def main():
"""
Drive GalSim to simulate the LSST.
"""
# Setup a parser to take command line arguments
parser = argparse.ArgumentParser()
parser.add_argument('file', help="The instance catalog")
parser.add_argument('-n',
'--numrows',
default=None,
type=int,
help="Read the first numrows of the file.")
parser.add_argument('--outdir',
type=str,
default='fits',
help='Output directory for eimage file')
parser.add_argument(
'--sensor',
type=str,
default=None,
help='Sensor to simulate, e.g., "R:2,2 S:1,1".' +
'If None, then simulate all sensors with sources on them')
parser.add_argument(
'--config_file',
type=str,
default=None,
help="Config file. If None, the default config will be used.")
parser.add_argument('--log_level',
type=str,
choices=['DEBUG', 'INFO', 'WARN', 'ERROR', 'CRITICAL'],
default='INFO',
help='Logging level. Default: "INFO"')
parser.add_argument(
'--psf',
type=str,
default='Kolmogorov',
choices=['DoubleGaussian', 'Kolmogorov'],
help="PSF model to use; either the double Gaussian "
"from LSE=40 (equation 30), or the Kolmogorov convolved "
"with a Gaussian proposed by David Kirkby at the "
"23 March 2017 SSims telecon")
parser.add_argument('--checkpoint_file',
type=str,
default=None,
help='Checkpoint file name.')
parser.add_argument('--nobj_checkpoint',
type=int,
default=1000,
help='# objects to process between checkpoints')
parser.add_argument('--seed',
type=int,
default=267,
help='integer used to seed random number generator')
arguments = parser.parse_args()
config = desc.imsim.read_config(arguments.config_file)
logger = desc.imsim.get_logger(arguments.log_level)
# Get the number of rows to read from the instance file. Use
# default if not specified.
numRows = arguments.numrows
if numRows is not None:
logger.info("Reading %i rows from the instance catalog %s.", numRows,
arguments.file)
else:
logger.info("Reading all rows from the instance catalog %s.",
arguments.file)
camera_wrapper = LSSTCameraWrapper()
catalog_contents = desc.imsim.parsePhoSimInstanceFile(arguments.file,
numRows=numRows)
obs_md = catalog_contents.obs_metadata
phot_params = catalog_contents.phot_params
sources = catalog_contents.sources
gs_object_arr = sources[0]
gs_object_dict = sources[1]
# Sub-divide the source dataframe into stars and galaxies.
if arguments.sensor is not None:
detector_list = [
make_galsim_detector(camera_wrapper, arguments.sensor, phot_params,
obs_md)
]
else:
detector_list = []
for det in camera_wrapper.camera:
det_type = det.getType()
if det_type != WAVEFRONT and det_type != GUIDER:
detector_list.append(
make_galsim_detector(camera_wrapper, det.getName(),
phot_params, obs_md))
# Add noise and sky background
# The simple code using the default lsst-GalSim interface would be:
#
# PhoSimStarCatalog.noise_and_background = ExampleCCDNoise(addNoise=True,
# addBackground=True)
#
# But, we need a more realistic sky model and we need to pass more than
# this basic info to use Peter Y's ESO sky model.
# We must pass obs_metadata, chip information etc...
noise_and_background \
= ESOSkyModel(obs_md, addNoise=True, addBackground=True)
bp_dict = BandpassDict.loadTotalBandpassesFromFiles(
bandpassNames=obs_md.bandpass)
gs_interpreter = GalSimInterpreter(obs_metadata=obs_md,
epoch=2000.0,
detectors=detector_list,
bandpassDict=bp_dict,
noiseWrapper=noise_and_background,
seed=arguments.seed)
gs_interpreter.checkpoint_file = arguments.checkpoint_file
gs_interpreter.nobj_checkpoint = arguments.nobj_checkpoint
gs_interpreter.restore_checkpoint(camera_wrapper, phot_params, obs_md)
# Add a PSF.
if arguments.psf.lower() == "doublegaussian":
# This one is taken from equation 30 of
# www.astro.washington.edu/users/ivezic/Astr511/LSST_SNRdoc.pdf .
#
# Set seeing from self.obs_metadata.
local_PSF = \
SNRdocumentPSF(obs_md.OpsimMetaData['FWHMgeom'])
elif arguments.psf.lower() == "kolmogorov":
# This PSF was presented by David Kirkby at the 23 March 2017
# Survey Simulations Working Group telecon
#
# https://confluence.slac.stanford.edu/pages/viewpage.action?spaceKey=LSSTDESC&title=SSim+2017-03-23
# equation 3 of Krisciunas and Schaefer 1991
airmass = 1.0 / np.sqrt(
1.0 - 0.96 *
(np.sin(0.5 * np.pi - obs_md.OpsimMetaData['altitude']))**2)
local_PSF = \
Kolmogorov_and_Gaussian_PSF(airmass=airmass,
rawSeeing=obs_md.OpsimMetaData['rawSeeing'],
band=obs_md.bandpass)
else:
raise RuntimeError("Do not know what to do with psf model: "
"%s" % arguments.psf)
gs_interpreter.setPSF(PSF=local_PSF)
if arguments.sensor is not None:
gs_objects_to_draw = gs_object_dict[arguments.sensor]
else:
gs_objects_to_draw = gs_object_arr
for gs_obj in gs_objects_to_draw:
if gs_obj.uniqueId in gs_interpreter.drawn_objects:
continue
gs_interpreter.drawObject(gs_obj)
desc.imsim.add_cosmic_rays(gs_interpreter, phot_params)
# Write out the fits files
outdir = arguments.outdir
if not os.path.isdir(outdir):
os.makedirs(outdir)
prefix = config['persistence']['eimage_prefix']
gs_interpreter.writeImages(nameRoot=os.path.join(outdir, prefix) +
str(obs_md.OpsimMetaData['obshistID']))
def test_checkpointing(self):
"Test checkpointing of .detectorImages data."
camera = camTestUtils.CameraWrapper().camera
camera_wrapper = GalSimCameraWrapper(camera)
phot_params = PhotometricParameters()
obs_md = ObservationMetaData(pointingRA=23.0,
pointingDec=12.0,
rotSkyPos=13.2,
mjd=59580.0,
bandpassName='r')
detectors = [
make_galsim_detector(camera_wrapper, dd.getName(), phot_params,
obs_md) for dd in camera_wrapper.camera
]
# Create a GalSimInterpreter object and set the checkpoint
# attributes.
gs_interpreter = GalSimInterpreter(detectors=detectors)
gs_interpreter.checkpoint_file = self.cp_file
nobj = 10
gs_interpreter.nobj_checkpoint = nobj
# Set the image data by hand.
key = "R00_S00_r.fits"
detname = "R:0,0 S:0,0"
detector = make_galsim_detector(camera_wrapper, detname, phot_params,
obs_md)
image = gs_interpreter.blankImage(detector=detector)
image += 17
gs_interpreter.detectorImages[key] = image
# Add some drawn objects and check that the checkpoint file is
# written at the right cadence.
for uniqueId in range(1, nobj + 1):
gs_interpreter.drawn_objects.add(uniqueId)
gs_interpreter.write_checkpoint()
if uniqueId < nobj:
self.assertFalse(os.path.isfile(self.cp_file))
else:
self.assertTrue(os.path.isfile(self.cp_file))
# Verify that the checkpointed data has the expected content.
with open(self.cp_file, 'rb') as input_:
cp_data = pickle.load(input_)
self.assertTrue(np.array_equal(cp_data['images'][key], image.array))
# Check the restore_checkpoint function.
new_interpreter = GalSimInterpreter(detectors=detectors)
new_interpreter.checkpoint_file = self.cp_file
new_interpreter.restore_checkpoint(camera_wrapper, phot_params, obs_md)
self.assertEqual(new_interpreter.drawn_objects,
gs_interpreter.drawn_objects)
self.assertEqual(set(new_interpreter.detectorImages.keys()),
set(gs_interpreter.detectorImages.keys()))
for det_name in new_interpreter.detectorImages.keys():
new_img = new_interpreter.detectorImages[det_name]
gs_img = gs_interpreter.detectorImages[det_name]
np.testing.assert_array_equal(new_img.array, gs_img.array)
self.assertEqual(new_img.bounds, gs_img.bounds)
self.assertEqual(new_img.wcs.crpix1, gs_img.wcs.crpix1)
self.assertEqual(new_img.wcs.crpix2, gs_img.wcs.crpix2)
self.assertEqual(new_img.wcs.crval1, gs_img.wcs.crval1)
self.assertEqual(new_img.wcs.crval2, gs_img.wcs.crval2)
self.assertEqual(new_img.wcs.detectorName, gs_img.wcs.detectorName)
for name in new_img.wcs.fitsHeader.names():
self.assertEqual(new_img.wcs.fitsHeader.get(name),
gs_img.wcs.fitsHeader.get(name))
