def test_optimization():
if os.name != 'nt':
return
ema_logging.log_to_stderr(ema_logging.INFO)
model = FluModel(r'../models', "fluCase")
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.parallel = True
pop_size = 8
nr_of_generations = 10
eps = np.array([1e-3, 1e6])
stats, pop = ensemble.perform_outcome_optimization(
obj_function=obj_function_multi,
algorithm=epsNSGA2,
reporting_interval=100,
weights=(MAXIMIZE, MAXIMIZE),
pop_size=pop_size,
nr_of_generations=nr_of_generations,
crossover_rate=0.8,
mutation_rate=0.05,
eps=eps)
def test_vensim_model(self):
#instantiate a model
wd = r'../models'
model = VensimExampleModel(wd, "simpleModel")
#instantiate an ensemble
ensemble = ModelEnsemble()
#set the model on the ensemble
ensemble.model_structure = model
nr_runs = 10
experiments, outcomes = ensemble.perform_experiments(nr_runs)
self.assertEqual(experiments.shape[0], nr_runs)
self.assertIn('TIME', outcomes.keys())
self.assertIn(model.outcomes[0].name, outcomes.keys())
def test_vensim_model(self):
#instantiate a model
wd = r'../models'
model = VensimExampleModel(wd, "simpleModel")
#instantiate an ensemble
ensemble = ModelEnsemble()
#set the model on the ensemble
ensemble.model_structure = model
nr_runs = 10
experiments, outcomes = ensemble.perform_experiments(nr_runs)
self.assertEqual(experiments.shape[0], nr_runs)
self.assertIn('TIME', outcomes.keys())
self.assertIn(model.outcomes[0].name, outcomes.keys())
def test_running_lookup_uncertainties(self):
'''
This is the more comprehensive test, given that the lookup
uncertainty replaces itself with a bunch of other uncertainties, check
whether we can successfully run a set of experiments and get results
back. We assert that the uncertainties are correctly replaced by
analyzing the experiments array.
'''
if os.name != 'nt':
return
model = LookupTestModel(r'../models/', 'lookupTestModel')
#model.step = 4 #reduce data to be stored
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.perform_experiments(10)
def test_running_lookup_uncertainties(self):
'''
This is the more comprehensive test, given that the lookup
uncertainty replaces itself with a bunch of other uncertainties, check
whether we can successfully run a set of experiments and get results
back. We assert that the uncertainties are correctly replaced by
analyzing the experiments array.
'''
if os.name != 'nt':
return
model = LookupTestModel( r'../models/', 'lookupTestModel')
#model.step = 4 #reduce data to be stored
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.perform_experiments(10)
def test_optimization():
if os.name != 'nt':
return
ema_logging.log_to_stderr(ema_logging.INFO)
model = FluModel(r'../models', "fluCase")
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.parallel=True
pop_size = 8
nr_of_generations = 10
eps = np.array([1e-3, 1e6])
stats, pop = ensemble.perform_outcome_optimization(obj_function = obj_function_multi,
algorithm=epsNSGA2,
reporting_interval=100,
weights=(MAXIMIZE, MAXIMIZE),
pop_size=pop_size,
nr_of_generations=nr_of_generations,
crossover_rate=0.8,
mutation_rate=0.05,
eps=eps)
"susceptible to immune population delay time region 1"),
ParameterUncertainty((0.5,2),
"susceptible to immune population delay time region 2"),
ParameterUncertainty((0.01, 5),
"root contact rate region 1"),
ParameterUncertainty((0.01, 5),
"root contact ratio region 2"),
ParameterUncertainty((0, 0.15),
"infection ratio region 1"),
ParameterUncertainty((0, 0.15),
"infection rate region 2"),
ParameterUncertainty((10, 100),
"normal contact rate region 1"),
ParameterUncertainty((10, 200),
"normal contact rate region 2")]
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = FluModel(r'./models/flu', "fluCase")
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.parallel = True #turn on parallel processing
nr_experiments = 1000
results = ensemble.perform_experiments(nr_experiments)
fh = r'./data/{} flu cases no policy.tar.gz'.format(nr_experiments)
save_results(results, fh)
a_mean = np.mean(a)
b_mean = np.mean(b)
if a_mean < 0.5 or b_mean < 0.5:
return (np.inf, ) * 2
else:
return a_mean, b_mean
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = DummyModel(r"", "dummy")
np.random.seed(123456789)
ensemble = ModelEnsemble()
ensemble.set_model_structure(model)
policy_levers = {
'Trigger a': {
'type': 'list',
'values': [0, 0.25, 0.5, 0.75, 1]
},
'Trigger b': {
'type': 'list',
'values': [0, 0.25, 0.5, 0.75, 1]
},
'Trigger c': {
'type': 'list',
'values': [0, 0.25, 0.5, 0.75, 1]
}
outcomes = [
Outcome('sheep', time=True),
Outcome('wolves', time=True),
Outcome('grass', time=True) # TODO patches not working in reporting
]
if __name__ == "__main__":
#turn on logging
ema_logging.log_to_stderr(ema_logging.INFO)
#instantiate a model
fh = r"./models/predatorPreyNetlogo"
model = PredatorPrey(fh, "simpleModel")
#instantiate an ensemble
ensemble = ModelEnsemble()
#set the model on the ensemble
ensemble.model_structure = model
#run in parallel, if not set, FALSE is assumed
ensemble.parallel = True
#perform experiments
results = ensemble.perform_experiments(100, reporting_interval=1)
plotting.lines(results, density=plotting_util.KDE)
plt.show()
"normal contact rate region 2")]
def model_init(self, policy, kwargs):
'''initializes the model'''
try:
self.model_file = policy['file']
except KeyError:
ema_logging.warning("key 'file' not found in policy")
super(FluModel, self).model_init(policy, kwargs)
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = FluModel(r'./models/flu', "flucase")
ensemble = ModelEnsemble()
ensemble.model_structure = model
#add policies
policies = [{'name': 'no policy',
'file': r'\FLUvensimV1basecase.vpm'},
{'name': 'static policy',
'file': r'\FLUvensimV1static.vpm'},
{'name': 'adaptive policy',
'file': r'\FLUvensimV1dynamic.vpm'}
]
ensemble.policies = policies
#turn on parallel processing
ensemble.parallel = True
else:
atomicBehavior.append([last, steps])
last = entry
steps = 0
atomicBehavior.append([last, steps])
behavior = []
behavior.append(atomicBehavior.pop(0))
for entry in atomicBehavior:
if entry[0] != behavior[-1][0] and entry[1] >2:
behavior.append(entry)
elif entry[1] <2:
continue
else:
behavior[-1][1] =+ entry[1]
behavior = [entry[0] for entry in behavior]
return behavior
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = ScarcityModel(r'..\data', "scarcity")
ensemble = ModelEnsemble()
ensemble.set_model_structure(model)
ensemble.parallel = True
results = ensemble.perform_experiments(100)
# determineBehavior(results)
susceptible_population_region_2 = susceptible_population_region_2_NEXT
immune_population_region_1 = immune_population_region_1_NEXT
immune_population_region_2 = immune_population_region_2_NEXT
deceased_population_region_1.append(deceased_population_region_1_NEXT)
deceased_population_region_2.append(deceased_population_region_2_NEXT)
#End of main code
return (runTime, deceased_population_region_1) #, Max_infected, Max_time)
if __name__ == "__main__":
np.random.seed(150) #set the seed for replication purposes
ema_logging.log_to_stderr(ema_logging.INFO)
fluModel = MexicanFlu(None, "mexicanFluExample")
ensemble = ModelEnsemble()
ensemble.parallel = True
ensemble.model_structure = fluModel
nr_experiments = 500
results = ensemble.perform_experiments(nr_experiments, reporting_interval=100)
lines(results, outcomes_to_show="deceased_population_region_1",
show_envelope=True, density=KDE, titles=None,
experiments_to_show=np.arange(0, nr_experiments, 10)
)
plt.show()
ParameterUncertainty((0.1,0.3),
"kk")]
#specification of the outcomes
outcomes = [Outcome("B4:B1076", time=True), #we can refer to a range in the normal way
Outcome("P_t", time=True)] # we can also use named range
#name of the sheet
sheet = "Sheet1"
#relative path to the Excel file
workbook = r'\excel example.xlsx'
if __name__ == "__main__":
ema_logging.log_to_stderr(level=ema_logging.INFO)
model = ExcelModel(r"./models/excelModel", "predatorPrey")
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.parallel = True #turn on parallel computing
pool = MultiprocessingPool(ensemble.model_structure,
nr_processes=2)
ensemble.pool = pool
#run 100 experiments
nr_experiments = 100
results = ensemble.perform_experiments(nr_experiments)
'''
This class represents a simple example of how one can extent the basic
ModelStructureInterface in order to do EMA on a simple model coded in
Python directly
'''
#specify uncertainties
uncertainties = [
ParameterUncertainty((0.1, 10), "x1"),
ParameterUncertainty((-0.01, 0.01), "x2"),
ParameterUncertainty((-0.01, 0.01), "x3")
]
#specify outcomes
outcomes = [Outcome('y')]
def model_init(self, policy, kwargs):
pass
def run_model(self, case):
"""Method for running an instantiated model structure """
self.output[
self.outcomes[0].name] = case['x1'] * case['x2'] + case['x3']
if __name__ == '__main__':
model = SimplePythonModel(None, 'simpleModel') #instantiate the model
ensemble = ModelEnsemble() #instantiate an ensemble
# ensemble.parallel = True
ensemble.set_model_structure(model) #set the model on the ensemble
results = ensemble.perform_experiments(1000) #run 1000 experiments
a_mean = np.mean(a)
b_mean = np.mean(b)
if a_mean < 0.5 or b_mean < 0.5:
return (np.inf,) * 2
else:
return a_mean, b_mean
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = DummyModel(r"", "dummy")
np.random.seed(123456789)
ensemble = ModelEnsemble()
ensemble.set_model_structure(model)
policy_levers = {'Trigger a': {'type':'list', 'values':[0, 0.25, 0.5, 0.75, 1]},
'Trigger b': {'type':'list', 'values':[0, 0.25, 0.5, 0.75, 1]},
'Trigger c': {'type':'list', 'values':[0, 0.25, 0.5, 0.75, 1]}}
cases = ensemble._generate_samples(10, UNION)[0]
ensemble.add_policy({"name":None})
experiments = [entry for entry in ensemble._generate_experiments(cases)]
for entry in experiments:
entry.pop("model")
entry.pop("policy")
cases = experiments
ParameterUncertainty, Outcome
class SimplePythonModel(ModelStructureInterface):
'''
This class represents a simple example of how one can extent the basic
ModelStructureInterface in order to do EMA on a simple model coded in
Python directly
'''
#specify uncertainties
uncertainties = [ParameterUncertainty((0.1, 10), "x1"),
ParameterUncertainty((-0.01,0.01), "x2"),
ParameterUncertainty((-0.01,0.01), "x3")]
#specify outcomes
outcomes = [Outcome('y')]
def model_init(self, policy, kwargs):
pass
def run_model(self, case):
"""Method for running an instantiated model structure """
self.output[self.outcomes[0].name] = case['x1']*case['x2']+case['x3']
if __name__ == '__main__':
model = SimplePythonModel(None, 'simpleModel') #instantiate the model
ensemble = ModelEnsemble() #instantiate an ensemble
# ensemble.parallel = True
ensemble.set_model_structure(model) #set the model on the ensemble
results = ensemble.perform_experiments(1000) #run 1000 experiments
[(-0.5, 0.35), (3, 5), (1, 10), (0.2, 0.4), (0, 120)],
"effect of hours worked on energy drain lookup",
self,
0,
3,
),
LookupUncertainty(
"hearne1",
[(0, 1), (0, 0.15), (1, 1.5), (0.75, 1.25)],
"effect of low energy on further depletion lookup",
self,
0,
1,
),
]
self._delete_lookup_uncertainties()
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = Burnout(r"./models/burnout", "burnout")
ensemble = ModelEnsemble()
ensemble.model_structure = model
# run policy with old cases
results = ensemble.perform_experiments(100)
lines(results, "Energy Level", density=BOXPLOT)
plt.show()
#note that this reference to the model should be relative
#this relative path will be combined with the workingDirectory
model_file = r'\model.vpm'
#specify outcomes
outcomes = [Outcome('a', time=True)]
#specify your uncertainties
uncertainties = [ParameterUncertainty((0, 2.5), "x11"),
ParameterUncertainty((-2.5, 2.5), "x12")]
if __name__ == "__main__":
#turn on logging
ema_logging.log_to_stderr(ema_logging.INFO)
#instantiate a model
wd = r'./models/vensim example'
vensimModel = VensimExampleModel(wd, "simpleModel")
#instantiate an ensemble
ensemble = ModelEnsemble()
#set the model on the ensemble
ensemble.model_structure = vensimModel
#run in parallel, if not set, FALSE is assumed
ensemble.parallel = True
#perform experiments
result = ensemble.perform_experiments(1000)
speed = kwargs.pop("lookup price substitute speed")
begin = kwargs.pop("lookup price substitute begin")
end = kwargs.pop("lookup price substitute end")
lookup = [self.priceSubstite(x, speed, begin, end) for x in range(0,100, 10)]
kwargs['relative price substitute lookup'] = lookup
scale = kwargs.pop("lookup returns to scale speed")
speed = kwargs.pop("lookup returns to scale scale")
lookup = [self.returnsToScale(x, speed, scale) for x in range(0, 101, 10)]
kwargs['returns to scale lookup'] = lookup
scale = kwargs.pop("lookup approximated learning speed")
speed = kwargs.pop("lookup approximated learning scale")
start = kwargs.pop("lookup approximated learning start")
lookup = [self.approxLearning(x, speed, scale, start) for x in range(0, 101, 10)]
kwargs['approximated learning effect lookup'] = lookup
super(ScarcityModel, self).run_model(kwargs)
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.DEBUG)
model = ScarcityModel(r'./models/scarcity', "scarcity")
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.parallel = True
results = ensemble.perform_experiments(2)
CategoricalUncertainty(("true", "true"), "grass?")
]
outcomes = [Outcome('sheep', time=True),
Outcome('wolves', time=True),
Outcome('grass', time=True) # TODO patches not working in reporting
]
if __name__ == "__main__":
#turn on logging
ema_logging.log_to_stderr(ema_logging.INFO)
#instantiate a model
fh = r"./models/predatorPreyNetlogo"
model = PredatorPrey(fh, "simpleModel")
#instantiate an ensemble
ensemble = ModelEnsemble()
#set the model on the ensemble
ensemble.model_structure = model
#run in parallel, if not set, FALSE is assumed
ensemble.parallel = True
#perform experiments
results = ensemble.perform_experiments(100, reporting_interval=1)
plotting.lines(results, density=plotting_util.KDE)
plt.show()
This class represents a simple example of how one can extent the basic
ModelStructureInterface in order to do EMA on a simple model coded in
Python directly
'''
#specify uncertainties
uncertainties = [
ParameterUncertainty((0.1, 10), "x1"),
ParameterUncertainty((-0.01, 0.01), "x2"),
ParameterUncertainty((-0.01, 0.01), "x3")
]
#specify outcomes
outcomes = [Outcome('y')]
def model_init(self, policy, kwargs):
pass
def run_model(self, case):
"""Method for running an instantiated model structure """
self.output[
self.outcomes[0].name] = case['x1'] * case['x2'] + case['x3']
if __name__ == '__main__':
ema_logging.log_to_stderr(ema_logging.INFO)
model = SimplePythonModel(None, 'simpleModel') #instantiate the model
ensemble = ModelEnsemble() #instantiate an ensemble
ensemble.parallel = True
ensemble.model_structure = model #set the model on the ensemble
results = ensemble.perform_experiments(1000) #run 1000 experiments
kwargs['relative price substitute lookup'] = lookup
scale = kwargs.pop("lookup returns to scale speed")
speed = kwargs.pop("lookup returns to scale scale")
lookup = [
self.returnsToScale(x, speed, scale) for x in range(0, 101, 10)
]
kwargs['returns to scale lookup'] = lookup
scale = kwargs.pop("lookup approximated learning speed")
speed = kwargs.pop("lookup approximated learning scale")
start = kwargs.pop("lookup approximated learning start")
lookup = [
self.approxLearning(x, speed, scale, start)
for x in range(0, 101, 10)
]
kwargs['approximated learning effect lookup'] = lookup
super(ScarcityModel, self).run_model(kwargs)
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.DEBUG)
model = ScarcityModel(r'./models/scarcity', "scarcity")
ensemble = ModelEnsemble()
ensemble.model_structure = model
ensemble.parallel = True
results = ensemble.perform_experiments(2)
"effect of high energy on further recovery lookup", self, 0,
1.25),
LookupUncertainty(
'hearne2', [(-2, 2), (-1, 1), (0, 100), (20, 120), (0.5, 1.5),
(0.5, 2)],
"effect of hours worked on energy recovery lookup", self, 0,
1.5),
LookupUncertainty('approximation', [(-0.5, 0.35), (3, 5), (1, 10),
(0.2, 0.4), (0, 120)],
"effect of hours worked on energy drain lookup",
self, 0, 3),
LookupUncertainty(
'hearne1', [(0, 1), (0, 0.15), (1, 1.5), (0.75, 1.25)],
"effect of low energy on further depletion lookup", self, 0, 1)
]
self._delete_lookup_uncertainties()
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = Burnout(r'./models/burnout', "burnout")
ensemble = ModelEnsemble()
ensemble.model_structure = model
#run policy with old cases
results = ensemble.perform_experiments(100)
lines(results, 'Energy Level', density=BOXPLOT)
plt.show()
#specify outcomes
outcomes = [Outcome('a', time=True)]
#specify your uncertainties
uncertainties = [
ParameterUncertainty((0, 2.5), "x11"),
ParameterUncertainty((-2.5, 2.5), "x12")
]
if __name__ == "__main__":
#turn on logging
ema_logging.log_to_stderr(ema_logging.INFO)
#instantiate a model
wd = r'./models/vensim example'
vensimModel = VensimExampleModel(wd, "simpleModel")
#instantiate an ensemble
ensemble = ModelEnsemble()
#set the model on the ensemble
ensemble.model_structure = vensimModel
#run in parallel, if not set, FALSE is assumed
ensemble.parallel = True
#perform experiments
result = ensemble.perform_experiments(1000)
'''
Created on Mar 15, 2012
.. codeauthor:: jhkwakkel <j.h.kwakkel (at) tudelft (dot) nl>
'''
from analysis import clusterer
from util import ema_logging
from core import ModelEnsemble
from test.scarcity_example import ScarcityModel
if __name__ == "__main__":
ema_logging.log_to_stderr(ema_logging.INFO)
model = ScarcityModel(r'..\..\src\test', "fluCase")
ensemble = ModelEnsemble()
ensemble.set_model_structure(model)
ensemble.parallel = True
results = ensemble.perform_experiments(200)
clusterer.cluster(data=results,
outcome='relative market price',
distance='gonenc',
cMethod='maxclust',
cValue=5,
plotDendrogram=False)
class SimplePythonModel(ModelStructureInterface):
'''
This class represents a simple example of how one can extent the basic
ModelStructureInterface in order to do EMA on a simple model coded in
Python directly
'''
#specify uncertainties
uncertainties = [ParameterUncertainty((0.1, 10), "x1"),
ParameterUncertainty((-0.01,0.01), "x2"),
ParameterUncertainty((-0.01,0.01), "x3")]
#specify outcomes
outcomes = [Outcome('y')]
def model_init(self, policy, kwargs):
pass
def run_model(self, case):
"""Method for running an instantiated model structure """
self.output[self.outcomes[0].name] = case['x1']*case['x2']+case['x3']
if __name__ == '__main__':
ema_logging.log_to_stderr(ema_logging.INFO)
model = SimplePythonModel(None, 'simpleModel') #instantiate the model
ensemble = ModelEnsemble() #instantiate an ensemble
ensemble.parallel = True
ensemble.model_structure = model #set the model on the ensemble
results = ensemble.perform_experiments(1000) #run 1000 experiments