-
Notifications
You must be signed in to change notification settings - Fork 0
/
reasech-multi-process.py
401 lines (311 loc) · 12.8 KB
/
reasech-multi-process.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
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
#! /usr/bin/python
# -*- coding: utf-8 -*-
# @author izhangxm
# Copyright 2017 [email protected]. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================
import arviz as az
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
import pymc as pm
from itertools import product
import os.path as osp
from scipy.optimize import leastsq
import time
from scipy.integrate import solve_ivp
from scipy.integrate import odeint
import math
import sunode
import sunode.wrappers.as_pytensor
from copy import deepcopy
from multiprocessing import Process, Queue, cpu_count
from tqdm import tqdm
from datetime import datetime
import pickle
az.style.use("arviz-darkgrid")
import os
from pathlib import Path
class MyDataset(object):
def __init__(self, dataset_path):
self.dataset_path = dataset_path
df = pd.read_csv(dataset_path)
self.df = df
self._setup()
def _setup(self):
df = self.df
# 数据初步处理
# 计算反应速率rate，初始速率固定设置为0
for c_i, (col_name, col_sr) in enumerate(df.items()):
if "error" in col_name or "time" in col_name or "rate" in col_name:
continue
rate_col_name = f"{col_name}_rate"
rates = []
pre_t = None
pre_v = None
for th, (index, value) in zip(df['time'], col_sr.items()):
if int(index) == 0:
rates.append(0.0)
pre_t = th
pre_v = value
continue
delta_t = th - pre_t
delta_value = value - pre_v
# print(col_name, index, pre_t, th, pre_v ,value)
rates.append(delta_value / delta_t)
pre_t = th
pre_v = value
df[rate_col_name] = rates
# 准备输出值 Y
self.cct_names = []
for x in self.df.columns:
if "time" in x or "error" in x or "rate" in x:
continue
self.cct_names.append(x)
self.rates_names = [f"{x}_rate" for x in self.cct_names]
self.error_names = [f"{x}-error" for x in self.cct_names]
self.cct = self.df[self.cct_names].values
self.rates = self.df[self.rates_names].values
self.errors = self.df[self.error_names].values
def set_as_sim_dataset(self, dcdt_fuc, t_eval, y0, args):
y = odeint(dcdt_fuc, y0=y0, t=t_eval, args=args)
# y.shape (size, 10)
df_new = pd.DataFrame(columns=['time'] + self.cct_names)
df_new['time'] = t_eval
for c_name, col_val in zip(self.cct_names, np.transpose(y, [1,0])):
df_new[c_name] = col_val
df_new[f"{c_name}-error"] = 0.001
self.df = df_new
self._setup()
def set_dataset(self, c_pred_df):
df = deepcopy(c_pred_df)
for c_name in self.cct_names:
c_error_name = f"{c_name}-error"
n_len = len(c_pred_df[c_name].values)
_error = np.repeat(0.0001, n_len)
c_pred_df[c_error_name] = _error
self.df = df
self._setup()
def get_rates(self):
return self.rates
def get_df(self):
return self.df
def get_errors(self):
return self.errors
def get_cct(self):
return self.cct
def get_var_col_names(self):
return self.cct_names, self.rates_names, self.error_names
def get_weights(self):
max_value = self.df[self.cct_names].describe().loc['max'].values.max()
vars_max = self.df[self.cct_names].describe().loc['max']
weights = (max_value / vars_max).values
return np.array(weights)
def get_vars_max(self):
vars_max = self.df[self.cct_names].describe().loc['max'].values
return vars_max
def get_format_time(f_s=None):
haomiao = str(time.time()).split('.')[-1]
if f_s is None:
f_s = "%Y-%-m-%d %H:%M:%S."
return datetime.now().strftime(f_s) + haomiao
return datetime.now().strftime(f_s)
def plot_dataset(dataset, dataset_pred=None):
df = dataset.get_df()
cct_names, error_names = dataset.get_var_col_names()
cols = 5
rows = math.ceil(len(cct_names) / cols)
fig, fig_axes = plt.subplots(ncols=cols, nrows=rows, figsize=(4.2 * cols, 4 * rows), dpi=100)
if isinstance(fig_axes, np.ndarray):
fig_axes = fig_axes.reshape(-1)
else:
fig_axes = [fig_axes]
for i, axes in enumerate(fig_axes):
if i >= len(cct_names):
axes.axis('off')
continue
y_name = cct_names[i]
Y = df[y_name].values
axes.plot(df['time'].values, Y, '*', label=f"ob")
axes.set_ylabel(f'cct_{y_name}')
axes.set_xlabel(f'time(h)')
# axes.plot(df['time'].values, df[rates_names[i]].values, '+', label=f"rate")
if dataset_pred:
_df_pred = dataset_pred.get_df()
t_eval = _df_pred['time'].values
axes.plot(t_eval, _df_pred[y_name].values, 'r', label=f"c(t)")
# axes.plot(t_eval, dcdt_df[y_name].values,'g', label=f"c'(t)")
axes.legend()
# axes.set_title(f"{y_name}", fontsize=14)
plt.tight_layout()
plt.show()
def get_dcdt_func(k_kinetics):
k_kinetics = np.array(k_kinetics).astype(np.uint8)
def _dcdt_func(t, c, p):
r1 = p.k1 * c.xN2 if k_kinetics[0] == 1 else p.k1
r2 = p.k2 * c.xNH3 if k_kinetics[1] == 1 else p.k2
r3 = p.k3 * c.xNO2 if k_kinetics[2] == 1 else p.k3
r4 = p.k4 * c.xNO3 if k_kinetics[3] == 1 else p.k4
r5 = p.k5 * c.xNO2 if k_kinetics[4] == 1 else p.k5
r6 = p.k6 * c.xNO2 * c.xNO3 if k_kinetics[5] == 1 else p.k6
r7 = p.k7 * c.xNO3 if k_kinetics[6] == 1 else p.k7
r8 = p.k8 * c.xNO3 if k_kinetics[7] == 1 else p.k8
r9 = p.k9 * c.xNH3 if k_kinetics[8] == 1 else p.k9
r10 = p.k10 * c.xNOrg if k_kinetics[9] == 1 else p.k10
r11 = p.k11 * c.xNOrg if k_kinetics[10] == 1 else p.k11
dc_xNH3 = 2 * r1 + r7 + r10 - r2 - r6 - r9
dc_xNO3 = r3 - r7 - r4 - r8 + r11
dc_xNO2 = r2 + r4 - r3 - r6 - 2 * r5
dc_xNOrg = r8 + r9 - r10 - r11
dc_xN2 = r5 + r6 - r1
dc_ANH3 = (2 * r1 * (c.AN2 - c.ANH3) + (c.ANO3 - c.ANH3) * r7 + (c.ANOrg - c.ANH3) * r10) / c.xNH3
dc_ANO3 = ((c.ANO2 - c.ANO3) * r2 + (c.ANOrg - c.ANO3) * r11) / c.xNO3
dc_ANO2 = ((c.ANH3 - c.ANO2) * r2 + (c.ANO3 - c.ANO2) * r4) / c.xNO2
dc_ANOrg = ((c.ANO3 - c.ANOrg) * r8 + (c.ANH3 - c.ANOrg) * r9) / c.xNOrg
dc_AN2 = ((c.ANO2 - c.AN2) * r5 + (c.ANO2 * c.ANH3 - c.AN2) * r6) / c.xN2
# dcdts = [dc_xNH3, dc_xNO3, dc_xNO2, dc_xNOrg, dc_xN2, dc_ANH3, dc_ANO3, dc_ANO2, dc_ANOrg, dc_AN2]
dcdts = {
'xNH3': dc_xNH3,
'xNO3': dc_xNO3,
'xNO2': dc_xNO2,
'xNOrg': dc_xNOrg,
'xN2': dc_xN2,
'ANH3': dc_ANH3,
'ANO3': dc_ANO3,
'ANO2': dc_ANO2,
'ANOrg': dc_ANOrg,
'AN2': dc_AN2,
}
return dcdts
return _dcdt_func
def dcdt_func_for_odeint(c, t, ks, k_kinetics):
# print(c, t, ks, k_kinetics)
# print()
c_xNH3, c_xNO3, c_xNO2, c_xNOrg, c_xN2, c_ANH3, c_ANO3, c_ANO2, c_ANOrg, c_AN2 = c
r1 = ks[0] * c_xN2 if k_kinetics[0] == 1 else ks[0]
r2 = ks[1] * c_xNH3 if k_kinetics[1] == 1 else ks[1]
r3 = ks[2] * c_xNO2 if k_kinetics[2] == 1 else ks[2]
r4 = ks[3] * c_xNO3 if k_kinetics[3] == 1 else ks[3]
r5 = ks[4] * c_xNO2 if k_kinetics[4] == 1 else ks[4]
r6 = ks[5] * c_xNO2 * c_xNO3 if k_kinetics[5] == 1 else ks[5]
r7 = ks[6] * c_xNO3 if k_kinetics[6] == 1 else ks[6]
r8 = ks[7] * c_xNO3 if k_kinetics[7] == 1 else ks[7]
r9 = ks[8] * c_xNH3 if k_kinetics[8] == 1 else ks[8]
r10 = ks[9] * c_xNOrg if k_kinetics[9] == 1 else ks[9]
r11 = ks[10] * c_xNOrg if k_kinetics[10] == 1 else ks[10]
dc_xNH3 = 2 * r1 + r7 + r10 - r2 - r6 - r9
dc_xNO3 = r3 - r7 - r4 - r8 + r11
dc_xNO2 = r2 + r4 - r3 - r6 - 2 * r5
dc_xNOrg = r8 + r9 - r10 - r11
dc_xN2 = r5 + r6 - r1
dc_ANH3 = (2 * r1 * (c_AN2 - c_ANH3) + (c_ANO3 - c_ANH3) * r7 + (c_ANOrg - c_ANH3) * r10) / c_xNH3
dc_ANO3 = ((c_ANO2 - c_ANO3) * r2 + (c_ANOrg - c_ANO3) * r11) / c_xNO3
dc_ANO2 = ((c_ANH3 - c_ANO2) * r2 + (c_ANO3 - c_ANO2) * r4) / c_xNO2
dc_ANOrg = ((c_ANO3 - c_ANOrg) * r8 + (c_ANH3 - c_ANOrg) * r9) / c_xNOrg
dc_AN2 = ((c_ANO2 - c_AN2) * r5 + (c_ANO2 * c_ANH3 - c_AN2) * r6) / c_xN2
dcdts = [dc_xNH3, dc_xNO3, dc_xNO2, dc_xNOrg, dc_xN2, dc_ANH3, dc_ANO3, dc_ANO2, dc_ANOrg, dc_AN2]
return np.array(dcdts)
# simulator function
def competition_model(rng, t_eval, y0, ks, k_kinetics, size=None):
# print(y0)
y = odeint(dcdt_func_for_odeint, y0=y0, t=t_eval, args=(ks, k_kinetics))
return y
def r2_loss(pred, y):
r2_loss = 1 - np.square(pred - y).sum() / np.square(y - np.mean(y)).sum()
return r2_loss
def get_model(dataset, k_kinetics, k_sigma_priors=0.01, kf_type=0):
df = dataset.get_df()
times = df['time'].values
errors = dataset.get_errors()
rates = dataset.get_rates()
cct_names, error_names = dataset.get_var_col_names()
# 定义参数优化模型
mcmc_model = pm.Model()
## 参数个数
params_n = 11
parames ={}
with mcmc_model:
for ki in range(1, params_n + 1):
if kf_type == 0:
p_dense = pm.HalfNormal(f"k{ki}", sigma=k_sigma_priors)
else:
p_dense = pm.Normal(f"k{ki}",mu=0, sigma=k_sigma_priors)
parames[f"k{ki}"] = (p_dense, ())
parames['extra']= np.zeros(1)
c0 = {}
with mcmc_model:
for c_name in cct_names:
_maxx = df[c_name].values.max()
c0[f"{c_name}"] = (pm.HalfNormal(f"{c_name}_s", sigma=_maxx), ())
y_hat, _, problem, solver, _, _ = sunode.wrappers.as_pytensor.solve_ivp(
y0=c0,
params=parames,
rhs=get_dcdt_func(k_kinetics),
tvals=times,
t0=times[0],
)
sd = pm.HalfNormal('sd')
for c_name in cct_names:
pm.Normal(f'{c_name}', mu=y_hat[f"{c_name}"], sigma=sd, observed=df[f"{c_name}"].values)
pm.Deterministic(f'{c_name}_mu', y_hat[f"{c_name}"])
return mcmc_model
def get_predict_ks(idata):
parames_summary = az.summary(idata, round_to=10)
ks_names = [f"k{x+1}" for x in range(11)]
predict_ks = []
for k_name in ks_names:
k_v = parames_summary["mean"][k_name]
predict_ks.append(k_v)
return np.array(predict_ks)
class NTraceModel(Process):
def __init__(self, p_name, dataset, k_queue:Queue):
super(NTraceModel, self).__init__()
self.p_name = p_name
self.dataset = dataset
self.k_queue = k_queue
def run(self):
while not self.k_queue.empty():
try:
k_kinetics = np.array(self.k_queue.get(1, timeout=100))
print(f"{self.p_name} {k_kinetics}")
k_str = "".join([f"{x}" for x in k_kinetics])
k_order = int(k_str,2)
save_file_path = f"save_idata/{k_order}-idata.dt"
if os.path.exists(save_file_path):
continue
Path(save_file_path).touch()
dataset = self.dataset
mcmc_model = get_model(dataset, k_kinetics, k_sigma_priors=0.01, kf_type=0)
idata = pm.sample(draws=1000, model=mcmc_model, chains=4, cores=4, progressbar=False)
pickle.dump(idata,open(save_file_path, 'wb'))
except Exception as e:
pass
def mutil_run():
kk_list_all = list(product([0,1], repeat=11))
q = Queue(10000)
for k_k in kk_list_all:
q.put(k_k)
dataset = MyDataset("dataset/data.csv")
p_list = []
n_cpu = int(cpu_count()/4)
print(cpu_count(), n_cpu)
for i in range(n_cpu):
p = NTraceModel(f"Process-{i}", dataset, q)
p.start()
p_list.append(p)
for p in p_list:
p.join()
print("all_finished")
if __name__ == '__main__':
mutil_run()