-
Notifications
You must be signed in to change notification settings - Fork 19
/
Copy pathdataset.py
200 lines (161 loc) · 6.1 KB
/
dataset.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
"""dataset.py"""
import os
import numpy as np
import itertools
import torch
from torch.utils.data import Dataset
from torch_geometric.data import Batch
class CouetteDataset(Dataset):
def __init__(self, sims, dset_dir):
'Initialization'
self.sims = sims
self.dset_dir = dset_dir
self.dims = {'z':5, 'q':2, 'q_0':0, 'n':2, 'f':0, 'g':2}
self.dt = 1/150
def __getitem__(self, index):
'Generates one sample of data'
# Select sample
Re, We = self.sims[index]
# Load data
name = os.path.join(self.dset_dir, 'couette_Re_{:.1f}_We_{:.1f}.pt'.format(Re,We))
data = torch.load(name)
return data
def __len__(self):
return len(self.sims)
def get_stats(self, device):
mean = 0
for sim in range(len(self.sims)):
batch = Batch.from_data_list(self[sim])
mean += batch.y.sum(0)
mean = mean/len(self.sims)/len(batch.batch)
var = 0
for sim in range(len(self.sims)):
batch = Batch.from_data_list(self[sim])
var += ((batch.y - mean)**2).sum(0)
var = var/len(self.sims)/len(batch.batch)
std = var**0.5
std[std==0] = 1
return {'mean': mean.to(device), 'std': std.to(device)}, None
class BeamDataset(Dataset):
def __init__(self, sims, dset_dir):
'Initialization'
self.sims = sims
self.dset_dir = dset_dir
self.dims = {'z':12, 'q':3, 'q_0':0, 'n':2, 'f':3, 'g':0}
self.dt = 1/20
def __getitem__(self, index):
'Generates one sample of data'
# Select sample
load = self.sims[index][0]
# Load data
name = os.path.join(self.dset_dir, 'beam_{}.pt'.format(load+1))
data = torch.load(name)
return data
def __len__(self):
return len(self.sims)
def get_stats(self, device):
mean_1 = 0
mean_2 = 0
for sim in range(len(self.sims)):
batch = Batch.from_data_list(self[sim])
mean_1 += batch.y.sum(0)
mean_2 += batch.f.sum(0)
mean_1 = mean_1/len(self.sims)/len(batch.batch)
mean_2 = mean_2/len(self.sims)/len(batch.batch)
var_1 = 0
var_2 = 0
for sim in range(len(self.sims)):
batch = Batch.from_data_list(self[sim])
var_1 += ((batch.y - mean_1)**2).sum(0)
var_2 += ((batch.f - mean_2)**2).sum(0)
var_1 = var_1/len(self.sims)/len(batch.batch)
var_2 = var_2/len(self.sims)/len(batch.batch)
std_1 = var_1**0.5
std_1[std_1==0] = 1
std_2 = var_2**0.5
std_2[std_2==0] = 1
return {'mean': mean_1.to(device), 'std': std_1.to(device)}, {'mean': mean_2.to(device), 'std': std_2.to(device)}
class CylinderDataset(Dataset):
def __init__(self, sims, dset_dir):
'Initialization'
self.sims = sims
self.dset_dir = dset_dir
self.dims = {'z':3, 'q':0, 'q_0':2, 'n':4, 'f':0, 'g':0}
self.dt = 1/100
def __getitem__(self, index):
'Generates one sample of data'
# Select sample
pos, v = self.sims[index]
# Load data
name = os.path.join(self.dset_dir, 'cylinder_{}_v_{:.2f}.pt'.format(pos,v))
data = torch.load(name)
return data
def __len__(self):
return len(self.sims)
def get_stats(self, device):
'Computes the statistics of the dataset'
# Mean
mean_1 = 0
mean_2 = 0
for sim in range(len(self.sims)):
batch = Batch.from_data_list(self[sim])
mean_1 += batch.y.sum(0)
mean_2 += batch.q_0.sum(0)
mean_1 = mean_1/len(self.sims)/len(batch.batch)
mean_2 = mean_2/len(self.sims)/len(batch.batch)
# Variance
var_1 = 0
var_2 = 0
for sim in range(len(self.sims)):
batch = Batch.from_data_list(self[sim])
var_1 += ((batch.y - mean_1)**2).sum(0)
var_2 += ((batch.q_0 - mean_2)**2).sum(0)
var_1 = var_1/len(self.sims)/len(batch.batch)
var_2 = var_2/len(self.sims)/len(batch.batch)
# Standard Deviation
std_1 = var_1**0.5
std_1[std_1==0] = 1
std_2 = var_2**0.5
std_2[std_2==0] = 1
return {'mean': mean_1.to(device), 'std': std_1.to(device)}, {'mean': mean_2.to(device), 'std': std_2.to(device)}
def load_dataset(args):
# Dataset directory path
sys_name = args.sys_name
dset_dir = os.path.join(args.dset_dir, 'database_' + sys_name)
# Create Dataset instances
if args.sys_name == 'couette':
# Cases: Reynolds + Weisemberg
Re = [0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1]
We = [1, 1.1, 1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2]
train_sims, val_sims, test_sims = split_dataset(Re,We)
train_set = CouetteDataset(train_sims, dset_dir)
val_set = CouetteDataset(val_sims, dset_dir)
test_set = CouetteDataset(test_sims, dset_dir)
elif args.sys_name == 'beam':
# Cases: Load
load = list(range(52))
train_sims, val_sims, test_sims = split_dataset(load)
train_set = BeamDataset(train_sims, dset_dir)
val_set = BeamDataset(val_sims, dset_dir)
test_set = BeamDataset(test_sims, dset_dir)
elif args.sys_name == 'cylinder':
# Cases: Position + Velocity
pos = [1, 2, 3, 4, 5, 6]
v = [1, 1.25, 1.5, 1.75, 2]
train_sims, val_sims, test_sims = split_dataset(pos,v)
train_set = CylinderDataset(train_sims, dset_dir)
val_set = CylinderDataset(val_sims, dset_dir)
test_set = CylinderDataset(test_sims, dset_dir)
return train_set, val_set, test_set
def split_dataset(*args):
# Train, validation and test simulations
indices = list(itertools.product(*args))
N_sims = len(indices)
# Random split
np.random.shuffle(indices)
train_sims = indices[:int(0.8*N_sims)]
val_sims = indices[int(0.8*N_sims):int(0.9*N_sims)]
test_sims = indices[int(0.9*N_sims):]
return train_sims, val_sims, test_sims
if __name__ == '__main__':
pass