def test_define_prior_parameters(self):
data = {
'points': np.array([[1]]),
'evaluations': np.array([1]),
'var_noise': None,
}
dimension = 1
result = Matern52.define_prior_parameters(data, dimension)
assert result == {
LENGTH_SCALE_NAME: [0.0],
}
data2 = {
'points': np.array([[1], [2]]),
'evaluations': np.array([1, 2]),
'var_noise': None,
}
dimension2 = 1
result2 = Matern52.define_prior_parameters(data2, dimension2)
assert result2 == {
LENGTH_SCALE_NAME: [1.5432098765432098],
}
def define_prior_parameters_using_data(data,
type_kernel,
dimensions,
sigma2=None,
**kernel_parameters):
"""
Defines value of the parameters of the prior distributions of the kernel's parameters.
:param data: {'points': np.array(nxm), 'evaluations': np.array(n),
'var_noise': np.array(n) or None}
:param type_kernel: [str]
:param dimensions: [int], It has only the n_tasks for the task_kernels, and for the
PRODUCT_KERNELS_SEPARABLE contains the dimensions of every kernel in the product, and
the total dimension of the product_kernels_separable too in the first entry.
:param: sigma2: float
:param kernel_parameters: additional kernel parameters,
- SAME_CORRELATION: (boolean) True or False. Parameter used only for task kernel.
:return: {
SIGMA2_NAME: float,
LENGTH_SCALE_NAME: [float],
LOWER_TRIANG_NAME: [float],
}
"""
# We assume that there is at most one task kernel, and mattern52 kernel in the product.
parameters_priors = {
SIGMA2_NAME: None,
LENGTH_SCALE_NAME: None,
LOWER_TRIANG_NAME: None,
}
index = 1
if TASKS_KERNEL_NAME in type_kernel:
same_correlation = kernel_parameters.get(SAME_CORRELATION, False)
index = type_kernel.index(TASKS_KERNEL_NAME)
index_tasks = 0
for i in xrange(1, index):
index_tasks += dimensions[i]
n_tasks = dimensions[index]
if n_tasks > 1:
tasks_index = data['points'][:, index_tasks]
task_data = data.copy()
task_data['points'] = tasks_index.reshape((len(tasks_index), 1))
else:
task_data = None
task_parameters = TasksKernel.define_prior_parameters(
task_data,
n_tasks,
same_correlation=same_correlation,
var_evaluations=sigma2)
parameters_priors[LOWER_TRIANG_NAME] = task_parameters[
LOWER_TRIANG_NAME]
if MATERN52_NAME in type_kernel:
m = data['points'].shape[1]
indexes = [i for i in range(m) if i != m - index + 1]
points_matern = data['points'][:, indexes]
matern_data = data.copy()
matern_data['points'] = points_matern
matern52_parameters = Matern52.define_prior_parameters(
matern_data, len(indexes))
parameters_priors[LENGTH_SCALE_NAME] = matern52_parameters[
LENGTH_SCALE_NAME]
if SCALED_KERNEL in type_kernel:
parameters = \
ScaledKernel.define_prior_parameters(data, len(indexes), var_evaluations=sigma2)
parameters_priors[SIGMA2_NAME] = parameters[SIGMA2_NAME]
return parameters_priors