def sampler(nr_samples, burnin=100, lag=100):
data = [[] for o in outputs]
for i in range(nr_samples * lag + burnin):
accept = False
while not accept:
index = numpy.random.randint(len(free_RVs))
accept = rr[index]()
if accept and i > burnin and (i - burnin) % lag == 0:
for d, o in zip(data, outputs):
# TODO: this can be optimized
if is_raw_rv(o):
d.append(
free_RVs_state[free_RVs.index(o)].get_value())
else:
full_observations = dict(observations)
full_observations.update(
dict([
(rv, s)
for rv, s in zip(free_RVs, free_RVs_state)
]))
d.append(
evaluate(evaluate_with_assignments(
o, full_observations),
givens=givens))
data = [numpy.asarray(d).squeeze() for d in data]
return data
def sampler(nr_samples, burnin = 100, lag = 100):
data = [[] for o in outputs]
for i in range(nr_samples*lag+burnin):
accept = False
while not accept:
index = numpy.random.randint(len(free_RVs))
accept = rr[index]()
if accept and i > burnin and (i-burnin) % lag == 0:
for d, o in zip(data, outputs):
# TODO: this can be optimized
if is_raw_rv(o):
d.append(free_RVs_state[free_RVs.index(o)].get_value())
else:
full_observations = dict(observations)
full_observations.update(dict([(rv, s) for rv, s in zip(free_RVs, free_RVs_state)]))
d.append(evaluate(evaluate_with_assignments(o, full_observations), givens=givens))
data = [numpy.asarray(d).squeeze() for d in data]
return data
def test(self):
print evaluate(self.sample_words([1,0,0,0,0]))
def test(self):
print evaluate(self.sample_words([1, 0, 0, 0, 0]))
import numpy, pylab
import theano
from theano import tensor
from rstreams import RandomStreams
import distributions
from sample import mh2_sample
from for_theano import evaluate
from rv import full_log_likelihood
s_rng = RandomStreams(23424)
fair_prior = 0.999
coin_weight = tensor.switch(
s_rng.binomial(1, fair_prior) > 0.5, 0.5,
s_rng.dirichlet([1, 1])[0])
make_coin = lambda p, size: s_rng.binomial(1, p, draw_shape=(size, ))
coin = lambda size: make_coin(coin_weight, size)
for size in [1, 3, 6, 10, 20, 30, 50, 70, 100]:
data = evaluate(make_coin(0.9, size))
sampler = mh2_sample(s_rng, [coin_weight], {coin(size): data})
print "nr of examples", size, ", estimated probability", sampler(
nr_samples=400, burnin=20000, lag=10)[0].mean()
import numpy, pylab
import theano
from theano import tensor
from rstreams import RandomStreams
import distributions
from sample import mh2_sample
from for_theano import evaluate
from rv import full_log_likelihood
s_rng = RandomStreams(23424)
fair_prior = 0.999
coin_weight = tensor.switch(s_rng.binomial(1, fair_prior) > 0.5, 0.5, s_rng.dirichlet([1, 1])[0])
make_coin = lambda p, size: s_rng.binomial(1, p, draw_shape=(size,))
coin = lambda size: make_coin(coin_weight, size)
for size in [1, 3, 6, 10, 20, 30, 50, 70, 100]:
data = evaluate(make_coin(0.9, size))
sampler = mh2_sample(s_rng, [coin_weight], {coin(size) : data})
print "nr of examples", size, ", estimated probability", sampler(nr_samples=400, burnin=20000, lag=10)[0].mean()
