def plot_fits(datafile, fitfile, fitp, output, numlines):
hdu = pyfits.open(datafile)[0]
data = hdu.data
header = hdu.header
cdelt = header['CDELT1']
crpix = header['CRPIX1']
crval = header['CRVAL1']
print cdelt
wave = (np.arange(data.shape[1]) + crpix-1)*cdelt + crval
fits = pyfits.open(fitfile)[0].data
d = i2p.parse_fitprofs(fitp,numlines)
if numlines == 3:
x = np.arange(rlist[1][0], rlist[1][1])
else:
x = np.arange(rlist[0][0], rlist[0][1])
idx = np.where((wave >= x.min()) & (wave <= x.max()))[0]
wave = wave[idx]
data = data[:,idx]
fits = fits[:,idx]
pp = PDF(output)
for ap in range(d[1].shape[0]):
print ap+1
y = np.zeros(wave.size)
ax = plt.figure().add_subplot(111)
ax.set_xlabel('Wavelength')
ax.set_ylabel('Flux')
ax.set_title('Ap {}'.format(ap+1))
back = []
for i in range(numlines):
back.append(d[6][ap,i])
tmp = np.exp(-0.5*(wave - d[1][ap,i])**2/(d[2][ap,i]/2.355)**2)
tmp *= d[7][ap,i]/np.max(tmp)
# tmp *= d[4][ap,i]/np.sum(tmp)
ax.plot(wave,tmp+back[-1],'r:')
y += tmp
if numlines == 3:
ratio = d[7][ap,2]/d[7][ap,1]
ax.text(0.8,0.8,'ratio = {:4.3f}'.format(ratio),transform=ax.transAxes)
ax.plot(wave,y+np.mean(back),'b',alpha=0.7)
ax.plot(wave,data[ap,:],'k')
ax.plot(wave,fits[ap,:]+np.mean(back),'r')
pp.savefig(ax.figure)
pp.close()
plt.close('all')
return
def get_results(output, threshold=3., filename=''):
"""Parse fitprof output and display results
The line centers are taken from the output of fitprof. For each line specified in the module header the average offset and stddev across all fibers in the IFU is computed. Output is a textfile and a plot of accuracy and stochasticity as a function of wavelength.
Parameters
----------
output : str
Name of the output text file. This file will contain the mean, offset, stddev, and number of rejected apertures.
threshold : float, optional
Threshold value for iterative sigma clipping in mean across IFU. The total number of rejected fibers will be recorded in the output file.
Returns
-------
None :
The result is a text file containing the results and a plot containing the accuracy as a function of wavelength.
Notes
-----
Right now the plot is hardcoded to be writting to WLC.png
"""
fig = plt.figure()
acax = fig.add_subplot(211)
acax.set_xticklabels([])
acax.set_ylabel('Mismatch [AA]')
stax = fig.add_subplot(212)
stax.set_xlabel('Wavelength')
stax.set_ylabel('IFU std')
with open(output,'a') as f:
f.write('# {}\n'.format(time.asctime()))
f.write('# {}\n'.format(filename))
for l, n, c in zip(llist,numlist,centlist):
proffile = '{}.fitp'.format(l)
d = i2p.parse_fitprofs(proffile,n)[1]
mean = np.mean(d,axis=0)
std = np.std(d,axis=0)
rejidx = np.where(np.abs(d - mean) > std*threshold)
i = 0
while rejidx[0].size != 0:
d = np.delete(d,rejidx[0],axis=0)
mean = np.mean(d,axis=0)
std = np.std(d,axis=0)
rejidx = np.where(np.abs(d - mean) > std*threshold)
i += rejidx[0].size
diff = mean - c
outstr = '{:} ({:}):\n\t{:>7}: {:}\n\t{:>7}: {:}\n\t{:>7}: {:}\n\t{:>7}: {:}\n'.\
format(l,c,'numrej',i,'mean',mean,'diff',diff,'std',std)
prtstr = ''
for j in range(len(c)):
prtstr += '{} {} {}\n'.format(c[j],diff[j],std[j])
print prtstr
f.write(outstr)
acax.plot(c,diff,'.k')
stax.plot(c,std,'.k')
f.write('\n\n')
fig.subplots_adjust(hspace=0.0001)
acax.set_xlim(*stax.get_xlim())
plt.savefig('WLC.png')
return
def make_balmer_model(Hafitp, location, velocities, output, tauV_coeffs=[-1.06,3.78], #for ma11
dispdata = '/d/monk/eigenbrot/WIYN/14B-0456/anal/disp/GP_disp_batch_avg_int.fits'):
#bc03 coeffs = [-0.91,4.15]
# tauV = np.loadtxt(balmerD, usecols=(1,), unpack=True)
tauV = np.poly1d(tauV_coeffs)
sizes, z = np.loadtxt(location, usecols=(1,5), unpack=True)
vels = np.loadtxt(velocities, usecols=(1,), unpack=True)
numap = vels.size
Ha_data = i2p.parse_fitprofs(Hafitp,3)
Ha_flux = Ha_data[4][:,0]/1e17 #- 1e-16
#It's been long enough I think I can just use these magic numbers
wave = np.arange(2011)*2.1 + 3340.
##wave = np.arange(1428)*2.1 + 3800. #for comparing to contsub data
m_wave = np.logspace(np.log10(3340), np.log10(7600), 5000)
mpix = np.mean(np.diff(m_wave)/m_wave[1:]*3e5) #size of 1 model pixel in km/s
balmer_flux = np.zeros((numap,m_wave.size))
#Read in disp data
disph = pyfits.open(dispdata)[0]
disp_head = disph.header
disp_data = disph.data
disp_numwave = disp_data.shape[1]
disp_wave = np.arange(disp_numwave)*disp_head['CDELT1'] + disp_head['CRVAL1']
disp_arr = np.zeros((5,m_wave.size))
for d in range(5):
disp_arr[d,:] = np.interp(m_wave,disp_wave,disp_data[d,:])/2.355
sized = {0.937: 0,
1.406: 1,
1.875: 2,
2.344: 3,
2.812: 4}
cents = [6563, 4861, 4341, 4102, 3970]
ratios = np.array([2.86, 1, 0.47, 0.25, 0.16])/2.86 #dividing here doesn't really matter
for c, r in zip(cents, ratios):
idx = np.argmin(np.abs(m_wave - c))
balmer_flux[:,idx] = r
final_model = np.zeros((numap,wave.size))
for i in range(numap):
print i
wave_red = wave * (1 + vels[i]/3e5)
if np.isnan(tauV[i]):
red = np.ones(m_wave.size)
else:
print 'z = {}; tau = {}'.format(z[i],tauV(np.abs(z[i])))
red = np.exp(-1 * tauV(np.abs(z[i]))*(m_wave/5500.)**(-0.7))
balmer_flux[i,:] *= red
did = sized[sizes[i]]
sigma_pix = disp_arr[did,:]/mpix
balmer_flux[i,:] = mconv(balmer_flux[i,:],sigma_pix)
# Haid = np.argmin(np.abs(m_wave - cents[0]))
# mHa_peak = balmer_flux[i,Haid]
Haid = np.where((m_wave > cents[0] - 30) & (m_wave < cents[0] + 30))[0]
mHa_flux = np.sum(balmer_flux[i,Haid])
print Ha_flux[i], mHa_flux
balmer_flux[i,:] *= Ha_flux[i]/mHa_flux
final_model[i,:] = np.interp(wave,m_wave,balmer_flux[i,:])
final_model[i,:] = np.interp(wave,wave_red,final_model[i,:])
final_hdu = pyfits.PrimaryHDU(final_model)
final_hdu.header.update('CRPIX1',1)
final_hdu.header.update('CRVAL1',3340.)
final_hdu.header.update('CDELT1',2.1)
final_hdu.writeto(output,clobber=True)
def get_results(pointing, output):
"""Parse fitprof output and display results
The line centers are taken from the output of fitprof. For each line specified in the module header the average offset and stddev across all fibers in the IFU is computed. Output is a textfile and a plot of accuracy and stochasticity as a function of wavelength.
Parameters
----------
output : str
Name of the output text file. This file will contain the mean, offset, stddev, and number of rejected apertures.
threshold : float, optional
Threshold value for iterative sigma clipping in mean across IFU. The total number of rejected fibers will be recorded in the output file.
Returns
-------
None :
The result is a text file containing the results and a plot containing the accuracy as a function of wavelength.
Notes
-----
Right now the plot is hardcoded to be writting to WLC.png
"""
print 'Consolidating measurements'
fig = plt.figure()
ax = fig.add_subplot(111)
ax.set_ylabel(r'$\tau_{\mathrm{V,balm}}$')
ax.set_xlabel('Aperture number')
ax.set_title(time.asctime())
ax.set_ylim(-10,15)
tHad = i2p.parse_fitprofs('P{}_Ha.fitp'.format(pointing),3)
Haf = tHad[4][:,0]
Hafe = tHad[5][:,0]
tHBd = i2p.parse_fitprofs('P{}_HB.fitp'.format(pointing),1)
HBf = tHBd[4][:,0]
HBfe = tHBd[5][:,0]
#Correct for error over/under-estimateion
datafile = 'P{}_contsub.ms.fits'.format(pointing)
hdu = pyfits.open(datafile)[0]
data = hdu.data
header = hdu.header
cdelt = header['CDELT1']
crpix = header['CRPIX1']
crval = header['CRVAL1']
wave = (np.arange(data.shape[1]) + crpix-1)*cdelt + crval
idx = np.where((wave > rlist[1][0]) & (wave < rlist[1][1]))[0]
tflux = np.mean(data[:,idx],axis=1)
ecorr = np.sqrt(0.289*tflux - 2.811)
print tflux, ecorr
# raw_input()
# Hafe *= ecorr
# HBfe *= ecorr
#Correct for underfit [NII] line
Hacorr = tHad[4][:,1] - tHad[4][:,2]/3.
Haf += Hacorr
ratio = Haf/HBf
ratio_e = np.sqrt((Hafe/HBf)**2 + (HBfe*Haf/HBf**2)**2)
# EBV = 1.97*np.log10(ratio/2.86)
# Av = 4.05*EBV
# tauV = Av/1.086
# Av_e = 1.97*4.05*ratio_e/(1.086*ratio*np.log(10))
tauV = 4.84*np.log(ratio/2.86)
tauV_e = 4.84/ratio*ratio_e
print 'Plotting'
ax.errorbar(np.arange(ratio.size)+1,tauV,yerr=tauV_e,fmt='.')
fig.savefig(output+'.png')
with open(output+'.txt','w') as f:
f.write('# Written on {}\n'.format(time.asctime()))
f.write('# With Charlot and Fall extinction law\n')
f.write('#\n#{:>3}{:>10}{:>10}\n\n'.format('Ap','Tau_V','TauV_err'))
for i in range(tauV.size):
f.write('{:4}{:10.2f}{:10.2f}\n'.format(i+1,tauV[i],tauV_e[i]))
return Haf, Hafe, HBf, HBfe, ratio, ratio_e
