def test_residuals_wideband_chi2(wideband_fake):
toas, model = wideband_fake
r = WidebandTOAResiduals(toas, model)
rn = Residuals(toas, model)
f = WidebandTOAFitter(toas, model)
assert_allclose(f.fit_toas(), r.chi2)
assert f.fit_toas() >= rn.chi2
def wb():
m = get_model(join(datadir, "NGC6440E.par"))
t = make_fake_toas(55000, 58000, 20, model=m, freq=1400 * u.MHz, dm=10 * pint.dmu)
wb = WidebandTOAFitter(t, m)
wb.fit_toas()
return wb
def test_wideband_chi2_null_updating(wideband_fake):
toas, model = wideband_fake
model.free_params = ["F0"]
f = WidebandTOAFitter(toas, model)
assert abs(f.fit_toas() - WidebandTOAResiduals(toas, model).chi2) > 1
c2 = WidebandTOAResiduals(toas, f.model).chi2
assert_allclose(f.fit_toas(), c2)
c2 = WidebandTOAResiduals(toas, f.model).chi2
assert_allclose(f.fit_toas(), c2)
def test_noise_design_matrix_index(self):
model = get_model("B1855+09_NANOGrav_12yv3.wb.gls.par")
toas = get_TOAs("B1855+09_NANOGrav_12yv3.wb.tim",
ephem="DE436",
bipm_version="BIPM2015")
fitter = WidebandTOAFitter(toas, model, additional_args={})
fitter.fit_toas(full_cov=False, debug=True)
# Test red noise basis
pl_rd = fitter.model.pl_rn_basis_weight_pair(fitter.toas)[0]
p0, p1 = fitter.resids.pl_red_noise_M_index
pl_rd_backwards = (fitter.resids.pl_red_noise_M[0] *
fitter.resids.norm[p0:p1][np.newaxis, :])
assert np.all(np.isclose(pl_rd, pl_rd_backwards[0:313, :]))
def calc_chi2(self, full_cov=False):
"""Return the weighted chi-squared for the model and toas.
If the errors on the TOAs are independent this is a straightforward
calculation, but if the noise model introduces correlated errors then
obtaining a meaningful chi-squared value requires a Cholesky
decomposition. This is carried out, here, by constructing a GLSFitter
and asking it to do the chi-squared computation but not a fit.
The return value here is available as self.chi2, which will not
redo the computation unless necessary.
The chi-squared value calculated here is suitable for use in downhill
minimization algorithms and Bayesian approaches.
Handling of problematic results - degenerate conditions explored by
a minimizer for example - may need to be checked to confirm that they
correctly return infinity.
"""
# Use GLS but don't actually fit
from pint.fitter import WidebandTOAFitter
m = copy.deepcopy(self.model)
m.free_params = []
f = WidebandTOAFitter(
self.toas, m, additional_args=dict(toa=dict(track_mode=self.toa.track_mode))
)
try:
return f.fit_toas(maxiter=1, full_cov=full_cov)
except LinAlgError as e:
log.warning(
"Degenerate conditions encountered when "
"computing chi-squared: %s" % (e,)
)
return np.inf
def test_wideband_chi2_updating(wideband_fake):
toas, model = wideband_fake
model.free_params = ["F0"]
model.F0.value += 1e-6
c2 = WidebandTOAResiduals(
toas, model, toa_resid_args=dict(track_mode="use_pulse_numbers")
).chi2
f2 = WidebandTOAFitter(
toas, model, additional_args=dict(toa=dict(track_mode="use_pulse_numbers"))
)
ftc2 = f2.fit_toas()
assert abs(ftc2 - c2) > 100
assert_allclose(f2.model.F0.value, 1)
assert 1e-3 > abs(WidebandTOAResiduals(toas, f2.model).chi2 - ftc2) > 1e-5
ftc2 = f2.fit_toas(maxiter=10)
assert_allclose(WidebandTOAResiduals(toas, f2.model).chi2, ftc2)
def test_fitting_no_full_cov(self):
fitter = WidebandTOAFitter([self.toas], self.model, additional_args={})
time_rms_pre = fitter.resids_init.toa.rms_weighted()
dm_rms_pre = fitter.resids_init.dm.rms_weighted()
fitter.fit_toas()
dm_rms_post = fitter.resids.dm.rms_weighted()
prefit_pint = fitter.resids_init.toa.time_resids
prefit_tempo = self.tempo_res[:, 0] * u.us
diff_prefit = (prefit_pint - prefit_tempo).to(u.ns)
# 50 ns is the difference of PINT tempo precession and nautation model.
assert np.abs(diff_prefit - diff_prefit.mean()).max() < 50 * u.ns
postfit_pint = fitter.resids.toa.time_resids
postfit_tempo = self.tempo_res[:, 1] * u.us
diff_postfit = (postfit_pint - postfit_tempo).to(u.ns)
assert np.abs(diff_postfit - diff_postfit.mean()).max() < 50 * u.ns
assert np.abs(dm_rms_pre - dm_rms_post) < 3e-8 * dm_rms_pre.unit
def test_wideband_fit_dmjump_all(wb_model, wb_toas_all):
wb_model.free_params = ["DMJUMP1"]
fitter = WidebandTOAFitter(wb_toas_all, wb_model)
fitter.fit_toas()
print(fitter.print_summary())
assert_allclose(fitter.model.DMJUMP1.value, -10, atol=1e-3)
def test_wideband_fit_dmjump(wb_model, wb_toas):
wb_model.free_params = ["DMJUMP1"]
fitter = WidebandTOAFitter(wb_toas, wb_model)
fitter.fit_toas()
assert_allclose(fitter.model.DMJUMP1.value, -10, atol=1e-3)
# %% toas.table[-1] # %% toas.table["flags"][0] # %% [markdown] # ## Do the fit # # As before, but now we need a fitter adapted to wideband TOAs. # %% fitter = WidebandTOAFitter(toas, model) # %% fitter.fit_toas() # %% [markdown] # ## What is new, compared to narrowband fitting? # %% [markdown] # ### Residual objects combine TOA and time data # %% type(fitter.resids) # %% [markdown] # #### If we look into the resids attribute, it has two independent Residual objects. # %% fitter.resids.toa, fitter.resids.dm
def test_fitting_full_cov(self):
fitter = WidebandTOAFitter([self.toas,], self.model, additional_args={})
rms_pre = fitter.resids_init.rms_weighted()
fitter.fit_toas(full_cov=True)
assert fitter.resids.rms_weighted() - rms_pre < 1e-9