def correct(self, cbvs=[1, 2], method='powell', options={}):
"""
Correct the SAP_FLUX by fitting a number of cotrending basis vectors
`cbvs`.
Parameters
----------
cbvs : list of ints
The list of cotrending basis vectors to fit to the data. For example,
[1, 2] will fit the first two basis vectors.
method : str
Numerical optimization method. See scipy.optimize.minimize for the
full list of methods.
options : dict
Dictionary of options to be passed to scipy.optimize.minimize.
"""
module, output = channel_to_module_output(self.lc_file.channel)
cbv_file = pyfits.open(self.get_cbv_url())
cbv_data = cbv_file['MODOUT_{0}_{1}'.format(module, output)].data
cbv_array = []
for i in cbvs:
cbv_array.append(
cbv_data.field(
'VECTOR_{}'.format(i))[self.lc_file.quality_mask])
cbv_array = np.asarray(cbv_array)
sap_lc = self.lc_file.SAP_FLUX
median_sap_flux = np.nanmedian(sap_lc.flux)
norm_sap_flux = sap_lc.flux / median_sap_flux - 1
norm_err_sap_flux = sap_lc.flux_err / median_sap_flux
def mean_model(*theta):
coeffs = np.asarray(theta)
return np.dot(coeffs, cbv_array)
prior = self.prior(mean=np.zeros(len(cbvs)), var=16.)
likelihood = self.likelihood(data=norm_sap_flux,
mean=mean_model,
var=norm_err_sap_flux)
x0 = likelihood.fit(x0=prior.mean, method=method, options=options).x
posterior = oktopus.Posterior(likelihood=likelihood, prior=prior)
self._opt_result = posterior.fit(x0=x0, method=method, options=options)
self._coeffs = self._opt_result.x
flux_hat = sap_lc.flux - median_sap_flux * mean_model(self._coeffs)
return LightCurve(time=sap_lc.time, flux=flux_hat.reshape(-1))
def correct(self, cbvs=(1, 2), method='powell', options=None):
"""
Correct the SAP_FLUX by fitting a number of cotrending basis vectors
`cbvs`.
Parameters
----------
cbvs : list of ints
The list of cotrending basis vectors to fit to the data. For example,
[1, 2] will fit the first two basis vectors.
method : str
Numerical optimization method. See scipy.optimize.minimize for the
full list of methods.
options : dict
Dictionary of options to be passed to scipy.optimize.minimize.
"""
if options is None:
options = {}
median_flux = np.nanmedian(self.lc.flux)
norm_flux = self.lc.flux / median_flux - 1
norm_err_flux = self.lc.flux_err / median_flux
# Trim down to the right number of cbvs
clip = np.in1d(np.arange(1, len(self.cbv_array) + 1), np.asarray(cbvs))
time_clip = np.in1d(self.cbv_cadenceno, self.lc.cadenceno)
def mean_model(*theta):
coeffs = np.asarray(theta)
return np.dot(coeffs, self.cbv_array[clip, :][:, time_clip])
prior = self.prior(mean=np.zeros(len(cbvs)), var=16.)
likelihood = self.likelihood(data=norm_flux,
mean=mean_model,
var=norm_err_flux)
x0 = likelihood.fit(x0=prior.mean, method=method, options=options).x
posterior = oktopus.Posterior(likelihood=likelihood, prior=prior)
self._opt_result = posterior.fit(x0=x0, method=method, options=options)
self._coeffs = self._opt_result.x
flux_hat = self.lc.flux - median_flux * mean_model(self._coeffs)
clc = self.lc.copy()
clc.flux = flux_hat.reshape(-1)
return clc
def correct(self, cbvs=[1, 2], method='powell', options={}):
"""
Correct the SAP_FLUX by fitting a number of cotrending basis vectors
`cbvs`.
Parameters
----------
cbvs : list of ints
The list of cotrending basis vectors to fit to the data. For example,
[1, 2] will fit the first two basis vectors.
method : str
Numerical optimization method. See scipy.optimize.minimize for the
full list of methods.
options : dict
Dictionary of options to be passed to scipy.optimize.minimize.
"""
cbv_array, _ = self._get_cbv_data(cbvs)
sap_lc = self.lc_file.SAP_FLUX
median_sap_flux = np.nanmedian(sap_lc.flux)
norm_sap_flux = sap_lc.flux / median_sap_flux - 1
norm_err_sap_flux = sap_lc.flux_err / median_sap_flux
def mean_model(*theta):
coeffs = np.asarray(theta)
return np.dot(coeffs, cbv_array)
prior = self.prior(mean=np.zeros(len(cbvs)), var=16.)
likelihood = self.likelihood(data=norm_sap_flux,
mean=mean_model,
var=norm_err_sap_flux)
x0 = likelihood.fit(x0=prior.mean, method=method, options=options).x
posterior = oktopus.Posterior(likelihood=likelihood, prior=prior)
self._opt_result = posterior.fit(x0=x0, method=method, options=options)
self._coeffs = self._opt_result.x
flux_hat = sap_lc.flux - median_sap_flux * mean_model(self._coeffs)
return LightCurve(time=sap_lc.time,
flux=flux_hat.reshape(-1),
flux_err=sap_lc.flux_err)