-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathutilities.py
248 lines (192 loc) · 8.6 KB
/
utilities.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
import torch
import numpy as np
import scipy.io
import h5py
import sklearn.metrics
from torch_geometric.data import Data
import torch.nn as nn
from scipy.ndimage import gaussian_filter
from torch_geometric.nn import GCNConv
import pdb
class MeshGenerator(object):
def __init__(self, real_space, mesh_size, attr_features=1,grid_input=np.array([])):
super(MeshGenerator, self).__init__()
self.d = len(real_space) #2
# self.m = sample_size #1000
self.attr_features = attr_features
assert len(mesh_size) == self.d
if self.d == 1:
self.n = mesh_size[0]
# self.grid = np.linspace(real_space[0][0], real_space[0][1], self.n).reshape((self.n, 1))
else:
self.n = 1
grids = []
for j in range(self.d):
# grids.append(np.linspace(real_space[j][0], real_space[j][1], mesh_size[j]))
# grids.append(np.linspace(real_space[j][0]+(0.5/mesh_size[j]), real_space[j][1]-(0.5/mesh_size[j]), mesh_size[j]))
self.n *= mesh_size[j]
self.idx = np.array(range(self.n))
self.grid=grid_input
self.grid_sample = self.grid
def sample(self, idx):
self.idx = torch.tensor(idx)
self.grid_sample = self.grid[self.idx]
return self.idx
def get_grid(self):
return torch.tensor(self.grid_sample, dtype=torch.float)
def deduplicate_rows(tensor):
unique_rows = []
seen_rows = set()
for row in tensor:
row_tuple = tuple(row.tolist())
if row_tuple not in seen_rows:
unique_rows.append(row)
seen_rows.add(row_tuple)
deduplicated_tensor = torch.stack(unique_rows)
return deduplicated_tensor
def ball_connectivity(self, is_forward=False,ns=10,tri_edge=None):
self.pwd = sklearn.metrics.pairwise_distances(self.grid_sample)
tri_edge = tri_edge.T
edge_index_1=np.array([])
edge_index_2=np.array([])
for i in range(self.grid_sample.shape[0]):
edge_index_1=np.append(edge_index_1,np.array([i]).repeat(ns+1))
edge_index_2=np.append(edge_index_2,np.argsort(self.pwd[i])[:ns+1])
self.edge_index = np.vstack([edge_index_1,edge_index_2])
self.edge_index = np.concatenate([self.edge_index,tri_edge],-1)
self.edge_index=torch.tensor(self.edge_index)
self.edge_index = torch.cat([self.edge_index, self.edge_index.flip(0)], dim=1)
self.edge_index = MeshGenerator.deduplicate_rows(self.edge_index.T).T
self.n_edges = self.edge_index.shape[1]
if is_forward:
print(self.edge_index.shape)
self.edge_index = self.edge_index[:, self.edge_index[0] >= self.edge_index[1]]
print(self.edge_index.shape)
self.n_edges = self.edge_index.shape[1]
return torch.tensor(self.edge_index, dtype=torch.long)
def attributes(self, theta=None):
# pwd = sklearn.metrics.pairwise_distances(self.grid_sample)
theta = theta[self.idx]
edge_attr = np.zeros((self.n_edges, 2 * self.d + 2*self.attr_features+1))
self.edge_index=torch.tensor(self.edge_index).to(torch.int64)
for p in range(self.n_edges):
edge_attr[p,6:7]=self.pwd[self.edge_index[0][p]][self.edge_index[1][p]]
edge_attr[:, 4:5] = theta[self.edge_index[0]].view(-1, self.attr_features)
edge_attr[:, 5:6] = theta[self.edge_index[1]].view(-1, self.attr_features)
edge_attr[:, 0:4] = self.grid_sample[self.edge_index.T].reshape((self.n_edges, -1))
return torch.tensor(edge_attr, dtype=torch.float)
# normalization, Gaussian
class GaussianNormalizer(object):
def __init__(self, x, eps=0.00001):
super(GaussianNormalizer, self).__init__()
self.mean = torch.mean(x)
self.std = torch.std(x)
self.eps = eps
def encode(self, x):
x = (x - self.mean) / (self.std + self.eps)
return x
def decode(self, x, sample_idx=None):
x = (x * (self.std + self.eps)) + self.mean
return x
# def cuda(self):
# self.mean = self.mean.cuda()
# self.std = self.std.cuda()
# def cpu(self):
# self.mean = self.mean.cpu()
# self.std = self.std.cpu()
def cuda(self,device):
self.mean = self.mean.to(device)
self.std = self.std.to(device)
def cpu(self):
self.mean = self.mean.cpu()
self.std = self.std.cpu()
#loss function with rel/abs Lp loss
class LpLoss(object):
def __init__(self, d=2, p=2, size_average=False, reduction=True):
super(LpLoss, self).__init__()
#Dimension and Lp-norm type are postive
assert d > 0 and p > 0
self.d = d
self.p = p
self.reduction = reduction
self.size_average = size_average
def abs(self, x, y):
num_examples = x.size()[0]
#Assume uniform mesh
h = 1.0 / (x.size()[1] - 1.0)
all_norms = (h**(self.d/self.p))*torch.norm(x.view(num_examples,-1) - y.view(num_examples,-1), self.p, 1)
if self.reduction:
if self.size_average:
return torch.mean(all_norms)
else:
return torch.sum(all_norms)
return all_norms
def rel(self, x, y):
num_examples = x.size()[0] #x.size()=[1,num_indomain]
diff_norms = torch.norm(x.reshape(num_examples,-1) - y.reshape(num_examples,-1), self.p, 1) #pred-gd 求L2范数
y_norms = torch.norm(y.reshape(num_examples,-1), self.p, 1)
if self.reduction:
if self.size_average:
return torch.mean(diff_norms/y_norms)
else:
return torch.sum(diff_norms/y_norms)
return diff_norms/y_norms
def __call__(self, x, y):
return self.rel(x, y)
class RandomMeshGenerator(object):
def __init__(self, real_space, mesh_size, sample_size):
super(RandomMeshGenerator, self).__init__()
self.d = len(real_space)
self.m = sample_size
assert len(mesh_size) == self.d
if self.d == 1:
self.n = mesh_size[0]
self.grid = np.linspace(real_space[0][0], real_space[0][1], self.n).reshape((self.n, 1))
else:
self.n = 1
grids = []
for j in range(self.d):
grids.append(np.linspace(real_space[j][0], real_space[j][1], mesh_size[j]))
self.n *= mesh_size[j]
self.grid = np.vstack([xx.ravel() for xx in np.meshgrid(*grids)]).T
if self.m > self.n:
self.m = self.n
self.idx = np.array(range(self.n))
self.grid_sample = self.grid
def sample(self):
perm = torch.randperm(self.n)
self.idx = perm[:self.m]
self.grid_sample = self.grid[self.idx]
return self.idx
def get_grid(self):
return torch.tensor(self.grid_sample, dtype=torch.float)
def ball_connectivity(self, r):
pwd = sklearn.metrics.pairwise_distances(self.grid_sample)
self.edge_index = np.vstack(np.where(pwd <= r))
self.n_edges = self.edge_index.shape[1]
return torch.tensor(self.edge_index, dtype=torch.long)
def gaussian_connectivity(self, sigma):
pwd = sklearn.metrics.pairwise_distances(self.grid_sample)
rbf = np.exp(-pwd**2/sigma**2)
sample = np.random.binomial(1,rbf)
self.edge_index = np.vstack(np.where(sample))
self.n_edges = self.edge_index.shape[1]
return torch.tensor(self.edge_index, dtype=torch.long)
def attributes(self, f=None, theta=None):
if f is None:
if theta is None:
edge_attr = self.grid[self.edge_index.T].reshape((self.n_edges, -1))
else:
theta = theta[self.idx]
edge_attr = np.zeros((self.n_edges, 3 * self.d))
edge_attr[:, 0:2 * self.d] = self.grid_sample[self.edge_index.T].reshape((self.n_edges, -1))
edge_attr[:, 2 * self.d] = theta[self.edge_index[0]]
edge_attr[:, 2 * self.d + 1] = theta[self.edge_index[1]]
else:
xy = self.grid_sample[self.edge_index.T].reshape((self.n_edges, -1))
if theta is None:
edge_attr = f(xy[:, 0:self.d], xy[:, self.d:])
else:
theta = theta[self.idx]
edge_attr = f(xy[:, 0:self.d], xy[:, self.d:], theta[self.edge_index[0]], theta[self.edge_index[1]])
return torch.tensor(edge_attr, dtype=torch.float)