachingEnabled(true);
}
// Get the results for the individual objectives.
// Note that we do not consider an estimate for each objective (as done in the unbounded phase) since the results from the previous epoch are already pretty close
for(auto objIndex : consideredObjectives) {
auto const& objectiveRewardVectorPS = PS.objectiveRewardVectors[objIndex];
auto const& objectiveSolutionVectorMS = MS.objectiveSolutionVectors[objIndex];
// compute rhs of equation system, i.e., PS.toMS * x + Rewards
// To safe some time, only do this for the obtained optimal choices
auto itGroupIndex = PS.toPS.getRowGroupIndices().begin();
auto itChoiceOffset = optimalChoicesAtCurrentEpoch.begin();
for(auto& bValue : linEq.b) {
uint_fast64_t row = (*itGroupIndex) + (*itChoiceOffset);
bValue = objectiveRewardVectorPS[row];
for(auto const& entry : PS.toMS.getRow(row)){
bValue += entry.getValue() * objectiveSolutionVectorMS[entry.getColumn()];
}
++itGroupIndex;
++itChoiceOffset;
}
linEq.solver->solveEquations(PS.objectiveSolutionVectors[objIndex], linEq.b);
}
}
}
template <class SparseMaModelType>
void SparseMaPcaaWeightVectorChecker<SparseMaModelType>::performMSStep(SubModel& MS, SubModel const& PS, storm::storage::BitVector const& consideredObjectives, std::vector<ValueType> const& weightVector) const {
MS.toMS.multiplyWithVector(MS.weightedSolutionVector, MS.auxChoiceValues);
storm::utility::vector::addVectors(MS.weightedRewardVector, MS.auxChoiceValues, MS.weightedSolutionVector);
MS.toPS.multiplyWithVector(PS.weightedSolutionVector, MS.auxChoiceValues);
storm::utility::vector::addVectors(MS.weightedSolutionVector, MS.auxChoiceValues, MS.weightedSolutionVector);
if(consideredObjectives.getNumberOfSetBits() == 1 && !storm::utility::isZero(weightVector[*consideredObjectives.begin()])) {
// In this case there is no need to perform the computation on the individual objectives
auto objIndex = *consideredObjectives.begin();
MS.objectiveSolutionVectors[objIndex] = MS.weightedSolutionVector;
if(!storm::utility::isOne(weightVector[objIndex])) {
storm::utility::vector::scaleVectorInPlace(MS.objectiveSolutionVectors[objIndex], storm::utility::one<ValueType>()/weightVector[objIndex]);
}
} else {
for(auto objIndex : consideredObjectives) {
MS.toMS.multiplyWithVector(MS.objectiveSolutionVectors[objIndex], MS.auxChoiceValues);
storm::utility::vector::addVectors(MS.objectiveRewardVectors[objIndex], MS.auxChoiceValues, MS.objectiveSolutionVectors[objIndex]);
MS.toPS.multiplyWithVector(PS.objectiveSolutionVectors[objIndex], MS.auxChoiceValues);
storm::utility::vector::addVectors(MS.objectiveSolutionVectors[objIndex], MS.auxChoiceValues, MS.objectiveSolutionVectors[objIndex]);
}
}
}
template class SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>;
template double SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::getDigitizationConstant<double>(std::vector<double> const& direction) const;
template void SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::digitize<double>(SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::SubModel& subModel, double const& digitizationConstant) const;
template void SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::digitizeTimeBounds<double>(SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::TimeBoundMap& lowerTimeBounds, SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::TimeBoundMap& upperTimeBounds, double const& digitizationConstant);
#ifdef STORM_HAVE_CARL
// template class SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<storm::RationalNumber>>;
// template storm::RationalNumber SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<storm::RationalNumber>>::getDigitizationConstant<storm::RationalNumber>(std::vector<double> const& direction) const;
// template void SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<storm::RationalNumber>>::digitize<storm::RationalNumber>(SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<storm::RationalNumber>>::SubModel& subModel, storm::RationalNumber const& digitizationConstant) const;
// template void SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<storm::RationalNumber>>::digitizeTimeBounds<storm::RationalNumber>(SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::TimeBoundMap& lowerTimeBounds, SparseMaPcaaWeightVectorChecker<storm::models::sparse::MarkovAutomaton<double>>::TimeBoundMap& upperTimeBounds, storm::RationalNumber const& digitizationConstant);
#endif
}
}
}