def test_with_cov(self):
simdict = Simulations()
meadict = Measurements()
covdict = Covariances()
# mock measurements
arr_a = np.random.rand(1, 4*mpisize)
comm.Bcast(arr_a, root=0)
mea = Observable(arr_a, 'measured')
meadict.append(name=('test', None, 4*mpisize, None),
data=mea, otype='plain')
# mock sims
arr_b = np.random.rand(5, 4*mpisize)
sim = Observable(arr_b, 'simulated')
simdict.append(name=('test', None, 4*mpisize, None),
data=sim, otype='plain')
# mock covariance
arr_c = np.random.rand(4, 4*mpisize)
cov = Observable(arr_c, 'covariance')
covdict.append(name=('test', None, 4*mpisize, None),
cov_data=cov)
# with covariance
lh = SimpleLikelihood(meadict, covdict)
# calc by likelihood
rslt = lh(simdict) # feed variable value, not parameter value
# calc by hand
full_b = np.vstack(comm.allgather(arr_b)) # global arr_b
diff = (np.mean(full_b, axis=0) - arr_a)
full_cov = np.vstack(comm.allgather(arr_c)) # global covariance
(sign, logdet) = np.linalg.slogdet(full_cov*2.*np.pi)
baseline = -0.5*(np.vdot(diff, np.linalg.solve(full_cov, diff.T))+sign*logdet)
assert np.allclose(rslt, baseline)
def test_without_simcov(self):
simdict = Simulations()
meadict = Measurements()
covdict = Covariances()
# mock measurements
arr_a = np.random.rand(1, 4*mpisize)
comm.Bcast(arr_a, root=0)
mea = Observable(arr_a, 'measured')
meadict.append(name=('test', None, 4*mpisize, None),
data=mea, otype='plain')
# mock covariance
arr_c = np.random.rand(4, 4*mpisize)
cov = Observable(arr_c, 'covariance')
covdict.append(name=('test', None, 4*mpisize, None),
cov_data=cov)
# mock observable with repeated single realisation
arr_b = np.random.rand(1, 4*mpisize)
comm.Bcast(arr_b, root=0)
arr_ens = np.zeros((2, 4*mpisize))
for i in range(len(arr_ens)):
arr_ens[i] = arr_b
sim = Observable(arr_ens, 'simulated')
simdict.append(name=('test', None, 4*mpisize, None),
data=sim, otype='plain')
# simplelikelihood
lh_simple = SimpleLikelihood(meadict, covdict)
rslt_simple = lh_simple(simdict)
# ensemblelikelihood
lh_ensemble = EnsembleLikelihood(meadict, covdict)
rslt_ensemble = lh_ensemble(simdict)
assert rslt_ensemble == rslt_simple
def test_without_cov(self):
simdict = Simulations()
meadict = Measurements()
# mock measurements
arr_a = np.random.rand(1, 12*mpisize**2)
comm.Bcast(arr_a, root=0)
mea = Observable(arr_a, 'measured')
meadict.append(name=('test', None, mpisize, None),
data=mea, otype='HEALPix')
# mock observable with repeated single realisation
arr_b = np.random.rand(1, 12*mpisize**2)
comm.Bcast(arr_b, root=0)
if not mpirank:
arr_ens = np.zeros((3, 12*mpisize**2))
else:
arr_ens = np.zeros((2, 12*mpisize**2))
for i in range(len(arr_ens)):
arr_ens[i] = arr_b
sim = Observable(arr_ens, 'simulated')
simdict.append(name=('test', None, mpisize, None),
data=sim, otype='HEALPix')
# simplelikelihood
lh_simple = SimpleLikelihood(meadict)
rslt_simple = lh_simple(simdict)
# ensemblelikelihood
lh_ensemble = EnsembleLikelihood(meadict)
rslt_ensemble = lh_ensemble(simdict)
assert rslt_ensemble == rslt_simple
def test_diag(self):
simdict = Simulations()
meadict = Measurements()
covdict = Covariances()
# mock measurements
arr_a = np.random.rand(1, 40)
mea = Observable(arr_a, 'measured')
meadict.append(name=('test', None, 40, None),
data=mea, otype='plain')
# mock (diagonal) covariance
arr_var = np.random.rand(40)
cov = Observable(np.diag(arr_var), 'covariance')
covdict.append(name=('test', None, 40, None),
cov_data=cov)
# mock observable
arr_ens = np.random.rand(10, 40)
sim = Observable(arr_ens, 'simulated')
simdict.append(name=('test', None, 40, None),
data=sim, otype='plain')
# ensemblelikelihood + diagonal_covcov
lh_ens = EnsembleLikelihood(meadict, covdict, cov_func=diagonal_mcov)
result_ens = lh_ens(simdict)
# EnsembleLikelihoodDiagonal
lh_diag = EnsembleLikelihoodDiagonal(meadict, covdict)
result_diag = lh_diag(simdict)
assert np.allclose(result_diag, result_ens)
def test_with_trace_approximation(self):
simdict = Simulations()
meadict = Measurements()
covdict = Covariances()
# mock measurements
arr_a = np.random.rand(1, 4*mpisize)
comm.Bcast(arr_a, root=0)
mea = Observable(arr_a, 'measured')
meadict.append(name=('test', None, 4*mpisize, None),
data=mea, otype='plain')
# mock covariance (NB for the trace approximation to work, the data
# covariance needs to be diagonal)
arr_c = np.diag(np.random.rand(4))
cov = Observable(arr_c, 'covariance')
covdict.append(name=('test', None, 4*mpisize, None),
cov_data=cov)
# mock observable with repeated single realisation
arr_b = np.random.rand(1, 4*mpisize)
comm.Bcast(arr_b, root=0)
arr_ens = np.zeros((2, 4*mpisize))
for i in range(len(arr_ens)):
arr_ens[i] = arr_b
sim = Observable(arr_ens, 'simulated')
simdict.append(name=('test', None, 4*mpisize, None),
data=sim, otype='plain')
# simplelikelihood
lh_simple = SimpleLikelihood(meadict, covdict)
result_simple = lh_simple(simdict)
# ensemblelikelihood
lh_ensemble = EnsembleLikelihood(meadict, covdict,
use_trace_approximation=True)
result_ensemble = lh_ensemble(simdict)
assert result_ensemble == result_simple
def test_measuredict_append_observable(self):
hrr = np.random.rand(1, 48)
obs1 = Observable(hrr, 'measured')
measuredict = Measurements()
measuredict.append(name=('test', None, 2, None),
data=obs1) # healpix Observable
assert np.allclose(measuredict[('test', None, 2, None)].data[0],
hrr[0])
arr = np.random.rand(1, 3)
obs2 = Observable(arr, 'measured')
measuredict.append(name=('test', None, 3, None),
data=obs2,
otype='plain') # plain Observable
assert np.allclose(measuredict[('test', None, 3, None)].data[0],
arr[0])
def test_measuredict_append_array(self):
arr = np.random.rand(1, 3)
measuredict = Measurements()
measuredict.append(name=('test', None, 3, None),
data=arr,
otype='plain') # plain array
local_arr = measuredict[('test', None, 3, None)].data
if mpirank == 0:
assert np.allclose(local_arr[0], arr[0])
hrr = np.random.rand(1, 48)
measuredict.append(name=('test', None, 2, None),
data=hrr,
otype='HEALPix') # healpix array
local_arr = measuredict[('test', None, 2, None)].data
if mpirank == 0:
assert np.allclose(local_arr[0], hrr[0])
def test_meadict_apply_mask(self):
msk = np.array([0, 1, 0, 1, 1]).reshape(1, 5)
mskdict = Masks()
comm.Bcast(msk, root=0)
mskdict.append(name=('test', None, 5, None), data=msk)
arr = np.array([0., 1., 2., 3., 4.]).reshape(1, 5)
meadict = Measurements()
meadict.append(name=('test', None, 5, None), data=arr, otype='plain')
meadict = mskdict(meadict)
assert np.allclose(meadict[('test', None, 3, None)].data[0], [1, 3, 4])
# HEALPix map
msk = np.random.randint(0, 2, 48).reshape(1, 48)
comm.Bcast(msk, root=0)
mskdict.append(name=('test', None, 2, None), data=msk)
arr = np.random.rand(1, 48)
meadict.append(name=('test', None, 2, None), data=arr, otype='HEALPix')
pix_num = msk.sum()
meadict = mskdict(meadict)
assert ('test', None, pix_num, None) in meadict.keys()
def test_Simulator_dependency_resolution():
dat = TabularDataset({
'data': [0],
'lat': 0,
'lon': 0,
'err': 0.1
},
name='nothing',
units=u.rad,
data_col='data',
err_col='err')
mea = Measurements()
mea.append(dat)
sim = DummySimulator(mea)
fields_list = list(map(lambda x: x(grid), (F, E, D, C, B, A)))
obs = sim(fields_list)
assert obs[('nothing', None, 'tab', None)].global_data[0][0] == 33.3
def test_without_cov(self):
simdict = Simulations()
meadict = Measurements()
# mock measurements
arr_a = np.random.rand(1, 48)
comm.Bcast(arr_a, root=0)
mea = Observable(arr_a, 'measured')
meadict.append(name=('test', None, 2, None), data=mea, otype='HEALPix')
# mock sims
arr_b = np.random.rand(3, 48)
sim = Observable(arr_b, 'simulated')
simdict.append(name=('test', None, 2, None), data=sim, otype='HEALPix')
# no covariance
lh = SimpleLikelihood(meadict)
# calc by likelihood
rslt = lh(simdict) # feed variable value, not parameter value
# calc by hand
full_b = np.vstack(comm.allgather(arr_b)) # global arr_b
diff = (np.mean(full_b, axis=0) - arr_a)
baseline = -float(0.5)*float(np.vdot(diff, diff))
# comapre
assert np.allclose(rslt, baseline)
def tutorial_one():
# Package imports
from imagine.fields import (
CosThermalElectronDensityFactory, NaiveGaussianMagneticFieldFactory,
UniformGrid)
from imagine.likelihoods import EnsembleLikelihood
from imagine.observables import Covariances, Measurements, TabularDataset
from imagine.pipelines import UltranestPipeline
from imagine.priors import FlatPrior
from imagine.simulators import TestSimulator
# IMAGINE-PRISM imports
from imagine_prism import PRISMPipeline
# Obtain data
data = get_mock_data()
# Create dataset
mock_dataset = TabularDataset(data, name='test', data_col='meas',
err_col='err')
# Create measurements and covariances objects
mock_data = Measurements(mock_dataset)
mock_cov = Covariances(mock_dataset)
# Create grid
grid = UniformGrid(box=[[0, 2*np.pi]*apu.kpc,
[0, 0]*apu.kpc,
[0, 0]*apu.kpc],
resolution=[30, 1, 1])
# Create factory for electron density field
ne_factory = CosThermalElectronDensityFactory(grid=grid)
ne_factory.default_parameters = {'a': 1*apu.rad/apu.kpc,
'beta': np.pi/2*apu.rad,
'gamma': np.pi/2*apu.rad}
# Create factory for magnetic field
B_factory = NaiveGaussianMagneticFieldFactory(grid=grid)
B_factory.active_parameters = ('a0', 'b0')
B_factory.priors = {'a0': FlatPrior(*[-5, 5]*apu.microgauss),
'b0': FlatPrior(*[0, 10]*apu.microgauss)}
# Combine factories together
factories = [ne_factory, B_factory]
# Create simulator
simulator = TestSimulator(mock_data)
# Create likelihood
likelihood = EnsembleLikelihood(mock_data, mock_cov)
# Create pipeline
img_pipe = UltranestPipeline(simulator=simulator,
factory_list=factories,
likelihood=likelihood,
ensemble_size=150)
# Create PRISMPipeline object
pipe = PRISMPipeline(img_pipe, root_dir='tests', working_dir='imagine_1')
# Return pipe
return(pipe)