-
Notifications
You must be signed in to change notification settings - Fork 0
/
StatAlignHist.py
302 lines (249 loc) · 10.8 KB
/
StatAlignHist.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
import argparse
import datetime
import os
import random
import sys
import numpy as np
from mpi4py import MPI
import src.Analysis.Analysis as srcAn
import src.C_Extensions.sample as c_ext_sam
import src.MCMC as MCMC
import src.MCMC.MCMC_MC3 as MCMC_3
import src.Utils.Data_class as srcUDC
import src.Utils.Read_Settings as srcRead
reload(sys)
sys.setdefaultencoding('utf8')
def write_settings_file(p, time):
# type: (str, datetime.datetime) -> None
"""
This function writes down the information for the current run such as to identify it
:param p: file name
:type p: str
:param time: call time
:type time: datetime
:rtype: None
"""
with open(p, "w") as outfile:
outfile.write("Settings for MCMC estimation\n")
outfile.write("Calltime:\n")
outfile.write(time.strftime("%Y-%m-%d %H:%M") + "\n")
outfile.write("-------Settings-------\n")
for entry in vars(args):
outfile.write("{0}\t{1}\n".format(str(entry), str(getattr(args, entry))))
def data_creation(data_file, sound_model, ldn, header, cc_sample, diag, check_consistency, pre_tree):
# type: (str, str, float, list, bool, bool, bool, str) -> srcUDC.DataClass
"""
Processes the data stored in data_file for the MCMC computation
cc_sample has precedence over ldn. is only taken into account if cc_sample is false
:param cc_sample: if cognate classes should be sampled or not
:type cc_sample: bool
:param data_file: the location of the data file
:type data_file: str
:param sound_model: asjp or ipa encoding
:type sound_model: str
:param ldn: Levenshtein distance for word pairs, all wordpairs below the threshold are not considered
:type ldn: float
:param header: specification of the relevant columns in the data file
:type header: list
:param diag:
:type diag: bool
:param check_consistency: check if consistencies with tree constraints should be checked
:type check_consistency: bool
:param pre_tree: specified starting point tree
:type pre_tree: str
:return: Class object holding the data for the computation
:rtype: srcUDC.DataClass
"""
return srcUDC.DataClass.create_data(data_file=data_file, header=header,
sound_model=sound_model, ldn=ldn, cc_sam=cc_sample, data_diag=diag,
check_consistency=check_consistency, pre_tree=pre_tree)
def set_random_seed(seed_val):
# type: (int) -> None
"""
This function sets the random seed for the computation
:param seed_val: random seed to use
:type seed_val: int
:rtype: None
"""
if seed_val is not None:
try:
np.random.seed(seed_val)
random.seed(seed_val)
c_ext_sam.set_srand_seed(seed_val)
except ValueError:
s = 1234
np.random.seed(s)
random.seed(s)
c_ext_sam.set_srand_seed(s)
print("could not transform given seed into integer. Using default seed: 1234")
else:
s = random.randint(0, 9999)
c_ext_sam.set_srand_seed(s)
def MCMC_setup():
"""
Perform setup for the MCMC from here on
"""
# set seed
set_random_seed(parameter_dict["Seed"])
# create output folder if not existent
if not os.path.isdir(parameter_dict["Output"]):
os.mkdir(parameter_dict["Output"])
# check if data file exists
if not os.path.exists(parameter_dict["Data"]):
raise Exception("Data file not found " + parameter_dict["Data"])
# write information about MCMC run
write_settings_file(parameter_dict["Output"] + "settings.log", time=now)
# if the header in the data file is not the default one, use these names for the respective columns
header = [parameter_dict["lang_col"],
parameter_dict["concept_col"],
parameter_dict["transcription_col"],
parameter_dict["cognate_class_col"]]
# process the data
if parameter_dict["folder"] is not None:
pre_tree, sm_d, tr_d = read_state_from_file(parameter_dict["folder"])
else:
pre_tree = None
sm_d = None
tr_d = None
# process the data
data = data_creation(data_file=parameter_dict["Data"],
header=header,
sound_model=parameter_dict["Sound Model"],
cc_sample=parameter_dict["sample_cognates"],
ldn=parameter_dict["ldn"],
diag=parameter_dict["bottom-up"],
check_consistency=parameter_dict["consistency_checker"],
pre_tree=pre_tree)
MCMC_mod = MCMC.MCMC.create_mcmc(data=data, parameters=parameter_dict, tr_params=tr_d, em_params=sm_d)
# set up the MCMC
return MCMC_mod
def swap_store_gen(swap_store):
max_index = len(swap_store)
index = 0
while index < max_index:
yield swap_store[index]
index += 1
def MCMC_MC3_setup(temperature, mpi_size, now):
"""
Perform setup for the MCMC from here on
"""
# set seed
set_random_seed(parameter_dict["Seed"] + int(temperature))
samples = parameter_dict["Iterations"] / 5
if temperature == 1:
# create output folder if not existent
if not os.path.isdir(parameter_dict["Output"]):
os.mkdir(parameter_dict["Output"])
# check if data file exists
if not os.path.exists(parameter_dict["Data"]):
raise Exception("Data file not found " + parameter_dict["Data"])
# write information about MCMC run
write_settings_file(parameter_dict["Output"] + "settings.log", time=now)
# create swap list
donor_recip = np.random.randint(mpi_size, size=(samples, 2))
swap_store = np.zeros((samples, 3), dtype=np.int)
swap_store[::, 0:2] = donor_recip
swap_store[::, 2] = np.sort(np.random.choice(parameter_dict["Iterations"], samples, replace=False))
else:
swap_store = np.empty((samples, 3), dtype=np.int)
comm.Barrier()
swap_store = comm.bcast(swap_store, root=0)
# if the header in the data file is not the default one, use these names for the respective columns
header = [parameter_dict["lang_col"],
parameter_dict["concept_col"],
parameter_dict["transcription_col"],
parameter_dict["cognate_class_col"]]
if parameter_dict["folder"] is not None:
pre_tree, sm_d, tr_d = read_state_from_file(parameter_dict["folder"])
parameter_dict["randomize"] = False
else:
pre_tree = None
sm_d = None
tr_d = None
# process the data
data = data_creation(data_file=parameter_dict["Data"],
header=header,
sound_model=parameter_dict["Sound Model"],
cc_sample=parameter_dict["sample_cognates"],
ldn=parameter_dict["ldn"],
diag=parameter_dict["bottom-up"],
check_consistency=parameter_dict["consistency_checker"],
pre_tree=pre_tree)
# set up the MCMC
MCMC_mod = MCMC_3.MCMC_MC3.create_mcmc(data=data, parameters=parameter_dict, temperature=temperature,
swap_store=swap_store, tr_params=tr_d, em_params=sm_d)
return MCMC_mod
def read_state_from_file(folder):
tr_filename = folder + "tr_mod.log"
sound_mod_filename = folder + "sound_mod.log"
sound_classes_filename = folder + "sound_mod.log_classes"
tree_filename = folder + "MCMC_test.trees.log"
with open(tree_filename, "r") as infile:
ct = infile.readlines()[-1]
tree = ct.strip().split()[-1]
tr_mod = srcAn.Evaluator.read_file(tr_filename)
tr_d = {k.split("_")[1]: v for k, v in zip(tr_mod.iloc[-1:].columns, tr_mod.iloc[-1:].values[0])}
sound_mod = srcAn.Evaluator.read_file(sound_mod_filename).iloc[-1:]
sc = srcAn.Evaluator.read_file(sound_classes_filename)
sm_d = sound_mod_dict(sound_mod, sc, "asjp")
return tree, sm_d, tr_d
def sound_mod_dict(sound_mod, sound_classes, dialect):
header = sound_mod.columns.tolist()
# header for evo class columns
evo_class_value_header = [i for i in header if i.split("_")[0] == "clv"]
# frequencies column names
freq_header = [i for i in header if i.split("_")[0] == "freq"]
# indices of the class value for the class
class_indices = sound_classes.values[0].tolist()
# get the names of the sounds
names = [i.split("_")[1] for i in freq_header]
names = [i.decode("utf-8") for i in names]
dct = {"names": names,
"freqs": np.array([sound_mod[i].values[0] for i in freq_header]),
"evo_map": class_indices,
"evo_vals": np.array([sound_mod[i].values[0] for i in evo_class_value_header]),
"model": dialect}
return dct
def calc_temp(my_rank, heat_scale=0.1):
# type: (int, float) -> float
"""
This function calculates the temperature of the heated chains.
:param my_rank: rank of the MCMC in the mpi setting
:type my_rank: float|int
:param heat_scale: scale factor for the heat
:type heat_scale: float
:return: the temperature of the chain
:rtype: float
"""
return 1.0 / (1.0 + heat_scale * my_rank)
if __name__ == '__main__':
# get time of start
now = datetime.datetime.now()
# get location of settings file from commandline parameter
parser = argparse.ArgumentParser(description="Model Setup for Historical Linguistics Statistical Alignment")
parser.add_argument("-s", dest="Settings", type=str, help="path to settings file", required=True)
args = parser.parse_args()
# get parameters from settings file
parameter_dict = srcRead.read_settings_file(args.Settings)
if parameter_dict["MC3"]:
comm = MPI.COMM_WORLD
rank = comm.Get_rank()
mpi_size = comm.Get_size()
temperature = calc_temp(my_rank=rank)
print("rank " + str(rank) + " starts model setup now.")
MCMC_mod = MCMC_MC3_setup(temperature, mpi_size, now)
print("rank " + str(rank) + " model setup done.")
comm.Barrier()
if parameter_dict["randomize"]:
# randomize starting point if desired
MCMC_mod.randomize_starting_point(parameter_dict["randomize steps"], window_size=parameter_dict["Window size"])
MCMC_mod.MC3_estimate(parameter_dict["Iterations"], parameter_dict["Thinning"], wsize=parameter_dict["Window size"])
comm.Barrier()
else:
MCMC_mod = MCMC_setup()
wsize = parameter_dict["Window size"]
# start the MCMC
if parameter_dict["randomize"]:
MCMC_mod.randomize_starting_point(parameter_dict["randomize steps"], window_size=wsize)
MCMC_mod.estimate(parameter_dict["Iterations"], parameter_dict["Thinning"], wsize=wsize)
print "we are done"