def __iter__(self):
large_set = choice_fast(self.universe_size, self.large_set_size,
self.random_state)
small_set = choice_fast(large_set, self.small_set_size,
self.random_state)
self.union_ids.update(set(large_set))
set_ids_list = ([large_set] * self.num_large_sets +
[small_set] * self.num_small_sets)
for i in self.set_indices:
yield set_ids_list[i]
return self
def __iter__(self):
for set_size in self.set_sizes:
set_ids = choice_fast(self.universe_size, set_size,
self.random_state)
self.union_ids = self.union_ids.union(set_ids)
yield set_ids
return self
def _select_ids(lb, ub, size):
lb = int(lb)
ub = int(ub)
candidate_ids = np.arange(lb, ub)
if size >= ub - lb:
return candidate_ids
return choice_fast(candidate_ids, size, self.random_state)
def test_choice_fast_choose_elements_from_list(self):
for i in range(50, 500):
# Get a random list of numbers from 0 to 5000 size i
elements = np.random.randint(0, 5000, i)
# Choose up to i elements from that list
chosen = random.choice_fast(elements, np.random.randint(1, i))
# Make sure chosen elements are actually from our original elements.
for element in chosen:
self.assertTrue(element in elements)
def __iter__(self):
for set_size, pmf in zip(self.set_sizes, self.pmf_list):
set_ids = choice_fast(self.universe_size, set_size, self.random_state)
freq_per_id = self.random_state.choice(len(pmf), size=set_size, p=pmf) + 1
multiset_ids = []
for i, freq in zip(set_ids, freq_per_id):
multiset_ids += [i] * freq
self.random_state.shuffle(multiset_ids)
yield multiset_ids
return self
def __iter__(self):
for i in range(len(self.set_size_list)):
overlap_size = self.overlap_size_list[i]
set_ids_overlapped = choice_fast(self.union_ids, overlap_size,
self.random_state)
set_size = self.set_size_list[i]
set_ids_non_overlapped = self.ids_pool[:(set_size - overlap_size)]
self.ids_pool = self.ids_pool[len(set_ids_non_overlapped):]
self.union_ids = np.concatenate(
[self.union_ids, set_ids_non_overlapped])
set_ids = np.concatenate(
[set_ids_overlapped, set_ids_non_overlapped])
yield set_ids
return self
def __iter__(self):
for set_size, gamma_params in zip(self.set_sizes, self.gamma_params):
set_ids = choice_fast(self.universe_size, set_size, self.random_state)
rate_parameters = self.random_state.gamma(shape=gamma_params[0],
scale=gamma_params[1],
size=set_size)
frequencies = self.random_state.poisson(lam=rate_parameters,
size=set_size) + 1
if self.freq_cap:
frequencies = np.minimum(frequencies, self.freq_cap)
multiset_ids = []
for i, freq in zip(set_ids, frequencies):
multiset_ids += [i] * freq
self.random_state.shuffle(multiset_ids)
yield multiset_ids
return self
def test_choice_fast_is_unique(self):
for i in range(50, 500):
chosen = random.choice_fast(500, i)
no_repeats = set(chosen)
self.assertTrue(len(chosen) == len(no_repeats))
def test_choice_fast_len_is_m(self):
for i in range(1000):
a = random.choice_fast(10000, i)
self.assertLen(a, i)
def test_choice_fast_same_random_state_same_output(self):
rs1 = np.random.RandomState(1)
rs2 = np.random.RandomState(1)
a = random.choice_fast(10000, 5000, rs1)
b = random.choice_fast(10000, 5000, rs2)
self.assertSameElements(a, b)
def __iter__(self):
self.union_ids.update(
choice_fast(self.universe_size, self.set_size, self.random_state))
for _ in range(self.num_sets):
yield list(self.union_ids)
return self