def main(msin, config_path):
o = options(config_path, option_list)
if o['pbimage'] is None:
die('pbimage must be specified')
# fix up the new list-type options
for i, cat in enumerate(o['list']):
try:
o[cat] = o['filenames'][i]
except:
pass
try:
o[cat + '_matchrad'] = o['radii'][i]
except:
pass
try:
o[cat + '_fluxfactor'] = o['fluxfactor'][i]
except:
pass
if o['logging'] is not None and not os.path.isdir(o['logging']):
os.mkdir(o['logging'])
# pybdsm source finding
sfind_image(o['catprefix'], o['pbimage'], o['sfind_pixel_fraction'])
# matching with catalogs
for cat in o['list']:
print 'Doing catalogue', cat
crossmatch_image(o['catprefix'] + '.cat.fits', cat)
filter_catalog(o['catprefix'] + '.cat.fits',
o['catprefix'] + '.cat.fits_' + cat + '_match.fits',
o['pbimage'],
o['catprefix'] + '.cat.fits_' + cat +
'_match_filtered.fits',
cat,
options=o)
# Filter catalogs (only keep isolated compact sources within 3deg of pointing centre)
# Astrometric plots
if 'FIRST' in o['list']:
report('Plotting position offsets')
plot_position_offset(
'%s.cat.fits_FIRST_match_filtered.fits' % o['catprefix'],
o['pbimage'],
'%s.cat.fits_FIRST_match_filtered_positions.png' % o['catprefix'],
'FIRST',
options=o)
t = Table.read(o['catprefix'] + '.cat.fits_FIRST_match_filtered.fits')
bsra = np.percentile(bootstrap(t['FIRST_dRA'], np.mean, 10000),
(16, 84))
bsdec = np.percentile(bootstrap(t['FIRST_dDEC'], np.mean, 10000),
(16, 84))
mdra = np.mean(t['FIRST_dRA'])
mddec = np.mean(t['FIRST_dDEC'])
print 'Mean delta RA is %.3f arcsec (1-sigma %.3f -- %.3f arcsec)' % (
mdra, bsra[0], bsra[1])
print 'Mean delta DEC is %.3f arcsec (1-sigma %.3f -- %.3f arcsec)' % (
mddec, bsdec[0], bsdec[1])
report('Plotting flux ratios')
# Flux ratio plots (only compact sources)
plot_flux_ratios(
'%s.cat.fits_FIRST_match_filtered.fits' % o['catprefix'],
o['pbimage'],
'%s.cat.fits_FIRST_match_filtered_fluxerrors.png' % o['catprefix'],
options=o)
report('Plotting flux scale comparison')
# Flux scale comparison plots
if 'TGSS' in o['list']:
plot_flux_errors(
'%s.cat.fits_TGSS_match_filtered.fits' % o['catprefix'],
o['pbimage'],
'%s.cat.fits_TGSS_match_filtered_fluxratio.png' % o['catprefix'],
'TGSS',
options=o)
t = Table.read(o['catprefix'] + '.cat.fits_TGSS_match_filtered.fits')
ratios = t['Total_flux'] / (t['TGSS_Total_flux'] /
o['TGSS_fluxfactor'])
bsratio = np.percentile(bootstrap(ratios, np.median, 10000), (16, 84))
print 'Median LOFAR/TGSS ratio is %.3f (1-sigma %.3f -- %.3f)' % (
np.median(ratios), bsratio[0], bsratio[1])
if 'NVSS' in o['list']:
t = Table.read(o['catprefix'] + '.cat.fits_NVSS_match_filtered.fits')
t = t[t['Total_flux'] > 10e-3]
ratios = t['Total_flux'] / t['NVSS_Total_flux']
bsratio = np.percentile(bootstrap(ratios, np.median, 10000), (16, 84))
print 'Median LOFAR/NVSS ratio is %.3f (1-sigma %.3f -- %.3f)' % (
np.median(ratios), bsratio[0], bsratio[1])
# Noise estimate
hdu = fits.open(o['pbimage'])
imagenoise = get_rms(hdu)
print 'An estimate of the image noise is %.3f muJy/beam' % (imagenoise *
1E6)
return 0
tgss_scale = np.median(ratios)
else:
tgss_scale = None
if 'NVSS' in o['list']:
t = Table.read(o['catprefix'] + '.cat.fits_NVSS_match_filtered.fits')
t = t[t['Total_flux'] > 10e-3]
ratios = t['Total_flux'] / t['NVSS_Total_flux']
bsratio = np.percentile(bootstrap(ratios, np.median, 10000), (16, 84))
print 'Median LOFAR/NVSS ratio is %.3f (1-sigma %.3f -- %.3f)' % (
np.median(ratios), bsratio[0], bsratio[1])
nvss_scale = np.median(ratios)
else:
nvss_scale = None
# Noise estimate
hdu = fits.open(o['pbimage'])
imagenoise = get_rms(hdu)
rms = imagenoise * 1e6
print 'An estimate of the image noise is %.3f muJy/beam' % rms
drs = do_dr_checker(o['catprefix'] + '.cat.fits',
o['pbimage'],
verbose=False,
peak=0.4)
dr = np.median(drs)
print 'Median dynamic range is', dr
# fit source counts
if o['fit_sourcecounts']:
from fit_sourcecounts import do_fit_sourcecounts
sc_norm, sc_index, scale = do_fit_sourcecounts(rms=imagenoise)
else:
sc_norm = sc_index = scale = None
print 'Median LOFAR/TGSS ratio is %.3f (1-sigma %.3f -- %.3f)' % (np.median(ratios),bsratio[0],bsratio[1])
tgss_scale=np.median(ratios)
else:
tgss_scale=None
if 'NVSS' in o['list']:
t=Table.read(o['catprefix']+'.cat.fits_NVSS_match_filtered.fits')
t=t[t['Total_flux']>30e-3]
ratios=t['Total_flux']/t['NVSS_Total_flux']
bsratio=np.percentile(bootstrap(ratios,np.median,10000),(16,84))
print 'Median LOFAR/NVSS ratio is %.3f (1-sigma %.3f -- %.3f)' % (np.median(ratios),bsratio[0],bsratio[1])
nvss_scale=np.median(ratios)
else:
nvss_scale=None
# Noise estimate
hdu=fits.open(o['pbimage'])
imagenoise = get_rms(hdu)
rms=imagenoise*1e6
print 'An estimate of the image noise is %.3f muJy/beam' % rms
drs=do_dr_checker(o['catprefix']+'.cat.fits',o['pbimage'],verbose=False,peak=0.4)
dr=np.median(drs)
print 'Median dynamic range is',dr
# fit source counts
if o['fit_sourcecounts']:
from fit_sourcecounts import do_fit_sourcecounts
sc_norm,sc_index,scale=do_fit_sourcecounts(rms=imagenoise)
else:
sc_norm=sc_index=scale=None
print rms,dr,catsources,first_ra,first_dec,tgss_scale,nvss_scale,sc_norm,sc_index,scale
#!/usr/bin/python
# find the central rms of an image by iteratively removing the outliers
from __future__ import print_function
from __future__ import absolute_import
import numpy as np
from astropy.io import fits
from auxcodes import get_rms
if __name__ == '__main__':
import sys
for name in sys.argv[1:]:
hdu = fits.open(name)
print(name, get_rms(hdu))
def make_mosaic(args):
if args.scale is not None:
if len(args.scale) != len(args.directories):
die('Scales provided must match directories')
if args.noise is not None:
if len(args.noise) != len(args.directories):
die('Noises provided must match directories')
if args.rootname:
rootname=args.rootname+'-'
else:
rootname=''
if args.exact:
reproj=reproject_exact
else:
reproj=reproject_interp_chunk_2d
if args.do_lowres:
intname='image_full_low_m.int.restored.fits'
appname='image_full_low_m.app.restored.fits'
elif args.use_shifted:
intname='image_full_ampphase_di_m.NS_shift.int.facetRestored.fits'
appname='image_full_ampphase_di_m.NS_shift.app.facetRestored.fits'
else:
intname='image_full_ampphase_di_m.NS.int.restored.fits'
appname='image_full_ampphase_di_m.NS.app.restored.fits'
# astromap blanking if required
bth=None
try:
bth=float(args.astromap_blank)
except:
pass
threshold=float(args.beamcut)
hdus=[]
app=[]
astromaps=[]
wcs=[]
print 'Reading files...'
noise=[]
name=[]
for d in args.directories:
name.append(d.split('/')[-1])
hdu=fits.open(d+'/'+intname)
if args.find_noise:
print 'Estimating noise for', d+'/' + intname
if args.do_lowres:
noise.append(get_rms(hdu,boxsize=1500))
else:
noise.append(get_rms(hdu))
hdus.append(flatten(hdu))
app.append(flatten(fits.open(d+'/'+appname)))
if bth:
astromaps.append(flatten(fits.open(d+'/astromap.fits')))
if args.find_noise:
args.noise=noise
print 'Noise values are:'
for t,n in zip(name,noise):
print t,n
print 'Computing noise/beam factors...'
for i in range(len(app)):
np.seterr(divide='ignore')
app[i].data=np.divide(app[i].data,hdus[i].data)
app[i].data[app[i].data<threshold]=0
# at this point this is the beam factor: we want 1/sigma**2.0, so divide by central noise and square
if args.noise is not None:
if args.scale is not None:
app[i].data/=args.noise[i]*args.scale[i]
else:
app[i].data/=args.noise[i]
app[i].data=app[i].data**2.0
if args.scale is not None:
hdus[i].data*=args.scale[i]
if args.shift:
print 'Finding shifts (NOTE THIS CODE IS OBSOLETE)...'
# shift according to the FIRST delta ra/dec from quality pipeline
dras=[]
ddecs=[]
for d in args.directories:
t=Table.read(d+'/image_full_ampphase1m.cat.fits_FIRST_match_filtered.fits')
dras.append(np.mean(t['FIRST_dRA']))
ddecs.append(np.mean(t['FIRST_dDEC']))
print 'Applying shifts:',dras,ddecs
for i in range(len(app)):
for hdu in [hdus[i],app[i]]:
ra=hdu.header['CRVAL1']
dec=hdu.header['CRVAL2']
hdu.header['CRVAL1']-=dras[i]/(3600.0*np.cos(np.pi*dec/180.0))
hdu.header['CRVAL2']-=ddecs[i]/3600.0
for i in range(len(app)):
wcs.append(WCS(hdus[i].header))
# astromap blanking
if bth:
print 'Blanking using astrometry quality maps with threshold',bth,'arcsec'
for i in range(len(app)):
outname=rootname+'astroblank-'+name[i]+'.fits'
if args.load and os.path.isfile(outname):
print 'Loading previously blanked image'
hdu=fits.open(outname)
hdus[i].data=hdu[0].data
else:
print 'Blanking image',i
dmaxy,dmaxx=hdus[i].data.shape
count=0
am=astromaps[i]
awcs=WCS(am.header)
maxy,maxx=am.data.shape
for y in range(maxy):
for x in range(maxx):
value=am.data[y,x]
if np.isnan(value):
if y<maxy-1:
value=am.data[y+1,x]
if value>bth:
ra,dec=[float(f) for f in awcs.wcs_pix2world(x,y,0)]
rx,ry=[int(p) for p in wcs[i].wcs_world2pix(ra,dec,0)]
rxp=rx+21 # astromap pix size, with margin
ryp=ry+21
if rx<0: rx=0
if ry<0: ry=0
if rxp>dmaxx: rxp=dmaxx
if ryp>dmaxy: ryp=dmaxy
hdus[i].data[ry:ryp,rx:rxp]=np.nan
count+=1
print '... blanked',count*900.0/3600,'square arcmin'
outname=rootname+'astroblank-'+name[i]+'.fits'
if args.save: hdus[i].writeto(outname,clobber=True)
app[i].data[np.isnan(hdus[i].data)]=np.nan
# If the header is directly passed in, use it
try:
header=args.header
xsize=header['NAXIS1']
ysize=header['NAXIS2']
print 'Mosaic using header passed from calling program'
except:
header=None
if header is None:
if args.load_layout:
with open(rootname+'mosaic-header.pickle') as f:
header=pickle.load(f)
xsize=header['NAXIS1']
ysize=header['NAXIS2']
print 'Mosaic using loaded header'
else:
print 'Creating the mosaic header'
ras=np.array([w.wcs.crval[0] for w in wcs])
decs=np.array([w.wcs.crval[1] for w in wcs])
mra=np.mean(ras)
mdec=np.mean(decs)
print 'Will make mosaic at',mra,mdec
# we make a reference WCS and use it to find the extent in pixels
# needed for the combined image
rwcs=WCS(naxis=2)
rwcs.wcs.ctype=wcs[0].wcs.ctype
rwcs.wcs.cdelt=wcs[0].wcs.cdelt
rwcs.wcs.crval=[mra,mdec]
rwcs.wcs.crpix=[1,1]
xmin=0
xmax=0
ymin=0
ymax=0
for a,w in zip(app,wcs):
ys,xs=np.where(a.data)
axmin=xs.min()
aymin=ys.min()
axmax=xs.max()
aymax=ys.max()
del(xs)
del(ys)
print 'non-zero',axmin,aymin,axmax,aymax
for x,y in ((axmin,aymin),(axmax,aymin),(axmin,aymax),(axmax,aymax)):
ra,dec=[float(f) for f in w.wcs_pix2world(x,y,0)]
#print ra,dec
nx,ny=[float (f) for f in rwcs.wcs_world2pix(ra,dec,0)]
print nx,ny
if nx<xmin: xmin=nx
if nx>xmax: xmax=nx
if ny<ymin: ymin=ny
if ny>ymax: ymax=ny
print 'co-ord range:', xmin, xmax, ymin, ymax
xsize=int(xmax-xmin)
ysize=int(ymax-ymin)
rwcs.wcs.crpix=[-int(xmin)+1,-int(ymin)+1]
print 'checking:', rwcs.wcs_world2pix(mra,mdec,0)
print rwcs
header=rwcs.to_header()
header['NAXIS']=2
header['NAXIS1']=xsize
header['NAXIS2']=ysize
with open(rootname+'mosaic-header.pickle','w') as f:
pickle.dump(header,f)
isum=np.zeros([ysize,xsize])
wsum=np.zeros_like(isum)
mask=np.zeros_like(isum,dtype=np.bool)
print 'now making the mosaic'
for i in range(len(hdus)):
print 'image',i,'(',name[i],')'
outname=rootname+'reproject-'+name[i]+'.fits'
if args.load and os.path.exists(outname):
print 'loading...'
hdu=fits.open(outname)
r=hdu[0].data
else:
print 'reprojecting...'
r, footprint = reproj(hdus[i], header, hdu_in=0, parallel=False)
r[np.isnan(r)]=0
hdu = fits.PrimaryHDU(header=header,data=r)
if args.save: hdu.writeto(outname,clobber=True)
print 'weights',i,'(',name[i],')'
outname=rootname+'weight-'+name[i]+'.fits'
if args.load and os.path.exists(outname):
print 'loading...'
hdu=fits.open(outname)
w=hdu[0].data
mask|=(w>0)
else:
print 'reprojecting...'
w, footprint = reproj(app[i], header, hdu_in=0, parallel=False)
mask|=~np.isnan(w)
w[np.isnan(w)]=0
hdu = fits.PrimaryHDU(header=header,data=w)
if args.save: hdu.writeto(outname,clobber=True)
print 'add to mosaic...'
isum+=r*w
wsum+=w
if not(args.no_write):
isum/=wsum
# mask now contains True where a non-nan region was present in either map
isum[~mask]=np.nan
for ch in ('BMAJ', 'BMIN', 'BPA'):
header[ch]=hdus[0].header[ch]
header['ORIGIN']='ddf-pipeline '+version()
hdu = fits.PrimaryHDU(header=header,data=isum)
hdu.writeto(rootname+'mosaic.fits',clobber=True)
hdu = fits.PrimaryHDU(header=header,data=wsum)
hdu.writeto(rootname+'mosaic-weights.fits',clobber=True)
def do_fit_sourcecounts(t=None, rms=None,do_plots=False,sfindarea=17.09):
global fluxnorm,ds
if t is None:
t=Table.read('image_full_ampphase_di_m.NS.cat.fits')
print 'Number of sources in full table is',len(t)
if rms is None:
rms=get_rms(fits.open('image_full_ampphase_di_m.NS_shift.int.facetRestored.fits'))
cutoff=rms*20
print 'Cutoff will be',cutoff,'Jy'
#cutoff=1e-3
maxflux=np.max(t['Total_flux'])
t=t[t['Total_flux']>cutoff]
print 'Number of sources after completeness cut is',len(t)
bins=np.logspace(np.log10(cutoff),np.log10(maxflux)*1.01,20)
cbins=0.5*(bins[:-1]+bins[1:])
ds=bins[1:]-bins[:-1]
hist,_=np.histogram(t['Total_flux'],bins=bins)
zval=None
for i in range(len(hist)):
if hist[i]==0:
zval=i
break
if zval is not None:
cbins=cbins[:zval]
hist=hist[:zval]
ds=ds[:zval]
nwalkers=10
ndim=2
fluxnorm=0.1 # find normalization at this flux in Jy
sampler = emcee.EnsembleSampler(nwalkers, ndim, lnpost,
args=(cbins,hist))
pos=[[1.6,0.86]+0.01*np.random.normal(size=ndim)
for i in range(nwalkers)]
sampler.run_mcmc(pos, 1000)
samples=sampler.chain[:, 200:, :].reshape((-1, ndim))
samplest=samples.transpose()
means=np.mean(samplest,axis=1)
errors=np.percentile(samplest,(16,84),axis=1)-means
for i in range(ndim):
print i,means[i],errors[0,i],errors[1,i]
fnorm=means[0]
falpha=means[1]
C=[3.5142,0.3738,-0.3138,-0.0717,0.0213,0.0097] # Intema+
fn=fluxnorm
totfactor=1.0
for i in range(10):
lf=np.log10(fn)
ncn=0
for i in range(6):
ncn+=C[i]*lf**i
print 'for %.3f Jy number count norm should be %f' % (fn,10**ncn)
measured_ncn=10**fnorm*fluxnorm*(fn**1.5)*3282.8/sfindarea # check precise area
print 'measured number count norm is',measured_ncn
scale=(measured_ncn/10**ncn)#**(1.0/1.5)
print 'scaling factor should be',scale
totfactor*=scale
print 'total factor is',totfactor
fn=fluxnorm/totfactor
print 'New flux norm value is',fn
if abs(scale-1.0)<1e-4:
print 'Converged, stopping'
break
if do_plots:
import matplotlib.pyplot as plt
import corner
plt.scatter(cbins,hist)
plt.xscale('log')
plt.yscale('log')
plt.xlim(cutoff,maxflux)
plt.ylim(0.5,np.max(hist)*1.3)
yv=model(cbins,fnorm,falpha)
plt.plot(cbins,yv)
fig = corner.corner(samples)
plt.show()
return fnorm,falpha,totfactor
def make_mosaic(args):
if args.scale is not None:
if len(args.scale) != len(args.directories):
die('Scales provided must match directories')
if args.noise is not None:
if len(args.noise) != len(args.directories):
die('Noises provided must match directories')
if args.rootname:
rootname = args.rootname + '-'
else:
rootname = ''
if args.exact:
reproj = reproject_exact
else:
reproj = reproject_interp_chunk_2d
if args.do_lowres:
intname = 'image_full_low_m.int.restored.fits'
appname = 'image_full_low_m.app.restored.fits'
elif args.use_shifted:
intname = 'image_full_ampphase_di_m.NS_shift.int.facetRestored.fits'
appname = 'image_full_ampphase_di_m.NS_shift.app.facetRestored.fits'
else:
intname = 'image_full_ampphase_di_m.NS.int.restored.fits'
appname = 'image_full_ampphase_di_m.NS.app.restored.fits'
# astromap blanking if required
bth = None
try:
bth = float(args.astromap_blank)
except:
pass
threshold = float(args.beamcut)
hdus = []
app = []
astromaps = []
wcs = []
print 'Reading files...'
noise = []
name = []
for d in args.directories:
name.append(d.split('/')[-1])
hdu = fits.open(d + '/' + intname)
if args.find_noise:
print 'Estimating noise for', d + '/' + intname
if args.do_lowres:
noise.append(get_rms(hdu, boxsize=1500))
else:
noise.append(get_rms(hdu))
hdus.append(flatten(hdu))
app.append(flatten(fits.open(d + '/' + appname)))
if bth:
astromaps.append(flatten(fits.open(d + '/astromap.fits')))
if args.find_noise:
args.noise = noise
print 'Noise values are:'
for t, n in zip(name, noise):
print t, n
print 'Computing noise/beam factors...'
for i in range(len(app)):
np.seterr(divide='ignore')
app[i].data = np.divide(app[i].data, hdus[i].data)
app[i].data[app[i].data < threshold] = 0
# at this point this is the beam factor: we want 1/sigma**2.0, so divide by central noise and square
if args.noise is not None:
if args.scale is not None:
app[i].data /= args.noise[i] * args.scale[i]
else:
app[i].data /= args.noise[i]
app[i].data = app[i].data**2.0
if args.scale is not None:
hdus[i].data *= args.scale[i]
if args.shift:
print 'Finding shifts (NOTE THIS CODE IS OBSOLETE)...'
# shift according to the FIRST delta ra/dec from quality pipeline
dras = []
ddecs = []
for d in args.directories:
t = Table.read(
d +
'/image_full_ampphase1m.cat.fits_FIRST_match_filtered.fits')
dras.append(np.mean(t['FIRST_dRA']))
ddecs.append(np.mean(t['FIRST_dDEC']))
print 'Applying shifts:', dras, ddecs
for i in range(len(app)):
for hdu in [hdus[i], app[i]]:
ra = hdu.header['CRVAL1']
dec = hdu.header['CRVAL2']
hdu.header['CRVAL1'] -= dras[i] / (3600.0 *
np.cos(np.pi * dec / 180.0))
hdu.header['CRVAL2'] -= ddecs[i] / 3600.0
for i in range(len(app)):
wcs.append(WCS(hdus[i].header))
# astromap blanking
if bth:
print 'Blanking using astrometry quality maps with threshold', bth, 'arcsec'
for i in range(len(app)):
outname = rootname + 'astroblank-' + name[i] + '.fits'
if args.load and os.path.isfile(outname):
print 'Loading previously blanked image'
hdu = fits.open(outname)
hdus[i].data = hdu[0].data
else:
print 'Blanking image', i
dmaxy, dmaxx = hdus[i].data.shape
count = 0
am = astromaps[i]
awcs = WCS(am.header)
maxy, maxx = am.data.shape
for y in range(maxy):
for x in range(maxx):
value = am.data[y, x]
if np.isnan(value):
if y < maxy - 1:
value = am.data[y + 1, x]
if value > bth:
ra, dec = [
float(f) for f in awcs.wcs_pix2world(x, y, 0)
]
rx, ry = [
int(p)
for p in wcs[i].wcs_world2pix(ra, dec, 0)
]
rxp = rx + 21 # astromap pix size, with margin
ryp = ry + 21
if rx < 0: rx = 0
if ry < 0: ry = 0
if rxp > dmaxx: rxp = dmaxx
if ryp > dmaxy: ryp = dmaxy
hdus[i].data[ry:ryp, rx:rxp] = np.nan
count += 1
print '... blanked', count * 900.0 / 3600, 'square arcmin'
outname = rootname + 'astroblank-' + name[i] + '.fits'
if args.save: hdus[i].writeto(outname, clobber=True)
app[i].data[np.isnan(hdus[i].data)] = np.nan
# If the header is directly passed in, use it
try:
header = args.header
xsize = header['NAXIS1']
ysize = header['NAXIS2']
print 'Mosaic using header passed from calling program'
except:
header = None
if header is None:
if args.load_layout:
with open(rootname + 'mosaic-header.pickle') as f:
header = pickle.load(f)
xsize = header['NAXIS1']
ysize = header['NAXIS2']
print 'Mosaic using loaded header'
else:
print 'Creating the mosaic header'
ras = np.array([w.wcs.crval[0] for w in wcs])
decs = np.array([w.wcs.crval[1] for w in wcs])
mra = np.mean(ras)
mdec = np.mean(decs)
print 'Will make mosaic at', mra, mdec
# we make a reference WCS and use it to find the extent in pixels
# needed for the combined image
rwcs = WCS(naxis=2)
rwcs.wcs.ctype = wcs[0].wcs.ctype
rwcs.wcs.cdelt = wcs[0].wcs.cdelt
rwcs.wcs.crval = [mra, mdec]
rwcs.wcs.crpix = [1, 1]
xmin = 0
xmax = 0
ymin = 0
ymax = 0
for a, w in zip(app, wcs):
ys, xs = np.where(a.data)
axmin = xs.min()
aymin = ys.min()
axmax = xs.max()
aymax = ys.max()
del (xs)
del (ys)
print 'non-zero', axmin, aymin, axmax, aymax
for x, y in ((axmin, aymin), (axmax, aymin), (axmin, aymax),
(axmax, aymax)):
ra, dec = [float(f) for f in w.wcs_pix2world(x, y, 0)]
#print ra,dec
nx, ny = [float(f) for f in rwcs.wcs_world2pix(ra, dec, 0)]
print nx, ny
if nx < xmin: xmin = nx
if nx > xmax: xmax = nx
if ny < ymin: ymin = ny
if ny > ymax: ymax = ny
print 'co-ord range:', xmin, xmax, ymin, ymax
xsize = int(xmax - xmin)
ysize = int(ymax - ymin)
rwcs.wcs.crpix = [-int(xmin) + 1, -int(ymin) + 1]
print 'checking:', rwcs.wcs_world2pix(mra, mdec, 0)
print rwcs
header = rwcs.to_header()
header['NAXIS'] = 2
header['NAXIS1'] = xsize
header['NAXIS2'] = ysize
with open(rootname + 'mosaic-header.pickle', 'w') as f:
pickle.dump(header, f)
isum = np.zeros([ysize, xsize])
wsum = np.zeros_like(isum)
mask = np.zeros_like(isum, dtype=np.bool)
print 'now making the mosaic'
for i in range(len(hdus)):
print 'image', i, '(', name[i], ')'
outname = rootname + 'reproject-' + name[i] + '.fits'
if args.load and os.path.exists(outname):
print 'loading...'
hdu = fits.open(outname)
r = hdu[0].data
else:
print 'reprojecting...'
r, footprint = reproj(hdus[i], header, hdu_in=0, parallel=False)
r[np.isnan(r)] = 0
hdu = fits.PrimaryHDU(header=header, data=r)
if args.save: hdu.writeto(outname, clobber=True)
print 'weights', i, '(', name[i], ')'
outname = rootname + 'weight-' + name[i] + '.fits'
if args.load and os.path.exists(outname):
print 'loading...'
hdu = fits.open(outname)
w = hdu[0].data
mask |= (w > 0)
else:
print 'reprojecting...'
w, footprint = reproj(app[i], header, hdu_in=0, parallel=False)
mask |= ~np.isnan(w)
w[np.isnan(w)] = 0
hdu = fits.PrimaryHDU(header=header, data=w)
if args.save: hdu.writeto(outname, clobber=True)
print 'add to mosaic...'
isum += r * w
wsum += w
if not (args.no_write):
isum /= wsum
# mask now contains True where a non-nan region was present in either map
isum[~mask] = np.nan
for ch in ('BMAJ', 'BMIN', 'BPA'):
header[ch] = hdus[0].header[ch]
header['ORIGIN'] = 'ddf-pipeline ' + version()
hdu = fits.PrimaryHDU(header=header, data=isum)
hdu.writeto(rootname + 'mosaic.fits', clobber=True)
hdu = fits.PrimaryHDU(header=header, data=wsum)
hdu.writeto(rootname + 'mosaic-weights.fits', clobber=True)
#!/usr/bin/python
# find the central rms of an image by iteratively removing the outliers
import numpy as np
from astropy.io import fits
from auxcodes import get_rms
if __name__=='__main__':
import sys
for name in sys.argv[1:]:
hdu=fits.open(name)
print name,get_rms(hdu)
def do_fit_sourcecounts(t=None, rms=None, do_plots=False, sfindarea=17.09):
global fluxnorm, ds
if t is None:
t = Table.read('image_full_ampphase_di_m.NS.cat.fits')
print('Number of sources in full table is', len(t))
if rms is None:
rms = get_rms(
fits.open(
'image_full_ampphase_di_m.NS_shift.int.facetRestored.fits'))
cutoff = rms * 20
print('Cutoff will be', cutoff, 'Jy')
#cutoff=1e-3
maxflux = np.max(t['Total_flux'])
t = t[t['Total_flux'] > cutoff]
print('Number of sources after completeness cut is', len(t))
bins = np.logspace(np.log10(cutoff), np.log10(maxflux) * 1.01, 20)
cbins = 0.5 * (bins[:-1] + bins[1:])
ds = bins[1:] - bins[:-1]
hist, _ = np.histogram(t['Total_flux'], bins=bins)
zval = None
for i in range(len(hist)):
if hist[i] == 0:
zval = i
break
if zval is not None:
cbins = cbins[:zval]
hist = hist[:zval]
ds = ds[:zval]
nwalkers = 10
ndim = 2
fluxnorm = 0.1 # find normalization at this flux in Jy
sampler = emcee.EnsembleSampler(nwalkers, ndim, lnpost, args=(cbins, hist))
pos = [[1.6, 0.86] + 0.01 * np.random.normal(size=ndim)
for i in range(nwalkers)]
sampler.run_mcmc(pos, 1000)
samples = sampler.chain[:, 200:, :].reshape((-1, ndim))
samplest = samples.transpose()
means = np.mean(samplest, axis=1)
errors = np.percentile(samplest, (16, 84), axis=1) - means
for i in range(ndim):
print(i, means[i], errors[0, i], errors[1, i])
fnorm = means[0]
falpha = means[1]
C = [3.5142, 0.3738, -0.3138, -0.0717, 0.0213, 0.0097] # Intema+
fn = fluxnorm
totfactor = 1.0
for i in range(10):
lf = np.log10(fn)
ncn = 0
for i in range(6):
ncn += C[i] * lf**i
print('for %.3f Jy number count norm should be %f' % (fn, 10**ncn))
measured_ncn = old_div(10**fnorm * fluxnorm * (fn**1.5) * 3282.8,
sfindarea) # check precise area
print('measured number count norm is', measured_ncn)
scale = (old_div(measured_ncn, 10**ncn)) #**(1.0/1.5)
print('scaling factor should be', scale)
totfactor *= scale
print('total factor is', totfactor)
fn = old_div(fluxnorm, totfactor)
print('New flux norm value is', fn)
if abs(scale - 1.0) < 1e-4:
print('Converged, stopping')
break
if do_plots:
import matplotlib.pyplot as plt
import corner
plt.scatter(cbins, hist)
plt.xscale('log')
plt.yscale('log')
plt.xlim(cutoff, maxflux)
plt.ylim(0.5, np.max(hist) * 1.3)
yv = model(cbins, fnorm, falpha)
plt.plot(cbins, yv)
fig = corner.corner(samples)
plt.show()
return fnorm, falpha, totfactor
#!/usr/bin/python
# find the central rms of an image by iteratively removing the outliers
import numpy as np
from astropy.io import fits
from auxcodes import get_rms
if __name__ == '__main__':
import sys
for name in sys.argv[1:]:
hdu = fits.open(name)
print name, get_rms(hdu)
def make_extended_mask(infile,
fullresfile,
rmsthresh=3.0,
sizethresh=2500,
maxsize=25000,
rootname=None,
verbose=False,
rmsfacet=False,
ds9region='image_dirin_SSD.tessel.reg'):
''' infile is the input low-res image, fullresfile is the full-resolution template image, sizethresh the minimum island size in pixels '''
if rootname is None:
prefix = ''
else:
prefix = rootname + '-'
hdu = fits.open(infile)
if rmsfacet == False:
rms = get_rms(hdu)
if rmsfacet == True:
get_rms_map2(infile, ds9region, prefix + 'rms-low.fits')
hdu2 = fits.open(prefix + 'rms-low.fits')
rms = hdu2[0].data[0, 0, :]
det = hdu[0].data[0, 0, :] > rmsthresh * rms
labels, count = nd.label(det)
print 'found', count, 'islands'
#label, counts = np.unique(labels, return_counts=True)
label = np.unique(labels)
counts = np.bincount(labels.flatten())
big = (counts > sizethresh) & (counts < maxsize)
big_regions = label[big]
print 'Found', len(big_regions) - 1, 'large islands'
if verbose: print counts[big]
mask = np.zeros_like(det, dtype=int)
for l in big_regions:
if l: mask += l * (labels == l)
slices = nd.find_objects(mask)
big_slices = [slices[i - 1] for i in big_regions if i]
kernel = np.ones((3, 3))
mask = convolve2d(mask, kernel, mode='same', fillvalue=0)
mask = (mask > 1)
w = WCS(hdu[0].header)
hdu[0].data = mask.astype(np.float32)
hdu.writeto(prefix + 'mask-low.fits', clobber=True)
if fullresfile is not None:
# regrid all the objects onto the full-res image, if a template is supplied
hduf = fits.open(fullresfile)
maskf = np.zeros_like(hduf[0].data[0, 0, :, :])
wf = WCS(hduf[0].header)
for slice in big_slices:
yslice = slice[0]
xslice = slice[1]
ymin = yslice.start
ymax = yslice.stop
xmin = xslice.start
xmax = xslice.stop
ppos = []
for x in [xmax, xmin]:
for y in [ymax, ymin]:
ppos.append([x, y, 0, 0])
worldlim = w.wcs_pix2world(ppos, 0)
wpos = []
for ra in [worldlim[:, 0].min(), worldlim[:, 0].max()]:
for dec in [worldlim[:, 1].min(), worldlim[:, 1].max()]:
wpos.append([ra, dec, 0, 0])
pixlim = wf.wcs_world2pix(wpos, 0)
xminf = int(pixlim[:, 0].min())
xmaxf = int(pixlim[:, 0].max())
yminf = int(pixlim[:, 1].min())
ymaxf = int(pixlim[:, 1].max())
xs = np.arange(xminf, xmaxf)
ys = np.arange(yminf, ymaxf)
x, y = np.meshgrid(xs, ys)
x = x.flatten()
y = y.flatten()
pix = np.array([x, y, np.zeros_like(x), np.zeros_like(x)]).T
world = wf.wcs_pix2world(pix, 0)
opix = w.wcs_world2pix(world, 0)
for xv, yv, op in zip(x, y, opix):
try:
if mask[int(op[1]), int(op[0])] > 0:
maskf[yv, xv] = 1
except IndexError:
# catch wcs mismatches or similar
pass
hduf[0].data = maskf.astype(np.float32)
hduf.writeto(prefix + 'mask-high.fits', clobber=True)
def make_extended_mask(infile,fullresfile,rmsthresh=3.0,sizethresh=2500,maxsize=25000,rootname=None,verbose=False,rmsfacet=False,ds9region='image_dirin_SSD_m_c.tessel.reg'):
''' infile is the input low-res image, fullresfile is the full-resolution template image, sizethresh the minimum island size in pixels '''
if rootname is None:
prefix=''
else:
prefix=rootname+'-'
hdu=fits.open(infile)
if rmsfacet == False:
rms=get_rms(hdu)
if rmsfacet == True:
get_rms_map2(infile,ds9region,prefix+'rms-low.fits')
hdu2=fits.open(prefix+'rms-low.fits')
rms=hdu2[0].data[0,0,:]
det=hdu[0].data[0,0,:]>rmsthresh*rms
labels, count = nd.label(det)
print 'found',count,'islands'
#label, counts = np.unique(labels, return_counts=True)
label=np.unique(labels)
counts=np.bincount(labels.flatten())
big=(counts>sizethresh) & (counts<maxsize)
big_regions=label[big]
print 'Found',len(big_regions)-1,'large islands'
if verbose: print counts[big]
mask=np.zeros_like(det,dtype=int)
for l in big_regions:
if l: mask+=l*(labels==l)
slices=nd.find_objects(mask)
big_slices=[slices[i-1] for i in big_regions if i]
kernel = np.ones((3,3))
mask = convolve2d(mask, kernel, mode='same', fillvalue=0)
mask = (mask>1)
w=WCS(hdu[0].header)
hdu[0].data=mask.astype(np.float32)
hdu.writeto(prefix+'mask-low.fits',clobber=True)
if fullresfile is not None:
# regrid all the objects onto the full-res image, if a template is supplied
hduf=fits.open(fullresfile)
maskf=np.zeros_like(hduf[0].data[0,0,:,:])
wf=WCS(hduf[0].header)
for slice in big_slices:
yslice=slice[0]
xslice=slice[1]
ymin=yslice.start
ymax=yslice.stop
xmin=xslice.start
xmax=xslice.stop
ppos=[]
for x in [xmax,xmin]:
for y in [ymax,ymin]:
ppos.append([x,y,0,0])
worldlim=w.wcs_pix2world(ppos,0)
wpos=[]
for ra in [worldlim[:,0].min(),worldlim[:,0].max()]:
for dec in [worldlim[:,1].min(),worldlim[:,1].max()]:
wpos.append([ra,dec,0,0])
pixlim=wf.wcs_world2pix(wpos,0)
xminf=int(pixlim[:,0].min())
xmaxf=int(pixlim[:,0].max())
yminf=int(pixlim[:,1].min())
ymaxf=int(pixlim[:,1].max())
xs=np.arange(xminf,xmaxf)
ys=np.arange(yminf,ymaxf)
x,y=np.meshgrid(xs,ys)
x=x.flatten()
y=y.flatten()
pix=np.array([x,y,np.zeros_like(x),np.zeros_like(x)]).T
world=wf.wcs_pix2world(pix,0)
opix=w.wcs_world2pix(world,0)
for xv,yv,op in zip(x,y,opix):
try:
if mask[int(op[1]),int(op[0])]>0:
maskf[yv,xv]=1
except IndexError:
# catch wcs mismatches or similar
pass
hduf[0].data=maskf.astype(np.float32)
hduf.writeto(prefix+'mask-high.fits',clobber=True)