def _value(self, obs: Tuple[str]):
obs = convert_to_container(value=obs, container=tuple)
values = self.get_iteration()
# TODO(Mayou36): add conversion to right dimension? (n_events, n_obs)? # check if 1-D?
if len(values.shape.as_list()) == 0:
values = tf.expand_dims(values, -1)
if len(values.shape.as_list()) == 1:
values = tf.expand_dims(values, -1)
perm_indices = self.space.axes if self.space.axes != tuple(
range(values.shape[-1])) else False
# permutate = perm_indices is not None
if obs:
if not frozenset(obs) <= frozenset(self.obs):
raise ValueError(
"The observable(s) {} are not contained in the dataset. "
"Only the following are: {}".format(
frozenset(obs) - frozenset(self.obs), self.obs))
perm_indices = self.space.get_axes(obs=obs)
# values = list(values[self.obs.index(o)] for o in obs if o in self.obs)
if perm_indices:
values = ztf.unstack_x(values)
values = list(values[i] for i in perm_indices)
values = ztf.stack_x(values)
# cast data to right type
if not values.dtype == self.dtype:
values = tf.cast(values, dtype=self.dtype)
return values
def func4_3deps(x):
if isinstance(x, np.ndarray):
a, b, c = x
else:
a, b, c = ztf.unstack_x(x)
return a**2 + b**3 + 0.5 * c
def unstack_x(self, obs: ztyping.ObsTypeInput = None):
"""Return the unstacked data: a list of tensors.
Args:
obs (): which observables to return
Returns:
List(tf.Tensor)
"""
return ztf.unstack_x(self._value_internal(obs=obs))
def gradients(self, x: ztyping.XType, norm_range: ztyping.LimitsType, params: ztyping.ParamsTypeOpt = None):
warnings.warn("Taking the gradient *this way* in TensorFlow is inefficient! Consider taking it with"
"respect to the loss function.")
if params is not None:
params = convert_to_container(params)
if params is None or isinstance(params[0], str):
params = self.get_params(only_floating=False, names=params)
probs = self.pdf(x, norm_range=norm_range)
gradients = [tf.gradients(prob, params) for prob in ztf.unstack_x(probs)]
return tf.stack(gradients)
def unstack_x(self,
obs: ztyping.ObsTypeInput = None,
always_list: bool = False):
"""Return the unstacked data: a list of tensors or a single Tensor.
Args:
obs (): which observables to return
always_list (bool): If True, always return a list (also if length 1)
Returns:
List(tf.Tensor)
"""
return ztf.unstack_x(self._value_internal(obs=obs))
def _sort_value(self, value, obs: Tuple[str]):
obs = convert_to_container(value=obs, container=tuple)
perm_indices = self.space.axes if self.space.axes != tuple(
range(value.shape[-1])) else False
# permutate = perm_indices is not None
if obs:
if not frozenset(obs) <= frozenset(self.obs):
raise ValueError(
"The observable(s) {} are not contained in the dataset. "
"Only the following are: {}".format(
frozenset(obs) - frozenset(self.obs), self.obs))
perm_indices = self.space.get_axes(obs=obs)
# values = list(values[self.obs.index(o)] for o in obs if o in self.obs)
if perm_indices:
value = ztf.unstack_x(value)
value = list(value[i] for i in perm_indices)
value = ztf.stack_x(value)
return value
def _func(self, x):
mu = self.params['mu']
sigma = self.params['sigma']
x = ztf.unstack_x(x)
return ztf.exp(-ztf.square((x - mu) / sigma))
def _unnormalized_pdf(self, x):
mu = self.params['mu']
sigma = self.params['sigma']
x = ztf.unstack_x(x)
return ztf.exp(-ztf.square((x - mu) / sigma))
def func2_pure(self, x):
x = ztf.unstack_x(x)
return param2 * x + param3
def func1_pure(self, x):
x = ztf.unstack_x(x)
return param1 * x
def func1_5deps(x):
a, b, c, d, e = ztf.unstack_x(x)
return a + b * c**2 + d**2 * e**3
def func3_2deps(x):
a, b = ztf.unstack_x(x)
return a**2 + b**2
