def loadStd(objectName,wvlBinEdges):
#import the known spectrum of the calibrator and rebin to the histogram parameters given
#must be imported into array with dtype float so division later does not have error
std = MKIDStd.MKIDStd()
a = std.load(objectName)
a = std.normalizeFlux(a)
x = a[:,0]
y = a[:,1]
realSpectra = y
realSpectraWvl = x
newa = rebin(x,y,wvlBinEdges)
binnedSpectra = newa[:,1]
return binnedSpectra
def cleanSpectrum(x,y,objectName, wvlBinEdges):
#locations and widths of absorption features in Angstroms
#features = [3890,3970,4099,4340,4860,6564,6883,7619]
#widths = [50,50,50,50,50,50,50,50]
#for i in xrange(len(features)):
# #check for absorption feature in std spectrum
# ind = np.where((x<(features[i]+15)) & (x>(features[i]-15)))[0]
# if len(ind)!=0:
# ind = ind[len(ind)/2]
# #if feature is found (flux is higher on both sides of the specified wavelength where the feature should be)
# if y[ind]<y[ind+1] and y[ind]<y[ind-1]:
# #cut out width[i] around feature[i]
# inds = np.where((x >= features[i]+widths[i]) | (x <= features[i]-widths[i]))
# x = x[inds]
# y = y[inds]
#fit a tail to the end of the spectrum to interpolate out to desired wavelength in angstroms
fraction = 2.1/3
newx = np.arange(int(x[fraction*len(x)]),20000)
slopeguess = (np.log(y[-1])-np.log(y[fraction*len(x)]))/(x[-1]-x[fraction*len(x)])
print "Guess at exponential slope is %f"%(slopeguess)
guess_a, guess_b, guess_c = float(y[fraction*len(x)]), x[fraction*len(x)], slopeguess
guess = [guess_a, guess_b, guess_c]
fitx = x[fraction*len(x):]
fity = y[fraction*len(x):]
exp_decay = lambda fx, A, x0, t: A * np.exp((fx-x0) * t)
params, cov = curve_fit(exp_decay, fitx, fity, p0=guess, maxfev=2000)
A, x0, t= params
print "A = %s\nx0 = %s\nt = %s\n"%(A, x0, t)
best_fit = lambda fx: A * np.exp((fx-x0)*t)
calcx = np.array(newx,dtype=float)
newy = best_fit(calcx)
#func = interpolate.splrep(x[fration*len(x):],y[fraction*len(x):],s=smooth)
#newx = np.arange(int(x[fraction*len(x)]),self.wvlBinEdges[-1])
#newy = interpolate.splev(newx,func)
wl = np.concatenate((x,newx[newx>max(x)]))
flux = np.concatenate((y,newy[newx>max(x)]))
#new method, rebin data to grid of wavelengths generated from a grid of evenly spaced energy bins
#R=7.0 at 4500
#R=E/dE -> dE = R/E
dE = 0.3936 #eV
start = 1000 #Angs
stop = 25000 #Angs
enBins = ObsFile.makeWvlBins(dE,start,stop)
rebinned = rebin(wl,flux,enBins)
re_wl = rebinned[:,0]
re_flux = rebinned[:,1]
#plt.plot(re_wl,re_flux,color='r')
re_wl = re_wl[np.isnan(re_flux)==False]
re_flux = re_flux[np.isnan(re_flux)==False]
start1 = wvlBinEdges[0]
stop1 = wvlBinEdges[-1]
#regrid downsampled data
new_wl = np.arange(start1,stop1)
#print re_wl
#print re_flux
#print new_wl
#weight=1.0/(re_flux)**(2/1.00)
print len(re_flux)
weight = np.ones(len(re_flux))
#decrease weights near peak
ind = np.where(re_flux == max(re_flux))[0]
weight[ind] = 0.3
for p in [1,2,3]:
if p==1:
wt = 0.3
elif p==2:
wt = 0.6
elif p==3:
wt = 0.7
try:
weight[ind+p] = wt
except IndexError:
pass
try:
if ind-p >= 0:
weight[ind-p] = wt
except IndexError:
pass
#change weights to set how tightly fit must match data points
#weight[-4:] = 1.0
weight = [0.7,0.7,0.7,0.7,0.7,0.7,0.7,0.7,0.7]
#print len(weight)
#weight = re_flux/min(re_flux)
#weight = 1.0/weight
#weight = weight/max(weight)
print weight
f = interpolate.splrep(re_wl,re_flux,w=weight,k=3,s=max(re_flux)**200)
new_flux = interpolate.splev(new_wl,f,der=0)
return new_wl, new_flux
def cleanSpectrum(x,y,objectName, wvlBinEdges):
#locations and widths of absorption features in Angstroms
#features = [3890,3970,4099,4340,4860,6564,6883,7619]
#widths = [50,50,50,50,50,50,50,50]
#for i in xrange(len(features)):
# #check for absorption feature in std spectrum
# ind = np.where((x<(features[i]+15)) & (x>(features[i]-15)))[0]
# if len(ind)!=0:
# ind = ind[len(ind)/2]
# #if feature is found (flux is higher on both sides of the specified wavelength where the feature should be)
# if y[ind]<y[ind+1] and y[ind]<y[ind-1]:
# #cut out width[i] around feature[i]
# inds = np.where((x >= features[i]+widths[i]) | (x <= features[i]-widths[i]))
# x = x[inds]
# y = y[inds]
#fit a tail to the end of the spectrum to interpolate out to desired wavelength in angstroms
fraction = 0 #4.0/5.0
newx = np.arange(int(x[fraction*len(x)]),20000)
#slopeguess = (np.log(y[-1])-np.log(y[fraction*len(x)]))/(x[-1]-x[fraction*len(x)])
#print "Guess at exponential slope is %f"%(slopeguess)
#guess_a, guess_b, guess_c = float(y[fraction*len(x)]), x[fraction*len(x)], slopeguess
#guess = [guess_a, guess_b, guess_c]
fitx = x[fraction*len(x)::]
fity = y[fraction*len(x)::]
#exp_decay = lambda fx, A, x0, t: A * np.exp((fx-x0) * t)
#params, cov = curve_fit(exp_decay, fitx, fity, p0=guess, maxfev=2000)
#A, x0, t= params
#print "A = %s\nx0 = %s\nt = %s\n"%(A, x0, t)
#best_fit = lambda fx: A * np.exp((fx-x0)*t)
#calcx = np.array(newx,dtype=float)
#newy = best_fit(calcx)
#normalizing
norm = fity.max()
fity/=norm
guess_a, guess_b = 1/(2*h*c**2/1e-9), 5600 #Constant, Temp
guess = [guess_a, guess_b]
blackbody = lambda fx, N, T: N * 2*h*c**2 / (fx)**5 * (exp(h*c/(k*T*(fx))) - 1)**-1 # Planck Law
#blackbody = lambda fx, N, T: N*2*c*k*T/(fx)**4 #Rayleigh Jeans tail
#blackbody = lambda fx, N, T: N*2*h*c**2/(fx**5) * exp(-h*c/(k*T*fx)) #Wein Approx
params, cov = curve_fit(blackbody, fitx*1.0e-8, fity, p0=guess, maxfev=2000)
N, T= params
print "N = %s\nT = %s\n"%(N, T)
best_fit = lambda fx: N * 2*h*c**2 / (fx)**5 * (exp(h*c/(k*T*(fx))) - 1)**-1 #Planck Law
#best_fit = lambda fx: N*2*c*k*T/(fx)**4 # Rayleigh Jeans Tail
#best_fit = lambda fx: N*2*h*c**2/(fx**5) * exp(-h*c/(k*T*fx)) #Wein Approx
calcx = np.array(newx,dtype=float)
bbfit = best_fit(calcx*1.0E-8)
calcx = np.array(newx,dtype=float)
newy = best_fit(calcx*1.0E-8)
fity*=norm
newy*=norm
plt.plot(calcx[3.0*len(fitx)/4.0::],newy[3.0*len(fitx)/4.0::]*1E15,linestyle='--',linewidth=2, color="black",alpha=0.5) #plot fit
#func = interpolate.splrep(x[fration*len(x):],y[fraction*len(x):],s=smooth)
#newx = np.arange(int(x[fraction*len(x)]),self.wvlBinEdges[-1])
#newy = interpolate.splev(newx,func)
wl = np.concatenate((x,newx[newx>max(x)]))
flux = np.concatenate((y,newy[newx>max(x)]))
#new method, rebin data to grid of wavelengths generated from a grid of evenly spaced energy bins
#R=7.0 at 4500
#R=E/dE -> dE = R/E
dE = 0.3936 #eV
start = 1000 #Angs
stop = 25000 #Angs
enBins = ObsFile.makeWvlBins(dE,start,stop)
rebinned = rebin(wl,flux,enBins)
re_wl = rebinned[:,0]
re_flux = rebinned[:,1]
plt.plot(re_wl,re_flux*1E15,linestyle="o", marker="o",markersize=6) #plot rebinned spectrum with exp tail
re_wl = re_wl[np.isnan(re_flux)==False]
re_flux = re_flux[np.isnan(re_flux)==False]
start1 = wvlBinEdges[0]
stop1 = wvlBinEdges[-1]
#regrid downsampled data
new_wl = np.arange(start1,stop1)
#print re_wl
#print re_flux
#print new_wl
#weight=1.0/(re_flux)**(2/1.00)
print len(re_flux)
weight = np.ones(len(re_flux))
#decrease weights near peak
ind = np.where(re_flux == max(re_flux))[0]
weight[ind] = 0.3
for p in [1,2,3]:
if p==1:
wt = 0.3
elif p==2:
wt = 0.6
elif p==3:
wt = 0.7
try:
weight[ind+p] = wt
except IndexError:
pass
try:
if ind-p >= 0:
weight[ind-p] = wt
except IndexError:
pass
#change weights to set how tightly fit must match data points
#weight[-4:] = 1.0
weight = 0.7*np.ones((len(re_wl)),dtype=float)
#weight = [0.7,0.7,0.7,0.7,0.7,0.7,0.7,0.7,0.7]
#print len(weight)
#weight = re_flux/min(re_flux)
#weight = 1.0/weight
#weight = weight/max(weight)
print weight
f = interpolate.splrep(re_wl,re_flux,w=weight,k=2,s=0)#max(re_flux)**300)
new_flux = interpolate.splev(new_wl,f,der=0)
return new_wl, new_flux
y = np.array(fdata[:,1])/scale #flux in ergs/s/cm^2/Angs
print y
y = y/(h*c/x) #flux in counts/s/cm^2/Angs
print y
print "Loaded standard spectrum of %s"%(objectName)
newwl, newflux = cleanSpectrum(x,y,objectName,wvlBinEdges)
print newwl
print newflux
#plot these to debug std spectrum fitting
#plt.plot(newwl,newflux)
#plt.plot(x,y)
#plt.show()
newa = rebin(newwl,newflux,wvlBinEdges)
x = newa[:,0]
y = newa[:,1]
#print x
#print y
ax.plot(wvls,curve)
ax.plot(x,y)
ax.set_title('Absolute Spectrum of '+FileName.split('/')[-1].split('_')[0])
ax.set_yscale('log')
plt.show()
fig = plt.figure()
ax = fig.add_subplot(111)
ax.set_xlim(3800,12500)
'''
#Plot SDSSJ0926 spectrum for comparison
scale = (1.0E17) #for .txt files strange unit scaling
fname = 'sdssj0926.txt'
fdata = np.loadtxt(fname,dtype=float)
x = np.array(fdata[:,0])
y = np.array(fdata[:,1])/scale #flux in ergs/s/cm^2/Angs
print y
y = y/(h*c/x) #flux in counts/s/cm^2/Angs
energyBinWidth = 0.3936
wvlStart = 3000
wvlStop = 13000
wvlBinEdges = ObsFile.makeWvlBins(energyBinWidth,wvlStart,wvlStop)
rebinned = rebin(x,y,wvlBinEdges)
plt.plot(rebinned[:,0],rebinned[:,1],color="black")
plt.plot(x,y,color='red')
plt.show()
'''
#put crab photometry points on plot
plt.errorbar(centers,counts,yerr=counterrors,fmt='ro')
#ax.set_yscale("log")
plt.show()
#output spectrum
outspec = synthSpectrum*corotScale
