def bolocam_mask():
bolocam = fits.open(config.OUTPUT + 'bolocam_master_template.fits')
header = bolocam[0].header
temphead = fits.PrimaryHDU()
bolohead = temphead.header
# Setting the tempheader for the bolocam data
bolohead.set('CRVAL1' , header['CRVAL1']) # DEC in deg of ref pixel
bolohead.set('CRVAL2' , header['CRVAL2']) # RA in deg of ref pixel
bolohead.set('CRPIX1' , header['CRPIX1']) # ref pixel x
bolohead.set('CRPIX2' , header['CRPIX2']) # ref pixel y
bolohead.set('CD1_1' , header['CD1_1']) # Deg/pixel
bolohead.set('CD1_2' , header['CD1_2']) # Deg/pixel
bolohead.set('CD2_1' , header['CD2_1']) # Deg/pixel
bolohead.set('CD2_2' , header['CD2_2']) # Deg/pixel
bolohead.set('EPOCH' , 2000)
bolohead.set('EQUINOX' , 2000)
bolohead.set('CTYPE1' , header['CTYPE1']) # coord type
bolohead.set('CTYPE2' , header['CTYPE2']) # coord type
bolo_shape = bolocam[0].data.shape
bolo = np.ones((bolo_shape[0],bolo_shape[1]))
hdu = fits.PrimaryHDU(bolo,bolohead)
bolo_psw = hcongrid(hdu.data,hdu.header,maps[0]['shead'])
bolo_pmw = hcongrid(hdu.data,hdu.header,maps[1]['shead'])
bolo_plw = hcongrid(hdu.data,hdu.header,maps[2]['shead'])
# We will use a gaussian kernel for simplicity, but in reality a proper # PSF kernel should be used (See, e.g., Aniano+ 2011). ## # First of all, all maps must be in surface brightness units. # PACS maps need hence to be converted. We choose the My/sr conversion # to keep consistency with SPIRE maps. if wind < 2: print "Now converting the flux into surface brightness units..." flx_sb = flx_zero/(1e6*kk**2*pixsz**2) # converted in MJy/sr from Jy/pix else: flx_sb = flx_zero print "Building the gaussian simulating the PSF" print '' # the following is the sigma of the gaussian, in pixels sigma = np.sqrt(psf[4]**2-psf[wind]**2)/pixsz*fw2sig kernel = Gaussian2DKernel(stddev=sigma,x_size=20,y_size=20) print '' print "Convolving..." convolved = convolve(flx_sb, kernel, mode='same') #, method='direct') print "Done!" # now regrids to the new header print '' print 'Regridding...' flxmap = hcongrid(convolved,header,header500) # Now converting from MJy/sr to Jy/px flx_reg = flxmap*1e6*kk**2*pixsz500**2 print '' print 'Writing the file with the convolved inmage...' hdu_s = fits.PrimaryHDU(flx_reg,header=header500) hdu_s.writeto(outfile)
print 'Done...!' nl = header1["NAXIS1"] map = cube[0, :, :] # Note the a convolution must be performed in surface brightness units # We first define the convolution kernel. # We will use a gaussian kernel for simplicity, but in reality a proper # PSF kernel should be used (See, e.g., Aniano+ 2011) # the sigma of the gaussian (PSF) should be defined in pixels pixsz1 = np.abs(header1["CDELT1"]) * 3600. pixsz2 = np.abs(header2["CDELT1"]) * 3600. # first of all, all maps must be in surface brightness units. print "I am converting the flux into surface brightness" flx_sb = map / (kk**2 * pixsz1**2) # converted in K/sr from K print "Building the gaussian simulating the PSF" # the following is the gaussian FWHM in pixels sigma = np.sqrt(psf2**2 - psf1**2) / pixsz1 #kernel = Gaussian2DKernel(stddev=sigma,x_size=xs,y_size=ys) kernel = Gaussian2DKernel(stddev=sigma, x_size=40, y_size=40) print "Convolving..." convolved = scipy_convolve(flx_sb, kernel, mode='same') #, method='direct') print "Done!" # now regrids to the new header flxmap = hcongrid(convolved, header1, header2) # Now converting from K/sr to K flx_reg = flxmap * kk**2 * pixsz2**2 outfile = 'convolved.fits' hdu_s = fits.PrimaryHDU(flx_reg, header=header2) hdu_s.writeto(outfile)
reshld = rbf(XI, YI)
#now add the values to the residual image
res[ee:ee + bxs, oo:oo + bxs] = reshld
tts = tts + 1
#get the median background
mbck = numpy.median(bck)
sbck = numpy.median(sbk)
#subtract the sky gradient and add back the median background
sub = bigimg - res
sub = sub + mbck
#align the image
algn = hcongrid(sub, header, rhead)
#update the header
header['CTYPE1'] = rhead['CTYPE1']
header['CTYPE2'] = rhead['CTYPE2']
header['CRVAL1'] = rhead['CRVAL1']
header['CRVAL2'] = rhead['CRVAL2']
header['CRPIX1'] = rhead['CRPIX1']
header['CRPIX2'] = rhead['CRPIX2']
header['CD1_1'] = rhead['CD1_1']
header['CD1_2'] = rhead['CD1_2']
header['CD2_1'] = rhead['CD2_1']
header['CD2_2'] = rhead['CD2_2']
#update the header
header['medback'] = mbck
reshld = rbf(XI, YI) #now add the values to the residual image res[ee:ee+bxs, oo:oo+bxs] = reshld tts = tts+1 #get the median background mbck = numpy.median(bck) sbck = numpy.median(sbk) #subtract the sky gradient and add back the median background sub = bigimg-res sub = sub + mbck #align the image algn = hcongrid(sub, header, rhead) #update the header header['CTYPE1'] = rhead['CTYPE1'] header['CTYPE2'] = rhead['CTYPE2'] header['CRVAL1'] = rhead['CRVAL1'] header['CRVAL2'] = rhead['CRVAL2'] header['CRPIX1'] = rhead['CRPIX1'] header['CRPIX2'] = rhead['CRPIX2'] header['CD1_1'] = rhead['CD1_1'] header['CD1_2'] = rhead['CD1_2'] header['CD2_1'] = rhead['CD2_1'] header['CD2_2'] = rhead['CD2_2'] #update the header header['medback'] = mbck
def clus_add_sziso(maps,
isim,
yin,
tin,
params,
verbose=0,
testflag=0,
saveplot=0,
nsim=0,
clusname=None):
errmsg = False
# Now the bolocam data is in the SPIRE format
# we can do a loop over the map adding in the false sz signal
mapsize = len(maps)
if verbose:
print('Fetching BOLOCAM maps')
''' This is just until Jack gives us more new Bolocam templates... '''
if clusname == 'rxj1347':
data_file = CLUSDATA + 'bolocam/new/' + maps[0]['name'] + '.fits'
# cluster_struct = fits.getdata(data_file)
hdul = fits.open(data_file)
header = hdul[0].header
naxis = hdul[0].data.shape
# BCAMNORM in units of MJy/sr
bolocam = clus_convert_bolocam(hdul[0].data,
norm=header['BCAMNORM'],
verbose=verbose,
clusname=clusname)
else:
#fetch bolocam data from .sav files in bolocam directory.
data_file = str('bolocam/data/' + maps[0]['name'] + '.sav')
data_dict = scipy.io.readsav(CLUSDATA + data_file, python_dict=True)
cluster_struct = list(data_dict.values())
bolocam, err = clus_convert_bolocam(cluster_struct,
verbose=verbose,
clusname=clusname)
if err:
errmsg = str('clus_convert_bolocam exited with error: ' + err)
if verbose:
print(errmsg)
return None, errmsg
# Set the size of the image for later use
naxis = bolocam[0]['deconvolved_image'][0].shape
naxis = np.array(naxis)
sziso = fits.PrimaryHDU()
temphead = sziso.header
if clusname == 'rxj1347':
# Setting the tempheader for the bolocam data
temphead.set('CRVAL1', header['CRVAL1']) # DEC in deg of ref pixel
temphead.set('CRVAL2', header['CRVAL2']) # RA in deg of ref pixel
temphead.set('CRPIX1', header['CRPIX1']) # ref pixel x
temphead.set('CRPIX2', header['CRPIX2']) # ref pixel y
temphead.set('CD1_1', header['CD1_1']) # Deg/pixel
temphead.set('CD1_2', header['CD1_2']) # Deg/pixel
temphead.set('CD2_1', header['CD2_1']) # Deg/pixel
temphead.set('CD2_2', header['CD2_2']) # Deg/pixel
temphead.set('EPOCH', 2000)
temphead.set('EQUINOX', 2000)
temphead.set('CTYPE1', header['CTYPE1']) # coord type
temphead.set('CTYPE2', header['CTYPE2']) # coord type
else:
# Set reference pizel position
crpix1 = int(naxis[0] / 2)
crpix2 = int(naxis[1] / 2)
# Setting the tempheader for the bolocam data
temphead.set(
'CRVAL1', bolocam[0]['deconvolved_image_ra_j2000_deg'][0][crpix1,
crpix2])
temphead.set(
'CRVAL2', bolocam[0]['deconvolved_image_dec_j2000_deg'][0][crpix1,
crpix2])
temphead.set('CRPIX1', crpix1)
temphead.set('CRPIX2', crpix2)
temphead.set(
'CD1_1',
-bolocam[0]['deconvolved_image_resolution_arcmin'][0] / 60.0)
temphead.set('CD1_2', 0)
temphead.set('CD2_1', 0)
temphead.set(
'CD2_2',
bolocam[0]['deconvolved_image_resolution_arcmin'][0] / 60.0)
temphead.set('EPOCH', 2000)
temphead.set('EQUINOX', 2000)
temphead.set('CTYPE1', 'RA---TAN')
temphead.set('CTYPE2', 'DEC--TAN')
# 14.5 = full size of the bolocam image in pixels ?
x = np.arange(naxis[0]) - 14.5
y = np.arange(naxis[1]) - 14.5
rad = np.zeros((naxis[0], naxis[1]))
for ix in range(naxis[1]):
rad[:, ix] = (np.tile(x[ix], naxis[0])**2 + y**2)**(1 / 2)
# Outer marks the indices of the outside of the circle used to add in the sz effect
n = 0
outer = []
for i in range(len(rad)):
for j in range(len(rad)):
if rad[i, j] > 10:
outer.append(n)
n += 1
# Set up the spectral shape of the sz effect to be appled to the 500um map
if testflag == 0:
if clusname == 'rxj1347':
szmap1 = hdul[0].data
else:
szmap1 = -1 * bolocam[0]['deconvolved_image'][
0] # -1 is needed because SZ amp flips below 217 GHz
szmap1 = (np.array(szmap1)).flatten()
sz_mean = np.mean(szmap1[outer])
szmap2 = [x - sz_mean for x in szmap1]
sz_max = max(szmap2)
szmap = [x / sz_max for x in szmap2]
final_dI = []
for imap in range(mapsize):
# Applying the effect to the 500 um and 350 um bands.
if imap == 2 or imap == 1:
if testflag == 1:
szmap, err = IB_model(maps[imap], params, verbose)
szmap = np.array(szmap)
plt.imshow(szmap)
plt.colorbar().set_label(['Jy'])
plt.title('Clus Add Sziso : IB Model for %s' %
(maps[imap]['name']))
plt.savefig(config.OUTPUT + 'add_sz/%s_ibmodel_%s_%s.png' %
(maps[imap]['name'], maps[imap]['band'], isim))
plt.clf()
naxis = szmap.shape
szmap = szmap.flatten()
nu = 3e5 / clus_get_lambdas((maps[imap]['band']))
if int(isim) == 0: # dI only needs to be calculated once...
dI, errmsg = clus_get_relsz(isim,
nu,
imap,
y=yin,
te=tin,
vpec=0.0) # dI = [MJy/sr]
np.save(
config.OUTPUT + 'add_sz/sim_dI_%s.npy' %
(maps[imap]['band']), dI)
else:
dI = np.load(config.OUTPUT + 'add_sz/sim_dI_%s.npy' %
(maps[imap]['band']))
if errmsg:
if verbose:
new_errmsg = 'Clus_get_relSZ exited with error' + errmsg
return None, new_errmsg
# Combine the spectral shape of the SZ effect, and combine with the peak intensity
# converted to Jy/beam ***Confirm these Units***
if int(isim) == 0:
szin = [x * dI / maps[imap]['calfac'] for x in szmap]
final_dI.append(dI / maps[imap]['calfac'])
else:
szin = [x * dI / maps[imap]['calfac'] for x in szmap]
final_dI.append(dI / maps[imap]['calfac'])
szin = np.reshape(szin, (naxis[0], naxis[1]))
# Have to interpolate to SPIRE map size
hdu = fits.PrimaryHDU(szin, temphead)
hdx = fits.PrimaryHDU(maps[imap]['signal'], maps[imap]['shead'])
if testflag == 0:
szinp = hcongrid(hdu.data, hdu.header, hdx.header)
# need to smooth output with SPIRE PSF
fwhm = maps[imap]['widtha']
pixscale = maps[imap]['pixsize']
retext = round(fwhm * 5.0 / pixscale)
if retext % 2 == 0:
retext += 1
bmsigma = fwhm / math.sqrt(8 * math.log(2))
beam = Gaussian2DKernel(bmsigma / pixscale,
x_size=retext,
y_size=retext,
mode='oversample',
factor=1)
beam *= 1.0 / beam.array.max()
out_map = convolve(szinp, beam, boundary='wrap')
# Combine the original signal with the sz effect
maps[imap]['signal'] = maps[imap]['signal'] + out_map
else:
# maps[imap]['signal'] = maps[imap]['signal'] + szin
'''THIS IS JUST FOR TESTING WITH SZ SIGNAL'''
maps[imap]['signal'] = maps[imap]['error'] + szin
# Used to check the alligned sz effect image
if saveplot:
filename = config.OUTPUT + 'sim_sz/' + maps[imap][
'name'] + '_sze_' + maps[imap]['band'] + '_' + str(
isim) + '.png'
plt.imshow(maps[imap]['signal'])
plt.title('Clus Add Sziso : SZE + Signal Map %s' %
(maps[imap]['band']))
plt.colorbar().set_label('[Jy]')
plt.savefig(filename)
plt.clf()
return maps, None, final_dI
def clus_subtract_xcomps(maps,
sgen=None,
verbose=1,
superplot=1,
nsim=0,
saveplot=1):
err = None
ncols = len(maps)
#check to see if the maps are in the correct order.
if maps[0]['band'] != 'PSW':
err = 'first element of map structure is not PSW, aborting.'
if verbose:
print(err)
return None, err
#make a noise mask and populate source removed map with nans where mask is 1 so they don't effect the fit.
for i in range(len(maps)):
# mask = clus_make_noise_mask(maps, i)
#maps[i]['mask'] = maps[i]['mask'] + mask
if sgen == None:
maps[i]['mask'] = fits.getdata(config.HOME +
'bolocam_mask_%s.fits' %
(maps[i]['band']))
for j in range(maps[i]['signal'].shape[0]):
for k in range(maps[i]['signal'].shape[1]):
if maps[i]['mask'][j, k] == 0:
maps[i]['srcrm'][j, k] = np.nan
if saveplot:
if sgen != None:
filename = config.OUTPUT + 'pcat_residuals/' + maps[i][
'name'] + '_mask_resid_' + maps[i]['band'] + '_' + str(
nsim) + '.png'
else:
filename = config.OUTPUT + 'pcat_residuals/' + maps[i][
'name'] + '_mask_resid_' + maps[i][
'band'] + '_real' + '.png'
plt.imshow(maps[i]['srcrm'])
plt.clim([-0.01, 0.01])
plt.colorbar().set_label('[Jy]')
plt.title('Clus Subtract Xcomps : Masked PCAT Residual %s' %
(maps[i]['band']))
plt.savefig(filename)
plt.clf()
''' For making hists for Jack '''
hda = fits.PrimaryHDU(maps[i]['srcrm'], maps[i]['shead'])
hda.writeto(config.OUTPUT + 'pcat_residuals/' + maps[i]['name'] +
'_mask_resid_' + maps[i]['band'] + '_' + str(nsim) +
'.fits',
overwrite=True)
for i in range(1, ncols):
if verbose:
print('On band %s' % (maps[i]['band']))
#create a new image for the PSW band that is the same shape and beamsize as the reference images
width = sqrt(maps[i]['widtha']**2 -
maps[0]['widtha']**2) / maps[0]['pixsize']
kern = makeGaussian(15, 15, fwhm=width, center=None)
kern = kern / np.sum(kern)
kern = padGaussian(maps[0]['srcrm'], kern)
inmap = convolve_fft(maps[0]['srcrm'], kern)
inmap = maps[0]['srcrm']
maps[0]['xclean'] = maps[0]['srcrm']
xmap = inmap
xmap_align = hcongrid(xmap, maps[0]['shead'], maps[i]['shead'])
# plt.title('HCONGRID')
# plt.imshow(xmap_align)
# plt.show()
plt.imshow(xmap_align)
plt.clim(0, 0.08)
plt.colorbar()
plt.savefig('/home/vaughan/dumb_folder/hcongrid.png')
plt.clf()
hda = fits.PrimaryHDU(xmap_align, maps[i]['shead'])
hda.writeto('/home/vaughan/hcongrid.fits', overwrite=True)
interp_map = interp_band_to_band(maps[1], maps[0])
plt.imshow(interp_map)
plt.clim(0, 0.08)
plt.colorbar()
plt.savefig('/home/vaughan/dumb_folder/interp.png')
plt.clf()
hda = fits.PrimaryHDU(interp_map, maps[i]['shead'])
hda.writeto('/home/vaughan/interp.fits', overwrite=True)
exit()
# plt.title('CUSTOM INTERP')
# plt.imshow(interp_map)
# plt.show()
# plt.imshow(xmap_align)
# plt.colorbar()
# plt.clim(-0.03,0.04)
# plt.savefig('xmap_align_%s.png' %(maps[i]['band']))
# plt.clf()
for j in range(xmap_align.shape[0]):
for k in range(xmap_align.shape[1]):
if np.isnan(xmap_align[j, k]) or np.isnan(maps[i]['srcrm'][j,
k]):
xmap_align[j, k] = 0
maps[i]['srcrm'][j, k] = 0
#now that we have our new PSW image flatten both images
PSW_array = xmap_align.flatten()
ref_array = maps[i]['srcrm'].flatten()
#find the linear fit for PSW vs ref
slope, intercept, r_value, p_value, std_err = linregress(
PSW_array, ref_array)
y = slope * PSW_array + intercept
if superplot or saveplot:
plt.plot(PSW_array, ref_array, 'x', label='Raw Intensities')
plt.plot(PSW_array, y, c='red', label='Linear Regression Fit')
plt.legend()
plt.title('Clus Subtract Xcomps : PSW vs. %s' % (maps[i]['band']))
plt.xlabel('PSW [Jy]')
plt.ylabel('%s [Jy]' % (maps[i]['band']))
if superplot:
plt.show()
elif saveplot:
if sgen != None:
filename = config.OUTPUT + 'corr_comps/' + maps[i][
'name'] + '_xcomps_' + maps[i]['band'] + '_' + str(
nsim) + '.png'
else:
filename = config.OUTPUT + 'corr_comps/' + maps[i][
'name'] + '_xcomps_' + maps[i]['band'] + '_real.png'
plt.savefig(filename)
plt.clf()
#subtract the correlated components from the image
maps[i]['xclean'] = np.empty(maps[i]['srcrm'].shape)
for j in range(maps[i]['xclean'].shape[0]):
for k in range(maps[i]['xclean'].shape[1]):
if maps[i]['srcrm'][j, k] == 0:
maps[i]['xclean'][j, k] = np.nan
else:
maps[i]['xclean'][j, k] = maps[i]['srcrm'][
j, k] - slope * xmap_align[j, k] + intercept
plt.imshow(maps[i]['xclean'])
plt.clim([-0.01, 0.01])
plt.colorbar().set_label('[Jy]')
plt.title('Clus Subtract Xcomp : Correlated Components Removed')
if superplot:
plt.show()
elif saveplot:
if sgen != None:
filename = config.OUTPUT + 'corr_comps/' + maps[i][
'name'] + '_xclean_' + maps[i]['band'] + '_' + str(
nsim) + '.png'
else:
filename = config.OUTPUT + 'corr_comps/' + maps[i][
'name'] + '_xclean_' + maps[i]['band'] + '_real.png'
plt.savefig(filename)
plt.clf()
#subtract the mean of the new map from itself. Why do we do this? we need to figure out what the purpose of this step is.
# for i in range(maps[0]['xclean'].shape[0]):
# for j in range(maps[0]['xclean'].shape[1]):
# maps[0]['xclean'][i,j] = maps[0]['xclean'][i,j] - np.mean(maps[0]['xclean'])
return maps, err
####input parameters#### infile = sys.argv[1] # input tfile = sys.argv[2] # target header outname = sys.argv[3] # output reso = float(sys.argv[4])# in arcsec ############################# #### Main code ############## ############################# hd = fits.getheader(infile) dat = fits.getdata(infile) projhd = fits.getheader(tfile) cornel = np.sqrt(reso**2.-(hd['BMAJ']*3600.)**2.) sigma = cornel/(hd['CDELT2']*3600.)/(2.*np.sqrt(2.*np.log(2.))) ### Gaussian smoothing the image ###### gimage=ndimage.filters.gaussian_filter(dat, sigma, order=0, output=None, mode='reflect', cval=0.0) ### project image #### new_image = hcongrid(gimage, hd, projhd) fits.writeto(outname, new_image, header=projhd,overwrite=True)
def clus_add_sziso_new(maps,
isim,
yin=0,
tin=0,
params=None,
verbose=0,
testflag=0,
saveplot=0,
clusname=None):
#------------------------------------- This part is just for testing
yin = 9.65 * 1e-4
tin = 10.88
errmsg = False
# Now the bolocam data is in the SPIRE format
# we can do a loop over the map adding in the false sz signal
mapsize = len(maps)
if verbose:
print('Fetching BOLOCAM maps')
''' This is just until Jack gives us more new Bolocam templates... '''
if clusname == 'rxj1347':
data_file = CLUSDATA + 'bolocam/new/' + maps[0]['name'] + '.fits'
cluster_struct = fits.getdata(data_file)
data = fits.open(data_file)
header = data[0].header
naxis = cluster_struct.shape
# BCAMNORM in units of MJy/sr
print(header['BCAMNORM'], 'bolo cam norm')
bolocam, err = clus_convert_bolocam(cluster_struct,
norm=header['BCAMNORM'],
verbose=verbose,
clusname=clusname)
else:
#fetch bolocam data from .sav files in bolocam directory.
data_file = str('bolocam/data/' + maps[0]['name'] + '.sav')
data_dict = scipy.io.readsav(CLUSDATA + data_file, python_dict=True)
cluster_struct = list(data_dict.values())
bolocam, err = clus_convert_bolocam(cluster_struct,
verbose=verbose,
clusname=clusname)
if err:
errmsg = str('clus_convert_bolocam exited with error: ' + err)
if verbose:
print(errmsg)
return None, errmsg
# Set the size of the image for later use
naxis = bolocam[0]['deconvolved_image'][0].shape
naxis = np.array(naxis)
sziso = fits.PrimaryHDU()
temphead = sziso.header
if clusname == 'rxj1347':
# Setting the tempheader for the bolocam data
temphead = header
else:
# Set reference pizel position
crpix1 = int(naxis[0] / 2)
crpix2 = int(naxis[1] / 2)
# Setting the tempheader for the bolocam data
temphead.set(
'CRVAL1', bolocam[0]['deconvolved_image_ra_j2000_deg'][0][crpix1,
crpix2])
temphead.set(
'CRVAL2', bolocam[0]['deconvolved_image_dec_j2000_deg'][0][crpix1,
crpix2])
temphead.set('CRPIX1', crpix1)
temphead.set('CRPIX2', crpix2)
temphead.set(
'CD1_1',
-bolocam[0]['deconvolved_image_resolution_arcmin'][0] / 60.0)
temphead.set('CD1_2', 0)
temphead.set('CD2_1', 0)
temphead.set(
'CD2_2',
bolocam[0]['deconvolved_image_resolution_arcmin'][0] / 60.0)
temphead.set('EPOCH', 2000)
temphead.set('EQUINOX', 2000)
temphead.set('CTYPE1', 'RA---TAN')
temphead.set('CTYPE2', 'DEC--TAN')
# 14.5 = full size of the bolocam image in pixels ?
x = np.arange(naxis[0]) - 14.5
y = np.arange(naxis[1]) - 14.5
rad = np.zeros((naxis[0], naxis[1]))
for ix in range(naxis[1]):
rad[:, ix] = (np.tile(x[ix], naxis[0])**2 + y**2)**(1 / 2)
# Outer marks the indices of the outside of the circle used to add in the sz effect
n = 0
outer = []
for i in range(len(rad)):
for j in range(len(rad)):
if rad[i, j] > 10:
outer.append(n)
n += 1
# Set up the spectral shape of the sz effect to be appled to the 500um map
if testflag == 0:
if clusname == 'rxj1347':
szmap1 = bolocam
else:
szmap1 = -1 * bolocam[0]['deconvolved_image'][
0] # -1 is needed because SZ amp flips below 217 GHz
szmap1 = (np.array(szmap1)).flatten()
sz_mean = np.mean(szmap1[outer])
szmap2 = [x - sz_mean for x in szmap1]
sz_max = max(szmap2)
szmap = [x / sz_max for x in szmap2]
final_dI = []
if clusname == 'rxj1347': # need to calculate the dI for Bolocam using new maps
# if os.path.isfile(config.HOME + 'lookup/rxj1347_bol_lookup.npy') : # use lookup file instead of running szpack every time
# dI_bolo = np.load(config.HOME + 'lookup/rxj1347_bol_lookup.npy')
# else :
szmap1 = bolocam
dI_bolo, thisx, JofXout, xout, errmsg = clus_get_relsz(
isim,
3e5 / clus_get_lambdas('BOLOCAM'),
'BOLOCAM',
y=yin,
te=tin,
vpec=0.0) # dI = [MJy/sr]
# np.save(config.HOME + 'lookup/rxj1347_bol_lookup.npy',dI_bolo)
szmap1 = np.array(szmap1).flatten()
szmap1 = [x / dI_bolo for x in szmap1]
for imap in range(mapsize):
# Applying the effect to the 500 um and 350 um bands.
# if imap == 2 or imap == 1:
if testflag == 1:
szmap, err = IB_model(maps[imap], params, verbose)
szmap = np.array(szmap)
plt.imshow(szmap, origin=0)
plt.colorbar().set_label(['Jy'])
plt.title('Clus Add Sziso : IB Model for %s' %
(maps[imap]['name']))
plt.savefig(config.OUTPUT + 'add_sz/%s_ibmodel_%s_%s.png' %
(maps[imap]['name'], maps[imap]['band'], isim))
plt.clf()
naxis = szmap.shape
szmap = szmap.flatten()
nu = 3e5 / clus_get_lambdas((maps[imap]['band']))
if int(isim) == 0: # dI only needs to be calculated once...
dI, thisx, JofXout, xout, errmsg = clus_get_relsz(
isim, nu, imap, y=yin, te=tin, vpec=0.0) # dI = [MJy/sr]
np.save(config.OUTPUT + 'add_sz/sim_dI_%s.npy' %
(maps[imap]['band']),
dI) #just the expected amplitude at spire wavelength
np.save(config.OUTPUT + 'add_sz/input_dI_%s.npy' %
(maps[imap]['band']), dI * -.94 /
dI_bolo) #input amplitude saved for comparison in fit_sz
else:
dI = np.load(config.OUTPUT + 'add_sz/sim_dI_%s.npy' %
(maps[imap]['band']))
if errmsg:
if verbose:
new_errmsg = 'Clus_get_relSZ exited with error' + errmsg
return None, new_errmsg
# Combine the spectral shape of the SZ effect, and combine with the peak intensity
# converted to Jy/beam
hdtemp = fits.PrimaryHDU(szmap1, temphead)
hdtemp.writeto(config.OUTPUT +
'sz_fit/bolo_fit_template%s.fits' % isim,
overwrite=True)
szin = [x * dI / maps[imap]['calfac'] for x in szmap1] #
final_dI.append(dI / maps[imap]['calfac'])
szin = np.reshape(szin, (naxis[0], naxis[1]))
# Have to interpolate to SPIRE map size
hdu = fits.PrimaryHDU(szin, temphead)
hdx = fits.PrimaryHDU(maps[imap]['signal'], maps[imap]['shead'])
if testflag == 0:
szinp = hcongrid(hdu.data, hdu.header, hdx.header)
# need to smooth output with SPIRE PSF
fwhm = maps[imap]['widtha']
pixscale = maps[imap]['pixsize']
retext = round(fwhm * 5.0 / pixscale)
if retext % 2 == 0:
retext += 1
bmsigma = fwhm / math.sqrt(8 * math.log(2))
beam = Gaussian2DKernel(bmsigma / pixscale,
x_size=retext,
y_size=retext,
mode='oversample',
factor=1)
beam *= 1.0 / beam.array.max()
out_map = convolve(szinp, beam, boundary='wrap')
# Combine the original signal with the sz effect
maps[imap]['signal'] = maps[imap]['signal'] + out_map
'''THIS IS JUST FOR TESTING WITH SZ SIGNAL'''
# maps[imap]['signal'] = out_map
elif testflag == 2:
szinp = hcongrid(hdu.data, hdu.header, hdx.header)
# need to smooth output with SPIRE PSF
fwhm = maps[imap]['widtha']
pixscale = maps[imap]['pixsize']
retext = round(fwhm * 5.0 / pixscale)
if retext % 2 == 0:
retext += 1
bmsigma = fwhm / math.sqrt(8 * math.log(2))
beam = np.asarray(
Gaussian2DKernel(bmsigma / pixscale,
x_size=retext,
y_size=retext,
mode='oversample',
factor=1))
n_beam = beam / np.sum(
beam
) #the input bolocam map is a surface brightness so we need to normalize over area
out_map = convolve(szinp, n_beam, boundary='wrap')
maps[imap]['signal'] = maps[imap]['signal'] + out_map
# maps[imap]['srcrm'] = out_map #for testing
# maps[imap]['signal'] = maps[imap]['error'] + out_map
# Used to check the alligned sz effect image
if saveplot:
filename = config.OUTPUT + 'sim_sz/' + maps[imap][
'name'] + '_sze+signal_' + maps[imap]['band'] + '_' + str(
isim) + '.png'
plt.imshow(maps[imap]['signal'], origin=0)
plt.title('Clus Add Sziso : SZE + Signal Map %s' %
(maps[imap]['band']))
plt.colorbar().set_label('[Jy]')
plt.clim(-0.1, 0.1)
plt.savefig(filename)
plt.clf()
filename = config.OUTPUT + 'sim_sz/' + maps[imap][
'name'] + '_sze_' + maps[imap]['band'] + '_' + str(
isim) + '.png'
plt.imshow(out_map, origin=0)
plt.title('Clus Add Sziso : SZE + Signal Map %s' %
(maps[imap]['band']))
plt.clim(-0.1, 0.1)
plt.colorbar().set_label('[Jy]')
plt.savefig(filename)
plt.clf()
savefile = config.OUTPUT + 'sim_sz/' + maps[imap][
'name'] + '_sze_' + maps[imap]['band'] + '_' + str(
isim) + '.fits'
hda = fits.PrimaryHDU(maps[imap]['signal'], maps[imap]['shead'])
hda.writeto(savefile, overwrite=True)
return maps, None, final_dI
def smooth_images_toresolution(target_resolution,
globs=["destripe*P[LMS]W*fits",
"destripe*blue*fits",
"destripe*red*fits"],
reject_regex='smooth|smregrid', verbose=True,
skip_existing=True, regrid=True,
target_header=None, regrid_order=1,
clobber=False, **kwargs):
"""
Smooth a series of images to the same resolution. The output files will be
of the form ``{inputfilename}_smooth.fits`` and
``{inputfilename}_smregrid.fits``
Parameters
----------
target_resolution : `~astropy.units.quantity.Quantity`
A degree-equivalent value that specifies the beam size in the output
image
globs : list
A list of strings to pass into `~glob.glob`. All files found will be
smoothed and possibly regridded.
reject_regex : str
A regular expression to apply to each discovered file to choose whether
to reject it. For example, if you've run this function once, you'll
have files named ``file.fits`` and ``file_smooth.fits`` that you don't
want to re-smooth and re-regrid.
verbose : bool
Print messages at each step?
skip_existing : bool
If the output smooth file is found and this is True, skip and move on
to the next
regrid : bool
Regrid the file? If True, ``target_header`` is also required
regrid_order : int
The order of the regridding operation. Regridding is performed with
interpolation, so 0'th order means nearest-neighbor and 1st order means
bilinear. Regridding is done with
`FITS_tools.hcongrid.hcongrid`
clobber : bool
Overwrite files if they exist?
kwargs : dict
Passed to `smooth_image_toresolution`
Raises
------
ValueError
If ``target_header`` is not specified but ``regrid`` is
Returns
-------
Nothing. All output is to disk
"""
for fn in [x for g in globs for x in glob.glob(g)
if not re.search(reject_regex, x)]:
if verbose:
print("Reading file {0}".format(fn),)
outnum = int(target_resolution.to(u.arcsec).value)
smoutfn = fn.replace(".fits", "_smooth{0:d}.fits".format(outnum))
if os.path.exists(smoutfn):
if skip_existing:
if verbose:
print("Skipping {0}".format(fn))
continue
smhduL = smooth_image_toresolution(fn, smoutfn, target_resolution,
verbose=verbose, clobber=clobber,
**kwargs)
smhdu = smhduL[0]
if regrid:
if target_header is None:
raise ValueError("Must specify a target header if regridding.")
newimage = hcongrid(smhdu.data, smhdu.header, target_header,
order=regrid_order)
rgoutfn = fn.replace(".fits", "_smregrid{0:d}.fits".format(outnum))
print("Regridding {0} to {1}".format(smoutfn, rgoutfn))
newhdu = fits.PrimaryHDU(data=newimage, header=target_header)
newhdu.writeto(rgoutfn, clobber=clobber)
def clus_new_fitsz(maps, saveplot=0, nsim=None):
band = ['PSW', 'PMW', 'PLW']
fit = [0] * 3
data_file = config.CLUSDATA + 'bolocam/new/' + maps[0]['name'] + '.fits'
base_model = fits.getdata(data_file)
data = fits.open(data_file)
header = data[0].header
ylims = [[-0.01, 0.01], [0, 0.1], [0.1, 0.4]]
for i in range(3):
#Error Analysis -------
error_map = maps[i]['error']
error_test = np.multiply(error_map, maps[i]['calfac'])
error = error_map.flatten()
noise_map = maps[i]['noise']
# plt.imshow(noise_map, origin='lower')
# plt.title('Bolocam Template_%s' % maps[i]['band'])
# plt.colorbar()
# plt.savefig(config.OUTPUT + 'sz_fit/fitsz_in_temp%s%s.png' % (maps[i]['band'], nsim))
# plt.clf()
# noise_map = np.random.normal(loc=0, scale=1.0,size=(maps[i]['signal'].shape[0],maps[i]['signal'].shape[1]))
noise = noise_map.flatten()
for j in range(len(noise)):
if not math.isnan(error[j]): #Jy/beam
noise[j] = maps[i]['calfac'] * noise[j]
noise_map = np.reshape(noise, maps[i]['signal'].shape)
bolo_temp = hcongrid(base_model, header, maps[i]['shead'])
fwhm = maps[i]['widtha']
pixscale = maps[i]['pixsize']
retext = round(fwhm * 5.0 / pixscale)
if retext % 2 == 0:
retext += 1
bmsigma = fwhm / math.sqrt(8 * math.log(2))
beam = Gaussian2DKernel(bmsigma / pixscale,
x_size=retext,
y_size=retext,
mode='oversample',
factor=1)
beam *= 1 / np.sum(beam.array)
model = convolve(bolo_temp, beam, boundary='wrap')
for j in range(maps[i]['signal'].shape[0]):
for k in range(maps[i]['signal'].shape[1]):
if maps[i]['mask'][j, k] == 0:
model[j, k] = np.nan
maps[i]['srcrm'][j, k] = np.nan
noise_map[j, k] = np.nan
else:
maps[i]['srcrm'][j, k] *= maps[i]['calfac']
image = maps[i]['srcrm']
#
# plt.imshow(maps[i]['srcrm'], origin='lower')
# plt.title('SZ without noise_ %s' % maps[i]['band'])
# plt.colorbar()
# plt.savefig(config.OUTPUT + 'sz_fit/sz_no_noise%s%s.png' % (maps[i]['band'], nsim))
# plt.clf()
#
# plt.imshow(noise_map, origin='lower')
# plt.title('Noise Map_%s' % maps[i]['band'])K
# plt.colorbar()
# plt.savefig(config.OUTPUT + 'sz_fit/noise_map%s%s.png' % (maps[i]['band'], nsim))
# plt.clf()
plt.imshow(image, origin='lower', clim=(-0.1, 0.1))
plt.title('Input SZ effect %s %s' % (maps[i]['band'], nsim))
plt.colorbar()
plt.savefig(config.OUTPUT + 'sz_fit/fitsz_in_map%s%s.png' %
(maps[i]['band'], nsim))
plt.clf()
plt.imshow(model, origin='lower')
plt.title('Bolocam Template %s' % maps[i]['band'])
plt.colorbar()
plt.savefig(config.OUTPUT + 'sz_fit/fitsz_in_temp%s%s.png' %
(maps[i]['band'], nsim))
plt.clf()
flat_im = image.flatten()
flat_mod = model.flatten()
x_data = [
flat_mod[i] for i in range(len(flat_mod)) if not np.isnan(noise[i])
]
y_data = [
flat_im[i] for i in range(len(flat_im)) if not np.isnan(noise[i])
]
# noise_data = [1 / n for n in noise_map.flatten() if not np.isnan(n)] #polyfit wants weights as 1 / sigma
sigma_data = [n for n in noise if not np.isnan(n)]
print(len(x_data), len(y_data), len(sigma_data))
# z, cov = np.polyfit(x_data,y_data,1, cov=True),# w = noise_data)
params, cov = curve_fit(fitting_func, x_data, y_data, sigma=sigma_data)
e = np.sqrt(np.diag(cov))
# p = np.poly1d(z)
intercept = params[1]
intercept_error = e[1]
slope = params[0]
slope_error = e[0]
y_fit = [slope * x + intercept for x in x_data]
fit[i] = params
chi_square = chi_square_test(y_data, y_fit, sigma_data)
red_chi_square = chi_square / len(y_data)
if saveplot:
input_DI = np.load(config.OUTPUT + 'add_sz/input_dI_%s.npy' %
(band[i]))
# plt.ylim(ylims[i])
plt.plot(
x_data,
y_fit,
c='red',
label=
'dI = %.4E +/- %.3E \n offset = %.1E +/- %.1E \n chi square: %.3E \n red chi square %.3E'
% (slope, slope_error, intercept, intercept_error, chi_square,
red_chi_square))
plt.scatter(x_data,
y_data,
alpha=0.5,
label='Image Data: Input DI : %.4E' % input_DI)
in_y = [input_DI * x for x in x_data]
# plt.plot(x_data, in_y, label='input dI = %.4E' % input_DI, c='green')
sigma = (slope - input_DI) / slope_error
print('uncertainty in fits', slope_error, intercept_error)
print('Slope is %.2E sigma from the expected value' % sigma)
#these next few lines is for testing only
plt.legend()
plt.xlabel('Model Flux [Unitless]')
plt.ylabel('Image Flux [MJy/Sr]')
plt.title('SZ Fit for %s %s' % (maps[i]['band'], nsim))
# plt.xlim(-0.005,0.005)
plt.savefig(
config.OUTPUT + 'sz_fit/rxj1347_%s_%s.png' %
(maps[i]['band'], nsim)) #rxj1347 placeholder for clusname.
plt.clf()
return fit