def fit_MM_MCPM_EMCEE(
files, files_formats, files_kwargs, skycoord, methods, MCPM_options,
starting_settings, parameters_to_fit, parameters_fixed,
min_values, max_values, emcee_settings, priors_gauss, priors_tabulated,
other_constraints, file_all_models, config_file_root, gamma_LD,
model_type=None, data_add_245=True):
"""
Fit the microlensing (MulensModel) and K2 photometry (MCPM) using
EMCEE method. The input is complicated - please see code below and
in read_config.py to find out details.
emcee_settings['PTSampler'] == True means no blobs (i.e. ground-based
fluxes are passed)
emcee_settings['file_posterior'] ending in ".npy" means we're saving 3D or
4D array, while other extensions mean we're saving text file with
flattened posterior
model_type: *None* or *str*
Can be *None* (i.e., MM parameters are used), 'wide', 'close_A',
or 'close_B'. If not None, then 't_0_pl', 'u_0_pl', and 't_E_pl'
parameters are translated to s, q, alpha.
"""
print("MM version: " + MM.__version__)
print("MCPM version: " + MCPM_version)
print("EMCEE version: " + emcee.__version__)
print("script version: " + __version__, flush=True)
utils.get_standard_parameters.model_type = model_type
n_params = len(parameters_to_fit)
if file_all_models is None:
file_all_models = config_file_root + ".models"
# read datasets
datasets = []
if skycoord is not None:
coords = MM.Coordinates(skycoord)
else:
coords = None
if files is not None:
for (file_, fmt, kwargs) in zip(files, files_formats, files_kwargs):
data = MM.MulensData(file_name=file_, add_2450000=data_add_245,
phot_fmt=fmt, coords=coords, **kwargs)
datasets.append(data)
# satellite datasets
cpm_sources = []
for campaign in MCPM_options['campaigns']:
cpm_source = CpmFitSource(
ra=skycoord.ra.deg, dec=skycoord.dec.deg,
campaign=campaign, channel=MCPM_options['channel'])
cpm_source.get_predictor_matrix(**MCPM_options['predictor_matrix'])
cpm_source.set_l2_l2_per_pixel(
l2=MCPM_options['l2'], l2_per_pixel=MCPM_options['l2_per_pixel'])
cpm_source.set_pixels_square(MCPM_options['half_size'])
cpm_source.select_highest_prf_sum_pixels(MCPM_options['n_select'])
cpm_sources.append(cpm_source)
# initiate model
starting = utils.generate_random_points(
starting_settings, parameters_to_fit,
emcee_settings['n_walkers'] * emcee_settings['n_temps'])
zip_ = zip(parameters_to_fit, starting[0])
parameters = {key: value for (key, value) in zip_}
parameters.update(parameters_fixed)
parameters_ = {**parameters}
for param in list(parameters_.keys()).copy():
pop_keys = ['f_s_sat', 'f_s_sat_over_u_0']
if param in pop_keys or param[:3] == 'q_f' or param[:7] == 'log_q_f':
parameters_.pop(param)
if 't_0_pl' in parameters_:
parameters_ = utils.get_standard_parameters(parameters_)
try:
model = MM.Model(parameters_, coords=coords)
except KeyError:
model = PixelLensingModel(parameters_, coords=coords)
for (m_key, m_value) in methods.items():
model.set_magnification_methods(m_value, m_key)
for (band, gamma) in gamma_LD.items():
model.set_limb_coeff_gamma(band, gamma)
if isinstance(model, MM.Model):
if 'f_s_sat' in parameters:
f_s_sat = parameters['f_s_sat']
else:
f_s_sat = parameters['f_s_sat_over_u_0'] * model.parameters.u_0
for cpm_source in cpm_sources:
times = cpm_source.pixel_time + 2450000.
times[np.isnan(times)] = np.mean(times[~np.isnan(times)])
if model.n_sources == 1:
if not isinstance(model, MM.Model):
model_flux = model.flux_difference(times)
else:
model_flux = f_s_sat * (_get_magnification(model, times) - 1.)
else:
if not isinstance(model, MM.Model):
raise NotImplementedError('not yet coded for pixel lensing')
if ('log_q_f' in parameters) or ('q_f' in parameters):
if 'log_q_f' in parameters:
q_f = 10**parameters['log_q_f']
else:
q_f = parameters['q_f']
model.set_source_flux_ratio(q_f)
model_magnification = _get_magnification(model, times)
else: # This is very simple solution.
model_magnification = _get_magnification(
model, times, separate=True)[0]
model_flux = f_s_sat * (model_magnification - 1.)
cpm_source.run_cpm(model_flux)
utils.apply_limit_time(cpm_source, MCPM_options)
mask = cpm_source.residuals_mask
if 'mask_model_epochs' in MCPM_options:
mask *= utils.mask_nearest_epochs(
cpm_source.pixel_time+2450000.,
MCPM_options['mask_model_epochs'])
sat_time = cpm_source.pixel_time[mask] + 2450000.
# sat_sigma = sat_time * 0. + MCPM_options['sat_sigma']
sat_sigma = np.sqrt(np.sum(
np.array([err[mask] for err in cpm_source.pixel_flux_err])**2,
axis=0))
if 'sat_sigma_scale' in MCPM_options:
sat_sigma *= MCPM_options['sat_sigma_scale']
data = MM.MulensData(
[sat_time, 0.*sat_time, sat_sigma],
phot_fmt='flux', ephemerides_file=MCPM_options['ephemeris_file'],
bandpass="******")
datasets.append(data)
# initiate event and minimizer
if isinstance(model, MM.Model):
event = MM.Event(datasets=datasets, model=model)
else:
event = PixelLensingEvent(datasets=datasets, model=model)
params = parameters_to_fit[:]
minimizer = Minimizer(event, params, cpm_sources)
if emcee_settings['PTSampler']:
minimizer.save_fluxes = False
# Somehow blobs are not allowed for PTSampler.
minimizer.file_all_models = file_all_models
minimizer.set_chi2_0()
if 'f_s_sat' in parameters_fixed or 'f_s_sat_over_u_0' in parameters_fixed:
minimizer.set_satellite_source_flux(f_s_sat)
if 'coeffs_fits_in' in MCPM_options:
minimizer.read_coeffs_from_fits(MCPM_options['coeffs_fits_in'])
if 'coeffs_fits_out' in MCPM_options:
minimizer.start_coeffs_cache()
if 'sat_sigma_scale' in MCPM_options:
minimizer.sigma_scale = MCPM_options['sat_sigma_scale']
if 'color_constraint' in MCPM_options:
cc = 'color_constraint'
ref_dataset = files.index(MCPM_options[cc][0])
if len(MCPM_options[cc]) == 3:
ref_mag = MM.utils.MAG_ZEROPOINT
else:
ref_mag = MCPM_options[cc][1]
if len(MCPM_options[cc]) in [3, 4]:
minimizer.add_color_constraint(
ref_dataset, ref_mag,
MCPM_options[cc][-2], MCPM_options[cc][-1])
elif len(MCPM_options[cc]) in [5, 6, 7, 8]:
minimizer.add_full_color_constraint(
ref_dataset,
files.index(MCPM_options[cc][1]),
files.index(MCPM_options[cc][2]),
*MCPM_options[cc][3:])
else:
raise ValueError('wrong size of "color_constraint" option')
key = 'min_blending_flux'
if key in other_constraints:
index = files.index(other_constraints[key][0])
other_constraints[key] = [datasets[index], other_constraints[key][1]]
minimizer.other_constraints = other_constraints
key = 'no_blending_files'
if key in MCPM_options:
indexes = [files.index(f) for f in MCPM_options['no_blending_files']]
for ind in indexes:
minimizer.fit_blending[ind] = False
if 'mask_model_epochs' in MCPM_options:
for i in range(minimizer.n_sat):
minimizer.model_masks[i] = utils.mask_nearest_epochs(
cpm_sources[i].pixel_time+2450000.,
MCPM_options['mask_model_epochs'])
# EMCEE fit:
if emcee_settings['PTSampler']:
print("EMCEE temps, walkers, steps, burn: {:} {:} {:} {:}".format(
emcee_settings['n_temps'], emcee_settings['n_walkers'],
emcee_settings['n_steps'], emcee_settings['n_burn']))
else:
print("EMCEE walkers, steps, burn: {:} {:} {:}".format(
emcee_settings['n_walkers'], emcee_settings['n_steps'],
emcee_settings['n_burn']))
minimizer.set_prior_boundaries(min_values, max_values)
for start_ in starting:
if minimizer.ln_prior(start_) <= -float('inf'):
raise ValueError('starting point is not in prior:\n' + str(start_))
minimizer.set_prior_gaussian(priors_gauss)
for (parameter, file_name) in priors_tabulated.items():
minimizer.set_prior_tabulated(parameter, file_name)
if emcee_settings['PTSampler']:
sampler = emcee.PTSampler(
emcee_settings['n_temps'], emcee_settings['n_walkers'], n_params,
minimizer.ln_like, minimizer.ln_prior)
shape = (emcee_settings['n_temps'], emcee_settings['n_walkers'],
n_params)
starting = np.array(starting).reshape(shape)
else:
sampler = emcee.EnsembleSampler(
emcee_settings['n_walkers'], n_params, minimizer.ln_prob)
acceptance_fractions = []
# run:
kwargs = {'initial_state': starting,
'iterations': emcee_settings['n_steps']}
for _ in tqdm(sampler.sample(**kwargs), total=emcee_settings['n_steps']):
acceptance_fractions.append(np.mean(sampler.acceptance_fraction))
# cleanup and close minimizer:
out_name = emcee_settings.get('file_acceptance_fractions', None)
if out_name is not None:
if len(out_name) == 0:
out_name = config_file_root + ".accept"
data_save = [str(i+1) + " " + str(af)
for (i, af) in enumerate(acceptance_fractions)]
with open(out_name, 'w') as file_out:
file_out.write('\n'.join(data_save))
n_burn = emcee_settings['n_burn']
if emcee_settings['PTSampler']:
samples = sampler.chain[0, :, n_burn:, :].reshape((-1, n_params))
n_fluxes = 0
else:
samples = sampler.chain[:, n_burn:, :].reshape((-1, n_params))
blob_sampler = np.transpose(np.array(sampler.blobs), axes=(1, 0, 2))
n_fluxes = blob_sampler.shape[-1]
if 'coeffs_fits_out' in MCPM_options:
minimizer.set_pixel_coeffs_from_samples(samples)
minimizer.save_coeffs_to_fits(MCPM_options['coeffs_fits_out'])
minimizer.stop_coeffs_cache()
minimizer.close_file_all_models()
# output
text = "Mean acceptance fraction"
fmt = ": {:.4f} +- {:.4f}"
print(text + fmt.format(
np.mean(sampler.acceptance_fraction),
np.std(sampler.acceptance_fraction)))
if emcee_settings['PTSampler']:
print(text + " of the lowest temperature walkers" + fmt.format(
np.mean(sampler.acceptance_fraction[0, :]),
np.std(sampler.acceptance_fraction[0, :])))
zip_ = zip(*np.percentile(samples, [16, 50, 84], axis=0))
results = map(lambda v: (v[1], v[2]-v[1], v[0]-v[1]), zip_)
for (param, r) in zip(parameters_to_fit, results):
if r[1] < 1.e-3 or r[2] > -1.e-3:
fmt = '{:7s} : {:.5f} {:+.4g} {:+.4g}'
else:
fmt = '{:7s} : {:.4f} {:+.4f} {:+.4f}'
print(fmt.format(param, *r))
if n_fluxes > 0:
blob_samples = blob_sampler[:, n_burn:, :].reshape((-1, n_fluxes))
percentiles = np.percentile(blob_samples, [16, 50, 84], axis=0)
blob_results = map(
lambda v: (v[1], v[2]-v[1], v[1]-v[0]), zip(*percentiles))
msg = 'flux_{:}_{:} : {:.4f} {:.4f} {:.4f}'
for (i, r) in zip(range(n_fluxes), blob_results):
print(msg.format(['S', 'B'][i % 2], i//2+1, *r))
if 'file_posterior' in emcee_settings:
if len(emcee_settings['file_posterior']) == 0:
emcee_settings['file_posterior'] = config_file_root + ".posterior"
if emcee_settings['file_posterior'][-4:] == '.npy':
if emcee_settings['PTSampler']:
all_samples = sampler.chain[:, :, n_burn:, :]
else: # XXX blobs here
all_samples = sampler.chain[:, n_burn:, :]
np.save(emcee_settings['file_posterior'], all_samples)
else:
# all_samples = samples
all_samples = sampler.chain[:, n_burn:, :]
if n_fluxes > 0:
all_samples = np.concatenate(
(all_samples, blob_samples), axis=1)
np.save(emcee_settings['file_posterior'], all_samples)
print('Best model:')
minimizer.print_min_chi2()
# Plots are below
if 'file_corner' in emcee_settings or 'file_trace' in emcee_settings:
data = sampler.chain[:, n_burn:, :]
if 't_0' in parameters_to_fit:
index = parameters_to_fit.index('t_0')
data[:, :, index] -= int(np.median(data[:, :, index]))
# The above line doesn't work properly for PTSampler XXX
if 'file_trace' in emcee_settings:
if len(emcee_settings['file_trace']) == 0:
emcee_settings['file_trace'] = config_file_root + "_trace.png"
alpha = 0.5
grid = gridspec.GridSpec(n_params, 1, hspace=0)
plt.figure(figsize=(7.5, 10.5))
plt.subplots_adjust(left=0.13, right=0.97, top=0.99, bottom=0.05)
plt.rcParams['font.size'] = 12
plt.rcParams['axes.linewidth'] = 1.4
for i in range(n_params):
if i == 0:
ax0 = plt.subplot(grid[i])
else:
plt.gcf().add_subplot(grid[i], sharex=ax0)
plt.ylabel(parameters_to_fit[i])
for j in range(data.shape[0]):
vector = data[j, :, i]
plt.plot(np.arange(len(vector)), vector, alpha=alpha)
plt.xlim(0, len(vector))
plt.gca().tick_params(axis='both', which='both', direction='in',
top=True, right=True)
if i != n_params - 1:
plt.setp(plt.gca().get_xticklabels(), visible=False)
plt.gca().set_prop_cycle(None)
plt.xlabel('step count')
plt.savefig(emcee_settings['file_trace'])
plt.rcParams['font.size'] = 10 # resetting to defaults
plt.rcParams['axes.linewidth'] = 0.8
if 'file_corner' in emcee_settings:
if len(emcee_settings['file_corner']) == 0:
emcee_settings['file_corner'] = config_file_root + "_corner.png"
kwargs = {'quantiles': [0.16, 0.50, 0.84], 'bins': 40,
'show_titles': True, 'labels': parameters_to_fit}
corner.corner(data.reshape(-1, n_params), **kwargs)
plt.savefig(emcee_settings['file_corner'])
model.set_source_flux_ratio(q_f)
model_magnification = model.magnification(times)
else:
model_magnification = model.magnification(
times, separate=True)[0] # This is very simple solution.
if 'f_s_sat' in parameters:
f_s_sat = parameters['f_s_sat']
else:
f_s_sat = parameters['f_s_sat_over_u_0'] * model.parameters.u_0
cpm_source.run_cpm(f_s_sat * (model_magnification - 1.))
utils.apply_limit_time(cpm_source, MCPM_options)
mask = cpm_source.residuals_mask
if 'mask_model_epochs' in MCPM_options:
mask *= utils.mask_nearest_epochs(cpm_source.pixel_time + 2450000.,
MCPM_options['mask_model_epochs'])
sat_time = cpm_source.pixel_time[mask] + 2450000.
# sat_sigma = sat_time * 0. + MCPM_options['sat_sigma']
sat_sigma = np.sqrt(
np.sum(np.array([err[mask] for err in cpm_source.pixel_flux_err])**2,
axis=0))
if 'sat_sigma_scale' in MCPM_options:
sat_sigma *= MCPM_options['sat_sigma_scale']
data = MM.MulensData([sat_time, 0. * sat_time, sat_sigma],
phot_fmt='flux',
ephemerides_file=MCPM_options['ephemeris_file'],
bandpass="******")
datasets.append(data)
# initiate event and minimizer
event = MM.Event(datasets=datasets, model=model)
def evaluate_MM_MCPM(
files, files_formats, files_kwargs, skycoord, methods, MCPM_options,
parameters_fixed, parameter_values, model_ids, plot_files, txt_files,
txt_files_prf_phot, txt_models, parameters_to_fit,
plot_epochs, plot_epochs_type, plot_settings, gamma_LD,
model_type=None, data_add_245=True):
"""
Evaluate MCPM model.
model_type: *None* or *str*
Can be *None* (i.e., MM parameters are used), 'wide', 'close_A',
or 'close_B'. If not None, then 't_0_pl', 'u_0_pl', and 't_E_pl'
parameters are translated to s, q, alpha.
"""
utils.get_standard_parameters.model_type = model_type
# read datasets
datasets = []
if skycoord is not None:
coords = MM.Coordinates(skycoord)
else:
coords = None
if files is not None:
for (file_, fmt, kwargs) in zip(files, files_formats, files_kwargs):
data = MM.MulensData(file_name=file_, add_2450000=data_add_245,
phot_fmt=fmt, coords=coords, **kwargs)
datasets.append(data)
# satellite datasets
cpm_sources = []
for campaign in MCPM_options['campaigns']:
cpm_source = CpmFitSource(ra=skycoord.ra.deg, dec=skycoord.dec.deg,
campaign=campaign,
channel=MCPM_options['channel'])
cpm_source.get_predictor_matrix(**MCPM_options['predictor_matrix'])
cpm_source.set_l2_l2_per_pixel(
l2=MCPM_options['l2'], l2_per_pixel=MCPM_options['l2_per_pixel'])
cpm_source.set_pixels_square(MCPM_options['half_size'])
if 'n_select' in MCPM_options:
cpm_source.select_highest_prf_sum_pixels(MCPM_options['n_select'])
cpm_sources.append(cpm_source)
# initiate model
model_begin = parameter_values[0]
parameters = {
key: value for (key, value) in zip(parameters_to_fit, model_begin)}
parameters.update(parameters_fixed)
parameters_ = {**parameters}
for param in list(parameters_.keys()).copy():
pop_keys = ['f_s_sat', 'f_s_sat_over_u_0']
if param in pop_keys or param[:3] == 'q_f' or param[:7] == 'log_q_f':
parameters_.pop(param)
if 't_0_pl' in parameters_:
parameters_ = utils.get_standard_parameters(parameters_)
try:
model = MM.Model(parameters_, coords=coords)
except KeyError:
model = PixelLensingModel(parameters_, coords=coords)
for (m_key, m_value) in methods.items():
model.set_magnification_methods(m_value, m_key)
for (band, gamma) in gamma_LD.items():
model.set_limb_coeff_gamma(band, gamma)
if isinstance(model, MM.Model):
if 'f_s_sat' in parameters:
f_s_sat = parameters['f_s_sat']
else:
f_s_sat = parameters['f_s_sat_over_u_0'] * model.parameters.u_0
for cpm_source in cpm_sources:
times = cpm_source.pixel_time + 2450000.
times[np.isnan(times)] = np.mean(times[~np.isnan(times)])
if model.n_sources == 1:
if not isinstance(model, MM.Model):
model_flux = model.flux_difference(times)
else:
model_flux = f_s_sat * (_get_magnification(model, times) - 1.)
else:
if not isinstance(model, MM.Model):
raise NotImplementedError('not yet coded for pixel lensing')
if ('log_q_f' in parameters) or ('q_f' in parameters):
if 'log_q_f' in parameters:
q_f = 10**parameters['log_q_f']
else:
q_f = parameters['q_f']
model.set_source_flux_ratio(q_f)
model_magnification = _get_magnification(model, times)
else: # This is very simple solution.
model_magnification = _get_magnification(
model, times, separate=True)[0]
model_flux = f_s_sat * (model_magnification - 1.)
cpm_source.run_cpm(model_flux)
utils.apply_limit_time(cpm_source, MCPM_options)
mask = cpm_source.residuals_mask
if 'mask_model_epochs' in MCPM_options:
mask *= utils.mask_nearest_epochs(
cpm_source.pixel_time+2450000.,
MCPM_options['mask_model_epochs'])
sat_time = cpm_source.pixel_time[mask] + 2450000.
# sat_sigma = sat_time * 0. + MCPM_options['sat_sigma']
err_masked = [err[mask] for err in cpm_source.pixel_flux_err]
sat_sigma = np.sqrt(np.sum(np.array(err_masked)**2, axis=0))
if 'sat_sigma_scale' in MCPM_options:
sat_sigma *= MCPM_options['sat_sigma_scale']
data = MM.MulensData(
[sat_time, 0.*sat_time, sat_sigma],
phot_fmt='flux', ephemerides_file=MCPM_options['ephemeris_file'],
bandpass="******", coords=coords)
datasets.append(data)
# initiate event
if isinstance(model, MM.Model):
event = MM.Event(datasets=datasets, model=model)
else:
event = PixelLensingEvent(datasets=datasets, model=model)
params = parameters_to_fit[:]
minimizer = Minimizer(event, params, cpm_sources)
if 'f_s_sat' in parameters_fixed or 'f_s_sat_over_u_0' in parameters_fixed:
minimizer.set_satellite_source_flux(f_s_sat)
if 'coeffs_fits_in' in MCPM_options:
minimizer.read_coeffs_from_fits(MCPM_options['coeffs_fits_in'])
if 'coeffs_fits_out' in MCPM_options:
raise ValueError("coeffs_fits_out cannot be set in this program")
if 'sat_sigma_scale' in MCPM_options:
minimizer.sigma_scale = MCPM_options['sat_sigma_scale']
if 'color_constraint' in MCPM_options:
cc = 'color_constraint'
ref_dataset = files.index(MCPM_options[cc][0])
if len(MCPM_options[cc]) == 3:
ref_mag = MM.utils.MAG_ZEROPOINT
else:
ref_mag = MCPM_options[cc][1]
if len(MCPM_options[cc]) in [3, 4]:
minimizer.add_color_constraint(
ref_dataset, ref_mag,
MCPM_options[cc][-2], MCPM_options[cc][-1])
elif len(MCPM_options[cc]) in [5, 6, 7, 8]:
minimizer.add_full_color_constraint(
ref_dataset, files.index(MCPM_options[cc][1]),
files.index(MCPM_options[cc][2]), *MCPM_options[cc][3:])
key = 'no_blending_files'
if key in MCPM_options:
indexes = [files.index(f) for f in MCPM_options['no_blending_files']]
for ind in indexes:
minimizer.fit_blending[ind] = False
if 'mask_model_epochs' in MCPM_options:
minimizer.model_masks[0] = utils.mask_nearest_epochs(
cpm_sources[0].pixel_time+2450000.,
MCPM_options['mask_model_epochs'])
# main loop:
zipped = zip(parameter_values, model_ids, plot_files, txt_files,
txt_files_prf_phot, txt_models, plot_epochs)
for zip_single in zipped:
(values, name, plot_file) = zip_single[:3]
(txt_file, txt_file_prf_phot, txt_model, plot_epochs_) = zip_single[3:]
minimizer.set_parameters(values)
chi2 = minimizer.chi2_fun(values)
print(name, chi2)
minimizer.set_satellite_data(values)
if txt_file_prf_phot is not None:
(y, y_mask) = cpm_source.prf_photometry()
x = cpm_source.pixel_time[y_mask]
err = (cpm_source.all_pixels_flux_err *
MCPM_options['sat_sigma_scale'])
y_model = minimizer._sat_models[0][y_mask]
# XXX We should not use private property above.
out = [x, y[y_mask], err[y_mask], y[y_mask]-y_model]
np.savetxt(txt_file_prf_phot, np.array(out).T)
if txt_file is not None:
y_mask = cpm_source.residuals_mask
x = cpm_source.pixel_time[y_mask]
y = minimizer.event.datasets[-1].flux
y_err = cpm_source.all_pixels_flux_err[y_mask]
y_err *= MCPM_options['sat_sigma_scale']
y_model = minimizer._sat_models[0][y_mask]
# XXX We should not use private property above.
np.savetxt(txt_file, np.array([x, y, y_err, y-y_model]).T)
if txt_model is not None:
y_mask = cpm_source.residuals_mask
x = cpm_source.pixel_time[y_mask]
y_model = minimizer._sat_models[0][y_mask]
# XXX We should not use private property above.
np.savetxt(txt_model, np.array([x, y_model]).T)
if plot_file is not None:
minimizer.set_satellite_data(values)
campaigns = MCPM_options['campaigns']
if 'xlim' in plot_settings:
(t_beg, t_end) = plot_settings.pop('xlim')
elif 91 in campaigns and 92 in campaigns:
(t_beg, t_end) = (7500.3, 7573.5)
elif 91 in campaigns:
(t_beg, t_end) = (7500.3, 7528.0)
elif 92 in campaigns:
(t_beg, t_end) = (7530., 7573.5)
else:
(t_beg, t_end) = (7425., 7670.)
ylim = plot_settings.pop('ylim', None)
ylim_residuals = plot_settings.pop('ylim_residuals', None)
adjust = dict(left=0.09, right=0.995, bottom=0.08, top=0.995)
if len(plot_settings) == 0:
minimizer.very_standard_plot(t_beg, t_end, ylim, title=name)
else:
minimizer.standard_plot(t_beg, t_end, ylim, title=name,
**plot_settings)
if 'fluxes_y_axis' in plot_settings:
adjust['right'] = 0.895
if ylim_residuals is not None:
plt.ylim(*ylim_residuals)
if ylim_residuals[0] > 0.1 and -ylim_residuals[1] > 0.1:
fmt = ticker.FormatStrFormatter('%0.1f')
plt.gca().yaxis.set_major_formatter(fmt)
plt.xlabel("BJD-2450000")
plt.subplots_adjust(**adjust)
if len(plot_file) == 0:
plt.show()
else:
plt.savefig(plot_file, dpi=400)
print("{:} file saved".format(plot_file))
plt.close()
if plot_epochs_ is not None:
if minimizer.n_sat != 1:
raise ValueError('.n_sat != 1 is not implemented')
for epoch in plot_epochs_:
args = [epoch - 2450000.]
if plot_epochs_type is not None:
args.append(plot_epochs_type)
minimizer.cpm_sources[0].plot_image(*args)
plt.show()
if len(datasets) > 1:
print("Non-K2 datasets (i, chi2, F_s, F_b):")
zip_ = zip(datasets, minimizer.fit_blending)
for (i, (dat, fb)) in enumerate(zip_):
chi2_data = event.get_chi2_for_dataset(i, fit_blending=fb)
print(i, chi2_data, event.fit.flux_of_sources(dat)[0],
event.fit.blending_flux(dat))
if i < len(files):
chi2 -= chi2_data
print("-----")
print("K2 chi2: ", chi2)
if len(cpm_sources) > 0:
if isinstance(model, MM.Model):
print("Satellite t_0, u_0, A_max:")
else:
print("Satellite t_0, delta_flux_max:")
print(*minimizer.satellite_maximum())
print()