def get_meta_graph_stat(g):
return MetaGraphStat(g, KWs)
trees = pkl.load(open('tmp/binary_rooted_tree_samples.pkl'))
U = 0.5
names = [# 'lst(round)',
'lst(ceil)', 'lst(floor)',
# 'greedy'
]
funcs = [lambda g, r, U, func=func: lst_dag(g, r, U,
fixed_point_func=func,
debug=False,
edge_weight_decimal_point=2)
for func in (# round,
ceil,
floor)]
# funcs.append(greedy_grow)
# t = trees[5]
# from lst import round_edge_weights_by_multiplying
# g_ceil, U = round_edge_weights_by_multiplying(t, U, 2, fixed_point_func=ceil)
# g_floor, U = round_edge_weights_by_multiplying(t, U, 2, fixed_point_func=floor)
# print(U)
# for u, v in g_ceil.edges():
# if 'dummy' not in g_ceil.node[v]:
)
parser.add_argument('--random_seed',
type=int,
default=None)
args = parser.parse_args()
random.seed(args.random_seed)
np.random.seed(args.random_seed)
dist_funcs = {'euclidean': euclidean, 'cosine': cosine}
dist_func = dist_funcs[args.dist]
lst = lambda g, r, U: lst_dag(g, r, U,
edge_weight_decimal_point=args.fixed_point,
debug=False)
variance_method = lambda g, r, U: dp_dag_general(
g, r,
int(U*(10**args.fixed_point)),
make_variance_cost_func(dist_func, 'topics',
args.fixed_point),
debug=False
)
quota_based_method = lambda g, r, U: binary_search_using_charikar(
g, r, U, args.charikar_level
)
methods = {'lst': lst,
'lst+dij': lst,
trees = pkl.load(open('tmp/binary_rooted_tree_samples.pkl'))
U = 0.5
names = [ # 'lst(round)',
'lst(ceil)',
'lst(floor)',
# 'greedy'
]
funcs = [
lambda g, r, U, func=func: lst_dag(g,
r,
U,
fixed_point_func=func,
debug=False,
edge_weight_decimal_point=2)
for func in ( # round,
ceil, floor)
]
# funcs.append(greedy_grow)
# t = trees[5]
# from lst import round_edge_weights_by_multiplying
# g_ceil, U = round_edge_weights_by_multiplying(t, U, 2, fixed_point_func=ceil)
# g_floor, U = round_edge_weights_by_multiplying(t, U, 2, fixed_point_func=floor)
# print(U)
# for u, v in g_ceil.edges():
parser.add_argument('--charikar_level',
type=int,
default=1,
help="the `level` parameter in charikar's algorithm")
parser.add_argument('--random_seed', type=int, default=None)
args = parser.parse_args()
random.seed(args.random_seed)
np.random.seed(args.random_seed)
dist_funcs = {'euclidean': euclidean, 'cosine': cosine}
dist_func = dist_funcs[args.dist]
lst = lambda g, r, U: lst_dag(
g, r, U, edge_weight_decimal_point=args.fixed_point, debug=False)
variance_method = lambda g, r, U: dp_dag_general(
g,
r,
int(U * (10**args.fixed_point)),
make_variance_cost_func(dist_func, 'topics', args.fixed_point),
debug=False)
quota_based_method = lambda g, r, U: binary_search_using_charikar(
g, r, U, args.charikar_level)
methods = {
'lst': lst,
'lst+dij': lst,
'variance': variance_method,
'greedy': greedy_grow_numpy,