def for_batch(self,
num_groups: int,
num_timesteps: int,
start_datetimes: Optional[np.ndarray] = None):
for_batch = super().for_batch(num_groups=num_groups,
num_timesteps=num_timesteps)
# determine the delta (integer time accounting for different groups having different start datetimes)
if start_datetimes is None:
start_datetimes = np.zeros(num_groups)
delta = self._dt_helper.make_delta_grid(start_datetimes, num_timesteps)
# determine season:
season = delta % self.seasonal_period
# generate the fourier tensor:
fourier_tens = fourier_tensor(time=Tensor(season),
seasonal_period=self.seasonal_period,
K=self.K)
for state_element in self.state_elements:
if state_element == 'position':
continue
r, c = (int(x) for x in state_element.split(sep=","))
for_batch._adjust_transition(from_element=state_element,
to_element='position',
adjustment=split_flat(
fourier_tens[:, :, r, c], dim=1))
return for_batch
def for_batch(
self,
num_groups: int,
num_timesteps: int,
start_datetimes: Optional[np.ndarray] = None) -> ProcessForBatch:
for_batch = super().for_batch(num_groups=num_groups,
num_timesteps=num_timesteps)
# determine the delta (integer time accounting for different groups having different start datetimes)
delta = self.dt_tracker.get_delta(for_batch.num_groups,
for_batch.num_timesteps,
start_datetimes=start_datetimes)
# determine season:
season = delta % self.seasonal_period
# generate the fourier tensor:
fourier_tens = fourier_tensor(time=Tensor(season),
seasonal_period=self.seasonal_period,
K=self.K)
for state_element in self.state_elements:
if state_element == 'position':
continue
r, c = (int(x) for x in state_element.split(sep=","))
for_batch.adjust_transition(from_element=state_element,
to_element='position',
adjustment=fourier_tens[:, :, r, c])
return for_batch
def for_batch(self,
num_groups: int,
num_timesteps: int,
start_datetimes: Optional[np.ndarray] = None):
for_batch = super().for_batch(num_groups=num_groups,
num_timesteps=num_timesteps)
# determine the delta (integer time accounting for different groups having different start datetimes)
if start_datetimes is None:
if self._dt_helper.start_datetime:
raise TypeError("Missing argument `start_datetimes`.")
start_datetimes = np.zeros(num_groups)
delta = self._dt_helper.make_delta_grid(start_datetimes, num_timesteps)
# determine season:
season = delta % self.seasonal_period
# generate the fourier tensor:
fourier_tens = fourier_tensor(time=Tensor(season),
seasonal_period=self.seasonal_period,
K=self.K)
for measure in self.measures:
for state_element in self.state_elements:
r, c = (int(x) for x in state_element.split(sep=","))
for_batch._adjust_measure(measure=measure,
state_element=state_element,
adjustment=split_flat(
fourier_tens[:, :, r, c], dim=1))
return for_batch
def for_batch(self,
num_groups: int,
num_timesteps: int,
start_datetimes: Optional[np.ndarray] = None):
for_batch = super().for_batch(num_groups=num_groups, num_timesteps=num_timesteps)
# determine the delta (integer time accounting for different groups having different start datetimes)
delta = self._get_delta(for_batch.num_groups, for_batch.num_timesteps, start_datetimes=start_datetimes)
# determine season:
season = delta % self.seasonal_period
# generate the fourier tensor:
fourier_tens = fourier_tensor(time=Tensor(season), seasonal_period=self.seasonal_period, K=self.K)
for measure in self.measures:
for state_element in self.state_elements:
r, c = (int(x) for x in state_element.split(sep=","))
for_batch._adjust_measure(
measure=measure,
state_element=state_element,
adjustment=split_flat(fourier_tens[:, :, r, c], dim=1)
)
return for_batch