def test_init(self):
#test default parameters
opt = SimpleBayesianOptimizer(None)
assert_equal(opt.initial_random_runs, 10)
assert_is_none(opt.acquisition_hyperparams)
assert_equal(opt.num_gp_restarts, 10)
assert_true(isinstance(opt.acquisition_function, ExpectedImprovement))
assert_dict_equal(opt.kernel_params, {})
assert_equal(opt.kernel, "matern52")
assert_false(opt.mcmc)
assert_equal(opt.num_precomputed, 10)
#test correct initialization
opt_arguments = {
"initial_random_runs": 5,
"acquisition_hyperparams": {},
"num_gp_restarts": 5,
"acquisition": ProbabilityOfImprovement,
"kernel_params": {},
"kernel": "matern52",
"mcmc": True,
"num_precomputed": 5
}
opt = SimpleBayesianOptimizer(opt_arguments)
assert_equal(opt.initial_random_runs, 5)
assert_dict_equal(opt.acquisition_hyperparams, {})
assert_equal(opt.num_gp_restarts, 5)
assert_true(isinstance(opt.acquisition_function, ProbabilityOfImprovement))
assert_dict_equal(opt.kernel_params, {})
assert_equal(opt.kernel, "matern52")
if import_if_exists("pymcmc")[0]:
assert_true(opt.mcmc)
else:
assert_false(opt.mcmc)
assert_equal(opt.num_precomputed, 5)
__author__ = 'Frederik Diehl'
from apsis.optimizers.optimizer import Optimizer
from apsis.optimizers.random_search import RandomSearch
from apsis.models.parameter_definition import *
from apsis.utilities.randomization import check_random_state
from apsis.models.candidate import Candidate
from apsis.optimizers.bayesian.acquisition_functions import *
from apsis.utilities.import_utils import import_if_exists
import GPy
import logging
mcmc_imported, pm = import_if_exists("pymcmc")
class SimpleBayesianOptimizer(Optimizer):
"""
This implements a simple bayesian optimizer.
It is simple because it only implements the simplest form - no freeze-thaw,
(currently) no multiple workers, only numeric parameters.
Attributes
----------
SUPPORTED_PARAM_TYPES : list of ParamDefs
The supported parameter types. Currently only numberic and position.
kernel : GPy Kernel
The Kernel to be used with the gp.
acquisition_function : acquisition_function
The acquisition function to use
acquisition_hyperparams :
__author__ = 'Frederik Diehl'
from apsis.optimizers.optimizer import Optimizer
from apsis.optimizers.random_search import RandomSearch
from apsis.models.parameter_definition import *
from apsis.utilities.randomization import check_random_state
from apsis.models.candidate import Candidate
from apsis.optimizers.bayesian.acquisition_functions import *
from apsis.utilities.import_utils import import_if_exists
import GPy
import logging
mcmc_imported, pm = import_if_exists("pymcmc")
class SimpleBayesianOptimizer(Optimizer):
"""
This implements a simple bayesian optimizer.
It is simple because it only implements the simplest form - no freeze-thaw,
(currently) no multiple workers, only numeric parameters.
Attributes
----------
SUPPORTED_PARAM_TYPES : list of ParamDefs
The supported parameter types. Currently only numberic and position.
kernel : GPy Kernel
The Kernel to be used with the gp.
acquisition_function : acquisition_function
The acquisition function to use
acquisition_hyperparams :
The acquisition hyperparameters.