def fit_line(df, channel): cutoff = get_cutoff(channel) idx = (df['sub_flux'] > 0) & (df['sub_flux'] < cutoff) X = df['hdr_flux'][idx] y = df['sub_flux'][idx] # model = sm.OLS(y,X) # result = model.fit() # slope = result.params[0] slope = ordinary_least_squares(y, X) return slope
def fit_line(df, channel):
cutoff = get_cutoff(channel)
idx = (df['sub_flux'] > 0) & (df['sub_flux'] < cutoff)
X = df['hdr_flux'][idx]
y = df['sub_flux'][idx]
# model = sm.OLS(y,X)
# result = model.fit()
# slope = result.params[0]
slope = ordinary_least_squares(y, X)
return slope
def combine_hdr_catalogs(catalog_filepaths_tuple):
"""
Takes a tuple containing the filepaths to the short and long exposure
single-channel catalogs for a given region and channel. The result is
a single catalog containing the union of all sources in both short and
long exposure catalogs, with the short exposure measurements being used
for the brighter sources, and the long exposure measurements used for
the fainter sources. The cutoff between the two is determined by the
parameter 'hdr_cutoff' in the metadata file and should be set to the
saturation limit for the long exposure data (there are actually 2
parameters, one for each channel: hdr_cutoff_ch1, hdr_cutoff_ch2).
"""
# read in the data
long_file, short_file = catalog_filepaths_tuple
work_dir = "/".join(short_file.split("/")[:-1])
meta = json.load(open(work_dir + "/metadata.json"))
header = "id ra dec flux unc n_obs"
names = header.split()
long_cat = np.recfromtxt(long_file, names=names)
short_cat = np.recfromtxt(short_file, names=names)
# fit a line to short ~ long
idx_s = short_cat.flux < meta["short_cutoff"]
idx_l = long_cat.flux < meta["long_cutoff"]
short_flux = short_cat.flux[idx_s]
long_flux = long_cat.flux[idx_l]
short_ra, short_dec = short_cat.ra[idx_s], short_cat.dec[idx_s]
long_ra, long_dec = long_cat.ra[idx_l], long_cat.dec[idx_l]
idx1, idx2, ds = spherematch(short_ra, short_dec, long_ra, long_dec, tolerance=1 / 3600.0)
y = short_flux[idx1]
X = long_flux[idx2]
slope = ordinary_least_squares(y, X)
# divide short flux/unc by the slope so that it agrees with the long flux
print("region {} correction value: {}".format(meta["name"], slope))
short_cat.flux /= slope
short_cat.unc /= slope
# get everything brighter than the cutoff in short and combine with long
idx_faint = long_cat.flux < meta["long_cutoff"]
idx_bright = short_cat.flux > meta["long_cutoff"]
# before concatenation of long and short subsets, check for any duplicates
# (if they exist they should tend to have flux very close to the cutoff)
ls, ss = long_cat[idx_faint], short_cat[idx_bright]
idx_s, idx_l, ds = spherematch(ss.ra, ss.dec, ls.ra, ls.dec, tolerance=1 / 3600.0)
dup_ids = []
for idx in idx_l:
if (ls.flux[idx] > 0.9 * meta["long_cutoff"]) & (ls.flux[idx] < meta["long_cutoff"]):
dup_ids.append(ls.id[idx])
# now use the ids of the duplicates to delete them from the long dataset
for idx in dup_ids:
ls = ls[ls.id != idx]
data = np.concatenate([ls, ss])
# eliminate sources with negative flux
good = data["flux"] >= 0
data = data[good]
# apply global sigma clip using the value from setup.yaml
snr = data["flux"] / data["unc"]
good = snr >= meta["sigma_clip"]
data = data[good]
# write to disk
header = "id ra dec flux unc n_obs"
data = data[header.split()]
idx = np.argsort(data["ra"])
data = data[idx]
data["id"] = np.arange(1, data.shape[0] + 1)
fmt = ["%i"] + ["%0.8f"] * 2 + ["%.4e"] * 2 + ["%i"]
out_name = "_".join([meta["name"], meta["channel"], "combined_hdr_catalog.txt"])
out_path = "/".join(["/".join(work_dir.split("/")[:-1]), out_name])
np.savetxt(out_path, data, fmt=fmt, header=header)
print("created file: " + out_path)
