def loss_Lagrangian_Re(_, y):
return tf.reduce_mean(lambla_Re * y)
def loss_Lagrangian_Im(_, y):
return tf.reduce_mean(lambla_Im * y)
def loss_volume(_, y):
return tf.math.square(tf.math.maximum(0.0, tf.reduce_mean(y) - GAMMA))
def loss_power(_, y):
return tf.reduce_mean(y)
def loss_V2(_, y):
return tf.reduce_mean(y) - GAMMA
def loss_V1(_, y):
if lambdaV > 0:
return tf.math.square(tf.reduce_mean(y) - GAMMA)
return loss_volume(None, y)
def loss_PDE3(_, y):
return tf.reduce_mean(lambla3 * y)
def loss_PDE2(_, y):
return tf.reduce_mean(lambla2 * y)
def loss_PDE1(_, y):
return tf.reduce_mean(lambla1 * y)