def test_mlfit_DSPL_LM_without_guess():
current_event = _create_event()
model = _create_model('DSPL')
model.model_dictionnary = {
'to1': 0,
'uo1': 1,
'delta_to': 2,
'uo2': 3,
'tE': 4,
'q_F_I': 5,
'fs_Test': 6,
'g_Test': 7
}
fancy_namedtuple = collections.namedtuple('Parameters',
model.model_dictionnary.keys())
model.pyLIMA_standard_parameters_to_fancy_parameters.return_value = fancy_namedtuple(
10.0, 0.1, 20, 0.05, 20, 0.1, 10, 5)
model.parameters_boundaries = [[0, 100], [0, 1], [0, 100], [0, 1],
[0, 300], [0, 1]]
fit = microlfits.MLFits(current_event)
fit.mlfit(model, 'LM')
assert fit.fit_covariance.shape == (6 + 2 * len(current_event.telescopes),
6 + 2 * len(current_event.telescopes))
assert len(fit.fit_results) == 6 + 2 * len(current_event.telescopes) + 1
def test_LM_Jacobian():
current_event = _create_event()
model = _create_model('FSPL')
model.model_dictionnary = {
'to': 0,
'uo': 1,
'tE': 2,
'rho': 3,
'fs_Test': 4,
'g_Test': 5
}
fit = microlfits.MLFits(current_event)
fit.model = model
to = 0.0
uo = 0.1
tE = 1.0
rho = 0.26
fs = 10
g = 1.0
parameters = [to, uo, tE, rho, fs, g]
Jacobian = fit.LM_Jacobian(parameters)
assert Jacobian.shape == (6,
len(current_event.telescopes[0].lightcurve_flux))
def test_chichi_telescopes():
current_event = _create_event()
model = _create_model('FSPL')
model.model_dictionnary = {
'to': 0,
'uo': 1,
'tE': 2,
'rho': 3,
'fs_Test': 4,
'g_Test': 5
}
fit = microlfits.MLFits(current_event)
fit.model = model
to = 0.0
uo = 0.1
tE = 1.0
rho = 0.26
fs = 10
g = 1.0
parameters = [to, uo, tE, rho, fs, g]
chichi_telescopes = fit.chichi_telescopes(parameters)
chichi = sum(fit.residuals_LM(parameters)**2)
assert len(chichi_telescopes) == 1
assert np.allclose(chichi, chichi_telescopes[0])
def test_mlfit_FSPL_LM_with_guess():
current_event = _create_event()
model = _create_model('FSPL')
model.parameters_guess = [10, 0.1, 20, 0.02]
model.model_dictionnary = {
'to': 0,
'uo': 1,
'tE': 2,
'rho': 3,
'fs_Test': 4,
'g_Test': 5
}
fancy_namedtuple = collections.namedtuple('Parameters',
model.model_dictionnary.keys())
model.pyLIMA_standard_parameters_to_fancy_parameters.return_value = fancy_namedtuple(
10.0, 0.1, 20, 0.05, 10, 5)
model.model_parameters = collections.namedtuple('parameters',
model.model_dictionnary)
model.parameters_boundaries = [[0, 100], [0, 1], [0, 300], [0, 1]]
fit = microlfits.MLFits(current_event)
fit.mlfit(model, 'LM')
print fit.fit_results
assert fit.fit_covariance.shape == (4 + 2 * len(current_event.telescopes),
4 + 2 * len(current_event.telescopes))
assert len(fit.fit_results) == 4 + 2 * len(current_event.telescopes) + 1
def test_check_fit_bad_rho():
current_event = _create_event()
model = _create_model('FSPL')
model.model_dictionnary = {
'to': 0,
'uo': 1,
'tE': 2,
'rho': 3,
'fs_Test': 4,
'g_Test': 5
}
fit = microlfits.MLFits(current_event)
fit.model = model
fit.fit_results = [0.0, 0.0, 0.0, -1.0, 1.0, 0.0]
flag = fit.check_fit()
assert flag == 'Bad Fit'
fit.fit_results = [0.0, 0.0, 0.0, 0.8, 1.0, 0.0]
flag = fit.check_fit()
assert flag == 'Bad Fit'
fit.fit_results = [0.0, 0.0, 0.0, 0.05, 1.0, 0.0]
flag = fit.check_fit()
assert flag == 'Good Fit'
def test_mlfit_PSPL_MCMC_with_guess():
current_event = _create_event()
model = _create_model('PSPL')
model.parameters_guess = [10, 0.1, 20]
fit = microlfits.MLFits(current_event)
fit.mlfit(model, 'MCMC')
assert fit.MCMC_chains.shape == (200, 100, 5)
def test_mlfit_PSPL_LM_without_guess():
current_event = _create_event()
model = _create_model('PSPL')
fit = microlfits.MLFits(current_event)
fit.mlfit(model, 'LM')
assert fit.fit_covariance.shape == (3 + 2 * len(current_event.telescopes), 3 + 2 * len(current_event.telescopes))
assert len(fit.fit_results) == 3 + 2 * len(current_event.telescopes) + 1
def fit(self,
model,
method,
DE_population_size=10,
flux_estimation_MCMC='MCMC',
fix_parameters_dictionnary=None,
grid_resolution=10,
computational_pool=None,
binary_regime=None):
"""Function to fit the event with a model and a method.
:param model: the model you want to fit. More details in the microlfits module
:param method: the fitting method you want to use. Has to be a string in the
available_methods parameter:
'LM' : Levenberg-Marquardt algorithm
'DE' : Differential Evolution algorithm
'MCMC' : Monte-Carlo Markov Chain algorithm
More details in the microlfits module
A microlfits object is added in the event.fits list. For example, if you request two fits,
you will obtain :
event.fits=[fit1,fit2]
More details in the microlfits module.
"""
available_kind = ['Microlensing']
available_methods = ['LM', 'DE', 'MCMC', 'GRIDS', 'TRF']
if self.kind not in available_kind:
print(
'ERROR : No possible fit yet for a non microlensing event, sorry :('
)
raise EventException('Can not fit this event kind')
if method not in available_methods:
print('ERROR : Wrong method request, has to be a string selected between ' + \
' or '.join(available_methods) + '')
raise EventException('Wrong fit method request')
fit = microlfits.MLFits(self)
fit.mlfit(model,
method,
DE_population_size=DE_population_size,
flux_estimation_MCMC=flux_estimation_MCMC,
fix_parameters_dictionnary=fix_parameters_dictionnary,
grid_resolution=grid_resolution,
computational_pool=computational_pool,
binary_regime=binary_regime)
self.fits.append(fit)
def test_check_fit_bad_covariance():
current_event = _create_event()
model = _create_model('PSPL')
fit = microlfits.MLFits(current_event)
fit.fit_covariance = np.array([[-1.0, 0.0], [0.0, 0.0]])
flag = fit.check_fit()
assert flag == 'Bad Fit'
def test_mlfit_FSPL_LM_with_guess():
current_event = _create_event()
model = microlmodels.create_model('FSPL',current_event)
model.parameters_boundaries = [[0, 100], [0, 1], [0, 300], [0, 1]]
fit = microlfits.MLFits(current_event)
fit.mlfit(model, 'LM')
assert fit.fit_covariance.shape == (4 + 2 * len(current_event.telescopes), 4 + 2 * len(current_event.telescopes))
assert len(fit.fit_results) == 4 + 2 * len(current_event.telescopes) + 1
def test_mlfit_PSPL_LM_with_guess():
current_event = _create_event()
model = microlmodels.create_model('PSPL',current_event)
model.parameters_guess = [10, 0.1, 20]
fit = microlfits.MLFits(current_event)
fit.mlfit(model, 'LM')
assert fit.fit_covariance.shape == (3 + 2 * len(current_event.telescopes), 3 + 2 * len(current_event.telescopes))
assert len(fit.fit_results) == 3 + 2 * len(current_event.telescopes) + 1
def test_check_fit_source_flux():
current_event = _create_event()
model = microlmodels.create_model('FSPL', current_event)
model.define_model_parameters()
fit = microlfits.MLFits(current_event)
fit.model = model
fit.fit_results = [0.0, 0.0, 0.0, 0.0, 1.0, 0.0]
flag = fit.check_fit()
assert flag == 'Good Fit'
fit.fit_results = [0.0, 0.0, 0.0, 0.8, -1.0, 0.0]
flag = fit.check_fit()
assert flag == 'Bad Fit'
def test_LM_Jacobian():
current_event = _create_event()
model = microlmodels.create_model('FSPL',current_event)
model.define_model_parameters()
fit = microlfits.MLFits(current_event)
fit.model = model
to = 0.0
uo = 0.1
tE = 1.0
rho = 0.26
fs = 10
g = 1.0
parameters = [to, uo, tE, rho, fs, g]
Jacobian = fit.LM_Jacobian(parameters)
Jacobian = Jacobian.T
assert Jacobian.shape == (6, len(current_event.telescopes[0].lightcurve_flux))
def test_chichi_telescopes():
current_event = _create_event()
model = microlmodels.create_model('FSPL', current_event)
model.define_model_parameters()
fit = microlfits.MLFits(current_event)
fit.model = model
to = 0.0
uo = 0.1
tE = 1.0
rho = 0.26
fs = 10
g = 1.0
parameters = [to, uo, tE, rho, fs, g]
chichi_telescopes = fit.chichi_telescopes(parameters)
chichi = sum(fit.residuals_LM(parameters)**2)
assert len(chichi_telescopes) == 1
assert np.allclose(chichi, chichi_telescopes[0])
def create_model(e, params, diagnostics=False):
f = microlfits.MLFits(e)
parallax_params = ['None', params['to']]
orbital_params = ['None', params['to']]
model_params = ['to', 'uo', 'tE']
if params['rho'] != None:
model_params.append('rho')
if params['logs'] != None and params['logq'] != None and params[
'alpha'] != None:
model_params.append('logs')
model_params.append('logq')
model_params.append('alpha')
if params['piEN'] != None and params['piEE'] != None:
parallax_params = ['Full', params['to']]
model_params.append('piEN')
model_params.append('piEE')
if params['sdsdt'] != None and params['dalphadt'] != None and params[
'sdszdt'] != None:
orbital_params = ['2D', params['to']]
model_params.append('sdsdt')
model_params.append('dalphadt')
model_params.append('sdszdt')
params['model_params'] = model_params
model = microlmodels.create_model(params['model_type'],
e,
parallax=parallax_params,
orbital_motion=orbital_params,
blend_flux_ratio=False)
if 'binary origin' in params.keys():
model.binary_origin = params['binary_origin']
model.define_model_parameters()
f.model = model
results = []
for key in model_params:
results.append(params[key])
params['fitted_model_params'] = results
results = results + params['phot_params']
results.append(params['chisq'])
f.fit_results = results
e.fits.append(f)
if diagnostics:
fig = microloutputs.LM_plot_lightcurves(f)
plt.show()
plt.close()
return e, params
def simulate_ffp():
spring = True
fall = False
u0 = 0.01
tE = 1.0
if spring:
t0 = 2460394.400000
start_jd = 2460389.500000 # 2024 March 20
end_jd = 2460399.500000
if fall:
t0 = 2460389.500000
start_jd = 2460573.500000 # 2024 Aug 20
end_jd = 2460583.500000 # 2024 Aug 30
ra = 270.0
dec = -27.0
piEN = 0.01
piEE = 0.01
blend_ratio = 0.2
source_flux = mag_to_flux(22.0,1.0,24.0)
blend_flux = source_flux * blend_ratio
model_params = [ t0, u0, tE, piEN, piEE ]
lsst_aperture = 6.68
lsst_read_noise = 10.0
wfirst_aperture = 2.4
wfirst_read_noise = 10.0
if spring:
horizons_file = 'wfirst_observer_table_spring.txt'
if fall:
horizons_file = 'wfirst_observer_table_fall.txt'
ffp_lsst = event.Event()
ffp_lsst.name = 'FFP'
ffp_lsst.ra = ra
ffp_lsst.dec = dec
lsst_lc = simulate_lightcurve(start_jd, end_jd, 0.5/24.0, lsst_aperture,
spring, fall, day_gaps=True )
lsst = telescopes.Telescope(name='LSST', camera_filter='i',
location='Earth',
light_curve_magnitude=lsst_lc)
ffp_lsst.telescopes.append(lsst)
model_lsst = microlmodels.create_model('PSPL', ffp_lsst,
parallax=['Full', t0])
model_lsst.define_model_parameters()
fit_lsst = microlfits.MLFits(ffp_lsst)
fit_lsst.model = model_lsst
fit_lsst.fit_results = model_params
ffp_lsst.fits.append(fit_lsst)
print('Generated event lightcurve from LSST')
print('Model parameters:')
print_fit_params(model_params)
ffp_wfirst = event.Event()
ffp_wfirst.name = 'FFP'
ffp_wfirst.ra = ra
ffp_wfirst.dec = dec
horizons_table = jplhorizons_utils.parse_JPL_Horizons_table(horizons_file_path=horizons_file,
table_type='OBSERVER')
spacecraft_positions = jplhorizons_utils.extract_spacecraft_positions(horizons_table,t0)
wfirst_lc = simulate_lightcurve(start_jd, end_jd, 0.25/24.0, wfirst_aperture, spring, fall)
wfirst = telescopes.Telescope(name='WFIRST', camera_filter='W149',
spacecraft_name = 'WFIRST',
location='Space',
light_curve_magnitude=wfirst_lc)
wfirst.spacecraft_positions = spacecraft_positions
ffp_wfirst.telescopes.append(wfirst)
model_wfirst = microlmodels.create_model('PSPL', ffp_wfirst,
parallax=['Full', t0])
model_wfirst.define_model_parameters()
fit_wfirst = microlfits.MLFits(ffp_wfirst)
fit_wfirst.model = model_wfirst
fit_wfirst.fit_results = model_params
ffp_wfirst.fits.append(fit_wfirst)
print('Generated event lightcurve from WFIRST')
print('Model parameters:')
print_fit_params(fit_wfirst)
parameters = [ t0, u0, tE ]
lsst_pylima_params = extract_event_parameters(ffp_lsst, fit_lsst, parameters, source_flux, blend_ratio)
wfirst_pylima_params = extract_event_parameters(ffp_wfirst, fit_wfirst, parameters, source_flux, blend_ratio)
lsst_lc = add_lensing_event_to_lightcurve(lsst_pylima_params, ffp_lsst,
fit_lsst, lsst_read_noise)
wfirst_lc = add_lensing_event_to_lightcurve(wfirst_pylima_params, ffp_wfirst,
fit_wfirst, wfirst_read_noise)
plot_fitted_lightcurves(lsst_lc, wfirst_lc, ffp_lsst, ffp_wfirst,
'LSST', 'WFIRST', 'ffp_sim_lightcurve.png',
t0=t0,tE=tE)
