def test_BaseModel_lnprior(self):
scat = Sphere(r=prior.Gaussian(1, 1), n=prior.Gaussian(1, 1),
center=[10, 10, 10])
mod = Model(scat, noise_sd=0.1)
# Desired: log(sqrt(0.5/pi))-1/2
desired_sigma = -1.4189385332
assert_obj_close(mod.lnprior({'n': 0, 'r': 0}), desired_sigma * 2)
def test_PerfectLensModelholo_likelihood():
holo = get_example_data('image0001')
sphere_center = (prior.Gaussian(5, 1), prior.Gaussian(5, 1),
prior.Gaussian(5, 1))
s = Sphere(n=prior.Gaussian(1.6, .1),
r=prior.Gaussian(.5, .1),
center=sphere_center)
model = PerfectLensModel(s, noise_sd=0.01, lens_angle=0.8)
assert_pickle_roundtrip(model)
def test_ExactModelholo_likelihood():
holo = get_example_data('image0001')
sphere_center = (prior.Gaussian(5, 1), prior.Gaussian(5, 1),
prior.Gaussian(5, 1))
s = Sphere(n=prior.Gaussian(1.6, .1),
r=prior.Gaussian(.5, .1),
center=sphere_center)
model = ExactModel(s, noise_sd=.01)
assert_pickle_roundtrip(model)
def test_multidim():
s = Sphere(n = {'r':par(0,[-1,1]),'g':par(0,0),'b':prior.Gaussian(0,1),'a':0}, r = xr.DataArray([prior.Gaussian(0,1),par(0,[-1,1]),par(0,0),0], dims='alph',coords={'alph':['a','b','c','d']}), center=[par(0,[-1,1]),par(0,0),0])
par_s = ParameterizedObject(s)
params = {'n_r':3,'n_g':4,'n_b':5,'n_a':6,'r_a':7,'r_b':8,'r_c':9,'r_d':10,'center[0]':7,'center[1]':8,'center[2]':9}
out_s = Sphere(n={'r':3,'g':0,'b':5,'a':0}, r = xr.DataArray([7,8,0,0], dims='alph',coords={'alph':['a','b','c','d']}), center=[7,0,0])
assert_obj_close(par_s.make_from(params), out_s)
m = Model(s, np.sum)
m._use_parameter(OrderedDict([('r',par(0,[-1,1])),('g',par(0,0)),('b',prior.Gaussian(0,1)),('a',0)]),'letters')
m._use_parameter(xr.DataArray([prior.Gaussian(0,1),par(0,[-1,1]),par(0,0),0],dims='numbers',coords={'numbers':['one','two','three','four']}),'count')
expected_params = [par(0,[-1,1],'letters_r'),par(0,0,'letters_g'),prior.Gaussian(0,1,'letters_b'),prior.Gaussian(0,1,'count_one'),par(0,[-1,1],'count_two'), par(0,0,'count_three')]
assert_equal(m.parameters[-6:], expected_params)
def test_TimeSeriesAlphaModel():
n = TimeIndependent(prior.Gaussian(5, .5))
assert_pickle_roundtrip(n)
st = Sphere(n=n,
r=TimeIndependent(prior.BoundedGaussian(1.6, .1, 0, np.inf)),
center=(prior.Gaussian(10, 1), prior.Gaussian(10, 1),
prior.BoundedGaussian(1.6, .1, 0, np.inf)))
assert_pickle_roundtrip(st)
noise_sd = .1
mt = TimeSeriesAlphaModel(st,
Mie,
noise_sd,
alpha=prior.Uniform(0, 1),
n_frames=2)
assert_pickle_roundtrip(mt)
def test_Emcee_Class():
np.random.seed(40)
scat = model.Parametrization(0,[prior.Gaussian(0,1)])
mod = model.BaseModel(scat)
e = mcmc.Emcee(mod,[],nwalkers=3)
assert_equal(e.nwalkers,3)
assert_obj_close(e.make_guess(),prior_dist)
def test_subset_tempering():
holo = normalize(get_example_data('image0001'))
scat = Sphere(r=0.65e-6,n=1.58,center=[5.5e-6,5.8e-6,14e-6])
mod = AlphaModel(scat,noise_sd=.1, alpha=prior.Gaussian(0.7,0.1))
with warnings.catch_warnings():
warnings.simplefilter('ignore')
inf = sample.tempered_sample(mod, holo, nwalkers=4, samples=10, stages=1, stage_len=10, threads=None, seed=40)
assert_obj_close(inf.MAP,gold_alpha, rtol=1e-3)
def test_subset_tempering():
np.random.seed(40)
holo = normalize(get_example_data('image0001.yaml'))
scat = Sphere(r=0.65e-6,n=1.58,center=[5.5e-6,5.8e-6,14e-6])
mod = AlphaModel(scat,Mie,noise_sd=.1,alpha=prior.Gaussian(0.7,0.1))
inf=mcmc.subset_tempering(mod,holo,final_len=10,nwalkers=4,stages=1,stage_len=10,threads=None, verbose=False,seed=40)
assert_obj_close(inf.most_probable_values(),gold_alpha)
assert_equal(inf.n_steps,gold_nsteps)
assert_obj_close(inf.acceptance_fraction,gold_frac)
assert_obj_close(float(inf.data_frame(burn_in=6)[1:2].alpha),gold_alpha)
def test_alpha_subset_tempering(self):
holo = normalize(get_example_data('image0001'))
scat = Sphere(r=0.65e-6, n=1.58, center=[5.5e-6, 5.8e-6, 14e-6])
mod = AlphaModel(scat, noise_sd=.1, alpha=prior.Gaussian(0.7, 0.1))
strat = TemperedStrategy(nwalkers=4, nsamples=10, stages=1,
stage_len=10, parallel=None, seed=40)
with warnings.catch_warnings():
warnings.simplefilter('ignore')
inference_result = strat.sample(mod, holo)
desired_alpha = np.array([0.650348])
assert_allclose(inference_result._parameters, desired_alpha, rtol=5e-3)
def test_AlphaModelholo_likelihood():
holo = get_example_data('image0001')
s = Sphere(
prior.Gaussian(.5, .1), prior.Gaussian(1.6, .1),
(prior.Gaussian(5, 1), prior.Gaussian(5, 1), prior.Gaussian(5, 1)))
model = AlphaModel(s, alpha=prior.Gaussian(.7, .1), noise_sd=.01)
assert_pickle_roundtrip(model)
def test_emcee():
holo = get_example_data('image0001.yaml')
s = Sphere(
prior.Gaussian(.5, .1), prior.Gaussian(1.6, .1),
(prior.Gaussian(5, 1), prior.Gaussian(5, 1), prior.Gaussian(5, 1)))
model = AlphaModel(s, Mie, alpha=prior.Gaussian(.7, .1), noise_sd=.01)
emcee = Emcee(model, holo)
assert_pickle_roundtrip(emcee)
def test_perfectlens_subset_tempering(self):
data = normalize(get_example_data('image0001'))
scatterer = Sphere(r=0.65e-6, n=1.58, center=[5.5e-6, 5.8e-6, 14e-6])
model = PerfectLensModel(
scatterer, noise_sd=.1, lens_angle=prior.Gaussian(0.7, 0.1))
strat = TemperedStrategy(nwalkers=4, nsamples=10, stages=1,
stage_len=10, parallel=None, seed=40)
with warnings.catch_warnings():
warnings.simplefilter('ignore')
inference_result = strat.sample(model, data)
desired_lens_angle = np.array([0.7197887])
is_ok = np.allclose(
inference_result._parameters, desired_lens_angle, rtol=1e-3)
self.assertTrue(is_ok)
def test_find_noise():
noise = 0.5
s = Sphere(n=prior.Uniform(1.5, 1.7), r=2, center=[1, 2, 3])
data_base = detector_grid(10, spacing=0.5)
data_noise = update_metadata(data_base, noise_sd=noise)
model_u = AlphaModel(s, alpha=prior.Uniform(0.7, 0.9))
model_g = AlphaModel(s, alpha=prior.Gaussian(0.8, 0.1))
pars = {'n': 1.6, 'alpha': 0.8}
assert_equal(model_u._find_noise(pars, data_noise), noise)
assert_equal(model_g._find_noise(pars, data_noise), noise)
assert_equal(model_u._find_noise(pars, data_base), 1)
assert_raises(MissingParameter, model_g._find_noise, pars, data_base)
pars.update({'noise_sd': noise})
assert_equal(model_g._find_noise(pars, data_base), noise)
def test_multidim():
par_s = Sphere(n={
'r': prior.Uniform(-1, 1),
'g': 0,
'b': prior.Gaussian(0, 1),
'a': 0
},
r=xr.DataArray(
[prior.Gaussian(0, 1),
prior.Uniform(-1, 1), 0, 0],
dims='alph',
coords={'alph': ['a', 'b', 'c', 'd']}),
center=[prior.Uniform(-1, 1), 0, 0])
params = {
'n:r': 3,
'n:g': 4,
'n:b': 5,
'n:a': 6,
'r:a': 7,
'r:b': 8,
'r:c': 9,
'r:d': 10,
'center.0': 7,
'center.1': 8,
'center.2': 9
}
out_s = Sphere(n={
'r': 3,
'g': 0,
'b': 5,
'a': 0
},
r={
'a': 7,
'b': 8,
'c': 0,
'd': 0
},
center=[7, 0, 0])
assert_obj_close(par_s.from_parameters(params), out_s)
m = ExactModel(out_s, np.sum)
parletters = {
'r': prior.Uniform(-1, 1),
'g': 0,
'b': prior.Gaussian(0, 1),
'a': 0
}
parcount = xr.DataArray(
[prior.Gaussian(0, 1),
prior.Uniform(-1, 1), 0, 0],
dims='numbers',
coords={'numbers': ['one', 'two', 'three', 'four']})
m._use_parameters({'letters': parletters, 'count': parcount})
expected_params = {
'letters:r': prior.Uniform(-1, 1, 0, 'letters:r'),
'letters:b': prior.Gaussian(0, 1, 'letters:b'),
'count:one': prior.Gaussian(0, 1, 'count:one'),
'count:two': prior.Uniform(-1, 1, 0, 'count:two')
}
assert_equal(m.parameters, expected_params)
def test_gaussian():
g = prior.Gaussian(1, 1)
assert_equal(g.guess, 1)
assert_obj_close(g.lnprob(0),gold_sigma)
def test_NoiseModel_lnprior():
scat=Sphere(r=prior.Gaussian(1,1),n=prior.Gaussian(1,1),center=[10,10,10])
mod=NoiseModel(scat,None,noise_sd=.1)
assert_obj_close(mod.lnprior([0,0]),gold_sigma*2)
def test_require_noise_if_nonuniform(self):
sphere = Sphere(r=prior.Gaussian(0, 1), n=prior.Uniform(1, 2))
model = AlphaModel(sphere)
schema = detector_grid(2, 2)
pars = [0.5, 0.5]
self.assertRaises(MissingParameter, model._find_noise, pars, schema)
def test_prior():
g = prior.Gaussian(1, 1)
assert_pickle_roundtrip(g)
assert_pickle_roundtrip(g.lnprob)
