import numpy
from sklearn.model_selection import GridSearchCV
from keras.models import Sequential
from keras.layers import Dense
from keras.wrappers.scikit_learn import KerasRegressor
from pyDOE import *
from sklearn.model_selection import PredefinedSplit
import time
from src import sampling
from functions import functions
start = time.time()
import airfoil.utility as u
test_dir = r'./test_DIR/'
u.make_sure_path_exists(test_dir)
funObj = functions.Airfoil(5, 0.2, 5e005, test_dir, False)
def create_model(layers=2, units=40, eps=1e-8, lr=1e-5):
model = Sequential()
dim = 6
## making the model graph, Stacking layers is done by .add():
model.add(Dense(units=units, input_dim=dim, activation='sigmoid'))
for i in range(layers - 1):
model.add(Dense(units=units, activation='sigmoid'))
model.add(Dense(units=1))
# optmiser = keras.optimizers.RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)
optimizer = keras.optimizers.Adam(lr=lr,
except OSError as exception:
if exception.errno != errno.EEXIST:
raise
make_sure_path_exists(path)
for i in range(1): # how many times end to end generalisation is done
i = 0
series = 0.2 + np.random.rand(5) * 0.2 #2=how many in a series of contexts
np.savez(path + 'series_{}.npz'.format(i), series)
start = time.time()
logdir = path + r's{}/context{}/'.format(i, 0)
u.make_sure_path_exists(logdir)
funObj = f.Airfoil(3, series[0], 5e005, logdir, graphics)
print('Starting iteration number : ', i)
print('initial number of samples: ', initial_samples)
model, X, Y = single_run(funObj,
initial_samples=initial_samples,
model_old=None,
X_old=None,
Y_old=None)
model.save(path + r"Airfoil_model_series_{}_context_{}".format(0, 0))
np.savez(path + r'/Airfoil_XY_series_{}_context_{}'.format(0, 0), X, Y)
for j in range(1, series.shape[0]):
logdir = path + r's{}/context{}/'.format(i, j)
