def test_bayesopt_batch(parameters, results):
gpyopt = GPyOpt(max_concurrent=10)
gpyopt.domain = gpyopt._initialize_domain(parameters)
gpyopt.lower_is_better = True
X, y, y_var = GPyOpt._prepare_data_for_bayes_opt(parameters, results)
domain = gpyopt._initialize_domain(parameters)
batch = gpyopt._generate_bayesopt_batch(X,
y,
lower_is_better=True,
domain=domain)
assert batch.shape == (10, 4)
def test_bayesopt_batch(parameters, results, transforms):
gpyopt = GPyOpt(max_concurrent=10)
domain = gpyopt._initialize_domain(parameters, transforms)
X, y = GPyOpt._prepare_data_for_bayes_opt(parameters, results, transforms)
batch = gpyopt._generate_bayesopt_batch(domain, X, y, lower_is_better=True)
assert batch.shape == (10, 5)
def test_transformation_to_gpyopt_domain_log_continuous(parameters):
domain = GPyOpt._initialize_domain(parameters)
for p, d in zip(parameters, domain):
assert d['name'] == p.name
assert d['type'] == 'continuous'
assert d['domain'] == tuple(
[numpy.log10(p.range[0]),
numpy.log10(p.range[1])])
def test_transformation_to_gpyopt_domain_with_multiple_parameters(parameters):
domain = GPyOpt._initialize_domain(parameters)
assert {
'name': 'dropout',
'type': 'continuous',
'domain': (0., 0.5)
} == domain[0]
assert {
'name': 'lr',
'type': 'continuous',
'domain': (-7, -1)
} == domain[1]
assert domain[2]['name'] == 'activation'
assert domain[2]['type'] == 'categorical'
assert numpy.array_equal(domain[2]['domain'], numpy.array([0, 1, 2]))
assert {
'name': 'num_hidden',
'type': 'discrete',
'domain': tuple(range(100, 301))
} == domain[3]
def test_domain(parameters, transforms):
domain = GPyOpt._initialize_domain(parameters, transforms)
assert {
'name': 'dropout',
'type': 'continuous',
'domain': (0., 0.5)
} in domain
assert {'name': 'lr', 'type': 'continuous', 'domain': (-7, -1)} in domain
assert {
'name': 'activation',
'type': 'discrete',
'domain': (0, 1, 2)
} in domain
assert {
'name': 'num_hidden',
'type': 'continuous',
'domain': (100, 300)
} in domain
assert {
'name': 'batch_size',
'type': 'continuous',
'domain': (0, 3)
} in domain
def test_transformation_to_gpyopt_domain_log_discrete():
parameters = [sherpa.Discrete('a', [1, 100], 'log')]
with pytest.warns(UserWarning, match='does not support log-scale'):
GPyOpt._initialize_domain(parameters)
def test_transformation_to_gpyopt_domain_discrete(parameters):
domain = GPyOpt._initialize_domain(parameters)
for p, d in zip(parameters, domain):
assert d['name'] == p.name
assert d['type'] == 'discrete'
assert d['domain'] == tuple(range(p.range[0], p.range[1] + 1))
