Fixed normalization of outputs.

JSWSC_Tremblay_et_al_2020/MURaM-ARs/d_1000km/train_deepvel.py

@@ -170,7 +170,7 @@ def __init__(self, root, noise, option, norm_filename='network/simulation_proper
  self.vy_d_1000km_stddev = tmp['vy_d_1000km_stddev']
 
  def residual(self, inputs):
- 
+
  x = Conv2D(self.n_filters, (self.kernel_size, self.kernel_size), strides=(1, 1), padding='same',
  kernel_initializer='he_normal')(inputs)
  x = BatchNormalization()(x)
@@ -179,41 +179,41 @@ def residual(self, inputs):
  kernel_initializer='he_normal')(x)
  x = BatchNormalization()(x)
  x = add([x, inputs])
- 
+
  return x
 
  def define_network(self):
  print("Setting up network...")
- 
+
  inputs = Input(shape=(self.nx, self.ny, self.n_inputs))
  conv = Conv2D(self.n_filters, (self.kernel_size, self.kernel_size), strides=(1, 1), padding='same',
  kernel_initializer='he_normal', activation='relu')(inputs)
- 
+
  x = self.residual(conv)
  for i in range(self.n_conv_layers):
  x = self.residual(x)
- 
+
  x = Conv2D(self.n_filters, (self.kernel_size, self.kernel_size), strides=(1, 1), padding='same',
  kernel_initializer='he_normal')(x)
  x = BatchNormalization()(x)
  x = add([x, conv])
- 
+
  final = Conv2D(self.n_outputs, (1, 1), strides=(1, 1), padding='same',
  kernel_initializer='he_normal', activation='linear')(x)
- 
+
  self.model = Model(inputs=inputs, outputs=final)
- 
+
  json_string = self.model.to_json()
  f = open('{0}_model.json'.format(self.root), 'w')
  f.write(json_string)
  f.close()
 
- def compile_network(self): 
+ def compile_network(self):
  self.model.compile(loss='mse', optimizer=Adam(lr=1e-4))
- 
+
  def read_network(self):
  print("Reading previous network...")
- 
+
  f = open('{0}_model.json'.format(self.root), 'r')
  json_string = f.read()
  f.close()
@@ -223,11 +223,11 @@ def read_network(self):
 
  def train(self, n_iterations):
  print("Training network...")
- 
+
  # Callbacks
  self.checkpointer = ModelCheckpoint(filepath="{0}_weights.hdf5".format(self.root), verbose=1,
  save_best_only=True)
- 
+
  # Load loss History
  n_val_loss = 0
  if self.option == 'continue':
@@ -273,6 +273,10 @@ def train(self, n_iterations):
  self.vx_d_1000km_min) / (self.vx_d_1000km_max - self.vx_d_1000km_min)
  output_train[:, :, :, 1] = (output_train[:, :, :, 1] -
  self.vy_d_1000km_min) / (self.vy_d_1000km_max - self.vy_d_1000km_min)
+ output_valid[:, :, :, 0] = (output_valid[:, :, :, 0] -
+ self.vx_d_1000km_min) / (self.vx_d_1000km_min - self.vx_d_1000km_min)
+ output_valid[:, :, :, 1] = (output_valid[:, :, :, 1] -
+ self.vy_d_1000km_min) / (self.vy_d_1000km_min - self.vy_d_1000km_min)
 
  # Training process
  self.model.fit(input_train, output_train, batch_size=self.batch_size, epochs=n_iterations, verbose=1,
@@ -298,7 +302,7 @@ def train(self, n_iterations):
 
 
 if __name__ == '__main__':
- 
+
  parser = argparse.ArgumentParser(description='Train DeepVel')
  parser.add_argument('-o', '--out', help='Output files')
  parser.add_argument('-e', '--epochs', help='Number of epochs', default=10)
@@ -308,21 +312,21 @@ def train(self, n_iterations):
  help='File containing the simulation properties for normalization',
  default='network/simulation_properties.npz')
  parsed = vars(parser.parse_args())
- 
+
  root = parsed['out']
  nEpochs = int(parsed['epochs'])
  option = parsed['action']
  noise = parsed['noise']
  norm_filename = parsed['properties']
- 
+
  out = train_deepvel(root, noise, option, norm_filename)
- 
+
  if option == 'start':
  out.define_network()
- 
+
  if option == 'continue':
  out.read_network()
- 
+
  if option == 'start' or option == 'continue':
  out.compile_network()
  out.train(nEpochs)

JSWSC_Tremblay_et_al_2020/MURaM-ARs/d_1000km/train_deepvelu.py

@@ -168,10 +168,10 @@ def __init__(self, root, noise, option, norm_filename='network/simulation_proper
  self.vy_d_1000km_var = tmp['vy_d_1000km_var']
  self.vy_d_1000km_median = tmp['vy_d_1000km_median']
  self.vy_d_1000km_stddev = tmp['vy_d_1000km_stddev']
- 
+
  def define_network(self):
  print("Setting up network...")
- 
+
  inputs = Input(shape=(self.nx, self.ny, self.n_inputs))
 
  x = inputs
@@ -256,20 +256,20 @@ def define_network(self):
 
  final = Conv2D(self.n_components, (1, 1), strides=(1, 1), padding='same',
  kernel_initializer='he_normal', activation='linear')(upconv1)
- 
+
  self.model = Model(inputs=inputs, outputs=final)
- 
+
  json_string = self.model.to_json()
  f = open('{0}_model.json'.format(self.root), 'w')
  f.write(json_string)
  f.close()
 
- def compile_network(self): 
+ def compile_network(self):
  self.model.compile(loss='mse', optimizer=Adam(lr=1e-4))
- 
+
  def read_network(self):
  print("Reading previous network...")
- 
+
  f = open('{0}_model.json'.format(self.root), 'r')
  json_string = f.read()
  f.close()
@@ -279,11 +279,11 @@ def read_network(self):
 
  def train(self, n_iterations):
  print("Training network...")
- 
+
  # Callbacks
  self.checkpointer = ModelCheckpoint(filepath="{0}_weights.hdf5".format(self.root), verbose=1,
  save_best_only=True)
- 
+
  # Load loss History
  n_val_loss = 0
  if self.option == 'continue':
@@ -329,6 +329,10 @@ def train(self, n_iterations):
  self.vx_d_1000km_min) / (self.vx_d_1000km_max - self.vx_d_1000km_min)
  output_train[:, :, :, 1] = (output_train[:, :, :, 1] -
  self.vy_d_1000km_min) / (self.vy_d_1000km_max - self.vy_d_1000km_min)
+ output_valid[:, :, :, 0] = (output_valid[:, :, :, 0] -
+ self.vx_d_1000km_min) / (self.vx_d_1000km_min - self.vx_d_1000km_min)
+ output_valid[:, :, :, 1] = (output_valid[:, :, :, 1] -
+ self.vy_d_1000km_min) / (self.vy_d_1000km_min - self.vy_d_1000km_min)
 
  # Training process
  self.model.fit(input_train, output_train, batch_size=self.batch_size, epochs=n_iterations, verbose=1,
@@ -354,7 +358,7 @@ def train(self, n_iterations):
 
 
 if __name__ == '__main__':
- 
+
  parser = argparse.ArgumentParser(description='Train DeepVel')
  parser.add_argument('-o', '--out', help='Output files')
  parser.add_argument('-e', '--epochs', help='Number of epochs', default=10)
@@ -364,21 +368,21 @@ def train(self, n_iterations):
  help='File containing the simulation properties for normalization',
  default='network/simulation_properties.npz')
  parsed = vars(parser.parse_args())
- 
+
  root = parsed['out']
  nEpochs = int(parsed['epochs'])
  option = parsed['action']
  noise = parsed['noise']
  norm_filename = parsed['properties']
- 
+
  out = train_deepvel(root, noise, option, norm_filename)
- 
+
  if option == 'start':
  out.define_network()
- 
+
  if option == 'continue':
  out.read_network()
- 
+
  if option == 'start' or option == 'continue':
  out.compile_network()
  out.train(nEpochs)

JSWSC_Tremblay_et_al_2020/MURaM-ARs/tau_0.001/train_deepvel.py

@@ -170,7 +170,7 @@ def __init__(self, root, noise, option, norm_filename='network/simulation_proper
  self.vy_d_1000km_stddev = tmp['vy_d_1000km_stddev']
 
  def residual(self, inputs):
- 
+
  x = Conv2D(self.n_filters, (self.kernel_size, self.kernel_size), strides=(1, 1), padding='same',
  kernel_initializer='he_normal')(inputs)
  x = BatchNormalization()(x)
@@ -179,41 +179,41 @@ def residual(self, inputs):
  kernel_initializer='he_normal')(x)
  x = BatchNormalization()(x)
  x = add([x, inputs])
- 
+
  return x
 
  def define_network(self):
  print("Setting up network...")
- 
+
  inputs = Input(shape=(self.nx, self.ny, self.n_inputs))
  conv = Conv2D(self.n_filters, (self.kernel_size, self.kernel_size), strides=(1, 1), padding='same',
  kernel_initializer='he_normal', activation='relu')(inputs)
- 
+
  x = self.residual(conv)
  for i in range(self.n_conv_layers):
  x = self.residual(x)
- 
+
  x = Conv2D(self.n_filters, (self.kernel_size, self.kernel_size), strides=(1, 1), padding='same',
  kernel_initializer='he_normal')(x)
  x = BatchNormalization()(x)
  x = add([x, conv])
- 
+
  final = Conv2D(self.n_outputs, (1, 1), strides=(1, 1), padding='same',
  kernel_initializer='he_normal', activation='linear')(x)
- 
+
  self.model = Model(inputs=inputs, outputs=final)
- 
+
  json_string = self.model.to_json()
  f = open('{0}_model.json'.format(self.root), 'w')
  f.write(json_string)
  f.close()
 
- def compile_network(self): 
+ def compile_network(self):
  self.model.compile(loss='mse', optimizer=Adam(lr=1e-4))
- 
+
  def read_network(self):
  print("Reading previous network...")
- 
+
  f = open('{0}_model.json'.format(self.root), 'r')
  json_string = f.read()
  f.close()
@@ -223,11 +223,11 @@ def read_network(self):
 
  def train(self, n_iterations):
  print("Training network...")
- 
+
  # Callbacks
  self.checkpointer = ModelCheckpoint(filepath="{0}_weights.hdf5".format(self.root), verbose=1,
  save_best_only=True)
- 
+
  # Load loss History
  n_val_loss = 0
  if self.option == 'continue':
@@ -273,6 +273,10 @@ def train(self, n_iterations):
  self.vx_tau_0001_min) / (self.vx_tau_0001_max - self.vx_tau_0001_min)
  output_train[:, :, :, 1] = (output_train[:, :, :, 1] -
  self.vy_tau_0001_min) / (self.vy_tau_0001_max - self.vy_tau_0001_min)
+ output_valid[:, :, :, 0] = (output_valid[:, :, :, 0] -
+ self.vx_tau_0001_min) / (self.vx_tau_0001_max - self.vx_tau_0001_min)
+ output_valid[:, :, :, 1] = (output_valid[:, :, :, 1] -
+ self.vy_tau_0001_min) / (self.vy_tau_0001_max - self.vy_tau_0001_min)
 
  # Training process
  self.model.fit(input_train, output_train, batch_size=self.batch_size, epochs=n_iterations, verbose=1,
@@ -298,7 +302,7 @@ def train(self, n_iterations):
 
 
 if __name__ == '__main__':
- 
+
  parser = argparse.ArgumentParser(description='Train DeepVel')
  parser.add_argument('-o', '--out', help='Output files')
  parser.add_argument('-e', '--epochs', help='Number of epochs', default=10)
@@ -308,21 +312,21 @@ def train(self, n_iterations):
  help='File containing the simulation properties for normalization',
  default='network/simulation_properties.npz')
  parsed = vars(parser.parse_args())
- 
+
  root = parsed['out']
  nEpochs = int(parsed['epochs'])
  option = parsed['action']
  noise = parsed['noise']
  norm_filename = parsed['properties']
- 
+
  out = train_deepvel(root, noise, option, norm_filename)
- 
+
  if option == 'start':
  out.define_network()
- 
+
  if option == 'continue':
  out.read_network()
- 
+
  if option == 'start' or option == 'continue':
  out.compile_network()
  out.train(nEpochs)