def newData(new_thetas, toexp=False):
model = monochain()
compiled_solver = VariableSSACSolver(model)
new_data = [
simulator(x, model, compiled_solver, toexp=toexp) for x in new_thetas
]
new_data, = compute(new_data)
new_data, = compute(new_data)
x = []
y = []
for e, i in enumerate(new_data):
if type(i) == float:
continue
elif np.max(i) > 2000:
continue
else:
x.append(i)
y.append(new_thetas[e])
new_data = np.asarray(x)
new_thetas = np.asarray(y)
new_data = np.reshape(
new_data, (new_data.shape[0], new_data.shape[2], new_data.shape[3]))
return (new_data, new_thetas)
def test_change_parameter(self):
model = Example()
initial_expression = model.listOfParameters['k1'].expression
solver = VariableSSACSolver(model)
results = model.run(solver=solver, variables={'k1': 0})
with self.subTest(msg='Test changed parameter simulation'):
self.assertEqual(results['Sp'][-1], results['Sp'][0])
with self.subTest(msg='Test changed parameter model integrity'):
self.assertEqual(model.listOfParameters['k1'].expression,
initial_expression)
def test_change_species(self):
model = Example()
initial_value = model.listOfSpecies['Sp'].initial_value
solver = VariableSSACSolver(model)
results = model.run(solver=solver, variables={'Sp': 3})
with self.subTest(msg='Test changed species simulation'):
self.assertEqual(results['Sp'][0], 3)
with self.subTest(msg='Test changed species model integrity'):
self.assertEqual(model.listOfSpecies['Sp'].initial_value,
initial_value)
def test_create(self):
model = Example()
solver = VariableSSACSolver(model)
def test_invalid_variable(self):
model = Example()
solver = VariableSSACSolver(model)
with self.assertRaises(SimulationError):
results = model.run(solver=solver, variables={'foobar': 0})
def test_run_example_precompiled(self):
model = Example()
solver = VariableSSACSolver(model)
results = model.run(solver=solver)
def test_file_with_directory_name_exists(self):
with self.assertRaises(DirectoryError):
temp = tempfile.NamedTemporaryFile()
model = Example()
solver = VariableSSACSolver(model, temp.name)
if transform:
# Extract only observed species
prey = res['prey']
predator = res['predator']
return np.vstack([prey, predator])
else:
return res
def wrapper(param):
return simulator(param, vilar_model, solver)
vilar_model = lotka_volterra()
solver = VariableSSACSolver(vilar_model)
class SummaryNet_large(nn.Module):
def __init__(self):
super().__init__()
# 2D convolutional layer
self.conv1 = nn.Conv1d(in_channels=2,
out_channels=20,
kernel_size=3,
padding=2)
# Maxpool layer that reduces YxY image to length (20,4)
self.pool = nn.MaxPool1d(kernel_size=5, stride=5)
# Fully connected layer taking as input the 6 flattened output arrays from the maxpooling layer
self.fc = nn.Linear(in_features=20 * 5 * 2, out_features=8)
self.add_parameter(Parameter(name="k3", expression=1.0))
# Species
self.add_species(Species(name='prey', initial_value = 100, mode = 'discrete'))
self.add_species(Species(name='predator', initial_value = 100, mode = 'discrete'))
# Reactions
self.add_reaction(Reaction(name="r1", reactants = {'prey' : 1}, products = {'prey' : 2}, rate = self.listOfParameters['k1']))
self.add_reaction(Reaction(name="r2", reactants = {'predator' : 1, 'prey' : 1}, products = {'predator' : 2}, rate = self.listOfParameters['k2']))
self.add_reaction(Reaction(name="r3", reactants = {'predator' : 1}, products = {}, rate = self.listOfParameters['k3']))
# Timespan
self.timespan(np.linspace(0, 50, 51))
model = lotka_volterra()
compiled_solver = VariableSSACSolver(model)
# # Data
#target_ts = np.load('target_ts.npy')
obs_data = np.load('target_original_shape_ts.npy')
# # Prior Distributions