def GetLocalConstraintsForRoot(self, file_prefix):
then_vtree_filename = "%s/%s_then_vtree.vtree" % (file_prefix,
self.name)
then_sdd_filename = "%s/%s_then_sdd.sdd" % (file_prefix, self.name)
constraint = {}
constraint["then_vtree"] = then_vtree_filename
constraint["then"] = [then_sdd_filename]
universe = []
# internal edges
for sub_region_edge_tup in self.sub_region_edges:
universe.append(sub_region_edge_tup)
GraphSet.set_universe(universe)
universe = GraphSet.universe()
paths = GraphSet()
child_names = self.children.keys()
for (i, j) in itertools.combinations(child_names, 2):
paths = paths.union(GraphSet.paths(i, j))
name_to_sdd_index = {}
zdd_to_sdd_index = [None] # for generating sdd from graphset
sdd_index = 0
for child in child_names:
sdd_index += 1
name_to_sdd_index["c%s" % child] = sdd_index
for sub_region_edge in universe:
corresponding_network_edges = self.sub_region_edges[
sub_region_edge]
coresponding_network_edges_sdd_index = []
for single_edge in corresponding_network_edges:
sdd_index += 1
name_to_sdd_index[str(single_edge)] = sdd_index
coresponding_network_edges_sdd_index.append(sdd_index)
zdd_to_sdd_index.append(coresponding_network_edges_sdd_index)
constraint["then_variable_mapping"] = name_to_sdd_index
rl_vtree = sdd.sdd_vtree_new(sdd_index, "right")
sdd_manager = sdd.sdd_manager_new(rl_vtree)
sdd.sdd_vtree_free(rl_vtree)
sdd.sdd_manager_auto_gc_and_minimize_off(sdd_manager)
# Construct simple path constraint
simple_path_constraint = generate_sdd_from_graphset(
paths, sdd_manager, zdd_to_sdd_index)
# non empty path in this region map
none_of_child = sdd.util.sdd_negative_term(
sdd_manager,
[name_to_sdd_index["c%s" % child] for child in self.children])
case_one = sdd.sdd_conjoin(none_of_child, simple_path_constraint,
sdd_manager)
# empty path in this region map
exactly_one_child = sdd.util.sdd_exactly_one(
sdd_manager,
[name_to_sdd_index["c%s" % child] for child in self.children])
empty_path_constraint = sdd.util.sdd_negative_term(
sdd_manager, sum(zdd_to_sdd_index[1:], []))
case_two = sdd.sdd_conjoin(exactly_one_child, empty_path_constraint,
sdd_manager)
total_constraint = sdd.sdd_disjoin(case_one, case_two, sdd_manager)
sdd.sdd_save(then_sdd_filename, total_constraint)
sdd.sdd_vtree_save(then_vtree_filename,
sdd.sdd_manager_vtree(sdd_manager))
sdd.sdd_manager_free(sdd_manager)
return constraint
def start_manager(var_count,order):
#vtree = sdd.sdd_vtree_new_with_var_order(var_count,order,"right")
vtree = sdd.sdd_vtree_new(var_count,"right")
#vtree = vtrees.right_linear_vtree(1,var_count+1)
manager = sdd.sdd_manager_new(vtree)
sdd.sdd_manager_auto_gc_and_minimize_off(manager)
sdd.sdd_vtree_free(vtree)
return manager
def start_manager(var_count,order):
#vtree = sdd.sdd_vtree_new_with_var_order(var_count,order,"right")
vtree = sdd.sdd_vtree_new(var_count,"right")
#vtree = vtrees.right_linear_vtree(1,var_count+1)
manager = sdd.sdd_manager_new(vtree)
sdd.sdd_manager_auto_gc_and_minimize_off(manager)
sdd.sdd_vtree_free(vtree)
return manager
def encode_graph(g,manager):
print "encoding graphs ..."
start = time()
sdd.sdd_manager_auto_gc_and_minimize_off(manager) # ACACAC
alpha = _encode_graph(g,manager)
print "done! (%.3fs)" % (time()-start)
print "model count..."
print "sdd mc :", sdd.sdd_model_count(alpha,manager)
print "sdd mc-g:", global_model_count(alpha,manager)
#print "minimizing ..."
#sdd.sdd_ref(alpha,manager)
#sdd.sdd_manager_minimize(manager)
#sdd.sdd_deref(alpha,manager)
print "sdd size:", sdd.sdd_size(alpha)
print "sdd nc :", sdd.sdd_count(alpha)
return alpha
def GetLocalConstraintsForInternalClusters(self, file_prefix):
if_vtree_filename = "%s/%s_if_vtree.vtree" % (file_prefix, self.name)
if_sdd_filename_prefix = "%s/%s_if_sdd" % (file_prefix, self.name)
then_vtree_filename = "%s/%s_then_vtree.vtree" % (file_prefix,
self.name)
then_sdd_filename_prefix = "%s/%s_then_sdd" % (file_prefix, self.name)
ifs = []
thens = []
if_variable_mapping = {}
if_sdd_index = 0
if_sdd_index += 1
if_variable_mapping[
"c%s" %
self.name] = if_sdd_index # cluster indicator for current cluster
for external_edge in self.external_edges:
if_sdd_index += 1
if_variable_mapping[str(external_edge)] = if_sdd_index
then_variable_mapping = {}
# variables for the child clusters
then_sdd_index = 0
zdd_to_sdd_index = [None]
for child in self.children:
then_sdd_index += 1
then_variable_mapping["c%s" % child] = then_sdd_index
universe = self.sub_region_edges.keys()
GraphSet.set_universe(universe)
universe = GraphSet.universe()
for node_pair in universe:
correponding_sdd_indexes = []
for internal_edge in self.sub_region_edges[node_pair]:
then_sdd_index += 1
then_variable_mapping[str(internal_edge)] = then_sdd_index
correponding_sdd_indexes.append(then_sdd_index)
zdd_to_sdd_index.append(correponding_sdd_indexes)
if_vtree, then_vtree = sdd.sdd_vtree_new(if_sdd_index,
"right"), sdd.sdd_vtree_new(
then_sdd_index, "right")
if_manager, then_manager = sdd.sdd_manager_new(
if_vtree), sdd.sdd_manager_new(then_vtree)
sdd.sdd_manager_auto_gc_and_minimize_off(if_manager)
sdd.sdd_manager_auto_gc_and_minimize_off(then_manager)
sdd.sdd_vtree_free(if_vtree)
sdd.sdd_vtree_free(then_vtree)
#none of the external edges are used and cluster indicator is off
case_index = 0
case_one_if = sdd.util.sdd_negative_term(if_manager,
range(1, if_sdd_index + 1))
case_one_then = sdd.util.sdd_negative_term(
then_manager, range(1, then_sdd_index + 1))
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
case_one_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
case_one_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
#none of the external edges are used and cluster indicator is on
case_index += 1
case_two_if = sdd.util.sdd_exactly_one_among(
if_manager, [if_variable_mapping["c%s" % self.name]],
range(1, if_sdd_index + 1))
#***Non empty path in this region map
none_of_child = sdd.util.sdd_negative_term(
then_manager,
[then_variable_mapping["c%s" % child] for child in self.children])
paths = GraphSet()
child_names = self.children.keys()
for c1, c2 in itertools.combinations(child_names, 2):
paths = paths.union(GraphSet.paths(c1, c2))
simple_path_constraint = generate_sdd_from_graphset(
paths, then_manager, zdd_to_sdd_index)
case_one = sdd.sdd_conjoin(simple_path_constraint, none_of_child,
then_manager)
#***Empty path in the region map
exactly_one_chlid = sdd.util.sdd_exactly_one(
then_manager,
[then_variable_mapping["c%s" % child] for child in self.children])
empty_path_constraint = sdd.util.sdd_negative_term(
then_manager, sum(zdd_to_sdd_index[1:], []))
case_two = sdd.sdd_conjoin(empty_path_constraint, exactly_one_chlid,
then_manager)
case_two_then = sdd.sdd_disjoin(case_one, case_two, then_manager)
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
case_two_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
case_two_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
#Exactly one of the external edge is used and cluster_indicator is off
aggregated_cases = {}
for external_edge in self.external_edges:
aggregated_cases.setdefault(self.external_edges[external_edge],
[]).append(external_edge)
for entering_node in aggregated_cases:
case_index += 1
cur_case_if = sdd.util.sdd_exactly_one_among(
if_manager, [
if_variable_mapping[str(e)]
for e in aggregated_cases[entering_node]
], range(1, if_sdd_index + 1))
paths = GraphSet()
for child in self.children:
if child == entering_node:
continue
paths = paths.union(GraphSet.paths(entering_node, child))
cur_case_then = generate_sdd_from_graphset(paths, then_manager,
zdd_to_sdd_index)
cur_case_then = sdd.sdd_disjoin(
cur_case_then,
sdd.util.sdd_negative_term(then_manager, [
then_variable_mapping[str(e)] for e in self.internal_edges
]), then_manager)
#conjoin that all the child indicator is off
cur_case_then = sdd.sdd_conjoin(
cur_case_then,
sdd.util.sdd_negative_term(then_manager, [
then_variable_mapping["c%s" % child]
for child in self.children
]), then_manager)
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
cur_case_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
cur_case_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
#Exactly two of the external edge is used and cluster_indicator is off
aggregated_cases = {}
for (i, j) in itertools.combinations(self.external_edges.keys(), 2):
entering_points = (self.external_edges[i], self.external_edges[j])
entering_points = (max(entering_points), min(entering_points))
aggregated_cases.setdefault(entering_points, []).append((i, j))
for entering_points in aggregated_cases:
case_index += 1
entering_edges = aggregated_cases[entering_points]
cur_case_if = generate_exactly_two_from_tuples(
if_manager,
[(if_variable_mapping[str(e1)], if_variable_mapping[str(e2)])
for (e1, e2) in entering_edges], range(1, if_sdd_index + 1))
if entering_points[0] == entering_points[1]:
cur_case_then = sdd.util.sdd_negative_term(
then_manager, range(1, then_sdd_index + 1))
else:
paths = GraphSet.paths(entering_points[0], entering_points[1])
cur_case_then = generate_sdd_from_graphset(
paths, then_manager, zdd_to_sdd_index)
cur_case_then = sdd.sdd_conjoin(
cur_case_then,
sdd.util.sdd_negative_term(then_manager, [
then_variable_mapping["c%s" % child]
for child in self.children
]), then_manager)
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
cur_case_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
cur_case_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
sdd.sdd_vtree_save(if_vtree_filename,
sdd.sdd_manager_vtree(if_manager))
sdd.sdd_vtree_save(then_vtree_filename,
sdd.sdd_manager_vtree(then_manager))
sdd.sdd_manager_free(if_manager)
sdd.sdd_manager_free(then_manager)
constraint = {}
constraint["if_vtree"] = if_vtree_filename
constraint["if"] = ifs
constraint["if_variable_mapping"] = if_variable_mapping
constraint["then_vtree"] = then_vtree_filename
constraint["then"] = thens
constraint["then_variable_mapping"] = then_variable_mapping
return constraint
def GetLocalConstraintsForLeaveClusters(self, file_prefix):
if_vtree_filename = "%s/%s_if_vtree.vtree" % (file_prefix, self.name)
if_sdd_filename_prefix = "%s/%s_if_sdd" % (file_prefix, self.name)
then_vtree_filename = "%s/%s_then_vtree.vtree" % (file_prefix,
self.name)
then_sdd_filename_prefix = "%s/%s_then_sdd" % (file_prefix, self.name)
ifs = []
thens = []
if_variable_mapping = {}
if_sdd_index = 0
if_sdd_index += 1
if_variable_mapping[
"c%s" %
self.name] = if_sdd_index # cluster indicator for current cluster
for external_edge in self.external_edges:
if_sdd_index += 1
if_variable_mapping[str(external_edge)] = if_sdd_index
then_variable_mapping = {}
zdd_to_sdd_index = [None]
universe = []
node_pair_to_edges = {}
for internal_edge in self.internal_edges:
if (internal_edge.x, internal_edge.y) not in node_pair_to_edges:
universe.append((internal_edge.x, internal_edge.y))
node_pair_to_edges.setdefault((internal_edge.x, internal_edge.y),
[]).append(internal_edge)
GraphSet.set_universe(universe)
universe = GraphSet.universe()
then_sdd_index = 0
for node_pair in universe:
correponding_sdd_indexes = []
for internal_edge in node_pair_to_edges[node_pair]:
then_sdd_index += 1
then_variable_mapping[str(internal_edge)] = then_sdd_index
correponding_sdd_indexes.append(then_sdd_index)
zdd_to_sdd_index.append(correponding_sdd_indexes)
if_vtree, then_vtree = sdd.sdd_vtree_new(if_sdd_index,
"right"), sdd.sdd_vtree_new(
then_sdd_index, "right")
if_manager, then_manager = sdd.sdd_manager_new(
if_vtree), sdd.sdd_manager_new(then_vtree)
sdd.sdd_manager_auto_gc_and_minimize_off(if_manager)
sdd.sdd_manager_auto_gc_and_minimize_off(then_manager)
sdd.sdd_vtree_free(if_vtree)
sdd.sdd_vtree_free(then_vtree)
#none of the external edges are used and cluster indicator is off
case_index = 0
case_one_if = sdd.util.sdd_negative_term(if_manager,
range(1, if_sdd_index + 1))
case_one_then = sdd.util.sdd_negative_term(
then_manager, range(1, then_sdd_index + 1))
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
case_one_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
case_one_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
#none of the external edges are used and cluster indicator is on
case_index += 1
case_two_if = sdd.util.sdd_exactly_one_among(
if_manager, [if_variable_mapping["c%s" % self.name]],
range(1, if_sdd_index + 1))
paths = GraphSet()
for (i, j) in itertools.combinations(self.nodes, 2):
paths = paths.union(GraphSet.paths(i, j))
case_two_then = generate_sdd_from_graphset(paths, then_manager,
zdd_to_sdd_index)
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
case_two_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
case_two_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
#exactly one of the external edge is used and cluster indicator is off
aggregated_cases = {}
for external_edge in self.external_edges:
aggregated_cases.setdefault(self.external_edges[external_edge],
[]).append(external_edge)
for entering_node in aggregated_cases:
case_index += 1
cur_case_if = sdd.util.sdd_exactly_one_among(
if_manager, [
if_variable_mapping[str(e)]
for e in aggregated_cases[entering_node]
], range(1, if_sdd_index + 1))
paths = GraphSet()
for node in self.nodes:
if node == entering_node:
continue
paths = paths.union(GraphSet.paths(entering_node, node))
cur_case_then = generate_sdd_from_graphset(paths, then_manager,
zdd_to_sdd_index)
# disjoin the empty path
cur_case_then = sdd.sdd_disjoin(
cur_case_then,
sdd.util.sdd_negative_term(then_manager,
range(1, then_sdd_index + 1)),
then_manager)
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
cur_case_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
cur_case_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
# exactly two of the external edge is used and cluster_indicator is off
aggregated_cases = {}
for (i, j) in itertools.combinations(self.external_edges.keys(), 2):
entering_points = (self.external_edges[i], self.external_edges[j])
entering_points = (max(entering_points), min(entering_points))
aggregated_cases.setdefault(entering_points, []).append((i, j))
for entering_points in aggregated_cases:
case_index += 1
entering_edges = aggregated_cases[entering_points]
cur_case_if = generate_exactly_two_from_tuples(
if_manager,
[(if_variable_mapping[str(e1)], if_variable_mapping[str(e2)])
for (e1, e2) in entering_edges], range(1, if_sdd_index + 1))
if entering_points[0] == entering_points[1]:
cur_case_then = sdd.util.sdd_negative_term(
then_manager, range(1, then_sdd_index + 1))
else:
paths = GraphSet.paths(entering_points[0], entering_points[1])
cur_case_then = generate_sdd_from_graphset(
paths, then_manager, zdd_to_sdd_index)
sdd.sdd_save("%s_%s" % (if_sdd_filename_prefix, case_index),
cur_case_if)
sdd.sdd_save("%s_%s" % (then_sdd_filename_prefix, case_index),
cur_case_then)
ifs.append("%s_%s" % (if_sdd_filename_prefix, case_index))
thens.append("%s_%s" % (then_sdd_filename_prefix, case_index))
sdd.sdd_vtree_save(if_vtree_filename,
sdd.sdd_manager_vtree(if_manager))
sdd.sdd_vtree_save(then_vtree_filename,
sdd.sdd_manager_vtree(then_manager))
sdd.sdd_manager_free(if_manager)
sdd.sdd_manager_free(then_manager)
constraint = {}
constraint["if_vtree"] = if_vtree_filename
constraint["if"] = ifs
constraint["if_variable_mapping"] = if_variable_mapping
constraint["then_vtree"] = then_vtree_filename
constraint["then"] = thens
constraint["then_variable_mapping"] = then_variable_mapping
return constraint
def run():
with timer.Timer("reading dataset"):
dataset = util.read_binary_dataset(test_filename)
domain = util.read_header(test_filename)
'''
if OPTIONS.majority_circuit_opt:
l = len(domain)
for k in xrange(num_trees):
domain["Tree_%d" % k] = l+k
'''
with timer.Timer("initializing manager"):
# start sdd manager
var_count = len(domain) - 1
vtree = sdd.sdd_vtree_new(var_count, "balanced")
manager = sdd.sdd_manager_new(vtree)
#sdd.sdd_manager_auto_gc_and_minimize_on(manager)
#sdd.sdd_manager_auto_gc_and_minimize_off(manager)
sdd_state = SddState(vtree, manager)
with timer.Timer("reading constraints"):
constraint_sdd, constraint_info = encode_logical_constraints(
constraint_filename, manager, domain)
sdd.sdd_ref(constraint_sdd, manager)
with timer.Timer("reading trees"):
tree_states = []
for filename in sorted(glob.glob(tree_basename.replace('%d', '*'))):
tree = pygv.AGraph(filename)
tree_state = TreeState(tree, domain, constraint_info)
tree_states.append(tree_state)
#tree.layout(prog='dot')
#tree.draw(filename+".png")
#num_trees = len(tree_states)
with timer.Timer("compiling trees"):
forest_sdds, _ = izip(*forest_sdds_iter(tree_states, sdd_state))
#forest_sdds = list(forest_sdds_iter(tree_states,sdd_state))
forest_sdds = [
(tree_state, tree_sdd)
for tree_state, tree_sdd in zip(tree_states, forest_sdds)
]
cmpf = lambda x, y: cmp(sdd.sdd_size(x[1]), sdd.sdd_size(y[1]))
forest_sdds.sort(cmp=cmpf)
tree_states = [tree_state for tree_state, tree_sdd in forest_sdds]
#ACACAC
sdd.sdd_manager_auto_gc_and_minimize_off(manager)
sdd.sdd_manager_minimize_limited(manager)
stats = SddSizeStats()
for tree_state, tree_sdd in forest_sdds:
stats.update(tree_sdd)
sdd.sdd_deref(tree_sdd, manager)
sdd.sdd_manager_garbage_collect(manager)
forest_sdds, used_vars_list = izip(
*forest_sdds_iter(tree_states, sdd_state))
print stats
with timer.Timer("compiling all", prefix="| "):
alpha = compile_all(forest_sdds, used_vars_list, num_trees, domain,
manager, constraint_sdd)
with timer.Timer("evaluating"):
msg = util.evaluate_dataset_all_sdd(dataset, alpha, manager)
print "| trees : %d" % num_trees
print "--- evaluating majority vote on random forest (compiled):"
print msg
print "| all size :", sdd.sdd_size(alpha)
print "| all count:", sdd.sdd_count(alpha)
print " model count:", sdd.sdd_global_model_count(alpha, manager)
with timer.Timer("checking monotonicity"):
result = is_monotone(alpha, manager)
print "Is monotone?", result
#for tree_sdd in forest_sdds: sdd.sdd_deref(tree_sdd,manager)
print "===================="
print "before garbage collecting..."
print "live size:", sdd.sdd_manager_live_count(manager)
print "dead size:", sdd.sdd_manager_dead_count(manager)
print "garbage collecting..."
sdd.sdd_manager_garbage_collect(manager)
print "live size:", sdd.sdd_manager_live_count(manager)
print "dead size:", sdd.sdd_manager_dead_count(manager)
vtree = sdd.sdd_manager_vtree(manager)
print "Writing sdd file %s and vtree file %s" % (sdd_filename,
vtree_filename)
sdd.sdd_save(sdd_filename, alpha)
sdd.sdd_vtree_save(vtree_filename, vtree)
print "Writing constraint sdd file %s and constraint vtree file %s" % (
constraint_sdd_filename, constraint_vtree_filename)
sdd.sdd_save(constraint_sdd_filename, constraint_sdd)
sdd.sdd_vtree_save(constraint_vtree_filename, vtree)
