def test_run_experiment(self):
mockMSI = mock.Mock(spec=Model)
mockMSI.name = 'test'
mockMSI.uncertainties = [RealParameter("a", 0, 10),
RealParameter("b", 0, 10)]
msis = NamedObjectMap(AbstractModel)
msis['test'] = mockMSI
runner = ExperimentRunner(msis)
experiment = Experiment('test',
mockMSI.name,
Policy('none'),
Scenario(a=1, b=2), 0)
runner.run_experiment(experiment)
sc, p = mockMSI.run_model.call_args[0]
self.assertEqual(sc.name, experiment.scenario.name)
self.assertEqual(p, experiment.policy)
mockMSI.reset_model.assert_called_once_with()
# assert handling of case error
mockMSI = mock.Mock(spec=Model)
mockMSI.name = 'test'
mockMSI.run_model.side_effect = Exception('some exception')
msis = NamedObjectMap(AbstractModel)
msis['test'] = mockMSI
runner = ExperimentRunner(msis)
experiment = Experiment('test',mockMSI.name,Policy('none'),
Scenario(a=1, b=2),0)
with self.assertRaises(EMAError):
runner.run_experiment(experiment)
# assert handling of case error
mockMSI = mock.Mock(spec=Model)
mockMSI.name = 'test'
mockMSI.run_model.side_effect = CaseError("message", {})
msis = NamedObjectMap(AbstractModel)
msis['test'] = mockMSI
runner = ExperimentRunner(msis)
experiment = Experiment('test',mockMSI.name,Policy('none'),
Scenario(a=1, b=2),0)
runner.run_experiment(experiment)
def test_run_model(self):
model_name = 'modelname'
function = mock.Mock()
model = Model(model_name, function)
model.uncertainties = [RealParameter('a', 0, 1)]
model.run_model(Scenario(**{'a': 0.1, 'b': 1}), Policy('test'))
function.assert_called_once_with(a=0.1)
# test complete translation of scenario
model = Model(model_name, function)
model.uncertainties = [
RealParameter('a', 0, 1, variable_name=['a', 'b'])
]
scenario = Scenario(**{'a': 0.1})
model.run_model(scenario, Policy('test'))
self.assertIn('a', scenario.keys())
self.assertIn('b', scenario.keys())
model = Model(model_name, function)
cats = [
Category('some name', [1, 2], multivalue=True),
Category('some other name', [3, 4], multivalue=True)
]
model.uncertainties = [
CategoricalParameter('a', cats, variable_name=['a', 'b'])
]
scenario = Scenario(**{'a': 'some name'})
model.run_model(scenario, Policy('test'))
self.assertIn('a', scenario.keys())
self.assertIn('b', scenario.keys())
self.assertEqual(scenario['a'], 1)
self.assertEqual(scenario['b'], 2)
scenario = Scenario(**{'a': 'some other name'})
model.run_model(scenario, Policy('test'))
self.assertIn('a', scenario.keys())
self.assertIn('b', scenario.keys())
self.assertEqual(scenario['a'], 3)
self.assertEqual(scenario['b'], 4)
def test_run_model(self):
model_name = 'modelname'
model_file = 'model_file'
with mock.patch('ema_workbench.em_framework.model.os') as patch:
patch.os.is_file.set_return_value(True)
model = FileModelTest(model_name, '.', model_file)
model.run_model(Scenario(a=1), Policy('test', b=2))
self.assertEqual(model.policy.name, 'test')
def test_store_results(self):
nr_experiments = 3
uncs = [RealParameter("a", 0, 1),
RealParameter("b", 0, 1)]
outcomes = [TimeSeriesOutcome("test")]
model = NamedObject('test')
experiment = Experiment(0, model, Policy('policy'), Scenario(a=1, b=0), 0)
# case 1 scalar shape = (1)
callback = DefaultCallback(uncs, [], outcomes,
nr_experiments=nr_experiments)
result = {outcomes[0].name: 1}
callback(experiment, result)
_, out = callback.get_results()
self.assertIn(outcomes[0].name, out.keys())
self.assertEqual(out[outcomes[0].name].shape, (3,))
# case 2 time series shape = (1, nr_time_steps)
callback = DefaultCallback(uncs, [], outcomes,
nr_experiments=nr_experiments)
result = {outcomes[0].name: np.random.rand(10)}
callback(experiment, result)
_, out = callback.get_results()
self.assertIn(outcomes[0].name, out.keys())
self.assertEqual(out[outcomes[0].name].shape, (3,10))
# case 3 maps etc. shape = (x,y)
callback = DefaultCallback(uncs, [], outcomes,
nr_experiments=nr_experiments)
result = {outcomes[0].name: np.random.rand(2,2)}
callback(experiment,result)
_, out = callback.get_results()
self.assertIn(outcomes[0].name, out.keys())
self.assertEqual(out[outcomes[0].name].shape, (3,2,2))
# case 4 assert raises EMAError
callback = DefaultCallback(uncs, [], outcomes,
nr_experiments=nr_experiments)
result = {outcomes[0].name: np.random.rand(2,2,2)}
self.assertRaises(EMAError, callback, experiment, result)
# KeyError
with mock.patch('ema_workbench.util.ema_logging.debug') as mocked_logging:
callback = DefaultCallback(uncs, [], outcomes,
nr_experiments=nr_experiments)
result = {'incorrect': np.random.rand(2,)}
callback(experiment, result)
for outcome in outcomes:
mocked_logging.assert_called_with("%s not specified as outcome in msi" % outcome.name)
def test_store_results(self):
nr_experiments = 3
uncs = [RealParameter("a", 0, 1), RealParameter("b", 0, 1)]
outcomes = [TimeSeriesOutcome("test")]
model = NamedObject('test')
experiment = Case(0, model, Policy('policy'), Scenario(a=1, b=0), 0)
# case 1 scalar shape = (1)
callback = DefaultCallback(uncs, [],
outcomes,
nr_experiments=nr_experiments)
model_outcomes = {outcomes[0].name: 1}
callback(experiment, model_outcomes)
_, out = callback.get_results()
self.assertIn(outcomes[0].name, out.keys())
self.assertEqual(out[outcomes[0].name].shape, (3, ))
# case 2 time series shape = (1, nr_time_steps)
callback = DefaultCallback(uncs, [],
outcomes,
nr_experiments=nr_experiments)
model_outcomes = {outcomes[0].name: np.random.rand(10)}
callback(experiment, model_outcomes)
_, out = callback.get_results()
self.assertIn(outcomes[0].name, out.keys())
self.assertEqual(out[outcomes[0].name].shape, (3, 10))
# case 3 maps etc. shape = (x,y)
callback = DefaultCallback(uncs, [],
outcomes,
nr_experiments=nr_experiments)
model_outcomes = {outcomes[0].name: np.random.rand(2, 2)}
callback(experiment, model_outcomes)
_, out = callback.get_results()
self.assertIn(outcomes[0].name, out.keys())
self.assertEqual(out[outcomes[0].name].shape, (3, 2, 2))
# case 4 assert raises EMAError
callback = DefaultCallback(uncs, [],
outcomes,
nr_experiments=nr_experiments)
model_outcomes = {outcomes[0].name: np.random.rand(2, 2, 2)}
self.assertRaises(EMAError, callback, experiment, model_outcomes)
def test_run_model(self):
model_name = 'modelname'
function = mock.Mock()
model = Model(model_name, function)
model.uncertainties = [RealParameter('a', 0 , 1)]
model.run_model(Scenario(**{'a':0.1, 'b':1}), Policy('test'))
function.assert_called_once_with(a=0.1)
# test complete translation of scenario
model = Model(model_name, function)
model.uncertainties = [RealParameter('a', 0 , 1, variable_name=['a', 'b'])]
scenario = Scenario(**{'a':0.1})
model.run_model(scenario, Policy('test'))
self.assertIn('a', scenario.keys())
self.assertIn('b', scenario.keys())
model = Model(model_name, function)
cats = [Category('some name', [1,2]),
Category('some other name', [3,4])]
model.uncertainties = [CategoricalParameter('a', cats,
variable_name=['a', 'b'], multivalue=True)]
scenario = Scenario(**{'a':cats[0].value})
model.run_model(scenario, Policy('test'))
self.assertIn('a', scenario.keys())
self.assertIn('b', scenario.keys())
self.assertEqual(scenario['a'], 1)
self.assertEqual(scenario['b'], 2)
scenario = Scenario(**{'a':cats[1].value})
model.run_model(scenario, Policy('test'))
self.assertIn('a', scenario.keys())
self.assertIn('b', scenario.keys())
self.assertEqual(scenario['a'], 3)
self.assertEqual(scenario['b'], 4)
def test_store_cases(self):
nr_experiments = 3
uncs = [
RealParameter("a", 0, 1),
RealParameter("b", 0, 1),
CategoricalParameter('c', [0, 1, 2]),
IntegerParameter("d", 0, 1)
]
outcomes = [TimeSeriesOutcome("test")]
case = {unc.name: random.random() for unc in uncs}
case["c"] = int(round(case["c"] * 2))
case["d"] = int(round(case["d"]))
model = NamedObject('test')
policy = Policy('policy')
scenario = Scenario(**case)
experiment = Experiment(0, model, policy, scenario, 0)
callback = DefaultCallback(uncs, [],
outcomes,
nr_experiments=nr_experiments,
reporting_interval=1)
result = {outcomes[0].name: 1}
callback(experiment, result)
experiments, _ = callback.get_results()
design = case
design['policy'] = policy.name
design['model'] = model.name
names = rf.get_names(experiments.dtype)
for name in names:
self.assertEqual(experiments[name][0], design[name])
# with levers
nr_experiments = 3
uncs = [RealParameter("a", 0, 1), RealParameter("b", 0, 1)]
levers = [RealParameter("c", 0, 1), RealParameter("d", 0, 1)]
outcomes = [TimeSeriesOutcome("test")]
case = {unc.name: random.random() for unc in uncs}
model = NamedObject('test')
policy = Policy('policy', c=1, d=1)
scenario = Scenario(**case)
experiment = Experiment(0, model, policy, scenario, 0)
callback = DefaultCallback(uncs,
levers,
outcomes,
nr_experiments=nr_experiments,
reporting_interval=1)
result = {outcomes[0].name: 1}
callback(experiment, result)
experiments, _ = callback.get_results()
design = case
design['c'] = 1
design['d'] = 1
design['policy'] = policy.name
design['model'] = model.name
names = rf.get_names(experiments.dtype)
print(experiments[0])
for name in names:
self.assertEqual(experiments[name][0], design[name])
def test_store_cases(self):
nr_experiments = 3
uncs = [
RealParameter("a", 0, 1),
RealParameter("b", 0, 1),
CategoricalParameter('c', [0, 1, 2]),
IntegerParameter("d", 0, 1)
]
outcomes = [TimeSeriesOutcome("test")]
case = {unc.name: random.random() for unc in uncs}
case["c"] = int(round(case["c"] * 2))
case["d"] = int(round(case["d"]))
model = NamedObject('test')
policy = Policy('policy')
scenario = Scenario(**case)
experiment = Case(0, model.name, policy, scenario, 0)
callback = DefaultCallback(uncs, [],
outcomes,
nr_experiments=nr_experiments,
reporting_interval=1)
model_outcomes = {outcomes[0].name: 1}
callback(experiment, model_outcomes)
experiments, _ = callback.get_results()
design = case
design['policy'] = policy.name
design['model'] = model.name
design['scenario'] = scenario.name
names = experiments.columns.values.tolist()
for name in names:
entry_a = experiments[name][0]
entry_b = design[name]
self.assertEqual(entry_a, entry_b, "failed for " + name)
# with levers
nr_experiments = 3
uncs = [RealParameter("a", 0, 1), RealParameter("b", 0, 1)]
levers = [RealParameter("c", 0, 1), RealParameter("d", 0, 1)]
outcomes = [TimeSeriesOutcome("test")]
case = {unc.name: random.random() for unc in uncs}
model = NamedObject('test')
policy = Policy('policy', c=1, d=1)
scenario = Scenario(**case)
experiment = Case(0, model.name, policy, scenario, 0)
callback = DefaultCallback(uncs,
levers,
outcomes,
nr_experiments=nr_experiments,
reporting_interval=1)
model_outcomes = {outcomes[0].name: 1}
callback(experiment, model_outcomes)
experiments, _ = callback.get_results()
design = case
design['c'] = 1
design['d'] = 1
design['policy'] = policy.name
design['model'] = model.name
design['scenario'] = scenario.name
names = experiments.columns.values.tolist()
for name in names:
self.assertEqual(experiments[name][0], design[name])
def test_worker(self, mocked_logging, mocked_runner):
if sys.version_info[0] < 3:
mocked_inqueue = mock.Mock(multiprocessing.queues.SimpleQueue())
mocked_outqueue = mock.Mock(multiprocessing.queues.SimpleQueue())
else:
mocked_inqueue = mock.Mock(multiprocessing.SimpleQueue())
mocked_outqueue = mock.Mock(multiprocessing.SimpleQueue())
function = mock.Mock()
mockMSI = MockMSI("test", function)
# task = None
mocked_inqueue.get.return_value = None
ema_parallel_multiprocessing.worker(mocked_inqueue,
mocked_outqueue,
[mockMSI])
mocked_logging.debug.assert_called_with('worker got sentinel -- exiting')
# EOFError, IOError
mocked_inqueue.get.side_effect = EOFError
ema_parallel_multiprocessing.worker(mocked_inqueue,
mocked_outqueue,
[mockMSI])
mocked_logging.debug.assert_called_with('worker got EOFError or IOError -- exiting')
mocked_inqueue.get.side_effect = IOError
ema_parallel_multiprocessing.worker(mocked_inqueue,
mocked_outqueue,
[mockMSI])
mocked_logging.debug.assert_called_with('worker got EOFError or IOError -- exiting')
# task = tuple of _, experiment dict
# - success
# - ema error
# - exception
# setup of test, we get a normal case
experiment = Experiment('try', mockMSI.name, Policy('none'),
Scenario(a=1), 0)
mocked_inqueue.get.return_value = (0, experiment)
mocked_inqueue.get.side_effect = None
# running experiment raises EMAError
mocked_runner().run_experiment.side_effect = EMAError
feeder_thread = threading.Thread(target=ema_parallel_multiprocessing.worker,
args=(mocked_inqueue,
mocked_outqueue,
[mockMSI]))
feeder_thread.deamon = True
feeder_thread.start()
time.sleep(0.001) # to avoid race conditions
mocked_inqueue.get.return_value = None
mocked_runner().run_experiment.assert_called_with(experiment)
# reset mocks
mocked_outqueue.reset_mock()
mocked_runner().reset_mock()
# running experiment raises EMAError
experiment = Experiment('try', mockMSI.name, Policy('none'),
Scenario(a=1), 0)
mocked_inqueue.get.return_value = (0, experiment)
mocked_inqueue.get.side_effect = None
mocked_runner().run_experiment.side_effect = Exception
feder_thread = threading.Thread(target=ema_parallel_multiprocessing.worker,
args=(mocked_inqueue,
mocked_outqueue,
[mockMSI]))
feder_thread.deamon = True
feder_thread.start()
time.sleep(0.001) # to avoid race conditions
mocked_inqueue.get.return_value = None
mocked_runner().run_experiment.assert_called_with(experiment)
# mocked_outqueue.put.assert_called_once()
# reset mocks
mocked_outqueue.reset_mock()
mocked_runner().reset_mock()
# running experiment works fine
experiment = Experiment('try', mockMSI.name, Policy('none'),
Scenario(a=1), 0)
mocked_inqueue.get.return_value = (0, experiment)
mocked_inqueue.get.side_effect = None
mocked_runner().run_experiment.side_effect = None
feder_thread = threading.Thread(target=ema_parallel_multiprocessing.worker,
args=(mocked_inqueue,
mocked_outqueue,
[mockMSI]))
feder_thread.deamon = True
feder_thread.start()
time.sleep(0.001) # to avoid race conditions
mocked_inqueue.get.return_value = None
mocked_runner().run_experiment.assert_called_with(experiment)
# mocked_outqueue.put.assert_called_once()
# reset mocks
mocked_outqueue.reset_mock()
mocked_runner().reset_mock()
RealParameter("r1", 0, 2),
RealParameter("r2", 0, 2),
RealParameter("w1", 0, 1)
]
#specify outcomes
lake_model.outcomes = [ScalarOutcome('max_P',
kind=ScalarOutcome.MINIMIZE), # @UndefinedVariable
ScalarOutcome('utility',
kind=ScalarOutcome.MAXIMIZE), # @UndefinedVariable
ScalarOutcome('inertia',
kind=ScalarOutcome.MINIMIZE), # @UndefinedVariable
ScalarOutcome('reliability',
kind=ScalarOutcome.MAXIMIZE)] # @UndefinedVariable
# override some of the defaults of the model
lake_model.constants = [Constant('alpha', 0.41),
Constant('nsamples', 100),
Constant('myears', 100)]
# reference is optional, but can be used to implement search for
# various user specified scenarios along the lines suggested by
# Watson and Kasprzyk (2017)
reference = Scenario('reference', b=0.4, q=2, mean=0.02, stdev=0.01)
with MultiprocessingEvaluator(lake_model) as evaluator:
evaluator.optimize(searchover='levers', nfe=1000,
epsilons=[0.1,]*len(lake_model.outcomes), reference=reference)
