def agn_b_scores(name, username, password, colors=False):
k = Kowalski(username=username, password=password, verbose=False)
q = {
"query_type": "find",
"query": {
"catalog": 'ZTF_alerts',
"filter": {
"objectId": name
},
"projection": {
"_id": 0,
"cutoutScience": 0,
"cutoutTemplate": 0,
"cutoutDifference": 0
},
}
}
r = k.query(query=q)
alerts = r['result_data']['query_result']
ra, dec = alerts[0]['candidate']['ra'], alerts[0]['candidate']['dec']
cc = SkyCoord(ra, dec, unit=(u.deg, u.deg))
table = Irsa.query_region(coordinates=cc,
catalog="allwise_p3as_psd",
spatial="Cone",
radius=2 * u.arcsec)
# AGN WISE
if len(table['w1mpro']) == 0:
agn = False
temp_points = 6
else:
w1, w1_err, w2, w2_err, w3, w3_err = table['w1mpro'], table[
'w1sigmpro'], table['w2mpro'], table['w2sigmpro'], table[
'w3mpro'], table['w3sigmpro']
if w1 - w2 > 0.8 + 0.1 and w2_err < 0.5 and w1_err < 0.5:
agn = True
temp_points = -2
elif w2 - w3 > 2.5 + 0.1 and w2_err < 0.5 and w3_err < 0.5:
agn = True
temp_points = -2
elif w1 - w2 > 0.8 and w2_err < 0.5 and w1_err < 0.5:
agn = True
temp_points = 0
elif w2 - w3 > 2.5 and w2_err < 0.5 and w3_err < 0.5:
agn = True
temp_points = 0
elif w1 - w2 > 0.8 - 0.2 and w2_err < 0.5 and w1_err < 0.5:
agn = False
temp_points = 2
elif w2 - w3 > 2.5 - 0.3 and w2_err < 0.5 and w3_err < 0.5:
agn = False
temp_points = 2
elif w1 - w2 > 0.8 - 0.5 and w2_err < 0.5 and w1_err < 0.5:
agn = False
temp_points = 4
elif w2 - w3 > 2.5 - 0.5 and w2_err < 0.5 and w3_err < 0.5:
agn = False
temp_points = 4
else:
agn = False
temp_points = 6
# low b
if np.abs(cc.galactic.b.value) < 15:
b_temp_points = -10
else:
b_temp_points = 0
if colors:
return temp_points, agn, [w1 - w2, w2 - w3]
else:
return temp_points, b_temp_points
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Check neighboring alerts')
parser.add_argument('names',
metavar='ZTF objectIds',
type=str,
nargs='+',
help='ZTF names of the candidates to check')
args = parser.parse_args()
# Read the secrets
secrets = ascii.read('./secrets.csv', format='csv')
username = secrets['kowalski_user'][0]
password = secrets['kowalski_pwd'][0]
kow = Kowalski(username=username, password=password)
if len(args.names) > 0:
print("Checking alerts...")
ind_check_alerts = []
problematic = []
for objid in args.names:
try:
index_check = alert_check_complete(kow, objid)
ind_check_alerts.append(index_check)
except IndexError:
ind_check_alerts = []
print(f"Problematic: {objid}")
problematic.append(objid)
ind_check_alerts = np.array(ind_check_alerts)
batch_size = opts.batch_size
Ncatalog = opts.Ncatalog
algorithm = opts.algorithm
catalogDir = os.path.join(outputDir, 'catalog', algorithm)
slurmDir = os.path.join(outputDir, 'slurm')
if not os.path.isdir(slurmDir):
os.makedirs(slurmDir)
logsDir = os.path.join(slurmDir, 'logs')
if not os.path.isdir(logsDir):
os.makedirs(logsDir)
if opts.doQuadrantScale:
kow = Kowalski(username=opts.user, password=opts.pwd)
if opts.lightcurve_source == "Kowalski":
if opts.source_type == "quadrant":
fields, ccds, quadrants = np.arange(1, 880), np.arange(1,
17), np.arange(
1, 5)
#ccds, quadrants = [1], [1]
fields1 = [683, 853, 487, 718, 372, 842, 359, 778, 699, 296]
fields2 = [841, 852, 682, 717, 488, 423, 424, 563, 562, 297, 700, 777]
fields3 = [851, 848, 797, 761, 721, 508, 352, 355, 364, 379]
fields4 = [1866, 1834, 1835, 1804, 1734, 1655, 1565]
fields_complete = fields1 + fields2 + fields3 + fields4
fields = np.arange(1600, 1700)
def test_authenticate(self, username, password):
# if authentication fails, exception will be raised
Kowalski(username=username, password=password)
def test_query(self, username, password, benchmark=False):
with Kowalski(username=username, password=password) as k:
assert k.check_connection()
# base query:
# qu = {"query_type": "general_search",
# "query": "db['ZTF_alerts'].find_one({}, {'_id': 1})",
# "kwargs": {}}
qu = {
"query_type": "find_one",
"query": {
"catalog": collection,
"query": {},
},
"kwargs": {}
}
timeout = 1 # seconds
# query: enqueue_only, save=True, save=False
qu['kwargs']['enqueue_only'] = True
resp = k.query(query=qu, timeout=timeout)
# print(resp)
assert 'query_id' in resp
assert resp['status'] == 'enqueued'
# fetch enqueued/saved query
time.sleep(timeout)
qid = resp['query_id']
result = k.get_query(query_id=qid, part='result')
# print(result)
assert 'result' in result
assert 'query_result' in result['result']
# delete saved query
result = k.delete_query(query_id=qid)
assert 'message' in result
assert result['message'] == 'success'
# little benchmark:
if benchmark:
# qu = {"query_type": "general_search",
# "query": "db['ZTF_alerts'].aggregate([{'$match': {'candidate.rb': {'$gt': 0.98}}}, {'$project': {'_id': 1}}, {'$sample': {'size': 1}}])",
# "kwargs": {}}
qu = {
"query_type": "aggregate",
"query": {
"catalog":
'ZTF_alerts',
"pipeline": [{
'$match': {
'candidate.rb': {
'$gt': 0.98
}
}
}, {
'$project': {
'_id': 1
}
}, {
'$sample': {
'size': 1
}
}],
},
"kwargs": {
"save": False
}
}
timeout = 5 # seconds
qu['kwargs'] = dict()
qu['kwargs']['save'] = False
times = []
for i in range(5):
tic = time.time()
result = k.query(query=qu, timeout=timeout)
toc = time.time()
assert 'result_data' in result
assert 'query_result' in result['result_data']
times.append(toc - tic)
print(times)
def query_kowalski(username,
password,
ra_center,
dec_center,
radius,
jd_trigger,
min_days,
max_days,
slices,
ndethist_min,
within_days,
after_trigger=True):
'''Query kowalski and apply the selection criteria'''
k = Kowalski(username=username, password=password, verbose=False)
# Initialize a set for the results
set_objectId_all = set([])
slices = slices + 1
for slice_lim, i in zip(
np.linspace(0, len(ra_center), slices)[:-1],
np.arange(len(np.linspace(0, len(ra_center), slices)[:-1]))):
try:
ra_center_slice = ra_center[int(slice_lim):int(
np.linspace(0, len(ra_center), slices)[:-1][i + 1])]
dec_center_slice = dec_center[int(slice_lim):int(
np.linspace(0, len(dec_center), slices)[:-1][i + 1])]
except IndexError:
ra_center_slice = ra_center[int(slice_lim):]
dec_center_slice = dec_center[int(slice_lim):]
coords_arr = []
for ra, dec in zip(ra_center_slice, dec_center_slice):
try:
# Remove points too far south for ZTF.
# Say, keep only Dec>-40 deg to be conservative
if dec < -40.:
continue
coords = SkyCoord(ra=float(ra) * u.deg, dec=float(dec) * u.deg)
coords_arr.append((coords.ra.deg, coords.dec.deg))
except ValueError:
print("Problems with the galaxy coordinates?")
continue
# Correct the minimum number of detections
ndethist_min_corrected = int(ndethist_min - 1)
# Correct the jd_trigger if the user specifies to query
# also before the trigger
if after_trigger is False:
jd_trigger = 0
try:
print(
f"slice: {int(slice_lim)}:{int(np.linspace(0,len(ra_center),slices)[:-1][i+1])}"
)
except:
print(f"slice: {int(slice_lim)}:{int(len(ra_center))}")
q = {
"query_type": "cone_search",
"object_coordinates": {
"radec": f"{coords_arr}",
"cone_search_radius": f"{radius}",
"cone_search_unit": "arcmin"
},
"catalogs": {
"ZTF_alerts": {
"filter": {
"candidate.jd": {
'$gt': jd_trigger
},
"candidate.drb": {
'$gt': 0.5
},
"candidate.ndethist": {
'$gt': ndethist_min_corrected
},
"candidate.jdstarthist": {
'$gt': jd_trigger,
'$lt': jd_trigger + within_days
}
},
"projection": {
"objectId": 1,
"candidate.rcid": 1,
"candidate.ra": 1,
"candidate.dec": 1,
"candidate.jd": 1,
"candidate.ndethist": 1,
"candidate.jdstarthist": 1,
"candidate.jdendhist": 1,
"candidate.jdendhist": 1,
"candidate.magpsf": 1,
"candidate.sigmapsf": 1,
"candidate.fid": 1,
"candidate.programid": 1,
"candidate.isdiffpos": 1,
"candidate.ndethist": 1,
"candidate.ssdistnr": 1,
"candidate.rb": 1,
"candidate.drb": 1,
"candidate.distpsnr1": 1,
"candidate.sgscore1": 1,
"candidate.srmag1": 1,
"candidate.distpsnr2": 1,
"candidate.sgscore2": 1,
"candidate.srmag2": 1,
"candidate.distpsnr3": 1,
"candidate.sgscore3": 1,
"candidate.srmag3": 1
}
}
},
"kwargs": {
"hint": "gw01"
}
}
# Perform the query
r = k.query(query=q)
print('Search completed for this slice.')
objectId_list = []
with_neg_sub = []
old = []
out_of_time_window = []
stellar_list = []
try:
if r['result_data']['ZTF_alerts'] == []:
continue
keys_list = list(r['result_data']['ZTF_alerts'].keys())
except AttributeError:
print("Error in the keys list?? Check 'r' ")
# Try one more time
print("Trying to query again the same slice")
try:
r = k.query(query=q)
keys_list = list(r['result_data']['ZTF_alerts'].keys())
except AttributeError:
print("The query failed again, skipping slice..")
continue
for key in keys_list:
all_info = r['result_data']['ZTF_alerts'][key]
for info in all_info:
if info['objectId'] in old:
continue
if info['objectId'] in stellar_list:
continue
if np.abs(info['candidate']['ssdistnr']) < 10:
continue
if info['candidate']['isdiffpos'] in ['f', 0]:
with_neg_sub.append(info['objectId'])
if (info['candidate']['jdendhist'] -
info['candidate']['jdstarthist']) < min_days:
continue
if (info['candidate']['jdendhist'] -
info['candidate']['jdstarthist']) > max_days:
old.append(info['objectId'])
if (info['candidate']['jdstarthist'] -
jd_trigger) > within_days:
old.append(info['objectId'])
# REMOVE! Only for O3a paper
# if (info['candidate']['jdendhist'] -
# info['candidate']['jdstarthist']) >= 72./24. and
# info['candidate']['ndethist'] <= 2. :
# out_of_time_window.append(info['objectId'])
if after_trigger is True:
if (info['candidate']['jdendhist'] -
jd_trigger) > max_days:
out_of_time_window.append(info['objectId'])
else:
if (info['candidate']['jdendhist'] -
info['candidate']['jdstarthist']) > max_days:
out_of_time_window.append(info['objectId'])
try:
if (np.abs(info['candidate']['distpsnr1']) < 1.5
and info['candidate']['sgscore1'] > 0.50):
stellar_list.append(info['objectId'])
except (KeyError, ValueError):
pass
try:
if (np.abs(info['candidate']['distpsnr1']) < 15.
and info['candidate']['srmag1'] < 15.
and info['candidate']['srmag1'] > 0.
and info['candidate']['sgscore1'] >= 0.5):
continue
except (KeyError, ValueError):
pass
try:
if (np.abs(info['candidate']['distpsnr2']) < 15.
and info['candidate']['srmag2'] < 15.
and info['candidate']['srmag2'] > 0.
and info['candidate']['sgscore2'] >= 0.5):
continue
except (KeyError, ValueError):
pass
try:
if (np.abs(info['candidate']['distpsnr3']) < 15.
and info['candidate']['srmag3'] < 15.
and info['candidate']['srmag3'] > 0.
and info['candidate']['sgscore3'] >= 0.5):
continue
except (KeyError, ValueError):
pass
objectId_list.append(info['objectId'])
set_objectId = set(objectId_list)
# Remove those objects with negative subtraction
for n in set(with_neg_sub):
try:
set_objectId.remove(n)
except (ValueError, KeyError):
pass
# Remove stellar objects
for n in set(stellar_list):
try:
set_objectId.remove(n)
except (ValueError, KeyError):
pass
# Remove those objects considered old
for n in set(old):
try:
set_objectId.remove(n)
except (ValueError, KeyError):
pass
# Remove those objects whole alerts go bejond jd_trigger+max_days
for n in set(out_of_time_window):
try:
set_objectId.remove(n)
except (ValueError, KeyError):
pass
print(set_objectId)
set_objectId_all = set_objectId_all | set_objectId
print("Cumulative:", set_objectId_all)
return set_objectId_all
from penquins import Kowalski
from time import time
import numpy as np
if __name__ == '__main__':
k = Kowalski(username='******', password='******', verbose=False)
batch_size = 100
num_batches = 100
# batch_size = 10
# num_batches = 100
times = []
for nb in range(num_batches):
qu = {
"query_type":
"general_search",
"query":
"db['ZTF_sources_20190412'].find({}, " +
"{'_id': 1, 'data.programid': 1, 'data.hjd': 1, " +
f"'data.mag': 1, 'data.magerr': 1}}).skip({nb*batch_size}).limit({batch_size})"
}
# print(qu)
tic = time()
r = k.query(query=qu)
toc = time()
times.append(toc - tic)
print(
f'Fetching batch {nb+1}/{num_batches} with {batch_size} sources/LCs took: {toc-tic:.3f} seconds'
absmagWD = gmag + 5 * (np.log10(np.abs(parallax)) - 2)
baseoutputDir = opts.outputDir
modelPaths = opts.modelPath.split(",")
if opts.doPlots:
kow = []
nquery = 10
cnt = 0
while cnt < nquery:
try:
TIMEOUT = 60
protocol, host, port = "https", "gloria.caltech.edu", 443
kow = Kowalski(username=opts.user,
password=opts.pwd,
timeout=TIMEOUT,
protocol=protocol,
host=host,
port=port)
break
except:
time.sleep(5)
cnt = cnt + 1
if cnt == nquery:
raise Exception('Kowalski connection failed...')
catalogPaths = []
for modelPath in modelPaths:
folders = glob.glob(os.path.join(modelPath, '*_*')) + glob.glob(
os.path.join(modelPath, '*.*'))
for folder in folders:
if opts.doField:
def query_kowalski(name, DEFAULT_AUTH_kowalski):
'''
Please see
https://kowalski.caltech.edu/docs/python_client
of description
'''
# generate directory to store data
cwd = os.getcwd()
targetdir = cwd + '/' + name + '/'
try:
os.stat(targetdir)
except:
os.mkdir(targetdir)
try:
os.stat(targetdir + 'lightcurves/')
except:
os.mkdir(targetdir + 'lightcurves/')
# kowalski query
k = Kowalski(username=DEFAULT_AUTH_kowalski[0],
password=DEFAULT_AUTH_kowalski[1],
verbose=False)
qu = {
"query_type": "general_search",
"query": "db['ZTF_alerts'].find({'objectId': {'$eq': '%s'}})" % name
}
r = k.query(query=qu)
if 'result_data' in r.keys():
rdata = r['result_data']
rrdata = rdata['query_result']
n = len(rrdata)
jds = np.zeros(n)
fids = np.zeros(n)
jdstarts = np.zeros(n)
jdends = np.zeros(n)
fieldids = np.zeros(n)
magpsfs = np.zeros(n)
sigmapsfs = np.zeros(n)
diffimfile = [' ' * 57 for x in range(n)]
for i in range(n):
rrr = rrdata[i]
# yyao: please see rrr['candidate'].keys() for what infomation you can get
# I only saved jd, filterid, fieldid, magpsf, sigmapsf, jdstartref, jdendref, and diffimfilename in this demo...
jds[i] = rrr['candidate']['jd']
fids[i] = rrr['candidate']['fid']
fieldids[i] = rrr['candidate']['field']
magpsfs[i] = rrr['candidate']['magpsf']
sigmapsfs[i] = rrr['candidate']['sigmapsf']
# yyao: sometimes these infomation is not in the keys (reference image not made?)
if 'jdstartref' in rrr['candidate'].keys():
jdstarts[i] = rrr['candidate']['jdstartref']
if 'jdendref' in rrr['candidate'].keys():
jdends[i] = rrr['candidate']['jdendref']
if 'pdiffimfilename' in rrr['candidate'].keys():
diffimfile[i] = rrr['candidate']['pdiffimfilename']
tb = Table([
jds, fids, fieldids, magpsfs, sigmapsfs, jdstarts, jdends,
diffimfile
],
names=[
'jd', 'fid', 'fieldid', 'magpsf', 'sigmapsf',
'jdstartref', 'jdendref', 'diffimfilename'
])
tb.write(targetdir + 'lightcurves/' + '/kowalski_lc_' + name + '.csv')
else:
print('query is not succesful for %s' % name)
def query_kowalski(username, password, clu, args):
'''Query kowalski and apply the selection criteria'''
k = Kowalski(username=username, password=password, verbose=False)
#Initialize a set for the results
set_objectId_all = set([])
for slice_lim,i in zip(np.linspace(0,len(clu),args.slices)[:-1], np.arange(len(np.linspace(0,len(clu),args.slices)[:-1]))):
try:
t = clu[int(slice_lim):int(np.linspace(0,len(clu),args.slices)[:-1][i+1])]
except IndexError:
t = clu[int(slice_lim):]
coords_arr = []
galaxy_names_arr = []
for galaxy,ra, dec in zip(t["name"],t["ra"], t["dec"]):
try:
coords = SkyCoord(ra=ra*u.deg, dec=dec*u.deg)
coords_arr.append((coords.ra.deg,coords.dec.deg))
except ValueError:
print("Problems with the galaxy coordinates?")
pdb.set_trace()
continue
galaxy_names_arr.append(galaxy)
try:
print(f"slice: {int(slice_lim)}:{int(np.linspace(0,len(clu),args.slices)[:-1][i+1])}" )
except:
print(f"slice: {int(slice_lim)}:{int(len(clu))}" )
q = {"query_type": "cone_search",
"object_coordinates": {
"radec": f"{coords_arr}",
"cone_search_radius": f"{args.radius}",
"cone_search_unit": "arcmin"
},
"catalogs": {
"ZTF_alerts": {
"filter": {
"candidate.ndethist": {'$gt': 1},
"candidate.rb": {'$gt': 0.2}
},
"projection": {
"objectId": 1,
"candidate.rcid": 1,
"candidate.ra": 1,
"candidate.dec": 1,
"candidate.jd": 1,
"candidate.ndethist": 1,
"candidate.jdstarthist": 1,
"candidate.jdendhist": 1,
"candidate.jdendhist": 1,
"candidate.magpsf": 1,
"candidate.sigmapsf": 1,
"candidate.fid": 1,
"candidate.programid": 1,
"candidate.isdiffpos": 1,
"candidate.ndethist": 1,
"candidate.ssdistnr": 1,
"candidate.rb": 1,
"candidate.drb": 1,
"candidate.distpsnr1": 1,
"candidate.sgscore1": 1,
"candidate.srmag1": 1,
"candidate.distpsnr2": 1,
"candidate.sgscore2": 1,
"candidate.srmag2": 1,
"candidate.distpsnr3": 1,
"candidate.sgscore3": 1,
"candidate.srmag3": 1
}
}
}
}
#Perform the query
r = k.query(query=q)
print('Search completed for this slice.')
# #Dump the results in a json file
# with open(f'results_clu25Mpc_1week_{i+1}.json', 'w') as j:
# json.dump(r, j)
#Identify 'candid' for all relevant candidates
objectId_list = []
with_neg_sub = []
old = []
stellar_list = []
try:
keys_list = list(r['result_data']['ZTF_alerts'].keys())
except:
print("Error in the keys list?? Check 'r' ")
pdb.set_trace()
for key in keys_list:
all_info = r['result_data']['ZTF_alerts'][key]
for info in all_info:
# #Stop at a certain candidId for debugging
# if info['objectId'] == 'ZTF19aanfkyc':
# pdb.set_trace()
if info['objectId'] in old:
continue
if info['objectId'] in stellar_list:
continue
try:
if info['candidate']['drb'] < 0.5:
continue
except:
do = 'do nothing.'
if np.abs(info['candidate']['ssdistnr']) < 10:
continue
if info['candidate']['isdiffpos'] in ['f',0]:
with_neg_sub.append(info['objectId'])
if (info['candidate']['jdendhist'] - info['candidate']['jdstarthist']) < args.min_days:
continue
if (info['candidate']['jdendhist'] - info['candidate']['jdstarthist']) > args.max_days:
old.append(info['objectId'])
try:
if (np.abs(info['candidate']['distpsnr1']) < 1. and info['candidate']['sgscore1'] > 0.0):
stellar_list.append(info['objectId'])
except:
do = 'do nothing.'
try:
if (np.abs(info['candidate']['distpsnr1']) < 15. and info['candidate']['srmag1'] < 15. and info['candidate']['srmag1'] > 0. and info['candidate']['sgscore1'] >= 0.5):
continue
except:
do = 'do nothing.'
try:
if (np.abs(info['candidate']['distpsnr2']) < 15. and info['candidate']['srmag2'] < 15. and info['candidate']['srmag2'] > 0. and info['candidate']['sgscore2'] >= 0.5):
continue
except:
do = 'do nothing.'
try:
if (np.abs(info['candidate']['distpsnr3']) < 15. and info['candidate']['srmag3'] < 15. and info['candidate']['srmag3'] > 0. and info['candidate']['sgscore3'] >= 0.5):
continue
except:
do = 'do nothing.'
objectId_list.append(info['objectId'])
set_objectId = set(objectId_list)
#Remove those objects with negative subtraction
for n in set(with_neg_sub):
try:
set_objectId.remove(n)
except:
do = 'do nothing'
#Remove stellar objects
for n in set(stellar_list):
try:
set_objectId.remove(n)
except:
do = 'do nothing'
#Remove those objects considered old
for n in set(old):
try:
set_objectId.remove(n)
except:
do = 'do nothing'
print(set_objectId)
set_objectId_all = set_objectId_all | set_objectId
print("Cumulative:", set_objectId_all)
'''
print('----stats-----')
print('Number of sources with negative sub: ', len(set(with_neg_sub)))
print('Number of sources with only pos subtraction: ', len(set_objectId))
print(f"Number of sources older than {args.max_days} days: {len(set(old))}, specifically {set(old)}")
'''
return set_objectId_all
from ...models import (
DBSession,
Obj,
Source,
Candidate,
User,
Token,
Group,
Spectrum,
CronJobRun,
)
_, cfg = load_env()
kowalski = Kowalski(
token=cfg["app.kowalski.token"],
protocol=cfg["app.kowalski.protocol"],
host=cfg["app.kowalski.host"],
port=int(cfg["app.kowalski.port"]),
)
class StatsHandler(BaseHandler):
@permissions(["System admin"])
def get(self):
"""
---
description: Retrieve basic DB statistics
tags:
- system_info
responses:
200:
content:
import csv
# load dataset
parser = argparse.ArgumentParser()
parser.add_argument("inputfile")
parser.add_argument("--id", type=int, default=1, help="group id on Fritz")
args = parser.parse_args()
with open(args.inputfile) as f:
data = csv.reader(f)
trainingset = pd.DataFrame(data)
# Kowalski
with open('password.txt', 'r') as f:
password = f.read().splitlines()
G = Kowalski(username=password[0],
password=password[1],
host='gloria.caltech.edu',
timeout=1000)
# get scores and data and combine
scores = get_highscoring_objects(G, otype='vnv', condition='$or')
index = scores.index
condition = ((scores["vnv_dnn"] > 0.95) &
(scores['vnv_xgb'] <= 0.1)) | ((scores["vnv_dnn"] <= 0.1) &
(scores['vnv_xgb'] > 0.95))
disagreements = index[condition]
disagreeing_scores = scores.iloc[disagreements, :]
stats = get_stats(G, disagreeing_scores['_id'])
data = pd.merge(disagreeing_scores, stats, left_on='_id', right_on='_id')
data['train'] = np.isin(data['_id'], trainingset['ztf_id'])