def run(make_hypothesis, make_data, data_size):
"""
This out on the DATA_RANGE amounts of data and returns all hypotheses in top count
"""
if LOTlib.SIG_INTERRUPTED:
return set()
return standard_sample(make_hypothesis,
lambda: make_data(data_size),
N=options.TOP_COUNT,
steps=options.STEPS,
show=False, save_top=None)
# #
# # print check_list[95][0]
# # print check_list[95][0].prior, check_list[95][0].likelihood, check_list[95][0].posterior_score
#
# cnt = 0
# for e in check_list:
# if not is_valid(e[1]): continue
# if cnt == 100: break
# print e[0], len(str(e[0])) - len('lambda recurse_: flatten2str(,sep="")')
# print np.exp(e[0].prior), e[0].likelihood, e[0].posterior_score
# cnt += 1
for i in xrange(10):
print 'estimating for data_size: %i' % i
data = make_data(i)
seen = set()
inf_sum = 0
all_sum = 0
for topN in out:
for h in topN:
if h not in seen:
seen.add(h)
poster_mass = h.compute_posterior(data)
if not is_valid(poster_mass): continue
poster_mass = np.exp(poster_mass)
all_sum += poster_mass
if len(str(h)) - len('lambda recurse_: flatten2str(,sep="")') > 46: inf_sum += poster_mass
prob = 0
if all_sum != 0: prob = inf_sum / all_sum
# Make a list of arguments to map over
args = list(itertools.product([make_hypothesis], [make_data], DATA_RANGE))
# run on MPI
results = MPI_map(run, args)
# collapse all returned sets
hypotheses = set()
for r in results:
hypotheses.update(r) # add the ith's results to the set
# Now go through each hypothesis and print out some summary stats
for data_size in DATA_RANGE:
evaluation_data = make_data(data_size)
# Now update everyone's posterior
for h in hypotheses:
h.compute_posterior(evaluation_data)
# compute the normalizing constant. This is the log of the sum of the probabilities
Z = logsumexp([h.posterior_score for h in hypotheses])
for h in hypotheses:
#compute the number of different strings we generate
generated_strings = set([h() for _ in xrange(1000)])
# print out some info. We can here use np.exp(h.posterior_score-Z) because Z is computed via logsumexp, so is more numerically stable
# This is the probability at this amount of data
print data_size, np.exp(h.posterior_score-Z), h.posterior_score, h.prior, h.likelihood, len(generated_strings), qq(h)
