directminimize = None
try:
import cma
except: # pragma: no cover
cma = None
import sobol_seq
from bayeso import acquisition
from bayeso import constants
from bayeso.gp import gp
from bayeso.gp import gp_common
from bayeso.utils import utils_common
from bayeso.utils import utils_covariance
from bayeso.utils import utils_logger
logger = utils_logger.get_logger('bo')
def get_grids(arr_ranges, int_grids):
"""
It returns grids of given `arr_ranges`, where each of dimension has `int_grids` partitions.
:param arr_ranges: ranges. Shape: (d, 2).
:type arr_ranges: numpy.ndarray
:param int_grids: the number of partitions per dimension.
:type int_grids: int.
:returns: grids of given `arr_ranges`. Shape: (`int_grids`:math:`^{\\text{d}}`, d).
:rtype: numpy.ndarray
:raises: AssertionError
# gp_common # author: Jungtaek Kim ([email protected]) # last updated: August 07, 2020 import numpy as np import scipy.linalg from bayeso import covariance from bayeso import constants from bayeso.utils import utils_gp from bayeso.utils import utils_logger logger = utils_logger.get_logger('gp_common') def get_kernel_inverse(X_train, hyps, str_cov, is_fixed_noise=constants.IS_FIXED_GP_NOISE, is_gradient=False, debug=False): """ This function computes a kernel inverse without any matrix decomposition techniques. :param X_train: inputs. Shape: (n, d) or (n, m, d). :type X_train: numpy.ndarray :param hyps: dictionary of hyperparameters for Gaussian process. :type hyps: dict. :param str_cov: the name of covariance function. :type str_cov: str. :param is_fixed_noise: flag for fixing a noise.
import os
import numpy as np
try:
import matplotlib.pyplot as plt
except:
plt = None
try:
import pylab
except:
pylab = None
from bayeso.utils import utils_common
from bayeso.utils import utils_logger
from bayeso import constants
logger = utils_logger.get_logger('utils_plotting')
@utils_common.validate_types
def _set_font_config(use_tex: bool) -> constants.TYPE_NONE: # pragma: no cover
"""
It sets a font configuration.
:param use_tex: flag for using latex.
:type use_tex: bool.
:returns: None.
:rtype: NoneType
"""
"""It defines the functions related to kernels for
Gaussian process regression."""
import time
import numpy as np
import scipy.optimize
from bayeso import covariance
from bayeso import constants
from bayeso.gp import gp_likelihood
from bayeso.utils import utils_gp
from bayeso.utils import utils_covariance
from bayeso.utils import utils_common
from bayeso.utils import utils_logger
logger = utils_logger.get_logger('gp_kernel')
@utils_common.validate_types
def get_optimized_kernel(
X_train: np.ndarray,
Y_train: np.ndarray,
prior_mu: constants.TYPING_UNION_CALLABLE_NONE,
str_cov: str,
str_optimizer_method: str = constants.STR_OPTIMIZER_METHOD_GP,
str_modelselection_method: str = constants.STR_MODELSELECTION_METHOD,
use_ard: bool = constants.USE_ARD,
fix_noise: bool = constants.FIX_GP_NOISE,
debug: bool = False) -> constants.TYPING_TUPLE_TWO_ARRAYS_DICT:
"""
This function computes the kernel matrix optimized by optimization
try:
from scipydirect import minimize as directminimize
except: # pragma: no cover
directminimize = None
try:
import cma
except: # pragma: no cover
cma = None
from bayeso import acquisition
from bayeso import constants
from bayeso.utils import utils_covariance
from bayeso.utils import utils_common
from bayeso.utils import utils_logger
logger = utils_logger.get_logger('utils_bo')
@utils_common.validate_types
def normalize_min_max(Y: np.ndarray) -> np.ndarray:
"""
It normalizes `Y` by min-max normalization.
:param Y: responses. Shape: (n, 1).
:type Y: numpy.ndarray
:returns: normalized responses. Shape: (n, 1).
:rtype: numpy.ndarray
:raises: AssertionError
# gp_gpytorch # author: Jungtaek Kim ([email protected]) # last updated: August 20, 2020 import time import numpy as np import torch import gpytorch from bayeso import constants from bayeso.gp import gp_common from bayeso.utils import utils_gp from bayeso.utils import utils_logger logger = utils_logger.get_logger('gp_gpytorch') class ExactGPModel(gpytorch.models.ExactGP): def __init__(self, str_cov, prior_mu, X_train, Y_train, likelihood): super(ExactGPModel, self).__init__(X_train, Y_train, likelihood) self.dim_X = X_train.shape[1] if prior_mu is None: self.mean_module = gpytorch.means.ConstantMean() else: raise NotImplementedError() if str_cov == 'eq' or str_cov == 'se': self.covar_module = gpytorch.kernels.ScaleKernel(gpytorch.kernels.RBFKernel(ard_num_dims=self.dim_X)) elif str_cov == 'matern32':
# gp # author: Jungtaek Kim ([email protected]) # last updated: August 21, 2020 import time import numpy as np import scipy.stats from bayeso import covariance from bayeso import constants from bayeso.gp import gp_common from bayeso.gp import gp_scipy from bayeso.utils import utils_gp from bayeso.utils import utils_logger logger = utils_logger.get_logger('gp') def sample_functions(mu, Sigma, num_samples=1): """ It samples `num_samples` functions from multivariate Gaussian distribution (mu, Sigma). :param mu: mean vector. Shape: (n, ). :type mu: numpy.ndarray :param Sigma: covariance matrix. Shape: (n, n). :type Sigma: numpy.ndarray :param num_samples: the number of sampled functions :type num_samples: int., optional :returns: sampled functions. Shape: (num_samples, n). :rtype: numpy.ndarray
#
"""It defines Student-:math:`t` process regression."""
import time
import numpy as np
import scipy.stats
from bayeso import covariance
from bayeso import constants
from bayeso.tp import tp_kernel
from bayeso.utils import utils_gp
from bayeso.utils import utils_covariance
from bayeso.utils import utils_common
from bayeso.utils import utils_logger
logger = utils_logger.get_logger('tp')
@utils_common.validate_types
def sample_functions(nu: float,
mu: np.ndarray,
Sigma: np.ndarray,
num_samples: int = 1) -> np.ndarray:
"""
It samples `num_samples` functions from multivariate Student-$t$ distribution (nu, mu, Sigma).
:param mu: mean vector. Shape: (n, ).
:type mu: numpy.ndarray
:param Sigma: covariance matrix. Shape: (n, n).
:type Sigma: numpy.ndarray
:param num_samples: the number of sampled functions
# last updated: August 10, 2020
import os
import time
import numpy as np
import tensorflow as tf
import tensorflow_probability as tfp
from bayeso import covariance
from bayeso import constants
from bayeso.gp import gp_common
from bayeso.utils import utils_gp
from bayeso.utils import utils_covariance
from bayeso.utils import utils_logger
logger = utils_logger.get_logger('gp_tensorflow')
gpus = tf.config.experimental.list_physical_devices('GPU')
if gpus: # pragma: no cover
try:
for gpu in gpus:
tf.config.experimental.set_memory_growth(gpu, True)
except RuntimeError as e:
print(e)
def get_optimized_kernel(X_train,
Y_train,
prior_mu,
str_cov,
is_fixed_noise=constants.IS_FIXED_GP_NOISE,
num_iters=1000,
# gp_scipy # author: Jungtaek Kim ([email protected]) # last updated: August 07, 2020 import time import numpy as np import scipy.linalg import scipy.optimize from bayeso import constants from bayeso.gp import gp_common from bayeso.utils import utils_gp from bayeso.utils import utils_covariance from bayeso.utils import utils_logger logger = utils_logger.get_logger('gp_scipy') def neg_log_ml(X_train, Y_train, hyps, str_cov, prior_mu_train, is_fixed_noise=constants.IS_FIXED_GP_NOISE, is_cholesky=True, is_gradient=True, debug=False): """ This function computes a negative log marginal likelihood. :param X_train: inputs. Shape: (n, d) or (n, m, d).