# generate training data, add to db survey='normal' for i in range(1000): aSurvey.generateCurve() tfe = np.column_stack((aSurvey.times[:,np.newaxis],aSurvey.fluxes[:,np.newaxis],aSurvey.errors[:,np.newaxis])) points_per_curve = len(aSurvey.times) source_class = aSurvey.class_name period = aSurvey.period_this curve_info = [points_per_curve,source_class,0,0,0,0,None,survey,0,period] curve_info_names = ["number_points","classification","c1","e1","c2","e2","raw_xml","survey","xml_filename","true_period"] print source_class print curve_info create_database.enter_record(curve_info,curve_info_names,tfe,cursor) ## create 'outlier' sources, put there class as "outlier" survey = 'outlier' ## create supernova remnants reload(synthetic_data) aCadence.generate_cadence() aSupernovaRemnant = synthetic_data.SupernovaRemnant() for i in range(4): aSupernovaRemnant.generateCurve() tfe = aSupernovaRemnant.curve_this(aCadence.cadence_this,aCadence.error_this) points_per_curve = tfe.size/3 source_class = "SN_remnant" period = 1
survey = "full"
points_per_curve = len(aCadence.error_this)
period = aRRLyraeFund.period_this
curve_info = [
points_per_curve, source_class, 0, 0, 0, 0, None, survey, 0, period
]
curve_info_names = [
"number_points", "classification", "c1", "e1", "c2", "e2", "raw_xml",
"survey", "xml_filename", "true_period"
]
## generate light curve using full cadence
errors = aCadence.error_this
aCadence.error_this = np.zeros(len(errors)) + .005
tfe = ComputeTfe(aRRLyraeFund, aCadence)
create_database.enter_record(curve_info, curve_info_names, tfe, cursor)
## return aCadence.error_this to original state
aCadence.error_this = errors
## construct 50 tfe (randomness is phase and error)
## for lightcurve trucated at 10,20, . . ., 80 measurements
trunc_points = 90 - np.arange(10, 90, 10)
for trunc in trunc_points:
aCadence.cadence_this = aCadence.cadence_this[0:trunc]
aCadence.error_this = aCadence.error_this[0:trunc]
survey = repr(trunc)
points_per_curve = len(aCadence.error_this)
curve_info = [
points_per_curve, source_class, 0, 0, 0, 0, None, survey, 0, period
]
for i in range(50):
for i in db_info:
tfe = create_database.get_measurements(i[0],cursor)
period = 1 / i[2]
smo = smoothers.supersmooth(tfe,period,normalize_times=False)
tfe_smoothed = np.concatenate((tfe[:,0].reshape((tfe[:,0].size,1)),
smo.reshape(smo.size,1),
tfe[:,2].reshape((tfe[:,2].size,1))),
axis=1)
residuals = np.mean(tfe[:,1]) + tfe[:,1] - smo
tfe_residual = np.concatenate((tfe[:,0].reshape((tfe[:,0].size,1)),
residuals.reshape(residuals.size,1),
tfe[:,2].reshape((tfe[:,2].size,1))),
axis=1)
curve_info = ['smoothed',i[1],i[0],tfe.shape[0],i[4]]
create_database.enter_record(curve_info,
curve_info_names,
tfe_smoothed,
cursor=cursor)
curve_info[0] = 'residual'
create_database.enter_record(curve_info,
curve_info_names,
tfe_residual,
cursor=cursor)
connection.commit()
## SANITY CHECK
## examine what we have collected
sql_cmd = """SELECT source_id,original_source_id,survey,number_points,classification FROM sources"""