def test_answer1(data, tol=0.00000001):
""" tests query answers without add_over_margins """
array_dims = data.shape
subset = (slice(1, 3), slice(0, 2), slice(2, 4))
query_no_sumover = cenquery.Query(array_dims, subset, None)
query_empty_sumover = cenquery.Query(array_dims, subset, ())
true_answer = data[subset]
assert numpy.abs(query_no_sumover.answer(data) - true_answer).sum() <= tol
assert numpy.abs(query_empty_sumover.answer(data) -
true_answer).sum() <= tol
def test_kron_rep1(data, tol=0.00000001):
""" tests query answers from kron representation without add_over_margins """
array_dims = data.shape
flat_data = data.flatten()
subset = (slice(1, 3), slice(0, 2), slice(2, 4))
query_no_sumover = cenquery.Query(array_dims, subset, None)
query_empty_sumover = cenquery.Query(array_dims, subset, ())
true_answer = data[subset].flatten()
assert numpy.abs(query_no_sumover.answer(flat_data) -
true_answer).sum() <= tol
assert numpy.abs(query_empty_sumover.answer(flat_data) -
true_answer).sum() <= tol
def hhgq(self, name="hhgq", subset=None):
add_over_margins = [1, 2, 3]
self.queries_dict[name] = cenquery.Query(
self.hist_shape,
subset=subset,
add_over_margins=add_over_margins,
name=name)
def minSchematize(node, array_dims, add_over_margins):
minSchemaQuery = cenquery.Query(array_dims=array_dims,
subset=None,
add_over_margins=add_over_margins)
node.raw = sparse.multiSparse(
minSchemaQuery.answer_original(node.raw.toDense()))
minSchema_shape = node.raw.shape
dims_keep = [
x
for x in set(range(len(array_dims))).difference(set(add_over_margins))
]
constraint_keys = node.cons.keys()
for key in constraint_keys:
node.cons[key].query.array_dims = minSchema_shape
node.cons[key].query.add_over_margins = tuple([
x for x in set(dims_keep).intersection(
set(node.cons[key].query.add_over_margins))
])
node.cons[key].query.subset_input = [
node.cons[key].query.subset_input[x] for x in dims_keep
]
node.cons[key].query.subset = np.ix_(
*tuple(node.cons[key].query.subset_input))
#axis_groupings = () ?? currently no axis groupings in constraints
print(node.cons[key].query)
node.cons[key].check_after_update()
return node
def union_hhgq_ub(self):
cats = self.cats
all_cat_combs = self.all_cat_combs
for cat_comb in all_cat_combs:
other_cats = list(set(range(8)).difference(set(cat_comb)))
name = "union_hhgq_ub." + ".".join([str(x) for x in cat_comb])
subset = (cat_comb, range(2), range(2), range(63))
add_over_margins = (0, 1, 2, 3)
if set(cats).issubset(set(cat_comb)):
rhs = self.invariants["tot"]
else:
gq_min_ext = self.invariants["gqhh_vect"]
gq_min_other = gq_min_ext[other_cats].sum(0)
rhs = self.invariants["tot"] - gq_min_other
sign = "le"
query = cenquery.Query(array_dims=self.hist_shape,
subset=subset,
add_over_margins=add_over_margins)
self.constraints_dict[name] = cenquery.Constraint(
query=query,
rhs=np.array(rhs).astype(int),
sign=sign,
name=name,
union=True,
union_type="union_hhgq_ub")
def total(self):
subset = None
add_over_margins = (0, 1, 2, 3)
query = cenquery.Query(array_dims=self.hist_shape,subset=subset, add_over_margins=add_over_margins)
rhs = self.invariants["tot"].astype(int)
sign = "="
self.constraints_dict["total"] = cenquery.Constraint(query=query, rhs=rhs, sign=sign, name="total")
def test_makeTabularGroupQuery(data):
import programs.engine.cenquery as cenquery_old
shape = data.shape
add_over_margins = (2,)
subset = (range(3), [1,2], range(5))
groupings = {1: [[1],[2]]}
axis_groupings = [ (1, ([0,1],[2])), (2, ([1,3],[0,2] ))]
groupings2 = {1: [[0,1],[2]], 2: [[1,3],[0,2]]}
q1 = cenquery_old.Query(shape, add_over_margins=add_over_margins).convertToQuerybase()
q2 = cenquery.QueryFactory.makeTabularGroupQuery(shape, add_over_margins=add_over_margins)
assert compare_arrays(q1.answer(data),q2.answer(data))
q3 = cenquery_old.Query(shape, add_over_margins=add_over_margins, subset=subset).convertToQuerybase()
q4 = cenquery.QueryFactory.makeTabularGroupQuery(shape, add_over_margins=add_over_margins, groupings=groupings)
assert compare_arrays(q3.answer(data),q4.answer(data))
q5 = cenquery_old.Query(shape, add_over_margins=(0,), axis_groupings = axis_groupings).convertToQuerybase()
q6 = cenquery.QueryFactory.makeTabularGroupQuery(shape, add_over_margins=(0,), groupings=groupings2)
assert compare_arrays(q5.answer(data),q6.answer(data))
def tot(self):
data = self.raw
add_over_margins = (0, 1, 2, 3)
subset = None
query = cenquery.Query(array_dims=data.shape,
subset=subset,
add_over_margins=add_over_margins)
self.invariants_dict["tot"] = np.array(query.answer(data)).astype(int)
def gq_vect(self):
data = self.raw_housing
add_over_margins = None
subset = (range(1, 8), )
query = cenquery.Query(array_dims=data.shape,
subset=subset,
add_over_margins=add_over_margins)
self.invariants_dict["gq_vect"] = np.array(
query.answer(data)).astype(int)
def number_of_races(self, name="number_of_races", subset=None):
add_over_margins = [0,1,2]
axis_groupings = [ (3, (range(0,7),
range(7,22),
range(22,42),
range(42,57),
range(57,63)
))]
self.queries_dict[name] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, subset=subset,
axis_groupings=axis_groupings, name=name)
def nurse_nva_0(self):
subset = ([3], [0], range(2), range(63))
add_over_margins = (0, 1, 2, 3)
query = cenquery.Query(array_dims=self.hist_shape,
subset=subset,
add_over_margins=add_over_margins)
rhs = np.array(0)
sign = "="
self.constraints_dict["nurse_nva_0"] = cenquery.Constraint(
query=query, rhs=rhs, sign=sign, name="nurse_nva_0")
def test_answer2(data, tol=0.00000001):
""" tests query answers with add_over_margins """
array_dims = data.shape
subset = (slice(1, 3), slice(0, 2), slice(2, 4))
axes = [(1, ), (0, 2), (0, 1, 2)]
true_answers = [data[subset].sum(axis=ax) for ax in axes]
queries = [cenquery.Query(array_dims, subset, ax) for ax in axes]
answers = [q.answer(data) for q in queries]
errors = [
numpy.abs(a - ta).sum() for (a, ta) in zip(answers, true_answers)
]
assert max(errors) < tol
def test_kron_consistency(data, tol=0.00000001):
""" checks whether kron rep is consistent with flattened answers """
array_dims = data.shape
subset = (slice(1, 3), slice(0, 2), slice(2, 4))
axes = [(1, ), (0, 2), (0, 1, 2)]
flattened_data = data.flatten()
queries = [cenquery.Query(array_dims, subset, ax) for ax in axes]
flatten_answers = [q.answer(data, flatten=True) for q in queries]
kron_answers = [q.answer(flattened_data) for q in queries]
errors = [
numpy.abs(fa - ka).sum()
for (fa, ka) in zip(flatten_answers, kron_answers)
]
assert max(errors) < tol
def hhgq_total_ub(self):
subset = None
add_over_margins = (1, 2, 3)
query = cenquery.Query(array_dims=self.hist_shape, subset=subset, add_over_margins=add_over_margins)
other_gqs = self.invariants["gqhh_tot"] - self.invariants["gqhh_vect"]
if "tot" in self.invariants.keys():
total = np.broadcast_to(np.array( self.invariants["tot"] ), (1,) )
total_ext= np.broadcast_to(total, (8,))
rhs = np.where(self.invariants["gqhh_vect"]>0, total_ext-other_gqs, np.zeros(8))
else:
rhs = self.invariants["gqhh_vect"]*self.hhgq_cap
rhs = rhs.astype(int)
sign = "le"
self.constraints_dict["hhgq_total_ub"] = cenquery.Constraint(query=query, rhs=rhs, sign=sign, name="hhgq_total_ub")
def hhgq_total_lb(self):
subset = None
add_over_margins = (1, 2, 3)
query = cenquery.Query(array_dims=self.hist_shape, subset=subset, add_over_margins=add_over_margins)
# gq_hh in other cats
other_gqs = self.invariants["gqhh_tot"] - self.invariants["gqhh_vect"]
if "tot" in self.invariants.keys():
total = np.broadcast_to(np.array(self.invariants["tot"]), (1,))
total_ext = np.broadcast_to(total, (8,))
rhs = np.where(other_gqs > 0, self.invariants["gqhh_vect"], total_ext)
else:
rhs = self.invariants["gqhh_vect"]
rhs = rhs.astype(int)
sign = "ge"
self.constraints_dict["hhgq_total_lb"] = cenquery.Constraint(query=query, rhs=rhs, sign=sign, name="hhgq_total_lb")
def union_hhgq_lb(self):
cats = self.cats
all_cat_combs = self.all_cat_combs
for cat_comb in all_cat_combs:
name = "union_hhgq_lb." + ".".join([str(x) for x in cat_comb])
subset = (cat_comb, range(2), range(2), range(6))
add_over_margins = (0, 1, 2, 3)
if set(cats).issubset(set(cat_comb)):
rhs = self.invariants["tot"]
else:
rhs = self.constraints_dict["hhgq_total_lb"].rhs[cat_comb].sum()
sign = "ge"
query = cenquery.Query(array_dims=self.hist_shape,
subset=subset, add_over_margins=add_over_margins)
self.constraints_dict[name] = cenquery.Constraint(query=query, rhs=np.array(rhs).astype(int),
sign=sign, name = name, union =True,
union_type="union_hhgq_lb")
def installPhase1Constraints(node_pair, config):
block_node = node_pair[0]
block_node_ms = node_pair[1]
add_over_margins = re.split(
das_utils.DELIM,
config["minimal_schema"]["minSchema.add_over_margins"])
add_over_margins = tuple([int(x) for x in add_over_margins])
query = cenquery.Query(array_dims=block_node.raw.shape,
add_over_margins=add_over_margins,
name="minSchema")
rhs = block_node_ms.raw.toDense()
ms_constraint = cenquery.Constraint(query, rhs, sign="=", name="minSchema")
cons = {}
cons["minSchema"] = ms_constraint
block_node.cons = cons
return block_node
def black_in_combo(self):
add_over_margins = [0, 1, 2, 3]
axis_groupings = [(3, ([1], [6], range(11, 15), range(21, 25), range(31, 37), range(41, 47), range(51, 55), range(56, 60), range(61,63)))]
self.queries_dict["black_in_combo"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, axis_groupings=axis_groupings, name="black_in_combo")
def hhgq_va_hisp(self):
add_over_margins = [3]
self.queries_dict["hhgq_va_hisp"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, name="hhgq_va_hisp")
def va_race(self):
add_over_margins = [0, 2]
self.queries_dict["va_race"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, name="va_race")
def race_ethnicity(self):
add_over_margins = [0, 1]
self.queries_dict["race_ethnicity"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins,
name="race_ethnicity")
def race(self, name="race", subset=None):
add_over_margins = [0,1,2]
self.queries_dict[name] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, subset=subset, name=name)
def hhgq_race(self):
add_over_margins = [1, 2]
self.queries_dict["hhgq_race"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, name="hhgq_race")
def racecomb(self):
add_over_margins = [0,1,2,3]
axis_groupings = [ (4, ([0,1,2],[3,4,5]))]
self.queries_dict["racecomb"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins,
axis_groupings=axis_groupings, name="racecomb")
def voting_age(self):
add_over_margins = [0,2,3]
self.queries_dict["voting_age"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins,
name="voting_age")
def race(self):
add_over_margins = [0,1,2,3]
self.queries_dict["race"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, name="race")
def weight_test(self):
add_over_margins = [0,2,3]
#weight_array = np.random.rand(self.hist_shape)
weight_array = np.full(self.hist_shape, 1.5)
self.queries_dict["weight_test"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins,
weight_array = weight_array, name="weight_test")
def nhpi_in_combo(self):
add_over_margins = [0, 1, 2, 3]
axis_groupings = [(3, ([4], [9], [13], [16], [18], [20], [23], [26], [28], [30], [32], [34], range(36, 38), range(39, 41), [42], [44], range(46, 48), range(49, 52), range(53, 57), range(58, 63)))]
self.queries_dict["nhpi_in_combo"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, axis_groupings=axis_groupings, name="nhpi_in_combo")
def asian_in_combo(self):
add_over_margins = [0, 1, 2, 3]
axis_groupings = [(3, ([3],[8],[12],[16], range(18,20), [23],[25], range(28,30), [31], range(34,36), range(37,39), range(40, 42), range(44, 46), range(47, 49), range(50, 53), range(54, 58), range(59,63)))]
self.queries_dict["asian_in_combo"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, axis_groupings=axis_groupings, name="asian_in_combo")
def aian_in_combo(self):
add_over_margins = [0, 1, 2, 3]
axis_groupings = [(3, ([2], [7], [11], range(15, 18), [21], range(25, 28), range(31, 34), range(37, 40), range(41, 44), range(47, 50), range(51, 54), range(55, 59), range(60, 63)))]
self.queries_dict["aian_in_combo"] = cenquery.Query(self.hist_shape, add_over_margins=add_over_margins, axis_groupings=axis_groupings, name="aian_in_combo")
