def _uniform_sample_indices_and_probabilities(self, n):
indices = list(
sample_n_k(len(self.data), max(0, n - len(self.data_inf))))
probabilities = [1 / len(self)] * len(indices)
while len(indices) < n:
i = len(self.data)
e = self._pop_random_data_inf()
self.data.append(e)
del self.priority_tree[i]
indices.append(i)
probabilities.append(None)
return indices, probabilities
def uniform_sample(self, n, remove):
assert n >= 0
ixs = list(sample_n_k(self.length, n))
vals = []
if n > 0:
for ix in ixs:
val = self._write(ix, 0.0)
vals.append(val)
if not remove:
for ix, val in zip(ixs, vals):
self._write(ix, val)
return ixs, vals
def _uniform_sample_indices_and_probabilities(self, n):
indices = list(sample_n_k(
len(self.data), n))
probabilities = [1 / len(self)] * len(indices)
return indices, probabilities
def sample(self, k):
return [self[i] for i in sample_n_k(len(self), k)]
def test_slow(self):
self.samples = [sample_n_k(self.n, self.k) for _ in range(100000)]
self.subtest_total_counts()
self.subtest_order_counts()
def test_fast(self):
self.samples = [sample_n_k(self.n, self.k) for _ in range(200)]
self.subtest_constraints()
def sample(self, k):
nf = len(self._queue_front)
return [
self._queue_front[i] if i < nf else self._queue_back[i - nf]
for i in sample_n_k(len(self), k)
]