def __init__(self, length, decay_rate, random_seed):
"""Creates an ExponentialSameKeyAggregator.
An ExponentialSameKeyAggregator includes three components:
1. An ExponentialBloomFilter for estimating the reach.
2. An AnySketch to track the unique key in each register.
3. Another AnySketch to track the frequency of each effective key.
Args:
length: The length of bit vector for the Exponential bloom filter.
decay_rate: The decay rate of Exponential distribution.
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
self.length = length
self.decay_rate = decay_rate
self.exponential_bloom_filter = ExponentialBloomFilter(
length=length, decay_rate=decay_rate, random_seed=random_seed)
self.unique_key_tracker = any_sketch.AnySketch(
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.ExponentialDistribution(length, decay_rate), 'exp')
], num_hashes=1, value_functions=[any_sketch.UniqueKeyFunction()]),
random_seed)
self.frequency_count_tracker = any_sketch.AnySketch(
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.ExponentialDistribution(length, decay_rate), 'exp')
], num_hashes=1, value_functions=[any_sketch.SumFunction()]),
random_seed)
def __init__(self, length, num_hashes=1, random_seed=None, probability=0.08):
"""Creates a BloomFilter.
Args:
length: The length of bit vector for the bloom filter
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
super().__init__(
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.GeometricDistribution(length, probability), "geometric")
], num_hashes=1, value_functions=[any_sketch.BitwiseOrFunction()]),
random_seed)
def __init__(self, length, num_hashes=1, random_seed=None):
"""Creates a BloomFilter.
Args:
length: The length of bit vector for the bloom filter
num_hashes: The number of hash functions to use.
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
any_sketch.AnySketch.__init__(
self,
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.UniformDistribution(length), "dimension_1")
], num_hashes, [any_sketch.BitwiseOrFunction()]), random_seed)
def __init__(self, length, random_seed=None):
"""Creates an LogarithmicBloomFilter.
Args:
length: The length of bit vector for the bloom filter.
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
AnyDistributionBloomFilter.__init__(
self,
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.LogBucketDistribution(length), "log")
], num_hashes=1, value_functions=[any_sketch.BitwiseOrFunction()]),
random_seed)
def __init__(self, length, random_seed=None):
"""Creates a Uniform Counting BloomFilter.
Args:
length: The length of bit vector for the bloom filter
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
super().__init__(
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.UniformDistribution(length), "uniformcbf")
],
num_hashes=1,
value_functions=[any_sketch.SumFunction()
]), random_seed)
def __init__(self, length, decay_rate, random_seed=None):
"""Creates an ExponentialBloomFilter.
Args:
length: The length of bit vector for the bloom filter.
decay_rate: The decay rate of Exponential distribution.
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
AnyDistributionBloomFilter.__init__(
self,
any_sketch.SketchConfig([
any_sketch.IndexSpecification(
any_sketch.ExponentialDistribution(length, decay_rate), "exp")
], num_hashes=1, value_functions=[any_sketch.BitwiseOrFunction()]),
random_seed)
self.decay_rate = decay_rate
def __init__(self, length, probability, random_seed=None):
"""Creates a BloomFilter.
Args:
length: The length of bit vector for the bloom filter
probability: p of geometric distribution, p should be small enough
that geom.cdf(length, probability) won't be 1 in the middle of the
array so all bits can be used
random_seed: An optional integer specifying the random seed for
generating the random seeds for hash functions.
"""
super().__init__(
any_sketch.SketchConfig(
[
any_sketch.IndexSpecification(
any_sketch.GeometricDistribution(length, probability),
"geometric")
],
num_hashes=1,
value_functions=[any_sketch.BitwiseOrFunction()]), random_seed)
