def read_light_curves_from_field(field,
ccd=None,
N_per_ccd=0,
clean=True,
randomize=False):
""" Read some number of light curves from a given PTF field
Parameters
----------
field : int, pdb.Field
The PTF field
ccd : int, pdb.CCD (optional)
The PTF CCD to grab light curves from. If None, uses all CCDs
N_per_ccd : int
The number of light curves to grab PER CCD. 0 means all.
clean : bool
Clean the light curves and require > min_number_good_observations.
randomize : bool
Grab random light curves vs. consecutive light curves.
"""
field = pdb.Field(field, "R")
if ccd != None:
ccds = [pdb.CCD(ccd, field, field.filter)]
else:
ccds = field.ccds.values()
all_light_curves = []
for ccd in ccds:
ccd_light_curves = []
chip = ccd.read()
sources = chip.sources.readWhere("(ngoodobs > {})".format(
pg.min_number_of_good_observations))
if randomize:
np.random.shuffle(sources)
for source in sources:
light_curve = ccd.light_curve(
source["matchedSourceID"], barebones=True,
clean=True) # clean applies a quality cut to the data
light_curve.db_indices = [
pu.source_index_name_to_pdb_index(source, index)
for index in indices
]
if light_curve == None or len(
light_curve) < pg.min_number_of_good_observations:
# If the light curve is not found, or has too few observations, skip this source_id
continue
ccd_light_curves.append(light_curve)
if N_per_ccd != 0 and len(ccd_light_curves) >= N_per_ccd: break
ccd.close()
all_light_curves += ccd_light_curves
return all_light_curves
def plot_best_light_curve(field_id, ccd_id):
field = pdb.Field(field_id, "R")
lc = get_best_light_curve(field, ccd_id)
fig = plt.figure(figsize=(15,5))
ax = fig.add_subplot(111)
lc.plot(ax)
#ax.yaxis.set_ticks([])
from pylab import ScalarFormatter
ax.yaxis.set_major_formatter(ScalarFormatter(False))
ax.set_xlabel("MJD", fontsize=20)
ax.set_ylabel("$R$ [mag]", fontsize=20)
fig.savefig("plots/new_detection_efficiency/example_light_curve_f{}_ccd{}.pdf".format(field.id, ccd_id), bbox_inches="tight")
args = parser.parse_args()
if args.verbose:
logger.setLevel(logging.DEBUG)
elif args.quiet:
logger.setLevel(logging.ERROR)
else:
logger.setLevel(logging.INFO)
if args.test:
test_magnitude_model()
test_ln_likelihood()
test_ln_prior()
sys.exit(0)
field = pdb.Field(args.field_id, "R")
ccd = field.ccds[args.ccd_id]
light_curve = ccd.light_curve(args.source_id,
clean=args.clean,
barebones=True)
if args.source_id == 5466:
idx = light_curve.mag > 18.25
light_curve.mjd = light_curve.mjd[idx]
light_curve.mag = light_curve.mag[idx]
light_curve.error = light_curve.error[idx]
sampler = fit_model_to_light_curve(light_curve,
nwalkers=args.walkers,
nsamples=args.steps,
nburn_in=args.burn_in)
end_chain = sampler.flatchain[-args.walkers * 100:]
logger.setLevel(logging.INFO)
#indices = ["eta","sigma_mu","j","k", "delta_chi_squared"]
indices = ["eta","delta_chi_squared", "j"]
config = dict()
config["number_of_fpr_light_curves"] = args.limit
config["number_of_fpr_simulations_per_light_curve"] = args.N
config["number_of_microlensing_light_curves"] = args.limit*10
config["number_of_microlensing_simulations_per_light_curve"] = args.N
np.random.seed(42)
if args.plot:
plot_best_light_curve(args.field_id, 2)
field = pdb.Field(int(args.field_id), filter="R")
simulated_microlensing_statistics, selected_distributions = detection_efficiency_for_field(field, \
config=config, \
overwrite=args.overwrite, \
indices=indices,
plot=args.plot)
sys.exit(0)
fig, axes = plt.subplots(1, 1, sharex=True, figsize=(15,10))
for ii,field_id in enumerate([4588, 100152, 3756]):
field = pdb.Field(field_id, "R")
lc = get_best_light_curve(field, 2)
axes.plot(lc.mjd, [ii]*len(lc.mjd), "ko", ms=6, alpha=0.15)
axes.text(55200, ii+0.1, str(field_id))
axes.set_ylim(-0.2, 2.2)
fig.savefig("plots/new_detection_efficiency/example_light_curves.png")
from __future__ import division
import sys
import time
import numpy as np
import ptf.db.photometric_database as pdb
field = pdb.Field(4588, "R")
ccd = field.ccds[0]
chip = ccd.read()
sourcedata = chip.sourcedata
def test_quality_cut(source_id):
a = time.time()
srcdata = pdb.quality_cut(sourcedata, source_id=source_id)
return time.time() - a
def test_where(source_id):
wheres = "(matchedSourceID == {})".format(source_id)
a = time.time()
data = [x.fetch_all_fields() for x in sourcedata.where(wheres)]
data = np.array(data, dtype=sourcedata.dtype)
data = data[(data["x_image"] > 15) & (data["x_image"] < 2033) & \
(data["y_image"] > 15) & (data["y_image"] < 4081) & \
(data["relPhotFlags"] < 4) & \
(data["mag"] > 14.3) & (data["mag"] < 21) & \
((data["sextractorFlags"] & 251) == 0) & \
((data["ipacFlags"] & 6077) == 0) & \
np.isfinite(data["mag"])]
len(field_ids)))
if args.field_range:
min_field_id, max_field_id = map(int, args.field_range.split("-"))
all_fields = np.load("data/survey_coverage/fields_observations_R.npy")
field_ids = all_fields[all_fields["num_exposures"] > \
min_number_of_good_observations]["field"]
field_ids = field_ids[(field_ids >= min_field_id) & \
(field_ids < max_field_id)]
for field_id in sorted(field_ids):
# Skip field 101001 because the data hasn't been reduced by the PTF pipeline?
if field_id == 101001: continue
field = pdb.Field(field_id, "R")
logger.info("Field: {}".format(field.id))
# There is some strangeness to getting the coordinates for a Field
# -- this just makes it stupidproof
try:
if field.ra == None or field.dec == None:
raise AttributeError()
except AttributeError:
logger.warn("Failed to get coordinates for this field!")
continue
# See if field is in database, remove it if we need to overwrite
if args.overwrite:
field_collection.remove({"_id" : field.id})
#light_curve_collection.remove({"field_id" : field.id})
def fields(self, min_num_observations):
""" Return a list of fields with more than the above number of observations """
rows = self._fields_exposures[self._fields_exposures["num_exposures"] >= min_num_observations]
fields = [pdb.Field(row["field"], self.filter, number_of_exposures=row["num_exposures"]) for row in rows]
return [f for f in fields if f.ra != None]
# If the light curve is not found, or has too few observations, skip this source_id
continue
ccd_light_curves.append(light_curve)
if N_per_ccd != 0 and len(ccd_light_curves) >= N_per_ccd: break
ccd.close()
all_light_curves += ccd_light_curves
return all_light_curves
# Configuration parameters
indices = ["eta"]
field = pdb.Field(3376, "R")
num_trials = 25
N_per_ccd = 1000
logfile = open(
os.path.join(
os.path.split(pg._base_path)[0], "tests",
"num_simulations_{0}.log".format(field.id)), "w")
# Start the test
all_light_curves = read_light_curves_from_field(field, N_per_ccd=N_per_ccd)
print("Read in {0} light curves from Field {1} on {2} CCDs".format(
len(all_light_curves), field.id, len(field.ccds)),
file=logfile)
all_eta_lower_criteria = dict()