def get_argdict(comp_data_dir, **args):
argdict = dict.fromkeys(args['sim'])
for sim in args['sim']:
arglist = []
# Get config and simulation files
config = 'IC79'
# config = comp.simfunctions.sim2cfg(sim)
# gcd, files = comp.simfunctions.get_level3_files(sim, testing=True)
files = glob.glob(
'/data/sim/IceTop/2010/filtered/level2a/CORSIKA-ice-top/{}/00000-00999/Level2a_*.i3.bz2'
.format(sim))
files = sorted(files)
# Default parameters
outdir = '{}/{}_sim/{}'.format(comp_data_dir, config, sim)
comp.checkdir(outdir + '/')
if args['test']:
args['n'] = 2
# List of existing files to possibly check against
existing_files = glob.glob('{}/Level2a_*.i3.bz2'.format(outdir))
existing_files.sort()
# # Split into batches
# batches = [files[i:i + args['n']]
# for i in range(0, len(files), args['n'])]
if args['test']:
files = files[:2]
# batches = batches[:2]
for fi, f in enumerate(files):
# for bi, batch in enumerate(batches):
# Name output hdf5 file
run = os.path.basename(f).split('.')[-3]
# out = '{}/Level2a_IC79_corsika_icetop.00{}.part{}-{}.i3.bz2'.format(outdir, sim, start, end)
out = '{}/Level3_IC79_{}_Run{:06d}.i3.gz'.format(
outdir, sim, int(run))
# Don't forget to insert GCD file at beginning of FileNameList
# batch.insert(0, gcd)
# batch = ' '.join(batch)
arg = '{} --isMC --do-inice --dataset {} --det IC79 --waveform -o {}'.format(
f, sim, out)
arglist.append(arg)
argdict[sim] = arglist
return argdict
def get_argdict(comp_data_dir, **args):
argdict = dict.fromkeys(args['sim'])
for sim in args['sim']:
arglist = []
# Get config and simulation files
config = comp.simfunctions.sim2cfg(sim)
gcd, files = comp.simfunctions.get_level3_files(sim, testing=True)
# Default parameters
outdir = '{}/{}_sim/test-files'.format(comp_data_dir, config)
comp.checkdir(outdir + '/')
if args['test']:
args['n'] = 2
# List of existing files to possibly check against
existing_files = glob.glob('{}/sim_{}_*.hdf5'.format(outdir, sim))
existing_files.sort()
# Split into batches
batches = [files[i:i + args['n']]
for i in range(0, len(files), args['n'])]
if args['test']:
batches = batches[:2]
for bi, batch in enumerate(batches):
# Name output hdf5 file
start_index = batch[0].find('Run') + 3
end_index = batch[0].find('.i3.gz')
start = batch[0][start_index:end_index]
end = batch[-1][start_index:end_index]
out = '{}/sim_{}_part{}-{}.hdf5'.format(outdir, sim, start, end)
# Don't forget to insert GCD file at beginning of FileNameList
batch.insert(0, gcd)
batch = ' '.join(batch)
arg = '--files {} -s {} -o {}'.format(batch, sim, out)
arglist.append(arg)
argdict[sim] = arglist
return argdict
def make_submit_script(executable, jobID, script_path, condor_dir):
comp.checkdir(script_path)
lines = ["universe = vanilla\n",
"getenv = true\n",
"executable = {}\n".format(executable),
"arguments = $(ARGS)\n",
"log = {}/logs/{}.log\n".format(condor_dir, jobID),
"output = /data/user/jbourbeau/composition/condor/outs/{}.out\n".format(jobID),
"error = /data/user/jbourbeau/composition/condor/errors/{}.error\n".format(jobID),
"notification = Never\n",
# "request_memory = 5000\n",
"queue \n"]
condor_script = script_path
with open(condor_script, 'w') as f:
f.writelines(lines)
return
def get_argdict(comp_data_dir, **args):
argdict = dict.fromkeys(args['sim'])
for sim in args['sim']:
arglist = []
# Get config and simulation files
config = 'IC79'
files = glob.glob('/data/sim/IceTop/2010/filtered/level2a/CORSIKA-ice-top/{}/*/Level2a_*.i3.bz2'.format(sim))
files = sorted(files)
# Default parameters
outdir = '{}/{}_sim/{}'.format(comp_data_dir, config, sim)
comp.checkdir(outdir + '/')
if args['test']:
args['n'] = 2
if args['test']:
files = files[:2]
for fi, f in enumerate(files):
# Name output hdf5 file
run = os.path.basename(f).split('.')[-3]
out = '{}/Level3_IC79_{}_Run{:06d}.i3.gz'.format(outdir, sim, int(run))
# If not overwriting, and outfile already exists, then continue onto next file
if not args['overwrite']:
if os.path.exists(out):
continue
arg = '{} --isMC --do-inice --dataset {} --det IC79 --waveform -o {}'.format(f, sim, out)
arglist.append(arg)
argdict[sim] = arglist
return argdict
return
def getjobID(jobID, condor_dir):
jobID += time.strftime('_%Y%m%d')
othersubmits = glob.glob(
'{}/submit_scripts/{}_??.submit'.format(condor_dir, jobID))
jobID += '_{:02d}'.format(len(othersubmits) + 1)
return jobID
if __name__ == "__main__":
# Setup global path names
mypaths = comp.Paths()
comp.checkdir(mypaths.comp_data_dir)
# Set up condor directory
condor_dir = '/scratch/{}/condor_composition'.format(getpass.getuser())
for directory in ['errors', 'logs', 'outs', 'submit_scripts']:
comp.checkdir(condor_dir + '/' + directory + '/')
simoutput = comp.simfunctions.getSimOutput()
default_sim_list = ['7006', '7579', '7241', '7263', '7791',
'7242', '7262', '7851', '7007', '7784']
p = argparse.ArgumentParser(
description='Runs level3_process.py on cluster en masse',
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog=simoutput)
p.add_argument('-s', '--sim', dest='sim', nargs='*',
choices=default_sim_list,
default=default_sim_list,
import time
import glob
import argparse
import os
from collections import defaultdict
from icecube.weighting.weighting import from_simprod
from icecube.weighting.fluxes import GaisserH3a, GaisserH4a
import composition as comp
if __name__ == "__main__":
# Setup global path names
mypaths = comp.Paths()
comp.checkdir(mypaths.comp_data_dir)
p = argparse.ArgumentParser(
description='Runs extra modules over a given fileList')
p.add_argument('-o', '--outfile', dest='outfile',
help='Output file')
args = p.parse_args()
dataframe_dict = defaultdict(list)
# Get simulation information
t_sim = time.time()
print('Loading simulation information...')
file_list = sorted(glob.glob(mypaths.comp_data_dir +
'/IT73_sim/files/sim_????.hdf5'))
value_keys = ['IceTopMaxSignal',
