-
Notifications
You must be signed in to change notification settings - Fork 9
/
sgp_regression.py
421 lines (351 loc) · 16 KB
/
sgp_regression.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
import pdb
import pickle
import sys
import os
import os.path
import collections
import torch
import argparse
import pandas as pd
from tqdm import tqdm
import itertools
from scipy.spatial.distance import pdist
import matplotlib.pyplot as plt
from search_methods.sparse_gp import SparseGP
import scipy.stats as sps
import numpy as np
import scipy.io
from scipy.io import loadmat
from scipy.stats import pearsonr
import time
from shutil import copy
from copy import deepcopy
sys.path.append('%s/../software/enas' % os.path.dirname(os.path.realpath(__file__)))
sys.path.append('%s/..' % os.path.dirname(os.path.realpath(__file__)))
sys.path.insert(0, '../')
from utils import *
from layers.models_ig import CktGNN, DVAE
from layers.dagnn_pyg import DAGNN
from layers.constants import *
'''Experiment settings'''
parser = argparse.ArgumentParser(description='SGP regression on Ckt-Bench-101.')
# must specify
parser.add_argument('--data-fold-name', default='CktBench101', help='dataset fold name')
parser.add_argument('--data-name', default='ckt_bench_101', help='circuit benchmark dataset name')
parser.add_argument('--save-appendix', default='_cktgnn', help='identifuy the encoder')
parser.add_argument('--checkpoint', type=int, default=300, help="load which epoch's model checkpoint")
parser.add_argument('--res-dir', default='res/',
help='where to save the Bayesian optimization results')
parser.add_argument('--reprocess', action='store_true', default=False,
help='if True, reprocess data instead of using prestored .pkl data')
# Model configuration
parser.add_argument('--model', default='CktGNN', help='model to use: DVAE, SVAE, \
DVAE_fast, DVAE_BN, SVAE_oneshot, DVAE_GCN')
parser.add_argument('--hs', type=int, default=301, metavar='N',
help='hidden size of GRUs')
parser.add_argument('--nz', type=int, default=66, metavar='N',
help='number of dimensions of latent vectors z')
parser.add_argument('--bidirectional', action='store_true', default=False,
help='whether to use bidirectional encoding')
parser.add_argument('--nvt', type=int, default=26, help='number of different node (subgraph) types')
parser.add_argument('--max_n', type=int, default=8, help='number of different node (subgraph) types')
parser.add_argument('--subg_nvt', type=int, default=10, help='number of subgraph nodes')
parser.add_argument('--subn_nvt', type=int, default=10, help='number of subgraph feat')
parser.add_argument('--ng', type=int, default=10000, help='number of circuits in the dataset')
parser.add_argument('--node_feat_type', type=str, default='discrete', help='node feature type: discrete or continuous')
parser.add_argument('--emb_dim', type=int, default=24, metavar='N', help='embdedding dimension')
parser.add_argument('--feat_emb_dim', type=int, default=8, metavar='N', help='embdedding dimension')
parser.add_argument('--dagnn_layers', type=int, default=2)
parser.add_argument('--dagnn_agg', type=str, default=NA_ATTN_H)
parser.add_argument('--dagnn_out_wx', type=int, default=0, choices=[0, 1])
parser.add_argument('--dagnn_out_pool_all', type=int, default=0, choices=[0, 1])
parser.add_argument('--dagnn_out_pool', type=str, default=P_MAX, choices=[P_ATTN, P_MAX, P_MEAN, P_ADD])
parser.add_argument('--dagnn_dropout', type=float, default=0.0)
# device setting
parser.add_argument('--cuda_id', type=int, default=1, metavar='N',
help='id of GPU')
parser.add_argument('--infer-batch-size', type=int, default=128, metavar='N',
help='batch size during inference')
parser.add_argument('--no-cuda', action='store_true', default=False,
help='disables CUDA training')
parser.add_argument('--all-gpus', action='store_true', default=False,
help='use all available GPUs')
parser.add_argument('--seed', type=int, default=1, metavar='S',
help='random seed (default: 1)')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
torch.manual_seed(args.seed)
if not args.cuda:
device = torch.device("cpu")
else:
torch.cuda.manual_seed(args.seed)
device = torch.device("cuda:{}".format(args.cuda_id))
np.random.seed(args.seed)
random.seed(args.seed)
# file dirs
args.file_dir = os.path.dirname(os.path.realpath('__file__'))
args.res_dir = os.path.join(args.file_dir, 'results/{}{}'.format(args.data_name,args.save_appendix))
if not os.path.exists(args.res_dir):
os.makedirs(args.res_dir)
test_results_name = os.path.join(args.res_dir, 'sgp_results.txt')
# loading data
args.data_dir = os.path.join(args.file_dir, 'OCB/{}'.format(args.data_fold_name))
data_name = args.data_name
if args.model.startswith('SVAE'):
data_type = 'tensor'
data_name += '_tensor'
elif args.model.startswith('DAGNN'):
data_type = 'pygraph'
data_name += '_pygraph'
else:
data_type = 'igraph'
pkl_name = os.path.join(args.data_dir, data_name + '.pkl')
with open(pkl_name, 'rb') as f:
all_datasets = pickle.load(f)
train_dataset = all_datasets[0]
test_dataset = all_datasets[1]
# determine data formats according to models, DVAE: igraph, SVAE: string (as tensors)
if args.model.startswith('CktGNN'):
train_data = [train_dataset[i][0] for i in range(len(train_dataset))]
test_data = [test_dataset[i][0] for i in range(len(test_dataset))]
elif args.model.startswith('DAGNN'):
train_data = [train_dataset[i] for i in range(len(train_dataset))]
test_data = [test_dataset[i] for i in range(len(test_dataset))]
else:
train_data = [train_dataset[i][1] for i in range(len(train_dataset))]
test_data = [test_dataset[i][1] for i in range(len(test_dataset))]
if args.model.startswith('CktGNN'):
nvt = 26
START_TYPE = 0
END_TYPE = 1
max_n = 8
max_pos = 8
subn_nvt = 40
subg = True
else:
nvt = 10
START_TYPE = 8
END_TYPE = 9
max_n = 24
subn_nvt=103
subg = False
def performance_readout(num_graphs, file_dir='circuit', name = 'ckt_simulation_summary_10000.txt'):
num_graphs = 10000
pbar = tqdm(range(num_graphs))
gain = []
bw = []
pm = []
fom = []
valid = []
#with open('ckt_simulation_summary_10000.txt', 'r') as f:
file_name = os.path.join(file_dir, name)
with open(file_name, 'r') as f:
for i in pbar:
row = f.readline().strip().split()
if not row[1] == 'Simulation':
g = float(row[1])/100.0
p = float(row[2])/-90.0
b = float(row[3])/1e9
gain.append(g)
pm.append(p)
bw.append(b)
fo = 1.2 * np.abs(g) + 1.6 * p + 10 * b
fom.append(fo)
valid.append(1)
else:
gain.append(0)
pm.append(0)
bw.append(0)
fom.append(0)
valid.append(0)
gain = np.array(gain) - np.min(gain) + 0.00001
pm = np.array(pm) - np.min(pm) + 0.00001
perform = {'valid':valid, 'gain':gain, 'pm':pm, 'bw':bw, 'fom':fom}
perform_df = pd.DataFrame(perform)
out_name = os.path.join(file_dir, 'perform.csv')
perform_df.to_csv(out_name)
return perform_df
def extract_latent(data, perform_df, start_idx=0):
model.eval()
Z = []
Y = []
Gain = []
BW = []
PM = []
g_batch = []
for i, g in enumerate(tqdm(data)):
if args.model.startswith('SVAE'):
g_ = g.to(device)
elif args.model.startswith('DAGNN'):
g_ = deepcopy(g)
else:
g_ = g.copy()
#if perform_df['valid'][start_idx + i] == 1:
g_batch.append(g_)
if len(g_batch) == args.infer_batch_size or i == len(data) - 1:
g_batch = model._collate_fn(g_batch)
mu, _ = model.encode(g_batch)
mu = mu.cpu().detach().numpy()
Z.append(mu)
g_batch = []
#if perform_df['valid'][start_idx + i] == 1:
y = perform_df['fom'][start_idx + i]
gain = perform_df['gain'][start_idx + i]
bw = perform_df['bw'][start_idx + i]
pm = perform_df['pm'][start_idx + i]
Y.append(y)
Gain.append(gain)
BW.append(bw)
PM.append(pm)
return np.concatenate(Z, 0), np.array(Y), np.array(Gain), np.array(BW), np.array(PM)
'''Extract latent representations Z'''
def save_latent_representations(epoch, perform_df):
Z_train, Y_train, Gain_train, BW_train, PM_train = extract_latent(train_data, perform_df, 0)
Z_test, Y_test, Gain_test, BW_test, PM_test = extract_latent(test_data, perform_df, 9000)
latent_pkl_name = os.path.join(args.res_dir, args.data_name +
'_latent_epoch{}.pkl'.format(epoch))
latent_mat_name = os.path.join(args.res_dir, args.data_name +
'_latent_epoch{}.mat'.format(epoch))
with open(latent_pkl_name, 'wb') as f:
pickle.dump((Z_train, Y_train, Z_test, Y_test), f)
print('Saved latent representations to ' + latent_pkl_name)
scipy.io.savemat(latent_mat_name,
mdict={
'Z_train': Z_train,
'Z_test': Z_test,
'Y_train': Y_train,
'Y_test': Y_test,
'Gain_train': Gain_train,
'Gain_test': Gain_test,
'BW_train':BW_train,
'BW_test':BW_test,
'PM_train':PM_train,
'PM_test':PM_test
}
)
# other BO hyperparameters
lr = 0.0005 # the learning rate to train the SGP model 0.0005
max_iter = 100 # how many iterations to optimize the SGP each time
#perform_df = performance_readout(args.ng, file_dir=args.data_dir)
perf_name = os.path.join(args.data_dir, 'perform101.csv')
perform_df = pd.read_csv(perf_name)
for rand_idx in range(1,10):
print('START FOLD: {}'.format(rand_idx))
save_dir = os.path.join(args.res_dir,'sgp_reg_{}_{}/'.format(args.save_appendix, rand_idx))
# set seed
random_seed = rand_idx
torch.manual_seed(random_seed)
torch.cuda.manual_seed(random_seed)
np.random.seed(random_seed)
if args.model.startswith('CktGNN'):
model = CktGNN(
max_n = max_n,
max_pos = max_pos,
nvt = nvt,
subn_nvt = subn_nvt,
START_TYPE = START_TYPE,
END_TYPE = END_TYPE,
emb_dim = args.emb_dim,
feat_emb_dim = args.feat_emb_dim,
hs=args.hs,
nz=args.nz,
pos=True
)
elif args.model.startswith('DAGNN'):
model = DAGNN(
emb_dim = 10,
hidden_dim = args.hs,
out_dim = args.hs,
max_n = max_n,
nvt = nvt,
START_TYPE = START_TYPE,
END_TYPE = END_TYPE,
hs=args.hs,
nz=args.nz,
num_nodes=nvt+2,
agg=args.dagnn_agg,
num_layers=args.dagnn_layers,
bidirectional=args.bidirectional,
out_wx=args.dagnn_out_wx > 0,
out_pool_all=args.dagnn_out_pool_all,
out_pool=args.dagnn_out_pool,
dropout=args.dagnn_dropout
)
else:
model = eval(args.model)(
max_n = max_n,
nvt = nvt,
feat_nvt = subn_nvt,
START_TYPE = START_TYPE,
END_TYPE = END_TYPE,
hs=args.hs,
nz=args.nz
)
model.to(device)
load_module_state(model, os.path.join(args.res_dir, 'model_checkpoint{}.pth'.format(args.checkpoint)), device=device)
X_train, Y_train, Gain_train, BW_train, PM_train = extract_latent(train_data, perform_df, 0)
X_test, Y_test, Gain_test, BW_test, PM_test = extract_latent(test_data, perform_df, 9000)
y_train = -Y_train.reshape((-1,1))
gain_train = -Gain_train.reshape((-1,1))
bw_train = -BW_train.reshape((-1,1))
pm_train = -PM_train.reshape((-1,1))
mean_y_train, std_y_train = np.mean(y_train), np.std(y_train)
mean_gain_train, std_gain_train = np.mean(gain_train), np.std(gain_train)
mean_bw_train, std_bw_train = np.mean(bw_train), np.std(bw_train)
mean_pm_train, std_pm_train = np.mean(pm_train), np.std(pm_train)
#print('Mean, std of y_train is ', mean_y_train, std_y_train)
y_train = (y_train - mean_y_train) / std_y_train
gain_train = (gain_train - mean_gain_train) / std_gain_train
bw_train = (bw_train - mean_bw_train) / std_bw_train
pm_train = (pm_train - mean_pm_train) / std_pm_train
y_test = -Y_test.reshape((-1,1))
y_test = (y_test - mean_y_train) / std_y_train
gain_test = -Gain_test.reshape((-1,1))
gain_test = (gain_test - mean_gain_train) / std_gain_train
bw_test = -BW_test.reshape((-1,1))
bw_test = (bw_test - mean_bw_train) / std_bw_train
pm_test = -PM_test.reshape((-1,1))
pm_test = (pm_test - mean_pm_train) / std_pm_train
'''SGP regression begins here'''
print("Average pairwise distance between train points = {}".format(np.mean(pdist(X_train))))
print("Average pairwise distance between test points = {}".format(np.mean(pdist(X_test))))
M = 500
### Predicting FoM
sgp_fom = SparseGP(X_train, 0 * X_train, y_train, M)
sgp_fom.train_via_ADAM(X_train, 0 * X_train, y_train, X_test, X_test * 0, y_test, minibatch_size = 2 * M, max_iterations = max_iter, learning_rate = lr)
pred_fom, uncert_fom = sgp_fom.predict(X_test, 0 * X_test)
error_fom= np.sqrt(np.mean((pred_fom - y_test)**2))
testll_fom = np.mean(sps.norm.logpdf(pred_fom - y_test, scale = np.sqrt(uncert_fom)))
pearson_fom = float(pearsonr(pred_fom.reshape(-1,), y_test.reshape(-1,))[0])
print('Fom RMSE: ', error_fom)
print('Fom Pearson r: ', pearson_fom)
### Predicting Gain
sgp_gain = SparseGP(X_train, 0 * X_train, gain_train, M)
sgp_gain.train_via_ADAM(X_train, 0 * X_train, gain_train, X_test, X_test * 0, gain_test, minibatch_size = 2 * M, max_iterations = max_iter, learning_rate = lr)
pred_gain, uncert_gain = sgp_gain.predict(X_test, 0 * X_test)
error_gain= np.sqrt(np.mean((pred_gain - gain_test)**2))
testll_gain = np.mean(sps.norm.logpdf(pred_gain - gain_test, scale = np.sqrt(uncert_gain)))
pearson_gain = float(pearsonr(pred_gain.reshape(-1,), gain_test.reshape(-1,))[0])
print('Gain RMSE: ', error_gain)
print('Gain Pearson r: ', pearson_gain)
### Predicting bw
sgp_bw = SparseGP(X_train, 0 * X_train, bw_train, M)
sgp_bw.train_via_ADAM(X_train, 0 * X_train, bw_train, X_test, X_test * 0, bw_test, minibatch_size = 2 * M, max_iterations = max_iter, learning_rate = lr)
pred_bw, uncert_bw = sgp_bw.predict(X_test, 0 * X_test)
error_bw= np.sqrt(np.mean((pred_bw - bw_test)**2))
testll_bw = np.mean(sps.norm.logpdf(pred_bw - bw_test, scale = np.sqrt(uncert_bw)))
pearson_bw = float(pearsonr(pred_bw.reshape(-1,), bw_test.reshape(-1,))[0])
print('BW RMSE: ', error_bw)
print('BW Pearson r: ', pearson_bw)
### Predicting pm
sgp_pm = SparseGP(X_train, 0 * X_train, pm_train, M)
sgp_pm.train_via_ADAM(X_train, 0 * X_train, pm_train, X_test, X_test * 0, pm_test, minibatch_size = 2 * M, max_iterations = max_iter, learning_rate = lr)
pred_pm, uncert_pm = sgp_pm.predict(X_test, 0 * X_test)
error_pm= np.sqrt(np.mean((pred_pm - pm_test)**2))
testll_pm = np.mean(sps.norm.logpdf(pred_pm - pm_test, scale = np.sqrt(uncert_pm)))
pearson_pm = float(pearsonr(pred_pm.reshape(-1,), pm_test.reshape(-1,))[0])
print('PM RMSE: ', error_pm)
print('PM Pearson r: ', pearson_pm)
with open(test_results_name, 'a') as result_file:
result_file.write(" Run: {} Fom rmse: {:.4f} Fom pearson: {:.4f} Gain rmse: {:.4f} Gain pearson: {:.4f} Bw rmse: {:.4f} Bw pearson: {:.4f} Pm rmse: {:.4f} Pm pearson: {:.4f} \n".format(rand_idx,
error_fom, pearson_fom, error_gain, pearson_gain, error_bw, pearson_bw, error_pm, pearson_pm))