var_energy_data = []
n = n_samples[1]
if Is_train == False:
f = open('./DATA/8by8lattices.pkl', 'rb')
if (f.read(2) == '\x1f\x8b'):
f.seek(0)
gzip.GzipFile(fileobj=f)
else:
f.seek(0)
training_data = pickle.load(f, encoding="latin1")
training_data = np.reshape(training_data,(320000, 64))
for i in range(0,n_samples[0],4):
Magnetization = get_parameters.get_mean_magnetization(gsample[i*n_samples[1]:(i+1)*n_samples[1]])
Magnetization_direction = get_parameters.get_magnetization_direction(gsample[i*n_samples[1]:(i+1)*n_samples[1]])
energy = get_parameters.get_energy(gsample[i*n_samples[1]:(i+1)*n_samples[1]])
print(i)
if Is_train == False:
fig1 = plt.figure(1)
plt.plot(zsample[:n_samples[1],1],Magnetization[0][0],label = (T_vals[i]))
fig2 = plt.figure(2)
plt.plot(zsample[:n_samples[1],1],Magnetization_direction,label = (T_vals[i]))
if Is_train == True:
lattices = np.array(training_data[i*10000:i*10000+n]).reshape(n,lattice_size,lattice_size)
energy_data = get_parameters.get_energy(lattices)
thetas_data = get_parameters.get_magnetization_direction(lattices)
[mag_data,mag_mean,mag_std] = get_parameters.get_mean_magnetization(lattices)
plt.subplot(3,1,1)
plt.hist(Magnetization[0][0][i*n_samples[1]:(i+1)*n_samples[1]],bins =20,color='b',range=[0, 1],alpha=0.5)
plt.hist(mag_data ,bins =20,color='g',range=[0, 1],alpha=0.5)
plt.ylabel('Magnetization ')
f = open('./DATA/16and16lattices.pkl', 'rb')
if (f.read(2) == '\x1f\x8b'):
f.seek(0)
gzip.GzipFile(fileobj=f)
else:
f.seek(0)
training_inputs = pickle.load(f, encoding="latin1")
training_inputs = np.reshape(training_inputs,(n_temps*10000, lattice_size**2))
for i in range(0,n_temps):
t = np.repeat(B_vals[i],n_z*n).reshape(n,n_z)
gsample = sess.run(generator(z,reuse=True,bs = n), feed_dict={z_rand:zsample,temp:t})
gsample = gsample.reshape(n,lattice_size,lattice_size)
print(B_vals[i], 360*np.mean(gsample[0]),360*np.std(gsample[0]))
Magnetization = get_parameters.get_mean_magnetization(gsample)
Magnetization_direction = get_parameters.get_magnetization_direction(gsample)
energy = get_parameters.get_energy(gsample)
if Is_train == k:
fig1 = plt.figure(1)
plt.plot(Magnetization[0][0],label = (B_vals[i]))
fig2 = plt.figure(2)
plt.plot(Magnetization_direction,label = (B_vals[i]))
if Is_train == l:
lattices = np.array(training_inputs[i*10000+8000:i*10000+n+8000]).reshape(n,lattice_size,lattice_size)
energy_data = get_parameters.get_energy(lattices)
thetas_data = get_parameters.get_magnetization_direction(lattices)
[mag_data,mag_mean,mag_std]=get_parameters.get_mean_magnetization(lattices)
mean_magnetization.append(Magnetization[1])
var_magnetization.append(Magnetization[2])
mean_magnetization_data.append(mag_mean)
var_magnetization_data.append(mag_std)
feed_dict={fully_connected_decoder1: Gsample})
gsample = sess.run(x_hat, feed_dict={fully_connected_decoder2: Gsample2})
# Gsamplemu = sess.run(x_mu, feed_dict={fully_connected_decoder2: Gsample2})
# Gsamplesig = sess.run(x_log_sigma_sq,feed_dict={fully_connected_decoder2:Gsample2})
# gsample = sess.run(tf.random_normal(shape = tf.shape(Gsamplemu) ,mean = Gsamplemu,stddev = tf.sqrt(tf.exp(Gsamplesig)), dtype = tf.float32 ))#tf.sqrt(tf.exp(Gsamplesig))
# print(360*Gsamplemu[:10], Gsamplesig[:10])
gsample = gsample.reshape(zsample.shape[0], lattice_size, lattice_size)
print("Specific Heat %f")
# for i in range(n_samples):
# print(get_parameters.get_specific_heat(gsample[i],zsample[i]))
print("Mean magnetization and its Standard Deviation")
mean_magnetization = []
Magnetization = get_parameters.get_mean_magnetization(gsample)
Magnetization_direction = get_parameters.get_magnetization_direction(
gsample)
energy = get_parameters.get_energy(gsample)
print(energy)
if n_z == 1:
plt.plot(zsample, Magnetization[0][0]) #all values
plt.xlabel('Latent Variable value', fontsize=12)
plt.ylabel('Magnetization of a single sample generated by the network',
fontsize=10)
plt.savefig('mag_vs_temp.png')
plt.show()
plt.plot(zsample, Magnetization_direction)
plt.show()
plt.hist(energy, bins=100)
plt.savefig('energy_vs_LV.png')
plt.show()
elif n_z == 2:
# LV[0] Magnetization LV[1] Temperature
gsample = gsample.reshape(zsample.shape[0], lattice_size, lattice_size)
print("Specific Heat %f")
print(360 * np.mean(Gsamplemu), 360 * np.std(Gsamplemu))
print(360 * np.mean(-1.0 * Gsamplesig))
print("Mean magnetization and its Standard Deviation")
# for i in range(n_samples):
# print(get_parameters.get_specific_heat(gsample[i],zsample[i]))
mean_magnetization = []
Magnetization = get_parameters.get_mean_magnetization(gsample)
Magnetization_direction = get_parameters.get_magnetization_direction(
gsample)
energy = get_parameters.get_energy(gsample)
if n_z == 1:
for d in range(n_samples[0]):
plt.plot(zsample[:n_samples[1], 1],
Magnetization[0][0][d * n_samples[1]:(d + 1) *
n_samples[1]],
label=d) #all values
plt.xlabel('Latent Variable value', fontsize=12)
plt.ylabel('Magnetization of a single sample generated by the network',
fontsize=10)
plt.legend()
plt.show()
for d in range(n_samples[0]):
plt.plot(
zsample[:n_samples[1], 1],
