stormpy.pomdp

API for the analysis of partially observable models (POMDPs).

Makes the functionality of storm::pomdp:: available, e.g., transformations of POMDPs, belief tracking, qualitative analysis, quantitative analysis and monitor verification.

Functions

apply_unknown_fsc(model[, mode])	Apply an unknown finite-state controller to the POMDP.
create_iterative_policy_search(model, ...[, ...])	Create a solver for qualitative reachability queries (almost-sure reachability) on POMDPs.
create_memory(pattern, nr_states)	Create a memory structure for finite-state controllers.
create_monitor_verifier(mc, monitor, risk, *)	Create the POMDP for verifying a monitor against a Markov chain.
create_nondeterministic_belief_tracker(model, *)	Create a tracker for the beliefs of the POMDP given a trace of observations.
create_observation_trace_unfolder(model, ...)	Create an unfolder which turns observation traces of the POMDP into MDPs.
make_canonic(model)	Make the representation of the POMDP canonic, i.e., order the choices of each state consistently for each observation.
make_simple(model, *[, keep_state_valuations])	Make the POMDP simple (aka alternating), i.e., each state has at most two actions, and if there is nondeterminism, then there is no probabilistic branching.
prepare_for_qualitative_search(model, property)	Prepare the POMDP for the qualitative analysis by making all states absorbing from which the target is not reached almost surely.
unfold_memory(model, memory, *[, ...])	Unfold the memory for a finite-state controller into the POMDP.
verify_with_belief_discretization(model, ...)	Approximate an unbounded reachability probability or expected reachability reward by discretizing the belief space (Freudenthal triangulation).
verify_with_belief_unfolding(model, property, *)	Approximate an unbounded reachability probability or expected reachability reward by unfolding the belief MDP.

Classes

BeliefBasedModelCheckerResult	Result of belief-based verification of a POMDP
BeliefBasedModelCheckerStatistics	Statistics of belief-based verification
BeliefSupportTracker	Tracker for belief supports
ExplorationQueueOrder(*values)	Order in which discovered beliefs are explored
IterativePolicySearch	Solver for POMDPs that solves qualitative queries.
MonitorVerifier	Container for monitor verifier POMDP with associated objects
NondeterministicBeliefTracker	Tracker for belief states (in sparse format) and uncontrollable actions
ObservationBasedFiniteStateController	Observation-based finite-state controller for a POMDP
ObservationTraceUnfolder	Unfolds observation traces in models
PomdpFscApplicationMode(*values)	Mode for applying an unknown FSC to a POMDP
PomdpMemory	Memory for POMDP policies
PomdpMemoryPattern(*values)	Memory pattern for POMDP memory
SparseBeliefState	Belief state in sparse format
WinningRegion	Winning region of a POMDP in terms of belief supports
WinningRegionQueryInterface	Queries on a winning region of a POMDP