def test_convolutional_model_log_gaussain_kernel_circular(self):
""" Test ck-motion w/ log gaussian kernel for circular orbit. """
# Arbitrary circular orbit.
windy_star.configure_params(
period=1000, eccentricity=0, rv_semi_amplitude=50,
argument_of_periastron=270, rv_offset=0,
log_sigma_time=0.5, mean_emissivity=30,
stellar_radius_a=1, wind_velocity_a=500)
windy_star.configure_epochs(
epochs=np.linspace(0, 2, 1000))
windy_star.configure_kernel(
kernel_mode='LogGaussian')
res_df = windy_star.convolutional_model_centroid_velocity
# Sanity checks.
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].max(), 50, places=0,
msg='Convolutional velocity unexpected max value.')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].min(), -50, places=0,
msg='Convolutional velocity unexpected min value.')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[0], 0, places=0,
msg='Convolutional velocity unexpected value at phase 0.')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[124], 50, places=0,
msg='Convolutional velocity unexpected value at phase 0.25')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[249], 0, places=0,
msg='Convolutional velocity unexpected value at phase 0.5')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[374], -50, places=0,
msg='Convolutional velocity unexpected value at phase 0.75')
def test_convolutional_model_top_hat_kernel_eccentric(self):
""" Test ck-motion w/ top hat kernel for eccentric orbit. """
# Arbitrary circular orbit.
windy_star.configure_params(
period=1000, eccentricity=0.9, rv_semi_amplitude=50,
argument_of_periastron=270, rv_offset=0,
ionisation_radius=50, photospheric_radius=5,
stellar_radius_a=1, wind_velocity_a=500)
windy_star.configure_epochs(
epochs=np.linspace(0, 2, 1000))
windy_star.configure_kernel(
kernel_mode='TopHat')
res_df = windy_star.convolutional_model_centroid_velocity
# Sanity checks.
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].max(), 50, places=0,
msg='Convolutional velocity unexpected max value.')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].min(), -50, places=0,
msg='Convolutional velocity unexpected min value.')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[0], 0, places=0,
msg='Convolutional velocity unexpected value at phase 0.')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[124], 50, places=0,
msg='Convolutional velocity unexpected value at phase 0.25')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[249], 0, places=0,
msg='Convolutional velocity unexpected value at phase 0.5')
self.assertNotAlmostEqual(
res_df['ConvolvedVelocity'].iloc[374], -50, places=0,
msg='Convolutional velocity unexpected value at phase 0.75')
def test_convolutional_model_rt_interp_kernel_eccentric(self):
""" Test ck-motion w/ RT interp kernel for eccentric orbit. """
# Arbitrary eccentric orbit.
emissivity = Emissivity(setup_mode=False, pickle_dir='test_jar')
windy_star.configure_params(
period=2022.7, eccentricity=0.9, rv_semi_amplitude=100,
argument_of_periastron=270, rv_offset=0,
m_dot_a=5e-4, wind_velocity_a=500)
windy_star.configure_epochs(epochs=np.linspace(0, 2, 1000))
windy_star.configure_kernel(kernel_mode='RTInterp',
kernel_line=3, interp_dims='Grid',
emissivity_interpolators=emissivity)
res_df = windy_star.convolutional_model_centroid_velocity
def _mcmc_lnlike_kepler_model(self, theta):
""" Logarithmic likelihood. """
# Current position of walkers in each dimension of parameter space.
t0, e, k, w, c = theta
# Configure params and epochs.
windy_star.__init__()
windy_star.configure_params(period=self.fixed_params['Period'],
eccentricity=e,
rv_semi_amplitude=k,
argument_of_periastron=w,
rv_offset=c)
windy_star.configure_epochs(epochs=self.observations['JD'].values,
time_of_periastron=t0)
# Calculate centroid velocity of a binary Keplerian orbit.
res_df = windy_star.keplerian_model_centroid_velocity
# Log likelihood: Chi-squared + ln(2 * pi * sigma^2).
ln_like = -0.5 * (
(((res_df['KeplerVelocity'] - self.observations['Velocity'])**2) /
(self.observations['VelocityError']**2)) +
np.log(2 * np.pi * (self.observations['VelocityError']**2))).sum()
if self.debug:
plt.scatter(windy_star.epochs,
self.observations['Velocity'],
c='red',
s=2)
plt.scatter(res_df['Phase'],
res_df['KeplerVelocity'],
c='blue',
s=2)
plt.title('ln(likelihood)={}'.format(round(ln_like, 5)),
fontsize=10)
plt.xlabel('Orbital Phase $\phi$', fontsize=10)
plt.ylabel('Velocity / $\\rm{km\,s^{-1}}$', fontsize=10)
plt.tight_layout()
plt.show()
return ln_like
def test_keplerian_eccentric_orbit_model(self):
""" Test Keplerian eccentric orbit model. """
# Arbitrary eccentric orbit.
windy_star.configure_params(period=1000,
eccentricity=0.9,
rv_semi_amplitude=50,
argument_of_periastron=270,
rv_offset=0)
windy_star.configure_epochs(epochs=np.linspace(0, 2, 1000))
res_df = windy_star.keplerian_model_centroid_velocity
# Sanity checks.
self.assertAlmostEqual(res_df['KeplerVelocity'].max(),
50,
places=0,
msg='Keplerian velocity unexpected max value.')
self.assertAlmostEqual(res_df['KeplerVelocity'].min(),
-50,
places=0,
msg='Keplerian velocity unexpected min value.')
self.assertAlmostEqual(
res_df['KeplerVelocity'].iloc[0],
0,
places=0,
msg='Keplerian velocity unexpected value at phase 0.')
self.assertNotAlmostEqual(
res_df['KeplerVelocity'].iloc[124],
50,
places=0,
msg='Keplerian velocity unexpected value at phase 0.25')
self.assertAlmostEqual(
res_df['KeplerVelocity'].iloc[249],
0,
places=0,
msg='Keplerian velocity unexpected value at phase 0.5')
self.assertNotAlmostEqual(
res_df['KeplerVelocity'].iloc[374],
-50,
places=0,
msg='Keplerian velocity unexpected value at phase 0.75')
obs = unpickle_observations(OBSERVATIONS_PATH,
OBSERVATIONS,
time_of_periastron=2555555,
period=1000,
n_periods=2)
# Load interpolation objects.
emissivity = Emissivity(setup_mode=False, pickle_dir='interp_jar/rt_code_v1')
# Configure params and epochs and kernel.
windy_star.configure_params(period=1000,
eccentricity=0.7,
rv_semi_amplitude=60,
argument_of_periastron=250,
rv_offset=-20)
windy_star.configure_epochs(epochs=np.linspace(0, 2, 1000))
windy_star.configure_kernel(kernel_mode='RTInterp',
kernel_line=4,
interp_dims='Point',
emissivity_interpolators=emissivity)
# Calculate centroid velocity of convolved/time averaged Keplerian orbit
res_df = windy_star.convolutional_model_centroid_velocity
# Plot.
fig = plt.figure(figsize=(15, 6))
ax1 = fig.add_subplot(1, 1, 1)
ax1.scatter(obs['Phase'], obs['Velocity'], s=2, c='#000000')
ax1.plot(res_df['Phase'], res_df['ConvolvedVelocity'], linewidth=2)
ax1.set_xlabel('Orbital Phase $\phi$', fontsize=14)
ax1.set_ylabel('Velocity / $\\rm{km\,s^{-1}}$', fontsize=14)
label='Observations')
[bar.set_alpha(0.2) for bar in bars]
[cap.set_alpha(0.2) for cap in caps]
# Plot model samples.
count = 0
for t0, e, k, w, c in sample_chain[
np.random.randint(len(sample_chain), size=N_CHAIN_SAMPLES), :]:
try:
windy_star.__init__()
windy_star.configure_params(period=fixed_params['Period'],
eccentricity=e,
rv_semi_amplitude=k,
argument_of_periastron=w,
rv_offset=c)
windy_star.configure_epochs(epochs=np.linspace(0, N_PERIODS, 2000 *
N_PERIODS))
res_df = windy_star.keplerian_model_centroid_velocity
adjusted_phase = res_df['Phase'] + (
(t0 - REFERENCE_PERIASTRON) / PERIOD)
if count == 0:
ax1.plot([], [],
color='#4891dc',
alpha=1,
linewidth=1,
label='MCMC Kepler solution samples')
ax1.legend(loc='upper right', fontsize=10)
else:
ax1.plot(adjusted_phase,
res_df['KeplerVelocity'],
color='#4891dc',
alpha=0.2,
# Plot model samples.
count = 0
for t0, e, k, w in sample_chain[
np.random.randint(len(sample_chain), size=N_CHAIN_SAMPLES), :]:
try:
windy_star.__init__()
windy_star.configure_params(
period=fixed_params['Period'],
eccentricity=e,
rv_semi_amplitude=k,
argument_of_periastron=w,
rv_offset=systemic[line_index - 1],
m_dot_a=fixed_params['MassLossRate'],
wind_velocity_a=fixed_params['WindTerminalVelocity'])
windy_star.configure_epochs(
epochs=np.linspace(0, N_PERIODS, EPOCH_DENSITY))
windy_star.configure_kernel(kernel_mode='RTInterp',
kernel_line=line,
interp_dims='Point',
emissivity_interpolators=emissivity)
res_df = windy_star.convolutional_model_centroid_velocity
adjusted_phase = res_df['Phase'] + (
(t0 - REFERENCE_PERIASTRON) / PERIOD)
if count == 0:
ax.plot([], [],
color='#4891dc',
alpha=1,
linewidth=1,
label='MCMC CK (RTInterp) solution samples')
ax.legend(loc='upper right', fontsize=8)
else: