def test_anue_free_spectra(tmp_path):
""" Test implementation of a model of antineutrino spectra with free
parameters in exponential parametrization.
"""
_enu = np.linspace(1.8, 8.0, 500, dtype='d')
Enu = C.Points(_enu, labels='anue energy')
indices = [
('i', 'isotope', ['U235', 'U238', 'Pu239', 'Pu241'])
]
expr = ['anuspec[i,r](enu())']
a = Expression_v01(expr, indices = NIndex.fromlist(indices))
a.parse()
lib = dict()
a.guessname(lib, save=True)
ns_anuexpr = env.globalns('anue_expr')
cfg = NestedDict(
anuspec = NestedDict(
bundle = dict(name='reactor_anu_spectra', version='v04'),
name = 'anuspec',
filename = ['data/reactor_anu_spectra/Huber/Huber_smooth_extrap_{isotope}_13MeV0.01MeVbin.dat',
'data/reactor_anu_spectra/Mueller/Mueller_smooth_extrap_{isotope}_13MeV0.01MeVbin.dat'],
# strategy = dict( underflow='constant', overflow='extrapolate' ),
varmode='log',
varname='anu_weight_{index}',
free_params=True,
ns_name='spectral_weights',
edges = np.concatenate( ( np.arange( 1.8, 8.7, 0.5 ), [ 12.3 ] ) ),
),
enu = NestedDict(
bundle = NestedDict(name='predefined', version='v01', major=''),
name = 'enu',
inputs = None,
outputs = Enu.single(),
),
)
context = ExpressionContext_v01(cfg, ns=ns_anuexpr)
a.build(context)
ns_anuexpr.printparameters(labels=True)
u235_spec = context.outputs.anuspec.U235
u235_spec.plot_vs(Enu.single(), label='default pars')
ns_anuexpr['spectral_weights.anu_weight_5'].set(0.3)
ns_anuexpr['spectral_weights.anu_weight_7'].set(-0.3)
plt.rcParams.update({'font.size': 14})
plt.rcParams.update({'text.usetex': True})
u235_spec.plot_vs(Enu.single(), label='update pars')
plt.yscale('log')
plt.xlabel(r'$E_{\nu}$, MeV')
plt.ylabel('Anue per MeV')
plt.legend()
plt.title('Antineutrino spectrum')
path = os.path.join(str(tmp_path), 'anuspec.png')
savefig(path, dpi=300)
allure_attach_file(path)
def test_MatrixProductDVDt(n1, n2, sym, tmp_path):
'''Matrix product of D V Dt'''
size1, size2 = n1 * n2, n2 * n2
left_n = np.arange(size1, dtype='d').reshape(n1, n2) + 1.0
square_n = 40.0 + size2 - np.arange(size2, dtype='d').reshape(n2, n2)
if sym:
square_n = 0.5 * (square_n + square_n.T)
left = C.Points(left_n)
square = C.Points(square_n)
dvdt = R.MatrixProductDVDt(left, square)
result = dvdt.product.product()
check = np.matmul(np.matmul(left_n, square_n), left_n.transpose())
diff = np.fabs(result - check)
# print('Max diff ({}, {})'.format(n1, n2), diff.max())
assert (result == check).all()
if sym:
checksym = result - result.T
assert not checksym.any()
path = os.path.join(str(tmp_path), 'graph_{:d}_{:d}.png'.format(n1, n2))
savegraph(dvdt.product, path, verbose=False)
allure_attach_file(path)
def test_offeq_correction_expression(tmp_path):
"""Same test but build bundle with expressions"""
_enu = np.linspace(1., 8.0, 500, dtype='d')
Enu = C.Points(_enu, labels='anue energy')
indices = [('i', 'isotope', ['U235', 'U238', 'Pu239', 'Pu241']),
('r', 'reactor', ['DB1', 'DB2', 'LA1', 'LA2', 'LA3', 'LA4'])]
expr = ['offeq_correction[i,r](enu())']
a = Expression_v01(expr, indices=NIndex.fromlist(indices))
a.parse()
lib = dict()
a.guessname(lib, save=True)
ns_offeq = env.globalns('offeq_expr')
cfg = NestedDict(
offeq_correction=NestedDict(
bundle=dict(name='reactor_offeq_spectra',
version='v03',
major='ir'),
offeq_data=
'./data/reactor_anu_spectra/Mueller/offeq/mueller_offequilibrium_corr_{isotope}.dat',
),
enu=NestedDict(
bundle=NestedDict(name='predefined', version='v01', major=''),
name='enu',
inputs=None,
outputs=Enu.single(),
),
)
context = ExpressionContext_v01(cfg, ns=ns_offeq)
a.build(context)
ns_offeq.printparameters(labels=True)
fig, ax = plt.subplots()
for iso in indices[0][2]:
corrected_spectra = context.outputs.offeq_correction[iso]['DB1'].data()
ax.plot(Enu.single().data(), corrected_spectra - 1., label=iso)
ax.set_title("Offequilibrium correction")
ax.legend(loc='best')
ax.grid()
ax.set_xlabel("Antineutrino energy, MeV")
ax.set_ylabel("(Corrected - nominal) / nominal")
suffix = 'correction'
path = os.path.join(str(tmp_path), suffix + '.png')
savefig(path, dpi=300)
savefig(path.replace('.png', '.pdf'), dpi=300)
allure_attach_file(path)
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph(context.outputs.offeq_correction['U235']['DB1'], path)
allure_attach_file(path)
def test_offeq_correction(tmp_path):
""" Test implementation of off-equilibrium contribution to antineutrino spectra
from nuclear fission in commercial reactors. The reason that contribution
appears is the fact the electron conversion spectra measured at ILL doesn't
take into account long-lived isotopes. That correction is required for
precision measurement of oscillation parameters and is uncorrelated
between reactors. It comes in a form of ratio of true spectra to nominal
Huber-Mueller.
"""
_enu = np.linspace(1., 8.0, 500, dtype='d')
Enu = C.Points(_enu, labels='anue energy')
indices = [('i', 'isotope', ['U235', 'U238', 'Pu239', 'Pu241']),
('r', 'reactor', ['DB1', 'DB2', 'LA1', 'LA2', 'LA3', 'LA4'])]
cfg = NestedDict(
bundle=dict(name='reactor_offeq_spectra', version='v03', nidx=indices),
offeq_data=
'./data/reactor_anu_spectra/Mueller/offeq/mueller_offequilibrium_corr_{isotope}.dat',
)
ns = env.globalns('offeq')
offeq, = execute_bundles(cfg, namespace=ns)
ns.printparameters(labels=True)
for iso in offeq.context.inputs.offeq_correction.values():
try:
for _input in iso.values():
Enu >> _input.values()
except AttributeError:
Enu >> iso
fig, ax = plt.subplots()
for iso in indices[0][2]:
corrected_spectra = offeq.context.outputs.offeq_correction[iso][
'DB1'].data()
ax.plot(Enu.single().data(), corrected_spectra - 1., label=iso)
ax.set_title("Offequilibrium correction")
ax.grid()
ax.legend(loc='best')
ax.set_xlabel("Antineutrino energy, MeV")
ax.set_ylabel("(Corrected - nominal) / nominal")
suffix = 'correction'
path = os.path.join(str(tmp_path), suffix + '.png')
savefig(path, dpi=300)
savefig(path.replace('.png', '.pdf'), dpi=300)
allure_attach_file(path)
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph(offeq.context.outputs.offeq_correction['U235']['DB1'], path)
allure_attach_file(path)
def savefig(self, *args):
path = '_'.join((self.tmp_path,)+tuple(args))+'.png'
savefig(path, dpi=300)
allure_attach_file(path)
def test_reactor_spectrum_unc_v01(tmp_path):
_enu = np.linspace(1.8, 10.0, 500, dtype='d')
Enu = C.Points(_enu, labels='anue energy')
indices = [('i', 'isotope', ['U235', 'U238', 'Pu239', 'Pu241']),
('r', 'reactor', ['DB1', 'LA1'])]
expr = ['anuspec[i](enu())', 'offeq_correction[i,r]| enu(), anuspec[i]()']
expr.append("corrected_spectrum[i,r]|enu(), anuspec[i]()")
a = Expression_v01(expr, indices=NIndex.fromlist(indices))
a.parse()
print(a.expressions)
lib = dict()
a.guessname(lib, save=True)
ns_anuexpr = env.globalns('anue_expr')
cfg = NestedDict(
anuspec=NestedDict(
bundle=dict(name='reactor_anu_spectra', version='v04'),
name='anuspec',
filename=[
'data/reactor_anu_spectra/Huber/Huber_smooth_extrap_{isotope}_13MeV0.01MeVbin.dat',
'data/reactor_anu_spectra/Mueller/Mueller_smooth_extrap_{isotope}_13MeV0.01MeVbin.dat'
],
varmode='log',
varname='anu_weight_{index}',
free_params=True,
ns_name='spectral_weights',
edges=np.concatenate((np.arange(1.8, 8.7, 0.5), [12.3])),
),
enu=NestedDict(
bundle=NestedDict(name='predefined', version='v01', major=''),
name='enu',
inputs=None,
outputs=Enu.single(),
),
corrected_spectrum=NestedDict(
bundle=dict(name='huber_mueller_spectra_uncertainty',
version='v01',
major='ir'),
reac_idx='r',
iso_idx='i',
ns_name='hm_uncertainties',
files_corr=[
'output/anue_spectra/Huber_corrrelated_unc_extrap_{isotope}_13.0_0.05_MeV.txt',
'output/anue_spectra/Mueller_corrrelated_unc_extrap_{isotope}_13.0_0.05_MeV.txt'
],
files_uncorr=[
'output/anue_spectra/Huber_uncorrrelated_unc_extrap_{isotope}_13.0_0.05_MeV.txt',
'output/anue_spectra/Mueller_uncorrrelated_unc_extrap_{isotope}_13.0_0.05_MeV.txt'
],
),
offeq_correction=NestedDict(
bundle=dict(name='reactor_offeq_spectra',
version='v04',
major='ir'),
offeq_data=
'./data/reactor_anu_spectra/Mueller/offeq/mueller_offequilibrium_corr_{isotope}.dat',
),
)
context = ExpressionContext_v01(cfg, ns=ns_anuexpr)
a.build(context)
ns_anuexpr.printparameters(labels=True)
u235_spec = context.outputs.corrected_spectrum.U235.DB1
suffix = 'hm_uncertainty_corrected_spec'
# savefig(path.replace('.png','.pdf'), dpi=300)
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph(u235_spec, path)
allure_attach_file(path)
path = os.path.join(str(tmp_path), suffix + '_graph.pdf')
savegraph(u235_spec, path)
allure_attach_file(path)
def test_offeq_correction_expression(tmp_path):
"""Tests for new version of offequilibrium correction with snapshoting
initial spectrum"""
_enu = np.linspace(1., 8.0, 500, dtype='d')
Enu = C.Points(_enu, labels='anue energy')
indices = [('i', 'isotope', ['U235', 'U238', 'Pu239', 'Pu241']),
('r', 'reactor', ['DB1', 'DB2', 'LA1', 'LA2', 'LA3', 'LA4'])]
expr = [
'enu()', 'anuspec[i](enu())',
'offeq_correction[i,r]| enu(), anuspec[i]()'
]
a = Expression_v01(expr, indices=NIndex.fromlist(indices))
a.parse()
lib = dict()
a.guessname(lib, save=True)
ns_offeq = env.globalns('offeq_expr_v04')
cfg = NestedDict(
offeq_correction=NestedDict(
bundle=dict(name='reactor_offeq_spectra',
version='v04',
major='ir'),
offeq_data=
'./data/reactor_anu_spectra/Mueller/offeq/mueller_offequilibrium_corr_{isotope}.dat',
),
enu=NestedDict(
bundle=NestedDict(name='predefined', version='v01', major=''),
name='enu',
inputs=None,
outputs=Enu.single(),
),
anuspec=NestedDict(
bundle=dict(name='reactor_anu_spectra', version='v04'),
name='anuspec',
filename=[
'data/reactor_anu_spectra/Huber/Huber_smooth_extrap_{isotope}_13MeV0.01MeVbin.dat',
'data/reactor_anu_spectra/Mueller/Mueller_smooth_extrap_{isotope}_13MeV0.01MeVbin.dat'
],
varmode='log',
varname='anu_weight_{index}',
free_params=True,
ns_name='spectral_weights',
edges=np.concatenate((np.arange(1.8, 8.7, 0.5), [12.3])),
),
)
context = ExpressionContext_v01(cfg, ns=ns_offeq)
a.build(context)
ns_offeq.printparameters(labels=True)
fig, ax = plt.subplots()
for iso in indices[0][2]:
corrected_spectra = context.outputs.offeq_correction[iso]['DB1'].data()
initial_spectra = context.outputs.anuspec[iso].data()
ax.plot(Enu.single().data(),
(corrected_spectra - initial_spectra) / initial_spectra,
label=iso)
ax.set_title("Offequilibrium corrected spectra")
ax.legend(loc='best')
ax.grid()
ax.set_xlabel("Antineutrino energy, MeV")
ax.set_ylabel("(Corrected - nominal) / nominal")
suffix = 'offeq'
path = os.path.join(str(tmp_path), suffix + '.png')
savefig(path.replace('.png', '.pdf'), dpi=300)
allure_attach_file(path)
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph(context.outputs.offeq_correction['U235']['DB1'], path)
allure_attach_file(path)
def test_covariated_prediction(syst1, syst2, tmp_path):
covbase_diag = not bool(syst1)
cov_diag = not bool(syst2) and covbase_diag
n = 10
start = 10
data = np.arange(start, start + n, dtype='d')
stat2 = data.copy()
fullcovmat = np.diag(stat2)
if syst1:
syst1 = np.ones((n, n), dtype='d') * 1.5
fullcovmat += syst1
cov = fullcovmat.copy()
covbase = cov
else:
cov = stat2
covbase = fullcovmat.copy()
Data = C.Points(data, labels='Data')
Covmat = C.Points(cov, labels='Input covariance matrix')
if syst2:
syst2 = np.ones((n, n), dtype='d') * 0.5
Syst = C.Points(syst2, labels='Input systematic matrix')
fullcovmat += syst2
else:
Syst = None
Cp = C.CovariatedPrediction(labels=[
'Concatenated prediction', 'Base covariance matrix',
'Full cov mat Cholesky decomposition'
])
Cp.append(Data)
Cp.covariate(Covmat, Data, 1, Data, 1)
if Syst:
Cp.addSystematicCovMatrix(Syst)
Cp.covbase.covbase >> Cp.cov.covbase
Cp.finalize()
Cp.printtransformations()
suffix = 'covariated_prediction_{}_{}'.format(
syst1 is not None and 'baseblock' or 'basediag',
syst2 is not None and 'syst' or 'nosyst')
fig = plt.figure()
ax = plt.subplot(111,
xlabel='X',
ylabel='Y',
title='Covariance matrix base')
ax.minorticks_on()
ax.grid()
if covbase_diag:
Cp.covbase.covbase.plot_hist(label='diag')
ax.legend()
else:
Cp.covbase.covbase.plot_matshow(colorbar=True)
path = os.path.join(str(tmp_path), suffix + '_covbase.png')
savefig(path, dpi=300)
allure_attach_file(path)
plt.close()
fig = plt.figure()
ax = plt.subplot(111,
xlabel='X',
ylabel='Y',
title='Covariance matrix full')
ax.minorticks_on()
ax.grid()
c = plt.matshow(np.ma.array(fullcovmat, mask=fullcovmat == 0.0),
fignum=False)
add_colorbar(c)
path = os.path.join(str(tmp_path), suffix + '_cov.png')
savefig(path, dpi=300)
allure_attach_file(path)
plt.close()
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph([Data.single(), Cp.covbase.covbase], path, verbose=False)
allure_attach_file(path)
data_o = Cp.prediction.prediction.data()
covbase_o = Cp.covbase.covbase.data()
if cov_diag:
L_o = Cp.cov.L.data()
L_expect = stat2**0.5
else:
L_o = np.tril(Cp.cov.L.data())
L_expect = np.linalg.cholesky(fullcovmat)
fig = plt.figure()
ax = plt.subplot(111,
xlabel='X',
ylabel='Y',
title='L: covariance matrix decomposition')
ax.minorticks_on()
ax.grid()
if cov_diag:
Cp.cov.L.plot_hist(label='diag')
ax.legend()
else:
c = plt.matshow(np.ma.array(L_o, mask=L_o == 0.0), fignum=False)
add_colorbar(c)
path = os.path.join(str(tmp_path), suffix + '_L.png')
savefig(path, dpi=300)
allure_attach_file(path)
plt.close()
assert (data == data_o).all()
if covbase_diag:
assert (covbase_o == data).all()
else:
assert (covbase == covbase_o).all()
assert np.allclose(L_o, L_expect)
def test_covariated_prediction_blocks(tmp_path):
ns = [1, 3, 4, 1]
start = 10
dataset = [np.arange(start, start + n, dtype='d') for n in ns]
stat2set = [data.copy() for data in dataset]
fullcovmat = [np.diag(stat2) for stat2 in stat2set]
it = 1.0
crosscovs = [[None] * len(ns) for i in range(len(ns))]
Crosscovs = [[None] * len(ns) for i in range(len(ns))]
for i in range(len(ns)):
for j in range(i + 1):
if i == j:
block = np.zeros((ns[i], ns[j]), dtype='d')
else:
block = np.ones((ns[i], ns[j]), dtype='d') * it * 0.5
it += 1
Crosscovs[i][j] = C.Points(block,
labels='Cross covariation %i %i' %
(i, j))
crosscovs[i][j] = block
if i != j:
crosscovs[j][i] = block.T
fulldata = np.concatenate(dataset)
fullsyst = np.block(crosscovs)
fullcovmat = np.diag(fulldata) + fullsyst
Dataset = [
C.Points(data, labels='Data %i' % i) for i, data in enumerate(dataset)
]
Statset = [
C.Points(data, labels='Stat %i' % i) for i, data in enumerate(dataset)
]
Cp = C.CovariatedPrediction(labels=[
'Concatenated prediction', 'Base covariance matrix',
'Full cov mat Cholesky decomposition'
])
for Data, Stat in zip(Dataset, Statset):
Cp.append(Data)
Cp.covariate(Stat, Data, 1, Data, 1)
for i in range(len(ns)):
for j in range(i):
Cp.covariate(Crosscovs[i][j], Dataset[i], 1, Dataset[j], 1)
Cp.covbase.covbase >> Cp.cov.covbase
Cp.finalize()
Cp.printtransformations()
suffix = 'covariated_prediction'
fig = plt.figure()
ax = plt.subplot(111,
xlabel='X',
ylabel='Y',
title='Covariance matrix base')
ax.minorticks_on()
ax.grid()
Cp.covbase.covbase.plot_matshow(colorbar=True, mask=0.0)
path = os.path.join(str(tmp_path), suffix + '_covbase.png')
savefig(path, dpi=300)
allure_attach_file(path)
plt.close()
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph([Cp.prediction, Cp.covbase.covbase], path, verbose=False)
allure_attach_file(path)
data_o = Cp.prediction.prediction.data()
covbase_o = Cp.covbase.covbase.data()
L_o = np.tril(Cp.cov.L.data())
L_expect = np.linalg.cholesky(fullcovmat)
fig = plt.figure()
ax = plt.subplot(111,
xlabel='X',
ylabel='Y',
title='L: covariance matrix decomposition')
ax.minorticks_on()
ax.grid()
c = plt.matshow(np.ma.array(L_o, mask=L_o == 0.0), fignum=False)
add_colorbar(c)
path = os.path.join(str(tmp_path), suffix + '_L.png')
savefig(path, dpi=300)
allure_attach_file(path)
plt.close()
assert (fulldata == data_o).all()
assert (fullcovmat == covbase_o).all()
assert np.allclose(L_o, L_expect)
def test_interpolators(interpolator, strategy, tmp_path):
'''Test various interpolators'''
segments = np.arange(1.0, 10.1, 1.5, dtype='d')
segments_t = C.Points(segments)
points = np.stack([
np.linspace(0.0 + i, 12. + i, 61, dtype='d')
for i in [0, -0.1, 0.1, 0.3, 0.5]
]).T
points_t = C.Points(points)
fcn = np.exp(-(segments - segments[0]) * 0.5)
fcn = np.exp(segments**(-0.5))
fcn_t = C.Points(fcn)
ie = interpolator()
ie.set_underflow_strategy(strategy)
ie.set_overflow_strategy(strategy)
ie.interpolate(segments_t, fcn_t, points_t)
seg_idx = ie.insegment.insegment.data()
res = ie.interp.interp.data()
fig = plt.figure()
ax = plt.subplot(111)
ax.minorticks_on()
ax.grid()
ax.set_xlabel('x')
ax.set_ylabel('y')
strategy_name = "Constant" if strategy is R.GNA.Interpolation.Strategy.Constant else "Extrapolate"
title = ie.__class__.__name__.strip(
'Interp') + ", " + "Strategy: {}".format(strategy_name)
ax.set_title(title)
ax.plot(segments,
fcn,
'o-',
markerfacecolor='none',
label='coarse function',
linewidth=0.5,
alpha=0.5)
markers = 'os*^vh'
for i, (p, r) in enumerate(zip(points.T, res.T)):
ax.plot(p,
r,
'.',
label='interpolation, col %i' % i,
marker=markers[i],
markersize=1.5)
break
ax.legend(loc='upper right')
suffix = '{}_{}'.format(ie.__class__.__name__, strategy_name)
path = os.path.join(str(tmp_path), suffix + '.png')
savefig(path, dpi=300)
allure_attach_file(path)
plt.close()
path = os.path.join(str(tmp_path), suffix + '_graph.png')
savegraph(points_t.points, path, verbose=False)
allure_attach_file(path)
plt.close()