# SPDX-License-Identifier: MPL-2.0
# Copyright (C) 2020- The University of Tokyo
#
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at https://mozilla.org/MPL/2.0/.
import numpy as np
import configparser
[ドキュメント]
class set_config:
def __init__(self, search_config=None, learning_config=None):
"""
Setting configuration for search and learning.
Parameters
----------
search_config: physbo.misc.search object
learning_config: physbo.misc.learning object
"""
if search_config is None:
search_config = search()
self.search = search_config
if learning_config is None:
learning_config = adam()
self.learning = learning_config
[ドキュメント]
def show(self):
"""
Showing information of search and learning objects.
Returns
-------
"""
self.search.show()
self.learning.show()
[ドキュメント]
def load(self, file_name="config.ini"):
"""
Loading information of configuration.
Parameters
----------
file_name: str
An input file name of configuration.
Returns
-------
"""
config = configparser.SafeConfigParser()
config.read(file_name)
search_config = search()
self.search = search_config
self.search.load(config)
temp_dict = config._sections["learning"]
method = temp_dict.get("method", "adam")
if method == "adam":
learning_config = adam()
self.learning = learning_config
self.learning.load(config)
if method in ("bfgs", "batch"):
learning_config = batch()
self.learning = learning_config
self.learning.load(config)
[ドキュメント]
class search:
def __init__(self):
self.multi_probe_num_sampling = 20
self.alpha = 1.0
[ドキュメント]
def load(self, config):
"""
Loading information of configuration from config._sectoins['search'].
Parameters
----------
config: physbo.misc.set_config object
Returns
-------
"""
temp_dict = config._sections["search"]
self.multi_probe_num_sampling = int(
temp_dict.get("multi_probe_num_sampling", 20)
)
self.alpha = np.float64(temp_dict.get("alpha", 1.0))
[ドキュメント]
def show(self):
"""
Showing information about search object.
Returns
-------
"""
print("(search)")
print("multi_probe_num_sampling: ", self.multi_probe_num_sampling)
print("alpha: ", self.alpha)
print("\n")
[ドキュメント]
class learning(object):
def __init__(self):
self.is_disp = True
self.num_disp = 10
self.num_init_params_search = 20
self.method = "adam"
[ドキュメント]
def show(self):
"""
Showing information about learning object.
Returns
-------
"""
print("( learning )")
print("method : ", self.method)
print("is_disp: ", self.is_disp)
print("num_disp: ", self.num_disp)
print("num_init_params_search: ", self.num_init_params_search)
[ドキュメント]
def load(self, config):
"""
Loading information of configuration from config._sectoins['learning'].
Parameters
----------
config: physbo.misc.set_config object
Returns
-------
"""
temp_dict = config._sections["learning"]
self.method = temp_dict.get("method", "adam")
self.is_disp = boolean(temp_dict.get("is_disp", True))
self.num_disp = int(temp_dict.get("num_disp", 10))
self.num_init_params_search = int(temp_dict.get("num_init_params_search", 20))
[ドキュメント]
class batch(learning):
def __init__(self):
super(batch, self).__init__()
self.method = "bfgs"
self.max_iter = 200
self.max_iter_init_params_search = 20
self.batch_size = 5000
[ドキュメント]
def show(self):
"""
Showing information about configuration about batch object.
Returns
-------
"""
super(batch, self).show()
print("max_iter: ", self.max_iter)
print("max_iter_init_params_search: ", self.max_iter_init_params_search)
print("batch_size: ", self.batch_size)
[ドキュメント]
def load(self, config):
"""
Loading information of configuration from config._sectoins['batch'].
Parameters
----------
config: physbo.misc.set_config object
Returns
-------
"""
super(batch, self).load(config)
temp_dict = config._sections["batch"]
self.max_iter = int(temp_dict.get("max_iter", 200))
self.max_iter_init_params_search = int(
temp_dict.get("max_iter_init_params_search", 20)
)
self.batch_size = int(temp_dict.get("batch_size", 5000))
[ドキュメント]
class online(learning):
def __init__(self):
super(online, self).__init__()
self.max_epoch = 500
self.max_epoch_init_params_search = 50
self.batch_size = 64
self.eval_size = 5000
[ドキュメント]
def show(self):
"""
Showing information about configuration about online object.
Returns
-------
"""
super(online, self).show()
print("max_epoch: ", self.max_epoch)
print("max_epoch_init_params_search: ", self.max_epoch_init_params_search)
print("batch_size: ", self.batch_size)
print("eval_size: ", self.eval_size)
[ドキュメント]
def load(self, config):
"""
Loading information of configuration from config._sectoins['online'].
Parameters
----------
config: physbo.misc.set_config object
Returns
-------
"""
super(online, self).load(config)
temp_dict = config._sections["online"]
self.max_epoch = int(temp_dict.get("max_epoch", 1000))
self.max_epoch_init_params_search = int(
temp_dict.get("max_epoch_init_params_search", 50)
)
self.batch_size = int(temp_dict.get("batch_size", 64))
self.eval_size = int(temp_dict.get("eval_size", 5000))
[ドキュメント]
class adam(online):
def __init__(self):
super(adam, self).__init__()
self.method = "adam"
self.alpha = 0.001
self.beta = 0.9
self.gamma = 0.999
self.epsilon = 1e-6
[ドキュメント]
def show(self):
"""
Showing information about configuration about adam object.
Returns
-------
"""
super(adam, self).show()
print("alpha = ", self.alpha)
print("beta = ", self.beta)
print("gamma = ", self.gamma)
print("epsilon = ", self.epsilon)
print("\n")
[ドキュメント]
def load(self, config):
"""
Loading information of configuration from config._sectoins['adam'].
Parameters
----------
config: physbo.misc.set_config object
Returns
-------
"""
super(adam, self).load(config)
temp_dict = config._sections["adam"]
self.alpha = np.float64(temp_dict.get("alpha", 0.001))
self.beta = np.float64(temp_dict.get("beta", 0.9))
self.gamma = np.float64(temp_dict.get("gamma", 0.9999))
self.epsilon = np.float64(temp_dict.get("epsilon", 1e-6))
[ドキュメント]
def boolean(str):
"""
Return boolean.
Parameters
----------
str: str or boolean
Returns
-------
True or False
"""
if str == "True" or str is True:
return True
else:
return False