def sigma_clip(time, flux, unc_flux, clip_at=6):
"""Perform sigma-clipping on the lightcurve.
Inputs
------
time, flux, unc_flux: array_like
clip_at: float (optional)
how many sigma to clip at. Defaults to 6.
Outputs
-------
clipped_time, clipped_flux, clipped_unc: arrays
to_keep: boolean mask of locations that were kept
"""
# Compute statistics on the lightcurve
med, stdev = utils.stats(flux, unc_flux)
# Sigma-clip the lightcurve
outliers = abs(flux-med)>(stdev*clip_at)
to_clip = np.where(outliers==True)[0]
to_keep = np.where(outliers==False)[0]
logging.debug("Sigma-clipping")
logging.debug(to_clip)
clipped_time = np.delete(time, to_clip)
clipped_flux = np.delete(flux, to_clip)
clipped_unc = np.delete(unc_flux, to_clip)
# Return clipped lightcurve
return clipped_time, clipped_flux, clipped_unc, to_keep
def prep_lc(time, flux, unc_flux, clip_at=3):
"""Trim, sigma-clip, and calculate stats on a lc.
Inputs
------
time, flux, unc_flux: array_like
clip_at: float (optional)
How many sigma to clip at. Defaults to 6.
Set to None for no sigma clipping
Outputs
-------
clean_time, clean_flux, clean_unc: arrays
"""
# Trim the lightcurve, remove bad values
t_time, t_flux, t_unc, t_kept = trim(time, flux, unc_flux)
# Run sigma-clipping if desired, repeat 2X
if clip_at is not None:
c_time, c_flux, c_unc, c_kept = smooth_and_clip(t_time, t_flux, t_unc,
clip_at=clip_at)
else:
c_time, c_flux, c_unc, c_kept = t_time, t_flux, t_unc, t_kept
all_kept = t_kept[c_kept]
# Calculate statistics on lightcurve
c_med, c_stdev = utils.stats(c_flux, c_unc)
# Return cleaned lightcurve and statistics
return c_time, c_flux, c_unc, c_med, c_stdev, all_kept
def fit_sine(time, flux, unc, period):
"""Fit a simple sine model fixed to the best-fit period."""
def _sine_model(t, amp, yoffset, tshift):
return amp * np.sin(2 * np.pi * t / period + tshift) + yoffset
# Phase by the period, then extend the arrays
# To fit two cycles instead of one
phased_time = utils.phase(time, period)
fit_time = np.append(phased_time, phased_time + period)
fit_flux = np.append(flux, flux)
fit_unc = np.append(unc, unc)
# initial amplitude and yoffset are stdev and median, respectively
p0 = np.append(utils.stats(flux, unc), 0.0)
popt, pcov = opt.curve_fit(_sine_model,
fit_time,
fit_flux,
sigma=fit_unc,
p0=p0)
perr = np.sqrt(np.diag(pcov))
logging.debug("amplitude, yoffset, tshift")
logging.debug(popt)
logging.debug(perr)
return phased_time, _sine_model(phased_time, *popt)
def sigma_clip(time, flux, unc_flux, clip_at=6):
"""Perform sigma-clipping on the lightcurve.
Inputs
------
time, flux, unc_flux: array_like
clip_at: float (optional)
how many sigma to clip at. Defaults to 6.
Outputs
-------
clipped_time, clipped_flux, clipped_unc: arrays
to_keep: boolean mask of locations that were kept
"""
# Compute statistics on the lightcurve
med, stdev = utils.stats(flux, unc_flux)
# Sigma-clip the lightcurve
outliers = abs(flux - med) > (stdev * clip_at)
to_clip = np.where(outliers == True)[0]
to_keep = np.where(outliers == False)[0]
logging.debug("Sigma-clipping")
logging.debug(to_clip)
clipped_time = np.delete(time, to_clip)
clipped_flux = np.delete(flux, to_clip)
clipped_unc = np.delete(unc_flux, to_clip)
# Return clipped lightcurve
return clipped_time, clipped_flux, clipped_unc, to_keep
def prep_lc(time, flux, unc_flux, clip_at=3):
"""Trim, sigma-clip, and calculate stats on a lc.
Inputs
------
time, flux, unc_flux: array_like
clip_at: float (optional)
How many sigma to clip at. Defaults to 6.
Set to None for no sigma clipping
Outputs
-------
clean_time, clean_flux, clean_unc: arrays
"""
# Trim the lightcurve, remove bad values
t_time, t_flux, t_unc, t_kept = trim(time, flux, unc_flux)
# Run sigma-clipping if desired, repeat 2X
if clip_at is not None:
c_time, c_flux, c_unc, c_kept = smooth_and_clip(t_time,
t_flux,
t_unc,
clip_at=clip_at)
else:
c_time, c_flux, c_unc, c_kept = t_time, t_flux, t_unc, t_kept
all_kept = t_kept[c_kept]
# Calculate statistics on lightcurve
c_med, c_stdev = utils.stats(c_flux, c_unc)
# Return cleaned lightcurve and statistics
return c_time, c_flux, c_unc, c_med, c_stdev, all_kept
def fit_sine(time, flux, unc, period):
"""Fit a simple sine model fixed to the best-fit period."""
def _sine_model(t, amp, yoffset, tshift):
return amp * np.sin(2 * np.pi * t / period + tshift) + yoffset
# Phase by the period, then extend the arrays
# To fit two cycles instead of one
phased_time = utils.phase(time, period)
fit_time = np.append(phased_time, phased_time + period)
fit_flux = np.append(flux, flux)
fit_unc = np.append(unc, unc)
# initial amplitude and yoffset are stdev and median, respectively
p0 = np.append(utils.stats(flux, unc), 0.0)
popt, pcov = opt.curve_fit(_sine_model, fit_time, fit_flux, sigma=fit_unc, p0=p0)
perr = np.sqrt(np.diag(pcov))
logging.debug("amplitude, yoffset, tshift")
logging.debug(popt)
logging.debug(perr)
return phased_time, _sine_model(phased_time, *popt)
