def test_austerity(self):
np.random.seed(SEED)
X = gen_X(SAMPLE_SIZE)
def vectorized_log_lik(X,theta):
return _vector_of_log_likelihoods(theta[0],theta[1],X)
def log_density_prior(theta):
return np.log(norm.pdf(theta[0],0, SIGMA_1)) + np.log(norm.pdf(theta[1],0, SIGMA_2))
sample,_ = austerity(vectorized_log_lik,log_density_prior, X,0.01,batch_size=50,chain_size=10, thinning=1, theta_t=np.random.randn(2))
assert_almost_equal(np.array([-0.2554517, 1.3805683]),sample[-1])
def test_austerity(self):
np.random.seed(SEED)
X = gen_X(SAMPLE_SIZE)
def vectorized_log_lik(X, theta):
return _vector_of_log_likelihoods(theta[0], theta[1], X)
def log_density_prior(theta):
return np.log(norm.pdf(theta[0], 0, SIGMA_1)) + np.log(
norm.pdf(theta[1], 0, SIGMA_2))
sample, _ = austerity(vectorized_log_lik,
log_density_prior,
X,
0.01,
batch_size=50,
chain_size=10,
thinning=1,
theta_t=np.random.randn(2))
assert_almost_equal(np.array([-0.2554517, 1.3805683]), sample[-1])
from time import time
from sgld_test.bimodal_SGLD import vSGLD, evSGLD
from sgld_test.gradients_of_likelihood import manual_grad, grad_log_prior
from sgld_test.mcmc_convergance.cosnt import NUMBER_OF_TESTS, NO_OF_SAMPELS_IN_TEST, CHAIN_SIZE, SEED, SGLD_CHAIN_SIZE, \
SAMPLE_SIZE
from sgld_test.likelihoods import gen_X, log_probability
import numpy as np
np.random.seed(SEED)
X = gen_X(SAMPLE_SIZE)
def vectorized_log_density(theta):
return log_probability(theta,X)
t1 = time()
sample = []
no_chains = NUMBER_OF_TESTS * NO_OF_SAMPELS_IN_TEST
for i in range(no_chains):
if i % (100) == 0:
print(float(i)*100.0/no_chains)
print(time()-t1)
sample.append(evSGLD(manual_grad, grad_log_prior, X, n=1, chain_size=SGLD_CHAIN_SIZE,theta = np.random.randn(2) ) )
from time import time
from sgld_test.bimodal_SGLD import vSGLD, evSGLD
from sgld_test.gradients_of_likelihood import manual_grad, grad_log_prior
from sgld_test.mcmc_convergance.cosnt import NUMBER_OF_TESTS, NO_OF_SAMPELS_IN_TEST, CHAIN_SIZE, SEED, SGLD_CHAIN_SIZE, \
SAMPLE_SIZE
from sgld_test.likelihoods import gen_X, log_probability
import numpy as np
np.random.seed(SEED)
X = gen_X(SAMPLE_SIZE)
def vectorized_log_density(theta):
return log_probability(theta, X)
t1 = time()
sample = []
no_chains = NUMBER_OF_TESTS * NO_OF_SAMPELS_IN_TEST
for i in range(no_chains):
if i % (100) == 0:
print(float(i) * 100.0 / no_chains)
print(time() - t1)
sample.append(
evSGLD(manual_grad,
grad_log_prior,
X,
n=1,
from sgld_test.gradients_of_likelihood import manual_grad, grad_log_prior
from sgld_test.likelihoods import gen_X
import numpy as np
import matplotlib.pyplot as plt
theta1 = np.arange(-2, 2, 0.025)
theta2 = np.arange(-2, 2, 0.025)
grid_dimension_size = len(theta1)
theta1, theta2 = np.meshgrid(theta1, theta2)
D_theta1 = np.copy(theta1)
D_theta2 = np.copy(theta1)
sample = gen_X(400)
for i in range(grid_dimension_size):
for j in range(grid_dimension_size):
th = np.array([theta1[i, j], theta2[i, j]])
# subsample = np.random.choice(sample, 40)
stoch_grad_log_lik = np.sum(manual_grad(th[0], th[1], sample),
axis=0) + grad_log_prior(th)
D_theta1[i, j] = stoch_grad_log_lik[0]
D_theta2[i, j] = stoch_grad_log_lik[1]
plt.figure()
CS = plt.streamplot(theta1, theta2, D_theta1, D_theta2, density=[0.5, 1])
plt.show()
import matplotlib.pyplot as plt
import numpy as np
import seaborn as sns;
from sgld_test.likelihoods import gen_X, _log_probability
sns.set(color_codes=True)
np.random.seed(1307)
N = 400
X = gen_X(N)
theta1 = np.arange(-2, 2, 0.25)
grid_size = len(theta1)
theta2 = np.arange(-2, 2, 0.25)
theta1, theta2 = np.meshgrid(theta1, theta2)
Z = np.copy(theta1)
for i in range(grid_size):
for j in range(grid_size):
probability = _log_probability(theta1[i, j], theta2[i, j], X)
Z[i, j] = probability
max = np.max(Z)+2
Z = np.exp(Z -max)
print(Z)
plt.figure()
from sgld_test.likelihoods import gen_X
import numpy as np
import matplotlib.pyplot as plt
theta1 = np.arange(-2, 2, 0.025)
theta2 = np.arange(-2, 2, 0.025)
grid_dimension_size = len(theta1)
theta1, theta2 = np.meshgrid(theta1, theta2)
D_theta1 = np.copy(theta1)
D_theta2 = np.copy(theta1)
sample = gen_X(400)
for i in range(grid_dimension_size):
for j in range(grid_dimension_size):
th = np.array([theta1[i, j], theta2[i, j]])
# subsample = np.random.choice(sample, 40)
stoch_grad_log_lik = np.sum(manual_grad(th[0], th[1], sample), axis=0) + grad_log_prior(th)
D_theta1[i, j] = stoch_grad_log_lik[0]
D_theta2[i, j] = stoch_grad_log_lik[1]
plt.figure()
CS = plt.streamplot(theta1, theta2, D_theta1, D_theta2, density=[0.5, 1])
plt.show()
