class ParticleSwarmPriorResample(ParticleSwarm):
"""
Like ParticleSwarm, but resamples from the prior
"""
def refresh(self):
"""
Resample by resampling those below the median from the prior.
"""
m = median(self.chainZ)
for i in range(self.nchains):
if self.chainZ[i] < m:
self.chains[i] = self.make_h0(**self.kwargs)
self.chainZ[i] = -Infinity # reset this
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
if __name__ == "__main__":
from LOTlib.Examples.Number.Global import generate_data, make_h0
data = generate_data(300)
ps = ParticleSwarm(make_h0, data)
for h in break_ctrlc(ps):
print h.posterior_score, h
if len(ps.seen) > 0:
print "#", sorted(ps.seen,
key=lambda x: x.posterior_score,
reverse=True)[0]
parser.add_option("--out", dest="OUT", type="string", help="Output prefix", default="output/proposal")
parser.add_option("--samples", dest="SAMPLES", type="int", default=100000, help="Number of samples to run")
parser.add_option("--chains", dest="CHAINS", type="int", default=10, help="Number of chains to run in parallel")
parser.add_option("--repetitions", dest="REPETITONS", type="int", default=100, help="Number of repetitions to run")
parser.add_option("--model", dest="MODEL", type="str", default="Number100", help="Which model to run on (Number, Galileo, RationalRules, SimpleMagnetism)")
parser.add_option("--print-every", dest="PRINTEVERY", type="int", default=1000, help="Evaluation prints every this many")
options, _ = parser.parse_args()
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# Define the test model
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
if options.MODEL == "Number100":
# Load the data
from LOTlib.Examples.Number.Global import generate_data, grammar, make_h0
data = synchronize_variable( lambda : generate_data(100) )
elif options.MODEL == "Number300":
# Load the data
from LOTlib.Examples.Number.Global import generate_data, grammar, make_h0
data = synchronize_variable( lambda : generate_data(300) )
elif options.MODEL == "Number1000":
# Load the data
from LOTlib.Examples.Number.Global import generate_data, grammar, make_h0
data = synchronize_variable( lambda : generate_data(1000) )
elif options.MODEL == "Galileo":
from LOTlib.Examples.SymbolicRegression.Galileo import data, grammar, make_h0
elif options.MODEL == "RationalRules":
from LOTlib.Examples.RationalRules.Model.Utilities import grammar, data, make_h0
class ParticleSwarmPriorResample(ParticleSwarm):
"""
Like ParticleSwarm, but resamples from the prior
"""
def refresh(self):
"""
Resample by resampling those below the median from the prior.
"""
m = median(self.chainZ)
for i in range(self.nchains):
if self.chainZ[i] < m:
self.chains[i] = self.make_h0(**self.kwargs)
self.chainZ[i] = -Infinity # reset this
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
if __name__ == "__main__":
from LOTlib.Examples.Number.Global import generate_data, grammar, make_h0
data = generate_data(300)
ps = ParticleSwarm(make_h0, data)
for h in lot_iter(ps):
print h.posterior_score, h
if len(ps.seen) > 0:
print "#", sorted(ps.seen, key=lambda x: x.posterior_score, reverse=True)[0]
class MemoizedMHSampler(MHSampler):
"""
Same as MHSampler, but the values of compute_posterior are cached via LRUCache
"""
def __init__(self, h0, data, memoize=Infinity, **kwargs):
MHSampler.__init__(self, h0, data, **kwargs)
# self.mem stores return of compute_posterior
self.mem = LRUCache(maxsize=memoize)
def compute_posterior(self, h, data):
if h in self.mem:
ret = self.mem[h]
h.posterior_score = ret # set this because it may not be set
return ret
else:
ret = MHSampler.compute_posterior(self, h, data)
self.mem[h] = ret
return ret
if __name__ == "__main__":
from LOTlib.Examples.Number.Global import generate_data, NumberExpression, grammar
data = generate_data(100)
h0 = NumberExpression(grammar)
sampler = MemoizedMHSampler(h0, data, steps=1000)
for h in sampler:
pass #print q(get_knower_pattern(h)), h.posterior_score, h.prior, h.likelihood, q(h), sampler.acceptance_count, sampler.acceptance_ratio()
"""
def __init__(self, h0, data, memoize=Infinity, **kwargs):
MHSampler.__init__(self, h0, data, **kwargs)
# self.mem stores return of compute_posterior
self.mem = LRUCache(maxsize=memoize)
def compute_posterior(self, h, data, shortcut=-Infinity):
if h in self.mem:
ret = self.mem[h]
h.posterior_score = ret # set this because it may not be set
return ret
else:
ret = MHSampler.compute_posterior(
self, h, data,
shortcut=-Infinity) # calls update to posterior counter
self.mem[h] = ret
return ret
if __name__ == "__main__":
from LOTlib.Examples.Number.Global import generate_data, NumberExpression, grammar
data = generate_data(100)
h0 = NumberExpression(grammar)
sampler = MemoizedMHSampler(h0, data, steps=1000)
for h in sampler:
pass #print q(get_knower_pattern(h)), h.posterior_score, h.prior, h.likelihood, q(h), sampler.acceptance_count, sampler.acceptance_ratio()