def test_scimodel(variable_x, variable_y, functional_fx, functional_gx):
xs = [variable_x, variable_y]
ys = [functional_fx, functional_gx]
assert isinstance(sn.SciModel(xs, ys), sn.SciModel)
assert isinstance(sn.SciModel(xs, ys, "mse", "adam"), sn.SciModel)
assert isinstance(sn.SciModel(variable_x, ys), sn.SciModel)
assert isinstance(sn.SciModel(xs, functional_fx), sn.SciModel)
def test_scimodel_keras_optimizers(variable_x, variable_y, functional_fx,
functional_gx):
xs = [variable_x, variable_y]
ys = [functional_fx, functional_gx]
assert isinstance(sn.SciModel(xs, ys, "mse", tf_optimizers.Adam()),
sn.SciModel)
assert isinstance(sn.SciModel(xs, ys, "mse", tf_optimizers.RMSprop()),
sn.SciModel)
def test_scimodel_optimizers(variable_x, variable_y, functional_fx,
functional_gx):
xs = [variable_x, variable_y]
ys = [functional_fx, functional_gx]
assert isinstance(sn.SciModel(xs, ys, "mse", "adam"), sn.SciModel)
assert isinstance(sn.SciModel(xs, ys, "mse", "rmsprop"), sn.SciModel)
assert isinstance(sn.SciModel(xs, ys, "mse", "scipy-l-bfgs-b"),
sn.SciModel)
assert isinstance(sn.SciModel(xs, ys, "mse", "scipy-newtoncg"),
sn.SciModel)
def model_fx_gx(variable_x, functional_fx, functional_gx):
return sn.SciModel(variable_x, [functional_fx, functional_gx])
def model_fx(variable_x, functional_fx):
return sn.SciModel(variable_x, functional_fx)
def test_eval_scimodel(test_data_xy, expected_hxy, variable_x, variable_y,
functional_hxy):
m = sn.SciModel([variable_x, variable_y], functional_hxy)
hxy_test = functional_hxy.eval(m, test_data_xy)
assert np.linalg.norm(hxy_test -
expected_hxy) < 1e-6 * np.linalg.norm(expected_hxy)
def test_scimodel_lossfunc_exceptions(variable_x, variable_y, functional_fx,
functional_gx):
xs = [variable_x, variable_y]
ys = [functional_fx, functional_gx]
with pytest.raises(ValueError):
assert isinstance(sn.SciModel(xs, ys, "to_fail", "adam"), sn.SciModel)
def test_scimodel_exceptions(variable_x, variable_y, functional_fx,
functional_gx):
with pytest.raises(ValueError):
sn.SciModel(variable_x)
with pytest.raises(ValueError):
sn.SciModel(functional_fx)
C1 = (1 - sign(t - TOL)) * (u - sin(np.pi * x) * sin(np.pi * y))
C2 = (1 - sign(t - TOL)) * (diff(u, t))
C3 = (1 - sign(x - TOL)) * u
C4 = (1 - sign(y - TOL)) * u
C5 = (1 + sign(x - (1 - TOL))) * u
C6 = (1 + sign(y - (1 - TOL))) * u
C7 = (1 - sign(x - TOL)) * (diff(u, x, order=2))
C8 = (1 - sign(y - TOL)) * (diff(u, y, order=2))
C9 = (1 + sign(x - (1 - TOL))) * (diff(u, x, order=2))
C10 = (1 + sign(y - (1 - TOL))) * (diff(u, y, order=2))
m = sn.SciModel(
[x, y, t],
[sn.PDE(L1), C1, C2, C3, C4, C5, C6, C7, C8, C9, C10],
# load_weights_from = 'plate-weights.hdf5'
)
inputs = [data['x'], data['y'], data['t']]
targets = [(data['dom_ids'], 'zeros')] \
+ 2*[(data['t0_ids'], 'zeros')] \
+ 8*[(data['bc_ids'], 'zeros')]
h = m.train(
inputs,
targets,
batch_size=BATCH,
learning_rate=0.001,
reduce_lr_after=20,
adaptive_weights={'freq': True},
u = sn.Functional('u', [x, y, t], 4 * [20], 'sin')
c = 1.0
L1 = c * (diff(u, x, order=2) + diff(u, y, order=2)) - diff(u, t, order=2)
TOL = 0.001
C1 = (1 - sign(t - TOL)) * (u - sin(np.pi * x) * sin(np.pi * y))
C2 = (1 - sign(t - TOL)) * (diff(u, t))
C3 = (1 - sign(x - TOL)) * u
C4 = (1 - sign(y - TOL)) * u
C5 = (1 + sign(x - (1 - TOL))) * u
C6 = (1 + sign(y - (1 - TOL))) * u
m = sn.SciModel(
[x, y, t],
[sn.PDE(L1), C1, C2, C3, C4, C5, C6],
# load_weights_from='membrane-weights.hdf5'
)
inputs = [data['x'], data['y'], data['t']]
targets = [(data['dom_ids'], 'zeros')] \
+ 2*[(data['t0_ids'], 'zeros')] \
+ 4*[(data['bc_ids'], 'zeros')]
h = m.train(inputs,
targets,
batch_size=BATCH,
learning_rate=0.001,
reduce_lr_after=50,
adaptive_weights={'freq': True},
epochs=EPOCHS)
import numpy as np
import matplotlib.pyplot as plt
import sciann as sn
omega = 2 * np.pi / 2.0
omega_bar = 2 * np.pi / 1.5
A = 1.0
beta = 1.0
tunits = 2
NDATA = 1000
t_data = np.linspace(0, 4 * np.pi, NDATA).reshape(-1, 1)
dt = np.diff(t_data.flatten()).mean()
t_data = t_data + np.linspace(0, dt * (tunits - 1), tunits).reshape(1, -1)
y_data = A * (np.sin(omega * t_data) - beta * np.sin(omega_bar * t_data))
# Add noise
# y_noise = 0.15*np.std(y_data)*np.random.randn(NDATA)
t = sn.functionals.RNNVariable(tunits, 't')
y = sn.functionals.RNNFunctional('y',
t, [1],
'sin',
recurrent_activation='sin')
mRNN = sn.SciModel(t, y)
mRNN.train(t_data, y_data, learning_rate=0.001, epochs=10000, batch_size=100)
Lambd11 = np.pi * np.sqrt(2)
u_data = np.sin(np.pi * x_data) * np.sin(np.pi * y_data) * np.cos(
Lambd11 * t_data)
x = sn.Variable('x', dtype='float64')
y = sn.Variable('y', dtype='float64')
t = sn.Variable('t', dtype='float64')
u = sn.Functional('u', [x, y, t], 4 * [20], 'sin')
c = sn.Parameter(np.random.rand(), inputs=[x, y, t], name='c')
L1 = c * (diff(u, x, order=2) + diff(u, y, order=2)) - diff(u, t, order=2)
m = sn.SciModel(
[x, y, t],
[sn.PDE(L1), sn.Data(u)],
# load_weights_from='membrane_inv-weights.hdf5'
)
inputs = [x_data, y_data, t_data]
targets = ['zeros', u_data]
h = m.train(inputs,
targets,
batch_size=BATCH,
learning_rate=0.001,
reduce_lr_after=50,
adaptive_weights={'freq': True},
epochs=EPOCHS,
log_parameters={
'parameters': c,