def test_liquid_density():
model = StollWerthSurrogate(30.069 * unit.gram / unit.mole) # C2H6
(
epsilon,
sigma,
bond_length,
bond_length_star,
quadrupole,
quadrupole_star_sqr,
) = generate_parameters()
temperatures = numpy.array([308.0])
temperatures_star = temperatures / epsilon
value = model.liquid_density(temperatures, epsilon, sigma, bond_length,
quadrupole)
assert numpy.isclose(value, 285.1592692)
reduced_gradient_function = autograd.grad(model.liquid_density_star,
(1, 2))
gradient_function = autograd.grad(model.liquid_density, (1, 2, 3, 4))
reduced_gradient = reduced_gradient_function(temperatures_star,
quadrupole_star_sqr,
bond_length_star)
gradient = gradient_function(temperatures, epsilon, sigma, bond_length,
quadrupole)
assert len(reduced_gradient) == 2 and not numpy.allclose(
reduced_gradient, 0.0)
assert numpy.allclose(
gradient,
numpy.array([28.12601669, 2044.99241265, -10856.5686318, 0.01421091]))
def test_critical_temperature():
model = StollWerthSurrogate(30.069 * unit.gram / unit.mole) # C2H6
(
epsilon,
sigma,
bond_length,
bond_length_star,
quadrupole,
quadrupole_star_sqr,
) = generate_parameters()
value = model.critical_temperature(epsilon, sigma, bond_length, quadrupole)
assert numpy.isclose(value, 310.99575)
reduced_gradient_function = autograd.grad(model.critical_temperature_star,
(0, 1))
gradient_function = autograd.grad(model.critical_temperature, (0, 1, 2, 3))
reduced_gradient = reduced_gradient_function(quadrupole_star_sqr,
bond_length_star)
gradient = gradient_function(epsilon, sigma, bond_length, quadrupole)
assert len(reduced_gradient) == 2 and not numpy.allclose(
reduced_gradient, 0.0)
assert numpy.allclose(
gradient,
numpy.array([3.17342591, 452.59372786, -1140.53645031, 0.00153665]))
def test_surface_tension():
model = StollWerthSurrogate(30.069 * unit.gram / unit.mole) # C2H6
(
epsilon,
sigma,
bond_length,
bond_length_star,
quadrupole,
quadrupole_star_sqr,
) = generate_parameters()
temperatures = numpy.array([308.0])
temperatures_star = temperatures / epsilon
value = model.surface_tension(temperatures, epsilon, sigma, bond_length,
quadrupole)
assert numpy.isclose(value, 0.00017652)
reduced_gradient_function = autograd.grad(model.surface_tension_star,
(1, 2))
gradient_function = autograd.grad(model.surface_tension, (1, 2, 3, 4))
reduced_gradient = reduced_gradient_function(temperatures_star,
quadrupole_star_sqr,
bond_length_star)
gradient = gradient_function(temperatures, epsilon, sigma, bond_length,
quadrupole)
assert len(reduced_gradient) == 2 and not numpy.allclose(
reduced_gradient, 0.0)
assert numpy.allclose(
gradient,
numpy.array([0.00023103, 0.03215253, -0.08337817, 9.44823625e-07]))
def build_model(model_name, data_set, property_types):
# Build the likelihood operator.
surrogate_model = StollWerthSurrogate(data_set.molecular_weight)
log_likelihood = StollWerthOp(data_set, property_types, surrogate_model)
# Define the potentially 'fixed' model constants.
bond_length = data_set.bond_length.to(unit.nanometer).magnitude
quadrupole = 0.0
# Build the model
with pymc3.Model() as model:
epsilon = pymc3.Bound(pymc3.Exponential, 0.0)("epsilon",
lam=1.0 / 400.0)
sigma = pymc3.Bound(pymc3.Exponential, 0.0)("sigma", lam=1.0 / 5.0)
if model_name == "AUA" or model_name == "AUA+Q":
bond_length = pymc3.Bound(pymc3.Exponential, 0.0)("bond_length",
lam=1.0 / 3.0)
if model_name == "AUA+Q":
quadrupole = pymc3.Bound(pymc3.Exponential, 0.0)("quadrupole",
lam=1.0)
theta = tt.as_tensor_variable(
[epsilon, sigma, bond_length, quadrupole])
pymc3.DensityDist(
"likelihood",
lambda v: log_likelihood(v),
observed={"v": theta},
)
return model
def test_saturation_pressure():
model = StollWerthSurrogate(30.069 * unit.gram / unit.mole) # C2H6
(
epsilon,
sigma,
bond_length,
bond_length_star,
quadrupole,
quadrupole_star_sqr,
) = generate_parameters()
temperatures = numpy.array([308.0])
temperatures_star = temperatures / epsilon
value = model.saturation_pressure(temperatures, epsilon, sigma,
bond_length, quadrupole)
assert numpy.isclose(value, 5027.57796073)
reduced_gradient_function = autograd.grad(model.saturation_pressure_star,
(1, 2))
gradient_function = autograd.grad(model.saturation_pressure, (1, 2, 3, 4))
reduced_gradient = reduced_gradient_function(temperatures_star,
quadrupole_star_sqr,
bond_length_star)
gradient = gradient_function(temperatures, epsilon, sigma, bond_length,
quadrupole)
assert len(reduced_gradient) == 2 and not numpy.allclose(
reduced_gradient, 0.0)
assert numpy.allclose(
gradient,
numpy.array(
[-234.84756081, -68931.5713522, 73156.01894348, -0.11010289]),
)
def _get_two_center_model():
data_set = NISTDataSet("C2H2")
property_types = [*data_set.data_types]
priors = {
"epsilon": ("exponential", [0.0, random() * 400.0]),
"sigma": ("exponential", [0.0, random() * 5.0]),
"L": ("exponential", [0.0, random() * 3.0]),
"Q": ("exponential", [0.0, random() * 1.0]),
}
fixed = {}
model = TwoCenterLJModel(
"AUA+Q",
priors,
fixed,
data_set,
property_types,
StollWerthSurrogate(data_set.molecular_weight),
)
return model
def test_evaluate():
model = StollWerthSurrogate(30.069 * unit.gram / unit.mole) # C2H6
epsilon, sigma, bond_length, _, quadrupole, _ = generate_parameters()
parameters = numpy.array([epsilon, sigma, bond_length, quadrupole])
temperatures = numpy.array([298.0, 300.0, 308.0])
gradient_function = autograd.jacobian(model.evaluate, 1)
density_gradients = gradient_function(NISTDataType.LiquidDensity,
parameters, temperatures)
pressure_gradients = gradient_function(NISTDataType.SaturationPressure,
parameters, temperatures)
tension_gradients = gradient_function(NISTDataType.SurfaceTension,
parameters, temperatures)
assert (density_gradients.shape == pressure_gradients.shape ==
tension_gradients.shape)
assert not numpy.allclose(density_gradients, 0.0)
assert not numpy.allclose(pressure_gradients, 0.0)
assert not numpy.allclose(tension_gradients, 0.0)
def get_model(model_name, data_set, property_types, simulation_params):
priors = {
"epsilon": simulation_params["priors"]["epsilon"],
"sigma": simulation_params["priors"]["sigma"],
}
fixed = {}
if model_name == "AUA":
priors["L"] = simulation_params["priors"]["L"]
fixed["Q"] = 0.0
elif model_name == "AUA+Q":
priors["L"] = simulation_params["priors"]["L"]
priors["Q"] = simulation_params["priors"]["Q"]
elif model_name == "UA":
fixed["L"] = data_set.bond_length.to(unit.nanometer).magnitude
fixed["Q"] = 0.0
else:
raise NotImplementedError()
model = TwoCenterLJModel(
model_name,
priors,
fixed,
data_set,
property_types,
StollWerthSurrogate(data_set.molecular_weight),
)
return model
def test_critical_density():
model = StollWerthSurrogate(30.069 * unit.gram / unit.mole) # C2H6
(
epsilon,
sigma,
bond_length,
bond_length_star,
quadrupole,
quadrupole_star_sqr,
) = generate_parameters()
reduced_gradient_function = autograd.grad(model.critical_density_star,
(0, 1))
gradient_function = autograd.grad(model.critical_density, (0, 1, 2, 3))
reduced_gradient = reduced_gradient_function(quadrupole_star_sqr,
bond_length_star)
gradient = gradient_function(epsilon, sigma, bond_length, quadrupole)
assert len(reduced_gradient) == 2 and not numpy.allclose(
reduced_gradient, 0.0)
assert len(gradient) == 4 and not numpy.allclose(gradient, 0.0)