def set_model(self, root, typ, apply_region=True):
"""
apply_region: Isolate the clean components of a model fits file to those under self.region
typ = init / best / regrid / ...
"""
if apply_region:
region_file = self.get_region()
for model_file in glob.glob(root+'*[0-9]-model.fits'):
lib_img.blank_image_reg(model_file, region_file, model_file, inverse=True, blankval=0.)
self.model[typ] = root
log='$nameMS_taql-c' + str(c) + '.log',
commandType='general')
### DONE
with w.if_todo('lowres_img_c%02i' % c):
# Making beam mask
logger.info('Preparing mask for low-res clean...')
lib_util.run_wsclean(s,
'wscleanLRmask.log',
MSs.getStrWsclean(),
name='img/tmp',
size=imgsizepix,
scale='30arcsec')
os.system('mv img/tmp-image.fits img/wide-lr-mask.fits')
lib_img.blank_image_reg('img/wide-lr-mask.fits',
beamReg,
blankval=0.)
lib_img.blank_image_reg('img/wide-lr-mask.fits',
beamReg,
blankval=1.,
inverse=True)
# reclean low-resolution
logger.info('Cleaning low-res...')
imagename_lr = 'img/wide-lr'
lib_util.run_wsclean(s,
'wscleanLR.log',
MSs.getStrWsclean(),
name=imagename_lr,
do_predict=False,
parallel_gridding=4,
'point': 'cross',
'font': '"helvetica 16 normal roman"'
})
sh.comment = 'color=red text="%s"' % (name + '.reg')
regions = pyregion.ShapeList([sh])
lib_util.check_rm(peel_region_file)
regions.write(peel_region_file)
# copy and blank models
logger.info('Peel - Cleanup model images...')
os.system('cp ' + imagenameM + '*model.fits peel-' + name)
imagename_peel = 'peel-' + name + '/' + imagenameM.split(
'/')[-1]
for model_file in glob.glob(imagename_peel + '*model.fits'):
lib_img.blank_image_reg(model_file,
peel_region_file,
blankval=0.,
inverse=True)
# predict the source to peel
logger.info('Peel - Predict init...')
s.add('wsclean -predict -name ' + imagename_peel + ' -j ' +
str(s.max_processors) + ' -channels-out 2 \
-reorder -parallel-reordering 4 ' + MSs.getStrWsclean(),
log='wsclean-pre.log',
commandType='wsclean',
processors='max')
s.run(check=True)
# add the source to peel back
logger.info(
'Peel - add model: CORRECTED_DATA = CORRECTED_DATA + MODEL_DATA...'
def clean(p, MSs, res='normal', size=[1, 1], empty=False, imagereg=None):
"""
p = patch name
mss = list of mss to clean
size = in deg of the image
"""
# set pixscale and imsize
pixscale = MSs.resolution
if res == 'normal':
pixscale = float('%.1f' % (pixscale / 2.5))
elif res == 'high':
pixscale = float('%.1f' % (pixscale / 3.5))
elif res == 'low':
pass # no change
imsize = [
int(size[0] * 1.5 / (pixscale / 3600.)),
int(size[1] * 1.5 / (pixscale / 3600.))
] # add 50%
imsize[0] += imsize[0] % 2
imsize[1] += imsize[1] % 2
if imsize[0] < 256: imsize[0] = 256
if imsize[1] < 256: imsize[1] = 256
logger.debug('Image size: ' + str(imsize) + ' - Pixel scale: ' +
str(pixscale))
if res == 'normal':
weight = 'briggs -0.3'
maxuv_l = None
elif res == 'high':
weight = 'briggs -0.6'
maxuv_l = None
elif res == 'low':
weight = 'briggs 0'
maxuv_l = 3500
else:
logger.error('Wrong "res": %s.' % str(res))
sys.exit()
if empty:
logger.info('Cleaning empty (' + str(p) + ')...')
imagename = 'img/empty-' + str(p)
lib_util.run_wsclean(s,
'wscleanE-' + str(p) + '.log',
MSs.getStrWsclean(),
name=imagename,
data_column='SUBTRACTED_DATA',
size=imsize,
scale=str(pixscale) + 'arcsec',
weight=weight,
niter=0,
no_update_model_required='',
minuv_l=30,
mgain=0,
baseline_averaging='')
else:
# clean 1
logger.info('Cleaning (' + str(p) + ')...')
imagename = 'img/extract-' + str(p)
lib_util.run_wsclean(s,
'wscleanA-' + str(p) + '.log',
MSs.getStrWsclean(),
name=imagename,
size=imsize,
scale=str(pixscale) + 'arcsec',
weight=weight,
niter=10000,
no_update_model_required='',
minuv_l=30,
maxuv_l=maxuv_l,
mgain=0.85,
baseline_averaging='',
parallel_deconvolution=512,
auto_threshold=5,
join_channels='',
fit_spectral_pol=3,
channels_out=ch_out,
deconvolution_channels=3)
# make mask
im = lib_img.Image(imagename + '-MFS-image.fits', userReg=userReg)
try:
im.makeMask(threshpix=10, rmsbox=(70, 5))
except:
logger.warning('Fail to create mask for %s.' % imagename +
'-MFS-image.fits')
return
if imagereg is not None:
lib_img.blank_image_reg(im.maskname,
imagereg,
inverse=True,
blankval=0.)
# clean 2
# TODO: add deconvolution_channels when bug fixed
logger.info('Cleaning w/ mask (' + str(p) + ')...')
imagenameM = 'img/extractM-' + str(p)
lib_util.run_wsclean(
s,
'wscleanB-' + str(p) + '.log',
MSs.getStrWsclean(),
name=imagenameM,
do_predict=True,
size=imsize,
scale=str(pixscale) + 'arcsec',
weight=weight,
niter=100000,
no_update_model_required='',
minuv_l=30,
maxuv_l=maxuv_l,
reuse_psf=imagename,
reuse_dirty=imagename,
mgain=0.85,
multiscale='',
baseline_averaging='',
parallel_deconvolution=512,
auto_threshold=0.7,
auto_mask=1.5,
fits_mask=im.maskname,
join_channels='',
fit_spectral_pol=3,
channels_out=ch_out) # , deconvolution_channels=3)
os.system('cat logs/wscleanB-' + str(p) +
'.log | grep "background noise"')
fit_spectral_pol=3,
channels_out=ch_out) # , deconvolution_channels=3)
os.system('cat logs/wscleanB-' + str(p) +
'.log | grep "background noise"')
with w.if_todo('predict_rest'):
# DDF predict+corrupt in MODEL_DATA of everything BUT the calibrator
indico = wideDD_image.root + '.DicoModel'
outdico = indico + '-' + target_reg_file.split('.')[
0] # use prefix of target reg
inmask = sorted(glob.glob(wideDD_image.root + '*_mask-ddcal.fits'))[-1]
outmask = outdico + '.mask'
lib_img.blank_image_reg(inmask,
target_reg_file,
outfile=outmask,
inverse=False,
blankval=0.)
s.add('MaskDicoModel.py --MaskName=%s --InDicoModel=%s --OutDicoModel=%s' %
(outmask, indico, outdico),
log='MaskDicoModel.log',
commandType='python',
processors='max')
s.run(check=True)
ddf_parms = {
'Data_MS': MSs.getStrDDF(),
'Data_ColName': 'CORRECTED_DATA',
'Data_Sort': 1,
'Output_Mode': 'Predict',
'Predict_InitDicoModel': outdico,
else:
# this dd-cal should not be in the data anymore but probably the source finder got some strong residuals
if d.peel_off:
logger.info('This sources has been peeled, skip.')
continue
# DDF predict+corrupt in MODEL_DATA of everything BUT the calibrator
indico = full_image.root + '.DicoModel'
outdico = indico + '-' + d.name
inmask = sorted(glob.glob(full_image.root + '.mask*.fits'))[-1]
outmask = outdico + '.mask'
lib_img.blank_image_reg(inmask,
d.get_region(),
outfile=outmask,
inverse=False,
blankval=0.)
s.add(
'MaskDicoModel.py --MaskName=%s --InDicoModel=%s --OutDicoModel=%s'
% (outmask, indico, outdico),
log='MaskDicoModel-' + logstring + '.log',
commandType='python',
processors='max')
s.run(check=True)
ddf_parms = {
'Data_MS':
MSs.getStrDDF(),
'Data_ColName':
'CORRECTED_DATA',
baseline_averaging='', parallel_deconvolution=512, local_rms='', auto_threshold=0.5, auto_mask=1., fits_mask=im.maskname, \
join_channels='', fit_spectral_pol=3, channels_out=9, deconvolution_channels=3)
lib_util.run_wsclean(s, 'wscleanBlow-'+str(p)+'.log', MSs.getStrWsclean(), name=imagename+'-low', size=imsize, scale=str(pixscale)+'arcsec', \
weight=weight, niter=100000, no_update_model_required='', minuv_l=30, maxuv_l=maxuv_l, taper_gaussian='30asec', mgain=0.85, \
multiscale='', multiscale_scale_bias=0.75, multiscale_scales='0,10,20,40', \
baseline_averaging='', parallel_deconvolution=512, local_rms='', auto_threshold=0.5, auto_mask=1., fits_mask=im.maskname, \
join_channels='', fit_spectral_pol=3, channels_out=9, deconvolution_channels=3)
os.system('cat logs/wscleanB-' + str(p) + '.log | grep "background noise"')
# Load facet mask and set target region to 0
mask_voro = 'ddcal/masks/facets%02i.fits' % lastcycle
os.system('cp %s facet/facets.fits' % mask_voro)
lib_img.blank_image_reg('facet/facets.fits', target_reg, blankval=0)
# mosaic the skymodel, Isl_patch_000 will be the target of interest
lsm = lsmtool.load(mosaic_image.skymodel_cut)
lsm.group('facet', facet='facet/facets.fits', root='Isl_patch')
lsm.setPatchPositions(method='mid') # center of the facets
directions = set(lsm.getColValues('patch'))
coord = [c.deg for c in lsm.getPatchPositions('Isl_patch_0')['Isl_patch_0']]
logger.info("Facet centre: " + str(coord))
# calculate region size (TODO: maybe better using regionfile?)
ramin = np.min(
lsm.getColValues('RA')[lsm.getColValues('Patch') == 'Isl_patch_0'])
ramax = np.max(
lsm.getColValues('RA')[lsm.getColValues('Patch') == 'Isl_patch_0'])
decmin = np.min(
else:
# these dd-cal are not in the data anymore
if d.peel_off:
logger.info('This sources has been peeled, skip.')
continue
# DDF predict+corrupt in MODEL_DATA of everything BUT the calibrator
indico = full_image.root + '.DicoModel'
outdico = indico + '-' + d.name
inmask = sorted(glob.glob(full_image.root + '.mask*.fits'))[-1]
outmask = outdico + '.mask'
lib_img.blank_image_reg(inmask,
d.get_region(),
outfile=outmask,
inverse=False,
blankval=0.)
s.add('MaskDicoModel.py --MaskName=%s --InDicoModel=%s --OutDicoModel=%s' % (outmask, indico, outdico), \
log='MaskDicoModel-'+logstring+'.log', commandType='python', processors='max')
s.run(check=True)
ddf_parms = {
'Data_MS':
MSs.getStrDDF(),
'Data_ColName':
'CORRECTED_DATA',
'Data_Sort':
1,
'Output_Mode':
'Predict',
# if next is a "cont" then I need the do_predict
logger.info('Cleaning shallow (cycle: ' + str(c) + ')...')
lib_util.run_wsclean(s, 'wsclean-c%02i.log' % c, MSs.getStrWsclean(), do_predict=True, name=imagename, \
parallel_gridding=4, baseline_averaging='', size=5000, scale='2.5arcsec', \
niter=1000, no_update_model_required='', minuv_l=30, mgain=0.4, nmiter=0, \
auto_threshold=5, local_rms='', local_rms_method='rms-with-min', \
join_channels='', fit_spectral_pol=2, channels_out=2, **kwargs1 )
# check if hand-made mask is available
im = lib_img.Image(imagename + '-MFS-image.fits')
im.makeMask(threshisl=5, rmsbox=(50, 5), atrous_do=True)
maskfits = imagename + '-mask.fits'
region = '%s/regions/%s.reg' % (parset_dir, target)
if os.path.exists(region):
lib_img.blank_image_reg(maskfits, beamReg, blankval=0.)
lib_img.blank_image_reg(maskfits, region, blankval=1.)
logger.info('Cleaning full (cycle: ' + str(c) + ')...')
lib_util.run_wsclean(s, 'wsclean-c%02i.log' % c, MSs.getStrWsclean(), do_predict=True, cont=True, name=imagename, \
parallel_gridding=4, size=5000, scale='2.5arcsec', \
niter=1000000, no_update_model_required='', minuv_l=30, mgain=0.4, nmiter=0, \
auto_threshold=0.5, auto_mask=2., local_rms='', local_rms_method='rms-with-min', fits_mask=maskfits, \
multiscale='', multiscale_scale_bias=0.8, \
join_channels='', fit_spectral_pol=2, channels_out=2, **kwargs2 )
os.system('cat logs/wsclean-c%02i.log | grep "background noise"' % c)
w.done('image-c%02i' % c)
### DONE
# Set CORRECTED_DATA = CORRECTED_DATA - MODEL_DATA