Heisenberg/NN_Energies.py at master · turleya/Heisenberg · GitHub

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
import numpy as np
import matplotlib.pylab as plt
import matplotlib
import h5py

from keras.models import Sequential, load_model
from keras.layers import Dense, Dropout
from keras.callbacks import EarlyStopping, ModelCheckpoint

import sklearn
from sklearn.model_selection import train_test_split

Heis_data =np.load('Energies_data.npz') #each row contains a feature vector
designmatrix = Heis_data['design_matrix']
Energies = Heis_data['Energies'] #Target vectors

font = {'family' : 'normal',
        'weight' : 'normal',
        'size'   : 12}
matplotlib.rc('font', **font)

#Histogram of Targets
plt.figure()
plt.title('Histogram of targets')
plt.hist(Energies, bins=20)
plt.xlabel("Energies")
plt.ylabel("Number of Samples")

#Neural Network (NN) - Regression
print("Number of Examples: %i\nNumber of Features: %i" % designmatrix.shape)

#Train/Validation/Test sets
designmatrix_tv, designmatrix_test, Energies_tv, Energies_test = train_test_split(designmatrix, Energies, test_size=0.2)

designmatrix_train, designmatrix_val, Energies_train, Energies_val = train_test_split(designmatrix_tv, Energies_tv, test_size=0.2)

print("\nTraining Set:\nNumber of Examples: %i\nNumber of Features: %i" % designmatrix_train.shape)
print("\nValidation Set:\nNumber of Examples: %i\nNumber of Features: %i" % designmatrix_val.shape)
print("\nTest Set:\nNumber of Examples: %i\nNumber of Features: %i" % designmatrix_test.shape)

#Standardization
designmatrix_mu = np.mean(designmatrix_train, axis=0)
designmatrix_std = np.std(designmatrix_train, axis=0)

designmatrix_train_std = (designmatrix_train - designmatrix_mu) / designmatrix_std
designmatrix_val_std = (designmatrix_val - designmatrix_mu) / designmatrix_std
designmatrix_test_std = (designmatrix_test - designmatrix_mu) / designmatrix_std

Energies_mu = np.mean(Energies_train, axis=0)
Energies_std = np.std(Energies_train, axis=0)

Energies_train_std = (Energies_train - Energies_mu) / Energies_std
Energies_val_std = (Energies_val - Energies_mu) / Energies_std
Energies_test_std = (Energies_test - Energies_mu) / Energies_std

plt.figure()
plt.title('Standardization of exchange parameters and energies')
plt.ylabel('Number of Samples')
plt.subplot(121)
plt.hist(Energies_train, bins=10)
plt.xlabel('Energies (before)')
plt.subplot(122)
plt.hist(Energies_train_std, bins=10)
plt.xlabel('Energies (after)')
#plt.ylabel("Number of Samples")


## Defining Model ##
model = Sequential() #defining model for Heisenberg Hamiltonian
model.add(Dense(input_dim=6, units=256, activation='relu'))
model.add(Dense(units=256, activation='relu'))
model.add(Dense(units=256, activation='relu'))
model.add(Dense(units=16))

## Compile ##
model.compile(optimizer='adam', loss='mse')
print(model.summary())

callback_list = [EarlyStopping(monitor='val_loss', min_delta=1e-7, patience=10, verbose=1), ModelCheckpoint(filepath='lattice.h5', monitor='val_loss', verbose=1, save_best_only=True)] #creates a HDF5 file 'my_model.h5'
n_epochs = 100

model_history = model.fit(x=designmatrix_train, y=Energies_train_std, validation_data=(designmatrix_val, Energies_val_std),
                batch_size=32, verbose=1,
                epochs=n_epochs, callbacks=callback_list, shuffle=True)

print("Epochs Taken:", len(model_history.history['loss']))

#Generalization Error
#Predictions
model_best = load_model('lattice.h5')
Energies_pred_train = model_best.predict(designmatrix_train)*Energies_std + Energies_mu
Energies_pred_val = model_best.predict(designmatrix_val)*Energies_std + Energies_mu


#plt.plot.print_out(y_true=Energies_train, y_pred=Energies_pred_train, setname='Train')
#plt.plot.print_out(y_true=Energies_val, y_pred=Energies_pred_val, setname='Val')
#plt.plot.plot_history(model_history)
plt.figure()
plt.scatter(Energies_val, Energies_pred_val)
plt.xlabel('True Energies')
plt.ylabel('Predicted Energies')
plt.plot([-2,2], [-2,2], linestyle='dashed', color='k')
plt.grid()

Log_loss = np.log10(model_history.history['loss'])
Log_val_loss = np.log10(model_history.history['val_loss'])

#Accuracy
plt.figure()
plt.plot(range(1, len(model_history.history['loss'])+1), Log_loss, label='Training set MSE')
plt.plot(range(1, len(model_history.history['val_loss'])+1), Log_val_loss, label='Validation set MSE')
plt.legend()
plt.xlabel('Number of Epochs')
plt.ylabel('log(Mean Squared Error)')

plt.show()