def update_L_alpha(self):
"""
Update the GP's L matrix and alpha vector without recalculating
the entire covariance matrix K.
"""
# Set L matrix and alpha if set_L_alpha has not been called yet
if self.l_mat is None or np.array(self.ky_mat) is np.array(None):
self.set_L_alpha()
return
_global_training_data[self.name] = self.training_data
_global_training_labels[self.name] = self.training_labels_np
ky_mat = get_ky_mat_update(self.ky_mat, self.hyps,
self.name,
self.kernel,
cutoffs=self.cutoffs,
hyps_mask=self.hyps_mask,
n_cpus=self.n_cpus,
n_sample=self.n_sample)
l_mat = np.linalg.cholesky(ky_mat)
l_mat_inv = np.linalg.inv(l_mat)
ky_mat_inv = l_mat_inv.T @ l_mat_inv
alpha = np.matmul(ky_mat_inv, self.training_labels_np)
self.ky_mat = ky_mat
self.l_mat = l_mat
self.alpha = alpha
self.ky_mat_inv = ky_mat_inv
def update_L_alpha(self):
"""
Update the GP's L matrix and alpha vector without recalculating
the entire covariance matrix K.
"""
# Set L matrix and alpha if set_L_alpha has not been called yet
if self.l_mat is None or np.array(self.ky_mat) is np.array(None):
self.set_L_alpha()
return
# Reset global variables.
self.sync_data()
ky_mat = get_ky_mat_update(self.ky_mat, self.n_envs_prev, self.hyps,
self.name, self.kernel, self.energy_kernel,
self.energy_force_kernel, self.energy_noise,
cutoffs=self.cutoffs,
hyps_mask=self.hyps_mask,
n_cpus=self.n_cpus,
n_sample=self.n_sample)
l_mat = np.linalg.cholesky(ky_mat)
l_mat_inv = np.linalg.inv(l_mat)
ky_mat_inv = l_mat_inv.T @ l_mat_inv
alpha = np.matmul(ky_mat_inv, self.all_labels)
self.ky_mat = ky_mat
self.l_mat = l_mat
self.alpha = alpha
self.ky_mat_inv = ky_mat_inv
self.n_envs_prev = len(self.training_data)
def update_L_alpha(self, expert_id: int):
"""
Update the GP's L matrix and alpha vector without recalculating
the entire covariance matrix K.
"""
# Set L matrix and alpha if set_L_alpha has not been called yet
if self.l_mat[expert_id] is None or np.array(
self.ky_mat[expert_id]) is np.array(None):
self.set_L_alpha_part(expert_id)
return
self.sync_experts_data(expert_id)
ky_mat = get_ky_mat_update(
self.ky_mat[expert_id],
self.n_envs_prev[expert_id],
self.hyps,
f"{self.name}_{expert_id}",
self.kernel,
self.energy_kernel,
self.energy_force_kernel,
self.energy_noise,
cutoffs=self.cutoffs,
hyps_mask=self.hyps_mask,
n_cpus=self.n_cpus,
n_sample=self.n_sample,
)
self.compute_experts_matrices(ky_mat, expert_id)
def update_L_alpha(self):
"""
Update the GP's L matrix and alpha vector without recalculating
the entire covariance matrix K.
"""
# Set L matrix and alpha if set_L_alpha has not been called yet
if self.l_mat is None:
self.set_L_alpha()
return
ky_mat = get_ky_mat_update(np.copy(self.ky_mat), self.training_data,
self.get_kernel_vector, self.hyps,
self.no_cpus)
l_mat = np.linalg.cholesky(ky_mat)
l_mat_inv = np.linalg.inv(l_mat)
ky_mat_inv = l_mat_inv.T @ l_mat_inv
alpha = np.matmul(ky_mat_inv, self.training_labels_np)
self.ky_mat = ky_mat
self.l_mat = l_mat
self.alpha = alpha
self.ky_mat_inv = ky_mat_inv
self.l_mat_inv = l_mat_inv