Browse Source

some tests

refactoring
hannah 5 years ago
committed by Matthias Volk
parent
commit
c0a1eb7a32
No known key found for this signature in database GPG Key ID: 83A57678F739FCD3
  1. 3
      src/storage/model.cpp
  2. 7
      src/storage/modelcomponents.cpp
  3. 65
      tests/storage/test_modelcomponents.py

3
src/storage/model.cpp

@ -212,7 +212,7 @@ void define_sparse_model(py::module& m) {
;
py::class_<SparsePomdp<double>, std::shared_ptr<SparsePomdp<double>>>(m, "SparsePomdp", "POMDP in sparse representation", mdp)
.def(py::init<SparsePomdp<double>>(), py::arg("other_model"))
.def(py::init<ModelComponents<double> const&, bool>(), py::arg("components"), py::arg("canonicFlag")=false) //todo tests
.def(py::init<ModelComponents<double> const&, bool>(), py::arg("components"), py::arg("canonic_flag")=false) //todo tests
.def("__str__", &getModelInfoPrinter)
.def("get_observation", &SparsePomdp<double>::getObservation, py::arg("state"))
.def_property_readonly("observations", &SparsePomdp<double>::getObservations)
@ -234,7 +234,6 @@ void define_sparse_model(py::module& m) {
;
py::class_<SparseRewardModel<double>>(m, "SparseRewardModel", "Reward structure for sparse models")
//todo init?
.def_property_readonly("has_state_rewards", &SparseRewardModel<double>::hasStateRewards)
.def_property_readonly("has_state_action_rewards", &SparseRewardModel<double>::hasStateActionRewards)
.def_property_readonly("has_transition_rewards", &SparseRewardModel<double>::hasTransitionRewards)

7
src/storage/modelcomponents.cpp

@ -25,15 +25,16 @@ template<typename ValueType> using SparseModelComponents = storm::storage::spars
void define_sparse_modelcomponents(py::module& m) {
py::class_<SparseModelComponents<double>>(m, "SparseModelComponents", "ModelComponents description..") //todo
// shared_ptr? todo
py::class_<SparseModelComponents<double>, std::shared_ptr<SparseModelComponents<double>>>(m, "SparseModelComponents", "ModelComponents description..") //todo
.def(py::init<SparseMatrix<double> const&, StateLabeling const&, std::unordered_map<std::string, SparseRewardModel<double>> const&,
bool, boost::optional<BitVector> const&, boost::optional<SparseMatrix<storm::storage::sparse::state_type>> const&>(),
py::arg("transition_matrix"), py::arg("state_labeling") = storm::models::sparse::StateLabeling(),
py::arg("transition_matrix") = SparseMatrix<double>(), py::arg("state_labeling") = storm::models::sparse::StateLabeling(),
py::arg("reward_models") = std::unordered_map<std::string, SparseRewardModel<double>>(), py::arg("rate_transitions") = false,
py::arg("markovian_states") = boost::none, py::arg("player1_matrix") = boost::none)
//.def(py::init<>()) // for rvalue ? todo
// General components (for all model types)
.def_readwrite("transition_matrix", &SparseModelComponents<double>::transitionMatrix)

65
tests/storage/test_modelcomponents.py

@ -0,0 +1,65 @@
import stormpy
import stormpy.logic
from helpers.helper import get_example_path
import pytest
class TestSparseModel:
def test_init_default(self):
components = stormpy.SparseModelComponents()
assert components.state_labeling.get_labels() == set()
assert components.reward_models == {}
assert components.transition_matrix.nr_rows == 0
assert components.transition_matrix.nr_columns == 0
assert components.markovian_states is None
assert components.player1_matrix is None
assert not components.rate_transitions
# def test_init(self):
# todo Build simple transition matrix etc
# transition_matrix =
def test_dtmc_modelcomponents(self):
program = stormpy.parse_prism_program(get_example_path("dtmc", "die.pm"))
model = stormpy.build_model(program)
components = stormpy.SparseModelComponents(transition_matrix=model.transition_matrix,
state_labeling=model.labeling,
reward_models=model.reward_models)
dtmc = stormpy.storage.SparseDtmc(components)
assert dtmc.model_type == stormpy.ModelType.DTMC
assert dtmc.initial_states == [0]
assert dtmc.nr_states == 13
for state in dtmc.states:
assert len(state.actions) <= 1
assert dtmc.labeling.get_labels() == {'init', 'deadlock', 'done', 'one', 'two', 'three', 'four', 'five', 'six'}
assert dtmc.nr_transitions == 20
assert len(dtmc.reward_models) == 1
assert not dtmc.reward_models["coin_flips"].has_state_rewards
assert dtmc.reward_models["coin_flips"].has_state_action_rewards
for reward in dtmc.reward_models["coin_flips"].state_action_rewards:
assert reward == 1.0 or reward == 0.0
assert not dtmc.reward_models["coin_flips"].has_transition_rewards
assert not dtmc.supports_parameters
def test_pmdp_modelcomponents(self):
program = stormpy.parse_prism_program(get_example_path("pmdp", "two_dice.nm"))
model = stormpy.build_parametric_model(program)
def test_ma_modelcomponents(self):
program = stormpy.parse_prism_program(get_example_path("ma", "simple.ma"), False, True)
formulas = stormpy.parse_properties_for_prism_program("Pmax=? [ F<=2 s=2 ]", program)
model = stormpy.build_model(program, formulas)
#todo create mc
assert model.nr_states == 4
assert model.nr_transitions == 7
assert model.model_type == stormpy.ModelType.MA
assert not model.supports_parameters
assert type(model) is stormpy.SparseMA
Loading…
Cancel
Save