def update_X(self, X, i, data, tinterp, len_t, objid, contextual_info,
otherinfo):
for j, pb in enumerate(self.passbands):
if pb not in data:
print("No", pb, "in objid:", objid)
continue
# Drop infinite values
data.replace([np.inf, -np.inf], np.nan)
# Get data
time = data[pb]['time'][0:self.nobs].dropna()
flux = data[pb]['flux'][0:self.nobs].dropna()
fluxerr = data[pb]['fluxErr'][0:self.nobs].dropna()
photflag = data[pb]['photflag'][0:self.nobs].dropna()
n = len(flux) # Get vector length (could be less than nobs)
if n > 1:
if flux.values[-1] > flux.values[
-2]: # If last values are increasing, then set fill_values to zero
f = interp1d(time,
flux,
kind='linear',
bounds_error=False,
fill_value=0.)
else:
f = interp1d(time,
flux,
kind='linear',
bounds_error=False,
fill_value='extrapolate'
) # extrapolate until all passbands finished.
fluxinterp = f(tinterp)
fluxinterp = np.nan_to_num(fluxinterp)
fluxinterp = fluxinterp.clip(min=0)
fluxerrinterp = np.zeros(len_t)
for interp_idx, fluxinterp_val in enumerate(fluxinterp):
if fluxinterp_val == 0.:
fluxerrinterp[interp_idx] = 0
else:
nearest_idx = helpers.find_nearest(
time, tinterp[interp_idx])
fluxerrinterp[interp_idx] = fluxerr[nearest_idx]
X[i][j][0:len_t] = fluxinterp
# X[i][j * 2 + 1][0:len_t] = fluxerrinterp
# Add contextual information
for jj, c_idx in enumerate(contextual_info, 1):
try:
X[i][j + jj][0:len_t] = otherinfo[c_idx] * np.ones(len_t)
except Exception as e:
X[i][j +
jj][0:len_t] = otherinfo[c_idx].values[0] * np.ones(len_t)
return X
def update_X(self, X, i, data, tinterp, len_t, objid, contextual_info,
meta_data):
for j, pb in enumerate(self.passbands):
# Drop infinite or nan values in any row
data.remove_rows(np.where(~np.isfinite(data['time']))[0])
data.remove_rows(np.where(~np.isfinite(data['flux']))[0])
data.remove_rows(np.where(~np.isfinite(data['fluxErr']))[0])
# Get data
pbmask = data['passband'] == pb
time = data[pbmask]['time'].data
flux = data[pbmask]['flux'].data
fluxerr = data[pbmask]['fluxErr'].data
photflag = data[pbmask]['photflag'].data
# Mask out times outside of mintime and maxtime
timemask = (time > self.mintime) & (time < self.maxtime)
time = time[timemask]
flux = flux[timemask]
fluxerr = fluxerr[timemask]
photflag = photflag[timemask]
n = len(flux) # Get vector length (could be less than nobs)
if n > 1:
# if flux[-1] > flux[-2]: # If last values are increasing, then set fill_values to zero
# f = interp1d(time, flux, kind='linear', bounds_error=False, fill_value=0.)
# else:
# f = interp1d(time, flux, kind='linear', bounds_error=False,
# fill_value='extrapolate') # extrapolate until all passbands finished.
f = interp1d(time,
flux,
kind='linear',
bounds_error=False,
fill_value=0.)
fluxinterp = f(tinterp)
fluxinterp = np.nan_to_num(fluxinterp)
fluxinterp = fluxinterp.clip(min=0)
fluxerrinterp = np.zeros(len_t)
for interp_idx, fluxinterp_val in enumerate(fluxinterp):
if fluxinterp_val == 0.:
fluxerrinterp[interp_idx] = 0
else:
nearest_idx = helpers.find_nearest(
time, tinterp[interp_idx])
fluxerrinterp[interp_idx] = fluxerr[nearest_idx]
X[i][j][0:len_t] = fluxinterp
# X[i][j * 2 + 1][0:len_t] = fluxerrinterp
# Add contextual information
for jj, c_info in enumerate(contextual_info, 1):
X[i][j + jj][0:len_t] = meta_data[c_info] * np.ones(len_t)
return X