def test_ihs_hyperparams_setter():
new_ihs = ihs.IHS()
try:
new_ihs.PAR_min = 'a'
except:
new_ihs.PAR_min = 0.5
try:
new_ihs.PAR_min = -1
except:
new_ihs.PAR_min = 0.5
assert new_ihs.PAR_min == 0.5
try:
new_ihs.PAR_max = 'b'
except:
new_ihs.PAR_max = 1.0
try:
new_ihs.PAR_max = -1
except:
new_ihs.PAR_max = 1.0
try:
new_ihs.PAR_max = 0
except:
new_ihs.PAR_max = 1.0
assert new_ihs.PAR_max == 1.0
try:
new_ihs.bw_min = 'c'
except:
new_ihs.bw_min = 1.0
try:
new_ihs.bw_min = -1
except:
new_ihs.bw_min = 1.0
assert new_ihs.bw_min == 1.0
try:
new_ihs.bw_max = 'd'
except:
new_ihs.bw_max = 10.0
try:
new_ihs.bw_max = -1
except:
new_ihs.bw_max = 10.0
try:
new_ihs.bw_max = 0
except:
new_ihs.bw_max = 10.0
assert new_ihs.bw_max == 10.0
def test_ihs_hyperparams():
hyperparams = {
'PAR_min': 0.5,
'PAR_max': 1,
'bw_min': 2,
'bw_max': 5
}
new_ihs = ihs.IHS(hyperparams=hyperparams)
assert new_ihs.PAR_min == 0.5
assert new_ihs.PAR_max == 1
assert new_ihs.bw_min == 2
assert new_ihs.bw_max == 5
def test_ihs_hyperparams_setter():
new_ihs = ihs.IHS()
new_ihs.HMCR = 0.7
assert new_ihs.HMCR == 0.7
new_ihs.PAR_min = 0.1
assert new_ihs.PAR_min == 0.1
new_ihs.PAR_max = 0.5
assert new_ihs.PAR_max == 0.5
new_ihs.bw_min = 1
assert new_ihs.bw_min == 1
new_ihs.bw_max = 10
assert new_ihs.bw_max == 10
def test_ihs_run():
def square(x):
return np.sum(x**2)
new_function = function.Function(pointer=square)
new_ihs = ihs.IHS()
search_space = search.SearchSpace(n_agents=2,
n_iterations=10,
n_variables=2,
lower_bound=[0, 0],
upper_bound=[10, 10])
history = new_ihs.run(search_space, new_function)
assert len(history.agents) > 0
assert len(history.best_agent) > 0
def test_ihs_run():
def square(x):
return np.sum(x**2)
def hook(optimizer, space, function):
return
new_function = function.Function(pointer=square)
new_ihs = ihs.IHS()
search_space = search.SearchSpace(n_agents=20, n_iterations=50,
n_variables=2, lower_bound=[0, 0],
upper_bound=[5, 5])
history = new_ihs.run(search_space, new_function, pre_evaluation_hook=hook)
assert len(history.agents) > 0
assert len(history.best_agent) > 0
best_fitness = history.best_agent[-1][1]
assert best_fitness <= constants.TEST_EPSILON, 'The algorithm ihs failed to converge.'
def test_ihs_rebuild():
new_ihs = ihs.IHS()
assert new_ihs.built == True