probfd::algorithms::acyclic_vi::AcyclicValueIteration< State, Action > Class Template Reference

#include "probfd/algorithms/acyclic_value_iteration.h"

Inheritance diagram for probfd::algorithms::acyclic_vi::AcyclicValueIteration< State, Action >:

Description

template<typename State, typename Action>
class probfd::algorithms::acyclic_vi::AcyclicValueIteration< State, Action >

Implements acyclic Value Iteration.

Performs value iteration on acyclic MDPs by performing a depth-first search ending at terminal states and applying Bellman updates during backtracking. Exactly one Bellman update is performed per encountered state.

Template Parameters

State	- The state type of the underlying MDP model.
Action	- The action type of the underlying MDP model.

Remarks

The search algorithm may only be used with acyclic state spaces. The planner aborts if it detects a cycle.

Public Member Functions
virtual std::unique_ptr< PolicyType >	compute_policy (MDPType &mdp, EvaluatorType &heuristic, param_type< State > state, ProgressReport progress, double maxtime)=0
	Computes a partial policy for the input state.
virtual Interval	solve (MDPType &mdp, EvaluatorType &heuristic, param_type< State > state, ProgressReport progress, double max_time)=0
	Runs the MDP algorithm for the initial state state with a maximum time limit.
virtual void	print_statistics (std::ostream &) const
	Prints algorithm statistics to the specified output stream.

Member Function Documentation

compute_policy()

template<typename State , typename Action >

virtual std::unique_ptr< PolicyType > probfd::MDPAlgorithm< State, Action >::compute_policy ( MDPType & mdp, EvaluatorType & heuristic, param_type< State > state, ProgressReport progress, double maxtime )

pure virtualinherited

Computes a partial policy for the input state.

solve()

template<typename State , typename Action >

virtual Interval probfd::MDPAlgorithm< State, Action >::solve ( MDPType & mdp, EvaluatorType & heuristic, param_type< State > state, ProgressReport progress, double max_time )

pure virtualinherited

Runs the MDP algorithm for the initial state state with a maximum time limit.

print_statistics()

template<typename State , typename Action >

virtual void probfd::MDPAlgorithm< State, Action >::print_statistics ( std::ostream & ) const

inlinevirtualinherited

Prints algorithm statistics to the specified output stream.

Reimplemented in probfd::algorithms::exhaustive_dfs::ExhaustiveDepthFirstSearch< State, Action, UseInterval >, probfd::algorithms::fret::FRET< typename, typename, typename, typename >, probfd::algorithms::heuristic_search::HeuristicSearchAlgorithm< State, Action, StateInfoT >, probfd::algorithms::heuristic_search::HeuristicSearchAlgorithm< State, Action, internal::PerStateInformation< Action, UseInterval > >, probfd::algorithms::interval_iteration::IntervalIteration< State, Action >, probfd::algorithms::ta_topological_vi::TATopologicalValueIteration< State, Action, UseInterval >, probfd::algorithms::topological_vi::TopologicalValueIteration< State, Action, UseInterval >, and probfd::algorithms::topological_vi::TopologicalValueIteration< QState, QAction, true >.