def generate_figure_samples(samples_per_frame, n_frames, burnin = int(1e4)):
""" Generates the figure
:param samples_per_frame: number of sample steps between each frame
:param n_frames: number of frames to draw
:returns: None
:rtype: None
"""
n_samples = samples_per_frame * n_frames
ndims = 36
nbasis = 72
rand_val = rand(ndims,nbasis/2,density=0.25)
W = np.concatenate([rand_val.toarray(), -rand_val.toarray()],axis=1)
logalpha = np.random.randn(nbasis, 1)
poe = ProductOfT(nbatch=1, W=W, logalpha=logalpha)
## NUTS uses a different number of grad evals for each update step!!
## makes it very hard to compare against others w/ same number of update steps
# # NUTS
# print "NUTS"
# nuts_init = poe.Xinit[:, 0]
# nuts_samples = nuts6(poe.reset(), n_samples, nuts_burnin, nuts_init)[0]
# nuts_frames = [nuts_samples[f_idx * samples_per_frame, :] for f_idx in xrange(0, n_frames)]
# x_init = nuts_samples[0, :].reshape(ndims, 1)
## burnin
print "MJHMC burnin"
x_init = poe.Xinit #[:, [0]]
mjhmc = MarkovJumpHMC(distribution=poe.reset(), **mjhmc_params)
mjhmc.state = HMCState(x_init.copy(), mjhmc)
mjhmc_samples = mjhmc.sample(burnin)
print mjhmc_samples.shape
x_init = mjhmc_samples[:, [0]]
# control HMC
print "Control"
hmc = ControlHMC(distribution=poe.reset(), **control_params)
hmc.state = HMCState(x_init.copy(), hmc)
hmc_samples = hmc.sample(n_samples)
hmc_frames = [hmc_samples[:, f_idx * samples_per_frame].copy() for f_idx in xrange(0, n_frames)]
# MJHMC
print "MJHMC"
mjhmc = MarkovJumpHMC(distribution=poe.reset(), resample=False, **mjhmc_params)
mjhmc.state = HMCState(x_init.copy(), mjhmc)
mjhmc_samples = mjhmc.sample(n_samples)
mjhmc_frames = [mjhmc_samples[:, f_idx * samples_per_frame].copy() for f_idx in xrange(0, n_frames)]
print mjhmc.r_count, hmc.r_count
print mjhmc.l_count, hmc.l_count
print mjhmc.f_count, hmc.f_count
print mjhmc.fl_count, hmc.fl_count
frames = [mjhmc_frames, hmc_frames]
names = ['MJHMC', 'ControlHMC']
frame_grads = [f_idx * samples_per_frame for f_idx in xrange(0, n_frames)]
return frames, names, frame_grads
def generate_figure_samples(samples_per_frame, n_frames, burnin=int(1e4)):
""" Generates the figure
:param samples_per_frame: number of sample steps between each frame
:param n_frames: number of frames to draw
:returns: None
:rtype: None
"""
n_samples = samples_per_frame * n_frames
ndims = 36
nbasis = 72
rand_val = rand(ndims, nbasis / 2, density=0.25)
W = np.concatenate([rand_val.toarray(), -rand_val.toarray()], axis=1)
logalpha = np.random.randn(nbasis, 1)
poe = ProductOfT(nbatch=1, W=W, logalpha=logalpha)
## NUTS uses a different number of grad evals for each update step!!
## makes it very hard to compare against others w/ same number of update steps
# # NUTS
# print "NUTS"
# nuts_init = poe.Xinit[:, 0]
# nuts_samples = nuts6(poe.reset(), n_samples, nuts_burnin, nuts_init)[0]
# nuts_frames = [nuts_samples[f_idx * samples_per_frame, :] for f_idx in xrange(0, n_frames)]
# x_init = nuts_samples[0, :].reshape(ndims, 1)
## burnin
print "MJHMC burnin"
x_init = poe.Xinit # [:, [0]]
mjhmc = MarkovJumpHMC(distribution=poe.reset(), **mjhmc_params)
mjhmc.state = HMCState(x_init.copy(), mjhmc)
mjhmc_samples = mjhmc.sample(burnin)
print mjhmc_samples.shape
x_init = mjhmc_samples[:, [0]]
# control HMC
print "Control"
hmc = ControlHMC(distribution=poe.reset(), **control_params)
hmc.state = HMCState(x_init.copy(), hmc)
hmc_samples = hmc.sample(n_samples)
hmc_frames = [hmc_samples[:, f_idx * samples_per_frame].copy() for f_idx in xrange(0, n_frames)]
# MJHMC
print "MJHMC"
mjhmc = MarkovJumpHMC(distribution=poe.reset(), resample=False, **mjhmc_params)
mjhmc.state = HMCState(x_init.copy(), mjhmc)
mjhmc_samples = mjhmc.sample(n_samples)
mjhmc_frames = [mjhmc_samples[:, f_idx * samples_per_frame].copy() for f_idx in xrange(0, n_frames)]
print mjhmc.r_count, hmc.r_count
print mjhmc.l_count, hmc.l_count
print mjhmc.f_count, hmc.f_count
print mjhmc.fl_count, hmc.fl_count
frames = [mjhmc_frames, hmc_frames]
names = ["MJHMC", "ControlHMC"]
frame_grads = [f_idx * samples_per_frame for f_idx in xrange(0, n_frames)]
return frames, names, frame_grads
def check_variance():
""" Simple benchmark that estimates the variance a ton of times and writes to file the results
:returns: Nothing
:rtype: None
"""
from mjhmc.misc.gen_mj_init import generate_initialization
np.random.seed(2015)
poe = ProductOfT(nbatch=1000, ndims=36, nbasis=36)
var_estimates = []
for trial_idx in xrange(100):
_, var_estimate = generate_initialization(poe.reset())
var_estimates.append(var_estimate)
with open("var_log.txt", 'a') as vlog:
vlog.write("Trial {} variance {}\n".format(trial_idx, var_estimate))
vlog.write("Variance of variance estimates so far {}".format(np.var(var_estimates)))
def sampler_speedometer():
""" Tests average sample speed of different samplers in different configurations
:returns: Just prints info
:rtype: None
"""
gaussian = Gaussian()
np.random.seed(2015)
pot = ProductOfT(ndims=36,nbasis=36)
mjhmc_gauss = MarkovJumpHMC(distribution=gaussian)
mjhmc_gauss_nr = MarkovJumpHMC(distribution=gaussian, resample=False)
control_gauss = ControlHMC(distribution=gaussian)
mjhmc_pot = MarkovJumpHMC(distribution=pot)
mjhmc_pot_nr = MarkovJumpHMC(distribution=pot, resample=False)
control_pot = ControlHMC(distribution=pot)
m_g_r_avg = time_per_sample(mjhmc_gauss)
m_g_nr_avg = time_per_sample(mjhmc_gauss_nr)
c_g_avg = time_per_sample(control_gauss)
m_p_r_avg = time_per_sample(mjhmc_pot)
m_p_nr_avg = time_per_sample(mjhmc_pot_nr)
c_p_avg = time_per_sample(control_pot)
print "Average times per samples..."
print "resampled MJHMC numpy gradient: {}".format(m_g_r_avg)
print "not resampled MJHMC numpy gradient: {}".format(m_g_nr_avg)
print "control HMC numpy gradient: {}".format(c_g_avg)
print "resampled MJHMC theano gradient: {}".format(m_p_r_avg)
print "not resampled MJHMC theano gradient: {}".format(m_p_nr_avg)
print "control HMC theano gradient: {}".format(c_p_avg)
def main(job_id, params):
ndims = 100
nbasis = 100
rand_val = rand(ndims, nbasis / 2, density=0.25)
print "job id: {}, params: {}".format(job_id, params)
return obj_func(MarkovJumpHMC,
ProductOfT(nbatch=25, ndims=ndims, nbasis=nbasis), job_id,
**params)
def main(job_id, params):
ndims = 256
nbasis = 72
rand_val = rand(ndims,nbasis/2,density=0.25)
W = np.concatenate([rand_val.toarray(), -rand_val.toarray()],axis=1)
logalpha = np.random.randn(nbasis, 1)
print "job id: {}, params: {}".format(job_id, params)
return obj_func(ControlHMC, ProductOfT(nbatch=250,ndims=ndims,nbasis=nbasis, W=W, logalpha=logalpha), job_id, **params)
def main(job_id, params):
ndims = 36
nbasis = 36
print "job id: {}, params: {}".format(job_id, params)
weights, lognu = init_weights(ndims, nbasis)
return obj_func(
ControlHMC,
ProductOfT(nbatch=25,
ndims=ndims,
nbasis=nbasis,
lognu=lognu,
W=weights), job_id, **params)
def ac_plot(n_samples=5000, **kwargs):
""" Plots the autocorrelation for the best found parameters of the 36
dimensional product of experts
:returns: None
:rtype: None
"""
from mjhmc.figures.ac_fig import plot_best
ndims = 36
nbasis = 36
np.random.seed(2015)
poe = ProductOfT(nbatch=25,ndims=ndims,nbasis=nbasis)
plot_best(poe, num_steps=n_samples, update_params=False, **kwargs)
from mjhmc.figures import ac_fig from mjhmc.misc.distributions import ProductOfT import numpy as np np.random.seed(2015) #Search for best hyper-parameters #Parameters for the distribution object ndims = 36 nbasis = 36 nbatch = 25 POT = ProductOfT(nbasis=nbasis,nbatch=nbatch,ndims=ndims) #Run a comparison ac_fig.plot_best(POT,num_steps=100000,update_params=True)