def initialize_objective(self):
self.C, self.c = (
T.variable(T.random_normal([self.ds, self.ds])),
T.variable(T.random_normal([self.ds])),
)
if self.learn_stdev:
self.stdev = T.variable(T.to_float(self.cost_stdev))
else:
self.stdev = T.to_float(self.cost_stdev)
def initialize_objective(self):
H, ds, da = self.horizon, self.ds, self.da
if self.time_varying:
A = T.concatenate([T.eye(ds), T.zeros([ds, da])], -1)
self.A = T.variable(A[None] + 1e-2 * T.random_normal([H - 1, ds, ds + da]))
self.Q_log_diag = T.variable(T.random_normal([H - 1, ds]) + 1)
self.Q = T.matrix_diag(T.exp(self.Q_log_diag))
else:
A = T.concatenate([T.eye(ds), T.zeros([ds, da])], -1)
self.A = T.variable(A + 1e-2 * T.random_normal([ds, ds + da]))
self.Q_log_diag = T.variable(T.random_normal([ds]) + 1)
self.Q = T.matrix_diag(T.exp(self.Q_log_diag))
def _sample(self, num_samples):
sigma, mu = self.natural_to_regular(self.regular_to_natural(self.get_parameters('regular')))
L = T.cholesky(sigma)
sample_shape = T.concat([[num_samples], T.shape(mu)], 0)
noise = T.random_normal(sample_shape)
L = T.tile(L[None], T.concat([[num_samples], T.ones([T.rank(sigma)], dtype=np.int32)]))
return mu[None] + T.matmul(L, noise[..., None])[..., 0]
def initialize_node(node, children):
if isinstance(node, Gaussian):
d = T.shape(node)
return Gaussian([T.eye(d[-1], batch_shape=d[:-1]), T.random_normal(d)])
elif isinstance(node, IW):
d = T.shape(node)
return IW([(T.to_float(d[-1]) + 1) * T.eye(d[-1], batch_shape=d[:-2]),
T.to_float(d[-1]) + 1])
def initialize_objective(self):
H, ds, da = self.horizon, self.ds, self.da
if self.time_varying:
A = T.concatenate(
[T.eye(ds, batch_shape=[H - 1]),
T.zeros([H - 1, ds, da])], -1)
self.A_prior = stats.MNIW([
2 * T.eye(ds, batch_shape=[H - 1]), A,
T.eye(ds + da, batch_shape=[H - 1]),
T.to_float(ds + 2) * T.ones([H - 1])
],
parameter_type='regular')
self.A_variational = stats.MNIW(list(
map(
T.variable,
stats.MNIW.regular_to_natural([
2 * T.eye(ds, batch_shape=[H - 1]),
A + 1e-2 * T.random_normal([H - 1, ds, ds + da]),
T.eye(ds + da, batch_shape=[H - 1]),
T.to_float(ds + 2) * T.ones([H - 1])
]))),
parameter_type='natural')
else:
A = T.concatenate([T.eye(ds), T.zeros([ds, da])], -1)
self.A_prior = stats.MNIW(
[2 * T.eye(ds), A,
T.eye(ds + da),
T.to_float(ds + 2)],
parameter_type='regular')
self.A_variational = stats.MNIW(list(
map(
T.variable,
stats.MNIW.regular_to_natural([
2 * T.eye(ds),
A + 1e-2 * T.random_normal([ds, ds + da]),
T.eye(ds + da),
T.to_float(ds + 2)
]))),
parameter_type='natural')
def glorot_normal(shape):
fan_in, fan_out = get_fans(shape)
s = np.sqrt(0.1 / (fan_in + fan_out))
return T.random_normal(shape, s)
def normal(shape, scale=0.05):
return T.random_normal(shape, mean=0.0, stddev=scale)