def test_join(self):
joined = join(self.sin_spline, self.exp_spline)
evaluation = joined.get_state(self.times)
evaluation = joined.get_state(self.times)
control = np.vstack((np.sin(self.times), np.exp(self.times))).T
control_2 = np.hstack((self.sin_evaluation, self.exp_evaluation))
self.has_matched_times(joined)
assert_allclose(control, control_2, atol=0.01)
assert_allclose(evaluation, control, atol=0.01)
def _init_noise_inputs(self, backend):
"""
Build noise / stimulus input to the model.
"""
if backend == "jitcdde":
init_func = lambda noise: noise.as_cubic_splines(
duration=self.params["duration"],
dt=self.params["sampling_dt"])
join_func = lambda x: join(*x)
elif backend == "numba":
init_func = lambda noise: noise.as_array(
duration=self.params["duration"], dt=self.params["dt"])
join_func = lambda x: np.hstack(x).T
else:
raise ValueError(f"Unknown backend {backend}")
# initialise each noise / stimulation process and join
return join_func(
[init_func(noise) for noise in self.model_instance.noise_input])
def test_run_node(self):
fhn_node = FitzHughNagumoNode()
model = MultiModel.init_node(fhn_node)
model.params["sampling_dt"] = 10.0
# run MultiModel
model.run()
# run model instance
inst_res = fhn_node.run(
duration=model.params["duration"],
dt=model.params["sampling_dt"],
noise_input=join(*[
noise.as_cubic_splines(model.params["duration"],
model.params["sampling_dt"])
for noise in fhn_node.noise_input
]),
backend=model.params["backend"],
)
for out_var in model.output_vars:
np.testing.assert_equal(model[out_var], inst_res[out_var].values.T)
def test_run_network(self):
DELAY = 13.0
fhn_net = FitzHughNagumoNetwork(np.random.rand(2, 2),
np.array([[0.0, DELAY], [DELAY, 0.0]]))
model = MultiModel(fhn_net)
model.params["sampling_dt"] = 10.0
# run MultiModel
model.run()
# run model instance
inst_res = fhn_net.run(
duration=model.params["duration"],
dt=model.params["sampling_dt"],
noise_input=join(*[
noise.as_cubic_splines(model.params["duration"],
model.params["sampling_dt"])
for noise in fhn_net.noise_input
]),
backend=model.params["backend"],
)
for out_var in model.output_vars:
np.testing.assert_equal(model[out_var], inst_res[out_var].values.T)