-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathCLOptimization.py
223 lines (170 loc) · 6.69 KB
/
CLOptimization.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
import tensorflow as tf
from tensorflow.keras import layers
from numpy import *
from scipy import *
import matplotlib.pyplot as plt
import os
import time
import sys
from random import randint
arguments = len(sys.argv)-1
pos = 1
while(arguments >= pos):
arg = sys.argv[pos]
is_restart = sys.argv[pos+1]
print ("Parameter %i: %s %s"%(pos,sys.argv[pos],is_restart))
pos = pos+2
#get one sample airfoil data and check for size
datafilename='./GenerateData/Airfoils/genairfoil%03d.txt'%0
data = loadtxt(datafilename)
npoints = size(data[:,0])
#get one sample airfoil data and check for size
datafilename='./GenerateData/NoiseInput/noiseinput%03d.txt'%0
data = loadtxt(datafilename)
noise_dim = size(data)
NOISE_DIM=noise_dim
TOTAL_SIZE=960
noise_input = zeros([TOTAL_SIZE,noise_dim])
for i in arange(0,TOTAL_SIZE,1):
datafilename='./GenerateData/NoiseInput/noiseinput%03d.txt'%i
data = loadtxt(datafilename)
noise_input[i,:] = data
CL_predicted = zeros([noise_dim])
datafilename='CL_new_airfoils.txt'
data = loadtxt(datafilename)
CL_predicted = data
def make_CLlearner_model():
model = tf.keras.Sequential()
model.add(layers.Conv2D(128, (5, 5), strides=(1, 1), padding='same',
input_shape=[noise_dim,]))
model.add(layers.LeakyReLU())
model.add(layers.Dropout(0.3))
model.add(layers.Conv2D(256, (5, 5), strides=(2, 2), padding='same'))
model.add(layers.LeakyReLU())
model.add(layers.Dropout(0.3))
model.add(layers.Flatten())
model.add(layers.Dense(1))
return model
def make_simpler_model():
model = tf.keras.models.Sequential()
model.add(tf.keras.layers.Dense(256, input_shape=[noise_dim,], activation='relu', kernel_initializer='he_uniform'))
model.add(tf.keras.layers.Dense(256, activation='relu', kernel_initializer='he_uniform'))
model.add(tf.keras.layers.Dense(256, activation='relu', kernel_initializer='he_uniform'))
model.add(layers.Flatten())
model.add(tf.keras.layers.Dense(1))
return model
#CLlearner = make_CLlearner_model()
CLlearner = make_simpler_model()
loss_fn = tf.keras.losses.MeanSquaredError()
CLlearner.compile(optimizer='adam',
loss=loss_fn)
checkpoint_dir = './Checkpoint_CL_Noise'
checkpoint = tf.train.Checkpoint(model=CLlearner)
status = checkpoint.restore(tf.train.latest_checkpoint(checkpoint_dir))
def make_generator_model():
model = tf.keras.Sequential()
model.add(layers.Dense(npoints*2*32, use_bias=False, input_shape=(NOISE_DIM,)))
model.add(layers.BatchNormalization())
model.add(layers.LeakyReLU())
model.add(layers.Reshape((npoints, 2, 32)))
assert model.output_shape == (None, npoints, 2, 32) # Note: None is the batch size
model.add(layers.Conv2DTranspose(16, (5, 5), strides=(1, 1), padding='same', use_bias=False))
assert model.output_shape == (None, npoints, 2, 16)
model.add(layers.BatchNormalization())
model.add(layers.LeakyReLU())
model.add(layers.Conv2DTranspose(16, (5, 5), strides=(1, 1), padding='same', use_bias=False))
assert model.output_shape == (None, npoints, 2, 16)
model.add(layers.BatchNormalization())
model.add(layers.LeakyReLU())
model.add(layers.Conv2DTranspose(1, (5, 5), strides=(1, 1), padding='same', use_bias=False, activation='tanh'))
assert model.output_shape == (None, npoints, 2, 1)
return model
generator = make_generator_model()
def make_discriminator_model():
model = tf.keras.Sequential()
model.add(layers.Conv2D(64, (5, 5), strides=(1, 1), padding='same',
input_shape=[npoints, 2, 1]))
model.add(layers.LeakyReLU())
model.add(layers.Dropout(0.3))
model.add(layers.Conv2D(128, (5, 5), strides=(2, 2), padding='same'))
model.add(layers.LeakyReLU())
model.add(layers.Dropout(0.3))
model.add(layers.Flatten())
model.add(layers.Dense(1))
return model
discriminator = make_discriminator_model()
generator_optimizer = tf.keras.optimizers.Adam(1e-4)
discriminator_optimizer = tf.keras.optimizers.Adam(1e-4)
checkpoint_dir = './Checkpoint_AirfoilGenerator'
checkpoint_prefix = os.path.join(checkpoint_dir, "ckpt")
checkpoint = tf.train.Checkpoint(generator_optimizer=generator_optimizer,
discriminator_optimizer=discriminator_optimizer,
generator=generator,
discriminator=discriminator)
status = checkpoint.restore(tf.train.latest_checkpoint(checkpoint_dir))
def generate_airfoil(iteration,seed,iter_vec,peri_vec,is_save):
generate_and_save_images(generator,iteration,seed,iter_vec,peri_vec,is_save)
niter = 300
ntries = 50
for tryno in arange(0,ntries,1):
filename = 'seed_for_opt_%03d.txt'%tryno
peri_vec = zeros(niter)
seed = tf.random.normal([1,noise_dim])
seed1 = seed
def generate_and_save_images(model,iteration,seed,iter_vec,peri_vec,is_save):
predictions1 = model(seed1, training=False)
predictions = model(seed, training=False)
plt.subplot(1, 2, 1)
plt.plot(predictions[0,:,0,0],predictions[0,:,1,0],linewidth=2)
plt.plot(predictions1[0,:,0,0],predictions1[0,:,1,0],'r')
plt.ylim([-0.15,0.15])
plt.xlim([0.0,1.0])
plt.axis('equal')
plt.subplot(1, 2, 2)
plt.plot(iter_vec[0:iteration],peri_vec[0:iteration],'o',markersize=10)
plt.suptitle('Iteration=%03d, CL=%f'%(iteration,peri_vec[iteration-1]),fontsize=20)
if(is_save==1):
imgfilename = './Images_Optimization/Optimization_Trial=%03d'%tryno
plt.savefig(imgfilename)
airfoil_filename='AirfoilData%03d.txt'%tryno
file=open(airfoil_filename,"w")
for i in arange(0,size(predictions[0,:,0,0]),1):
file.write('%f %f\n'%(predictions[0,i,0,0],predictions[0,i,1,0]))
file.close()
plt.draw()
plt.pause(0.001)
plt.clf()
# Do the optimization
eps = 1e-6
iter_vec = zeros(niter)
def eps_array(indx):
eps_array = zeros(noise_dim)
eps_array[indx] = eps
return eps_array
f_x_temp = 0.0
wait = 10
count=0
for iteration in arange(0,niter,1):
iter_vec[iteration] = iteration
grad_f = zeros(noise_dim)
#seed = tf.random.normal([1,noise_dim])+0.1*tf.random.normal([1,noise_dim])
f_x = CLlearner.predict(seed)
if(abs(f_x_temp-f_x)<=1e-6):
count = count + 1
if(abs(f_x_temp-f_x)<=1e-6 and count > wait):
print f_x
file = open(filename,'w')
for i in arange(0,noise_dim,1):
file.write('%f\n'%seed1[0,i])
file.write('%f\n'%f_x)
file.close()
generate_airfoil(iteration,seed,iter_vec,peri_vec,1)
break
f_x_temp = f_x
peri_vec[iteration] = f_x
generate_airfoil(iteration,seed,iter_vec,peri_vec,0)
for ptindx in arange(0,noise_dim,1):
seedpluseps = seed + eps_array(ptindx)
f_xpluseps = CLlearner.predict(seedpluseps)
grad_f[ptindx] = (f_xpluseps-f_x)/eps
seed = seed + 0.05*grad_f