def test_mcmc(core, likelihood_coeval, default_params, tmpdirec):
chain = mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TEST",
continue_sampling=False,
datadir=str(tmpdirec),
params=default_params,
walkersRatio=2,
burninIterations=0,
sampleIterations=2,
threadCount=1,
)
samples_from_chain = mcmc.get_samples(chain)
samples_from_file = mcmc.get_samples(tmpdirec / "TEST")
# make sure reading from file is the same as the chain.
assert samples_from_chain.iteration == samples_from_file.iteration
assert np.all(samples_from_file.accepted == samples_from_chain.accepted)
assert np.all(
samples_from_file.get_chain() == samples_from_chain.get_chain())
assert all(c in ["HII_EFF_FACTOR", "ION_Tvir_MIN"]
for c in samples_from_chain.param_names)
assert samples_from_chain.has_blobs
assert samples_from_chain.param_guess["HII_EFF_FACTOR"] == 30.0
assert samples_from_chain.param_guess["ION_Tvir_MIN"] == 4.7
def test_load_chain(core, likelihood_coeval, default_params, tmpdirec):
mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTLOADCHAIN",
continue_sampling=False,
datadir=tmpdirec,
params=default_params,
walkersRatio=2,
burninIterations=0,
sampleIterations=1,
threadCount=1,
)
lcc = mcmc.load_primitive_chain("TESTLOADCHAIN", direc=tmpdirec)
assert lcc.getCoreModules()[0].redshift == core.redshift
def test_multinest():
model_name = "LuminosityLikelihood"
redshifts = [6, 7, 8, 10]
F_STAR10 = [-1.3, -3, 0, 1.0]
ALPHA_STAR = [0.5, -0.5, 1.0, 1.0]
M_TURN = [8.69897, 8, 10, 1.0]
t_STAR = [0.5, 0.01, 1, 0.3]
mcmc_options = {
"n_live_points": 10,
"max_iter": 10,
}
mcmc.run_mcmc(
[
mcmc.CoreLuminosityFunction(redshift=z, sigma=0, name="lfz%d" % z)
for z in redshifts
],
[
mcmc.LikelihoodLuminosityFunction(name="lfz%d" % z)
for z in redshifts
],
model_name=model_name,
params={
"F_STAR10": F_STAR10,
"ALPHA_STAR": ALPHA_STAR,
"M_TURN": M_TURN,
"t_STAR": t_STAR,
},
use_multinest=True,
**mcmc_options,
)
import pymultinest
nest = pymultinest.Analyzer(4,
outputfiles_basename="./MultiNest/%s" %
model_name)
data = nest.get_data()
assert data.shape[1] == 6
def test_bad_continuation(core, likelihood_coeval, default_params, tmpdirec):
"check if trying to continue a chain that isn't compatible with previous chain raises an error"
mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=False,
datadir=tmpdirec,
params=default_params,
walkersRatio=2,
burninIterations=0,
sampleIterations=1,
threadCount=1,
)
with pytest.raises(RuntimeError):
# core with different redshift
core = mcmc.CoreCoevalModule(
redshift=8,
user_params={
"HII_DIM": 35,
"DIM": 70
},
cache_mcmc=False,
cache_init=False,
)
mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=True,
datadir=tmpdirec,
params=default_params,
walkersRatio=2,
burninIterations=0,
sampleIterations=2,
threadCount=1,
)
def test_continue_burnin(core, likelihood_coeval, default_params, tmpdirec):
with pytest.raises(
AssertionError): # needs to be sampled for at least 1 iteration!
mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=False,
datadir=str(tmpdirec),
params=default_params,
walkersRatio=2,
burninIterations=1,
sampleIterations=0,
threadCount=1,
)
chain = mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=False,
datadir=str(tmpdirec),
params=default_params,
walkersRatio=2,
burninIterations=1,
sampleIterations=1,
threadCount=1,
)
# HAVE TO SAVE THE CHAIN TO MEMORY HERE, BECAUSE THE OBJECT ACCESS THE FILE ON EVERY CALL,
# WHICH MEANS IT CONSTANTLY UPDATES
chain_b_chain = mcmc.get_samples(chain, burnin=True).get_chain()
chain_s_chain = mcmc.get_samples(chain).get_chain()
chain2 = mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=True,
datadir=tmpdirec,
params=default_params,
walkersRatio=2,
burninIterations=2,
sampleIterations=1,
threadCount=1,
)
burnin2 = mcmc.get_samples(chain2, burnin=True)
chain2_b_chain = burnin2.get_chain()
chain2_s_chain = mcmc.get_samples(chain).get_chain()
assert likelihood_coeval._simulate is False # because we're continuing sampling
assert burnin2.iteration == 2
assert np.all(chain2_b_chain[:1] ==
chain_b_chain) # first 5 iteration should be unchanged
# The actual samples *should* have been deleted, because they have different burnin times.
assert not np.all(chain_s_chain == chain2_s_chain)
chain3 = mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=True,
datadir=tmpdirec,
params=default_params,
walkersRatio=2,
burninIterations=2,
sampleIterations=2,
threadCount=1,
)
samples3 = chain3.samples
assert samples3.iteration == 2
chain3_b_chain = mcmc.get_samples(chain3, burnin=True).get_chain()
assert np.all(chain3_b_chain == chain2_b_chain)
chain3_s_chain = mcmc.get_samples(chain3).get_chain()
assert np.all(chain2_s_chain == chain3_s_chain[:1])
with pytest.raises(
ValueError
): # don't run if we already have all samples, and let the user know!
mcmc.run_mcmc(
core,
likelihood_coeval,
model_name="TESTBURNIN",
continue_sampling=True,
datadir=tmpdirec,
params=default_params,
walkersRatio=2,
burninIterations=2,
sampleIterations=2,
threadCount=1,
)
# We set the _simulate back to True to have no side-effects.
likelihood_coeval._simulate = True