def init(args, params, whole_params):
info = {}
for k, w in params.items():
info[k] = w
info["time"] = ct.get_time_str()
info["whole_params"] = whole_params
info["conf_name"] = args.conf
info["res_home"] = os.path.join(os.path.join(RES_PATH, args.conf),
info["time"])
info["data_path"] = DATA_PATH
info["home_path"] = ROOT_PATH
info["network_path"] = os.path.join(DATA_PATH, params["network_path"])
ct.check_attr(info, "mem_interval", 0.001)
# if not exists, then mkdir
ct.mkdir(info["res_home"])
log_path = os.path.join(LOG_PATH, info["time"] + ".log")
info["log_path"] = log_path
info["log"] = ct.get_logger(log_path)
ct.symlink(log_path, os.path.join(LOG_PATH, "new_log"))
ct.symlink(info["res_home"], os.path.join(RES_PATH, "new_res"))
random.seed(info["random_seed"])
np.random.seed(info["np_seed"])
return info
def params_handler(params, info, pre_res, **kwargs):
res = {}
if ct.check_attr(params, "save_path", info["res_home"]):
params["save_path"] = os.path.join(info["home_path"],
params["save_path"])
params["folder_path"] = os.path.join(info["data_path"],
info["network_folder"]["name"])
ct.mkdir(params["save_path"])
res["save_path"] = params["save_path"]
return res
def params_handler(params, info, pre_res, **kwargs):
res = {}
if ct.check_attr(params, "save_path", info["res_home"]):
params["save_path"] = os.path.join(info["home_path"],
params["save_path"])
params["folder_path"] = os.path.join(info["data_path"],
info["network_folder"]["name"])
ct.check_attr(params, "is_direct", False)
res["train_path"] = os.path.join(params["save_path"], "train")
res["test_path"] = os.path.join(params["save_path"], "test")
ct.mkdir(res["train_path"])
ct.mkdir(res["test_path"])
return res
def split_graph(params, info, res, **kwargs):
params_handler(params, info, res)
n = params["num_nodes"]
k = params["num_community"]
#print n, k
#print np.argpartition(params["degree_distrib"], n-k)
params["top_set"] = set(
np.argpartition(params["degree_distrib"], n - k)[-k:])
uf = UnionFind(params)
ret = {}
#print params["top_set"]
with open(params["network_path"], "r") as f:
for line in f:
line = line.strip()
if len(line) == 0:
continue
items = line.split()
if len(items) != 2:
continue
u, v = [int(i) for i in items]
flag = uf.union(u, v)
#print u, v, flag
ret["community_info"] = uf.refine()
home_path = os.path.join(info["res_home"], "split_graph")
com_info = ret["community_info"]
ct.mkdir(home_path)
wf = {}
for u in com_info:
com_info[u]["file_path"] = os.path.join(home_path, str(u) + ".dat")
wf[u] = open(com_info[u]["file_path"], "w")
with open(params["network_path"], "r") as f:
for line in f:
line = line.strip()
if len(line) == 0:
continue
items = line.split()
if len(items) != 2:
continue
u = uf.find(int(items[0]))
v = uf.find(int(items[1]))
if u != v:
continue
wf[u].write(items[0] + "\t" + items[1] + "\n")
for it in wf:
wf[it].close()
return ret
def params_handler(params, info, pre_res, **kwargs):
params["num_nodes"] = info["num_nodes"]
params["community_size"] = info["community_size"]
params["num_community"] = info["num_community"]
params["num_top"] = info["num_top"]
params["res_path"] = info["res_home"]
params["network_path"] = info["network_path"]
params["dim"] = info["embedding_size"]
params["tmp_path"] = os.path.join(info["res_home"], "tmp")
ct.mkdir(params["tmp_path"])
ct.check_attr(params, "is_directed", False)
if "data_path" in params:
params["data_path"] = os.path.join(info["home_path"], params["data_path"])
else:
params["data_path"] = params["res_path"]
return {}