def __init__(self,
shape_list,
domain_shape,
size,
seed=9001,
pretty_name='random range'):
self.init_params = util.init_params_from_locals(locals())
self.shape_list = copy.deepcopy(shape_list)
self.seed = seed
self.pretty_name = pretty_name
self._size = size
prng = numpy.random.RandomState(seed)
if shape_list == None:
shapes = randomQueryShapes(domain_shape, prng)
else:
prng.shuffle(self.shape_list)
shapes = itertools.cycle(
self.shape_list) # infinite iterable over shapes in shape_list
queries = []
for i in range(size):
lb, ub = placeRandomly(next(shapes), domain_shape,
prng) # seed must be None or repeats
queries.append(ndRangeUnion().addRange(lb, ub, 1.0))
super(RandomRange, self).__init__(queries, domain_shape)
def __init__(self, index='DEFAULT1D', short_name='AHP*'):
self.init_params = util.init_params_from_locals(locals())
self._vSEP = Lfunction[index]['vSEP']
self._vRatio = Lfunction[index]['vRatio']
self._vEta = Lfunction[index]['vEta']
self.short_name = short_name
def __init__(self, domain, stripe_dim, ratio, approx):
self.init_params = util.init_params_from_locals(locals())
self.ratio = ratio
self.domain = domain
self.stripe_dim = stripe_dim
self.approx = approx
super().__init__()
def __init__(self, short_name="GreedyH"):
self.init_params = util.init_params_from_locals(locals())
self._max_block_size = None
self._partition_engine = None
self._estimate_engine = greedyH.greedyH_engine()
self.short_name = short_name
def __init__(self, domain_shape, ratio, approx, workload_based=False):
self.init_params = util.init_params_from_locals(locals())
self.domain_shape = domain_shape
self.ratio = ratio
self.approx = approx
self.workload_based = workload_based
super().__init__()
def __init__(self, domain_shape, workload_based=False):
self.init_params = util.init_params_from_locals(locals())
self.domain_shape = domain_shape
self.workload_based = workload_based
super().__init__()
assert len(domain_shape) in [1, 2], "HB only works for 1D and 2D domains"
def __init__(self, fileName, reduce_to_dom_shape=None):
self.init_params = util.init_params_from_locals(locals())
self.fname = fileName
#assert nickname in filenameDict, 'Filename parameter not recognized: %s' % nickname
hist = load(self.fname)
super(DatasetFromFile, self).__init__(hist, reduce_to_dom_shape, None)
def __init__(self, domain_shape, eta, ratio, workload_based=False):
self.init_params = util.init_params_from_locals(locals())
self.n = np.prod(domain_shape)
self.eta = eta
self.ratio = ratio
self.workload_based = workload_based
super().__init__()
def __init__(self, config):
self.init_params = util.init_params_from_locals(locals())
self.config = config
self._df = None
self.hist = None
self.edges = None
def __init__(self, data_scale, alpha=0.5, c=10, c2=5):
self.init_params = util.init_params_from_locals(locals())
self.alpha = alpha
self.c = c
self.c2 = c2
self.data_scale = data_scale
super().__init__()
def __init__(self, ratio=0.25, short_name='DAWA'):
# set ratio to 0 if no partition needed => GREEDYH_LINEAR,else DAWA_LINEAR.
self.init_params = util.init_params_from_locals(locals())
self._ratio = ratio
self._partition_engine = l1partition.l1partition_approx_engine()
self._estimate_engine = greedyH.greedyH_engine()
self.short_name = short_name
def __init__(self, ratio, rounds, data_scale, domain_shape, total_noise_scale):
self.init_params = util.init_params_from_locals(locals())
self.ratio = ratio
self.rounds = rounds
self.data_scale = data_scale
self.domain_shape = domain_shape
self.total_noise_scale = total_noise_scale
super().__init__()
def __init__(self, ratio, rounds, data_scale, domain_shape, use_history):
self.init_params = util.init_params_from_locals(locals())
self.ratio = ratio
self.rounds = rounds
self.data_scale = data_scale
self.domain_shape = domain_shape
self.use_history = use_history
super().__init__()
def __init__(self, domain_shape, stripe_dim, ratio, rounds, approx):
self.init_params = util.init_params_from_locals(locals())
self.domain_shape = domain_shape
self.stripe_dim = stripe_dim
self.ratio = ratio
self.rounds = rounds
self.approx = approx
super().__init__()
def __init__(self, domain_shape, ratio, approx, workload_based=False):
self.init_params = util.init_params_from_locals(locals())
self.ratio = ratio
self.approx = approx
self.domain_shape = domain_shape
self.workload_based = workload_based
super().__init__()
assert len(domain_shape) in [1, 2], "DAWA only works for 1D and 2D domains"
def __init__(self, domain_shape_int, pretty_name='prefix 1D'):
self.init_params = util.init_params_from_locals(locals())
self.pretty_name = pretty_name
queries = [
ndRangeUnion().add1DRange(0, c, 1.0)
for c in range(domain_shape_int)
]
super(self.__class__, self).__init__(queries, (domain_shape_int, ))
def __init__(self, branch=2, granu=100):
"""Setup the branching factor and granularity in numerical search.
branch(2) - the branching factor of the hierarchy
granu(100) - the granularity in numerical search
"""
self.init_params = util.init_params_from_locals(locals())
self._branch = branch
self._granu = granu
def __init__(self, df, bins, domains, normed=False, weights=None):
self.init_params = util.init_params_from_locals(locals())
self.df = df
self.bins = bins
self.domains = domains
self.normed = normed
self.weights = weights
self.hist = None
self.edges = None
def __init__(self, domain_shape, weight=1.0, pretty_name='identity'):
self.init_params = util.init_params_from_locals(locals())
self.weight = weight
self.pretty_name = pretty_name
if type(domain_shape) is int:
domain_shape = (domain_shape, )
indexes = itertools.product(*[list(range(i)) for i in domain_shape
]) # generate index tuples
queries = [ndRangeUnion().addRange(i, i, weight) for i in indexes]
super(self.__class__, self).__init__(queries, domain_shape)
def __init__(self, ratio=0.5, max_block_size=None, short_name="DAWA"):
""" c,c2 are the constant parameter controlling the number of cells in each level
alpha is the ratio of budget split on the first level beside total budget"""
self.init_params = util.init_params_from_locals(locals())
self._ratio = ratio
self._max_block_size = max_block_size
self._partition_engine = l1partition.l1partition_approx_engine()
self._estimate_engine = greedyH.greedyH_engine()
self.short_name = short_name
def __init__(self, domain_shape, k, pretty_name='k-way marginals'):
self.init_params = util.init_params_from_locals(locals())
self.pretty_name = pretty_name
if type(domain_shape) is int:
domain_shape = (domain_shape, )
assert 0 <= k <= len(domain_shape), 'invalid k'
self.k = k
D = len(domain_shape)
idents = [Identity.oneD(n) for n in domain_shape]
totals = [Total(n) for n in domain_shape]
workloads = []
for c in itertools.combinations(list(range(D)), k):
oned = [idents[i] if i in c else totals[i] for i in range(D)]
workloads.append(Kronecker(oned))
super(self.__class__, self).__init__(workloads)
def __init__(self,
relation,
nickname,
normed=False,
weights=None,
reduce_to_dom_shape=None):
self.init_params = util.init_params_from_locals(locals())
self.fname = nickname
histogram = Histogram(relation.df, relation.bins, relation.domains,
normed, weights).generate()
self.statistics = relation.statistics()
self.statistics.update(histogram.statistics())
if reduce_to_dom_shape != None:
self.statistics['domain_valid'] = False
else:
self.statistics['domain_valid'] = True
super(DatasetFromRelation,
self).__init__(histogram.hist.copy(order='C'),
reduce_to_dom_shape)
def __init__(self,
domain_shape,
prefix_axes=(0, ),
pretty_name='prefix marginals'):
self.init_params = util.init_params_from_locals(locals())
self.pretty_name = pretty_name
self.prefix_axes = prefix_axes
if type(domain_shape) is int:
domain_shape = (domain_shape, )
if len(domain_shape) - len(prefix_axes) > min(prefix_axes):
import warnings
warnings.warn(
'It is not recommended to construct PrefixMarginals this way. Transpose the data so that the prefix workloads are on the last dimensions'
)
oned = []
for i, n in enumerate(domain_shape):
if i in self.prefix_axes:
oned.append(Prefix1D(n))
else:
oned.append(Identity.oneD(n) + Total(n))
super(self.__class__, self).__init__(oned)
def __init__(self, uniformity, dom_shape, scale, seed=None):
'''
Generate synthetic data of varying uniformity
uniformity: parameter in [0,1] where 1 produces perfectly uniform data, 0 is maximally non-uniform
All cells set to zero except fraction equal to 'uniformity' value.
All non-zero cells are set to same value, then shuffled randomly.
'''
self.init_params = util.init_params_from_locals(locals())
self.u = uniformity
assert 0 <= uniformity and uniformity <= 1
n = numpy.prod(dom_shape) # total domain size
hist = numpy.zeros(n)
num_nonzero = max(1, int(uniformity * n))
hist_value = util.old_div(scale, num_nonzero)
hist[0:num_nonzero] = hist_value
prng = numpy.random.RandomState(seed)
prng.shuffle(hist)
super(DatasetUniformityVaryingSynthetic,
self).__init__(hist.reshape(dom_shape),
reduce_to_domain_shape=None,
dist=None)
def __init__(self, ratio=0.85, eta=0.35, short_name='AHP'):
self.init_params = util.init_params_from_locals(locals())
self._ratio = ratio
self._eta = eta
self.short_name = short_name
def __init__(self, domain, impl='MM', stripe_dim=-1):
self.init_params = util.init_params_from_locals(locals())
self.domain = domain
self.impl = impl
self.stripe_dim = stripe_dim
super().__init__()
def __init__(self, domain, stripe_dim):
self.init_params = util.init_params_from_locals(locals())
self.domain = domain
self.stripe_dim = stripe_dim
super().__init__()
def __init__(self, data_scale):
self.init_params = util.init_params_from_locals(locals())
self.data_scale = data_scale
super().__init__()
def __init__(self, eta, ratio, workload_based=False):
self.init_params = util.init_params_from_locals(locals())
self.eta = eta
self.ratio = ratio
self.workload_based = workload_based
super().__init__()
def __init__(self, theta, domain):
self.init_params = util.init_params_from_locals(locals())
self.theta = theta
self.domain = domain
super().__init__()