def test_sampling_no_replacement_single(self):
elements = ['a']
weights = [33.3]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
element, weight = s.sample(replace=False)
assert element == 'a' and weight == 33.3 and len(s) == 0
def test_python_generator_empty():
elements = range(1, 10)
weights = range(1, 10)
s = SamplableSet(1, 10, zip(elements, weights))
s.clear()
expected = []
assert list(expected) == list(s.__iter__())
def test_sampling_no_replacement_generator_error(self):
elements = ['a', 'b']
weights = [33.3, 50.]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
sample_list = []
with pytest.raises(KeyError):
for sample in s.sample(n_samples=3, replace=False):
sample_list.append(sample)
def test_sampling_no_replacement_generator(self):
elements = ['a', 'b']
weights = [33.3, 50.]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
sample_list = []
for sample in s.sample(n_samples=2, replace=False):
sample_list.append(sample)
assert sample_list[0][0] in elements and sample_list[1][0] in elements
assert len(s) == 0
def test_sampling_no_replacement_generator(self):
elements = ['a']
weights = [33.3]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
sample_list = []
for sample in s.sample(n_samples=5, replace=False):
sample_list.append(sample)
assert sample_list == [('a', 33.3), None, None, None, None
] and len(s) == 0
def test_sampling_generator(self):
elements = ['a']
weights = [33.3]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
element_list = []
weight_list = []
for element, weight in s.sample(n_samples=5):
element_list.append(element)
weight_list.append(weight)
assert element_list == ['a'] * 5 and weight_list == [33.3] * 5 and len(
s) == 1
def test_next_error(self):
elements = {1: 10, 2: 50}
s = SamplableSet(1, 100, elements)
s.init_iterator()
s.next()
with pytest.raises(StopIteration):
s.next()
def test_clear(self):
elements = ['a']
weights = [33.3]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
s.clear()
assert s.total_weight() == 0 and len(s) == 0 and s.empty()
with pytest.raises(KeyError):
s.sample()
def test_next_get_None(self):
elements = {1: 10, 2: 50}
s = SamplableSet(1, 100, elements)
s.init_iterator()
s.next()
s.next()
assert s.get_at_iterator() is None
def test_next_1(self):
#elements in same group
elements = {1: 10, 2: 10}
s = SamplableSet(1, 100, elements)
s.init_iterator()
s.next()
assert s.get_at_iterator() == (2, 10.)
def test_next_2(self):
#elements in different groups
elements = {1: 10, 2: 50}
s = SamplableSet(1, 100, elements)
s.init_iterator()
s.next()
assert s.get_at_iterator() == (2, 50.)
def test_clear(self):
elements = ['a']
weights = [33.3]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
s.clear()
assert s.total_weight() == 0 and len(s) == 0 and s.sample() is None
def test_throw_error_2(self):
with pytest.raises(ValueError):
s = SamplableSet(1, np.inf)
def test_throw_error_1(self):
with pytest.raises(ValueError):
s = SamplableSet(0, 100)
def test_iterable_init(self):
elements = ['a', 'b']
weights = [33.3, 66.6]
elements_weights = zip(elements, weights)
s = SamplableSet(1, 100, elements_weights)
assert 'a' in s and 'b' in s
def test_dict_init(self):
elements_weights = {3: 33.3, 6: 66.6}
s = SamplableSet(1, 100, elements_weights)
assert 3 in s and 6 in s
def test_empty_init_1(self):
s = SamplableSet(1, 100)
assert s.cpp_type is None
s['a'] = 2.
assert s.cpp_type == 'str'
assert len(s) == 1 and s['a'] == 2.
def test_get_weight(self):
s = SamplableSet(1, 10)
s['a'] = 2.
assert s['a'] == 2.
def test_throw_error_4(self):
with pytest.raises(RuntimeError):
s = SamplableSet(1, 2)
s.size()
import numpy as np
from SamplableSet import SamplableSet
from time import time
N = int(2e4)
n = np.arange(N)
np.random.shuffle(n)
weights = np.random.random(N) + 0.1
_min, _max = np.amin(weights), np.amax(weights)
weighted_items = [(int(_n), float(_w)) for _n, _w in zip(n, weights)]
print("#========= CREATING ========")
start = time()
smpset = SamplableSet(0.1, 1.1, weighted_items)
end = time()
print("Orig:", end - start, "seconds")
start = time()
mckset = MockSamplableSet(0.1, 1.1, weighted_items)
end = time()
print("Mock:", end - start, "seconds")
print("#========= ADDING ========")
n = np.arange(N) + N
np.random.shuffle(n)
weights = np.random.random(N) + 0.1
def test_throw_error_5(self):
with pytest.raises(RuntimeError):
s = SamplableSet(1, 2)
s.total_weight()
def test_insert(self):
s = SamplableSet(1, 10)
s['a'] = 2.
assert len(s) == 1 and not s.empty()
def test_set_weight(self):
s = SamplableSet(1, 10)
s['a'] = 2.
s['a'] = 3.
assert s['a'] == 3. and len(s) == 1 and s.total_weight() == 3.
def test_get_weight_no_item(self):
s = SamplableSet(1, 10)
s['a'] = 2.
with pytest.raises(KeyError):
s['b']
def test_empty_init_2(self):
s = SamplableSet(1, 100, {})
assert s.cpp_type is None
s[1] = 5.
assert s.cpp_type == 'int'
assert len(s) == 1 and s[1] == 5.
def test_weight_out_of_bound_3(self):
with pytest.raises(ValueError):
s = SamplableSet(1, 10)
s['a'] = 2.
s['a'] = 11
def test_erase(self):
s = SamplableSet(1, 10)
s['a'] = 2.
del s['a']
assert len(s) == 0 and s.empty()
choice_time_list = []
for N in Nlist:
weights = np.arange(1, N + 1, dtype=np.float64)
weights /= np.sum(weights)
elements = list(range(1, N + 1))
#test random choice --- each time it is as if the distribution was new
t1 = time()
for i in range(5 * 10**4):
np.random.choice(elements, p=weights)
t2 = time()
choice_time_list.append(t2 - t1)
#test SamplableSet
t1 = time()
s = SamplableSet(min(weights), max(weights), zip(elements, weights))
for i in range(5 * 10**4):
e, w = s.sample()
s[e] = w * 1.
t2 = time()
ss_time_list.append(t2 - t1)
#Plot results
font_size = 8
plt.style.use('seaborn-paper')
plt.rc('text', usetex=True)
plt.rc('font', family='serif', serif='Computer Modern')
plt.rc('xtick', labelsize=font_size)
plt.rc('ytick', labelsize=font_size)
plt.rc('axes', labelsize=font_size)
plt.rc('legend', fontsize=font_size - 1)
def __init__(self, seed=None):
"""Empty constructor."""
if not seed is None:
sset.seed(seed)