-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtestProb.py
188 lines (158 loc) · 6.92 KB
/
testProb.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
import numpy as np
import torch
import torchvision.transforms as T
import pandas as pd
import sys
# import matplotlib.pyplot as plt
import time
sys.path.append('../')
from models.unet import UNet
from common.data import my_read_csv,ProtoDataset,get_flow_stats,single_frame_from_cases,shifted_paths
from common.sde import forward_rollout_batch, get_fraction_ones_batch,evolve_batch,get_fraction_ones_batch,find_centroid_batch
import random
from argparse import ArgumentParser
parser = ArgumentParser()
# Trainer arguments
parser.add_argument("--dt", type=int, default=4)
parser.add_argument("--nt_sde", type=int, default=1)
parser.add_argument("--dt_allsde", type=float, default=8e-6)
parser.add_argument("--classes", type=int, default=2)
parser.add_argument("--batch_size", type=int, default=32)
parser.add_argument("--ckpt_path", type=str, default="../ckpt/pbatch/unet_default/model-epoch=45-val_acc=0.9976.ckpt")
parser.add_argument("--num_nodes", type=int, default=1)
parser.add_argument("--num_workers", type=int, default=20)
parser.add_argument("--tune",action='store_true')
parser.add_argument("--profiler",type=str,default=None)
parser.add_argument("--timeit", action='store_true')
parser.add_argument("--gpu",type=int,default=4)
parser.add_argument("--precision",type=int,default=16)
parser.add_argument("--case_name",type=str,default='tmp')
parser.add_argument("--test_path",type=str,default='./all_sheet.csv')
parser.add_argument("--nseed",type=int,default=10)
parser.add_argument("--start_seed",type=int,default=0)
parser.add_argument("--output_path",type=str,default='./outputs/')
parser.add_argument("--prefactor",type=int,default=10)
parser.add_argument("--data_path",type=str,default='../ign_bin/')
parser.add_argument("--ori_shape",type=str,default='217')
parser.add_argument("--stats_mode",type=str,default='default')
parser.set_defaults(tune=False)
parser.set_defaults(timeit=True)
parser.set_defaults(evolve_mean=False)
# Parse the user inputs and defaults (returns a argparse.Namespace)
args = parser.parse_args()
dt = args.dt
nt_sde = args.nt_sde
dt_allsde = args.dt_allsde
classes = args.classes
batch_size = args.batch_size
checkpoint_path = args.ckpt_path
num_nodes = args.num_nodes
tune = args.tune
profiler = args.profiler
timeit = args.timeit
gpu = args.gpu
precision = args.precision
case_name = args.case_name
test_path = args.test_path
nseed = args.nseed
start_seed = args.start_seed
output_path = args.output_path
num_workers = args.num_workers
prefactor = args.prefactor
data_path = args.data_path
ori_shape = args.ori_shape
stats_mode = args.stats_mode
if __name__ == '__main__':
dt_sde = dt_allsde/nt_sde
cfl = 400*dt_sde/min(45e-3/256,14.6e-3/160)
print("CFL: ", cfl)
inert_les = get_flow_stats('../donatella_inert/')
model = UNet(9,classes,firstC=64)
state_dict= torch.load(checkpoint_path,map_location=torch.device('cpu'))['state_dict']
new_state_dict = {}
for k in state_dict.keys():
new_state_dict[k[6:]] = state_dict[k]
model.load_state_dict(new_state_dict)
model.eval()
model = model.to('cuda')
test_dict = my_read_csv(test_path)
test_cases = test_dict['Case #']
# %%
if 'init' in data_path:
test_paths = shifted_paths(ori_shape=ori_shape)
else:
test_paths = single_frame_from_cases(test_cases,data_path=data_path)
validation_transformations = T.Compose([T.Resize((256,160))])
target_transformations = T.Compose([T.Resize((256,160),interpolation=0)])
test_ds = ProtoDataset(path_list=test_paths,
transform=validation_transformations,mode='prob',
target_transform=target_transformations,
classes=classes,data_path=data_path,sim_path='../donatella_inert/',stats_mode=stats_mode)
test_loader = torch.utils.data.DataLoader(test_ds,batch_size=batch_size,
shuffle=False,num_workers=num_workers)
print('Testing...')
if timeit:
start_time = time.time()
count = 0
test_iterator = iter(test_loader)
results_list= []
while True:
try:
print("Count: ",count)
X0,_ = next(test_iterator)
max_iter= (256-8)//dt
seed_list = list(range(start_seed,start_seed+nseed))
for seed in seed_list:
print("Seed: ",seed)
torch.manual_seed(seed)
random.seed(seed)
np.random.seed(seed)
pred_0d = []
X = X0.clone()
centroid = []
pred_0d.append(get_fraction_ones_batch(X[:,0].detach().numpy()))
centroid.append(find_centroid_batch(X))
for i in range(max_iter):
#evolve chemistry
X1pred = forward_rollout_batch(X,model,'cuda')
#evolve velocity
X = evolve_batch(X1pred, nt=nt_sde,dt_all=dt_allsde,p=prefactor)
pred_0d.append(get_fraction_ones_batch(X[:,0].detach().numpy()))
centroid.append(find_centroid_batch(X))
pred_no_sde = []
centroid_no_sde = []
X1 = X0.clone()
pred_no_sde.append(get_fraction_ones_batch(X1[:,0].detach().numpy()))
centroid_no_sde.append(find_centroid_batch(X1))
for i in range(max_iter):
X1 = forward_rollout_batch(X1,model,'cuda')
pred_no_sde.append(get_fraction_ones_batch(X1[:,0].detach().numpy()))
centroid_no_sde.append(find_centroid_batch(X1))
pred_0d = np.array(pred_0d)
pred_no_sde = np.array(pred_no_sde)
#stack list of 2d arrays along batch dimension
centroid = np.stack(centroid,axis=-1)
centroid_no_sde = np.stack(centroid_no_sde,axis=-1)
for ib in range(X0.shape[0]):
results_list.append({'id':test_paths[count+ib][0],
'init':test_paths[count+ib][1],
'seed':seed,
'pred':pred_0d[:,ib],
'pred_no_sde':pred_no_sde[:,ib],
'centroid':centroid[ib,:,:],
'centroid_no_sde':centroid_no_sde[ib,:,:]
})
count += X0.shape[0]
except StopIteration:
break
if 'init' in data_path:
initcase = str(ori_shape)
else:
initcase = ''
results_df = pd.DataFrame(results_list)
results_df.to_json(output_path+initcase+'ign_prob_seed'+str(nseed)+'_prefactor'+str(prefactor)+'CFL'+str(int(cfl))
+'seed_'+str(seed_list[0])+'-'+str(seed_list[-1])
+'stats_'+stats_mode+'.json')
if timeit:
end_time = time.time()
print('Test time: ' + str((end_time-start_time)/60))