def main():
x1 = [0.51374991, 0.18869174, 7.1395660212000003, 0.34985208915100002, 0.23553298503299999, 0.39732301091299999, 9.2615599123899998e+44, 5.1084213145500001]
x1 = [0.51374991, 0.18869174, 8.5049162065200008, -0.15442891679199999, 0.23553298503299999, 0.063145416345300007, -3.50757872893e+50, 111.002838225]
label1 = 'ld real' #standard errors
#x1 = [0.51374991, 0.18869174, 7.5354499164600002, -0.095576620693199998, 0.22283249803800001, 0.54247514116999995, 1.20815021896e+45, 3.5895451398099998]
#label1 = 'ld real' #small errors
x2 = [2.16157235,-1.72168638,6.94563836,-0.02163093,0.23282932,0.53659169,9.87500318e+44,4.09058055]
x2 = [2.3793539848950003, -2.3551000484399998, 7.6581408896499994, 1.0522893867450001, 0.23553298503299999, 0.28236854826800001, -3.6695715340100001e+50, 9.5370933805650004]
x2 = [2.3793539848950003, -2.3551000484399998, 7.6581408896499994, 1.0522893867450001, 0, 0.28236854826800001, -3.6695715340100001e+50, 9.5370933805650004]
x2 = [0.51374991, 0.18869174, 4.17524364486, 4.17524364486, 0.23553298503299999, 0.063145416345300007, -3.50757872893e+50, 111.002838225]
label2 = 'ld free'
#x2 = [1.76975025,-0.76728519,7.66635193,-0.09572382,0.21485677,0.58513335,1.21818912e+45,3.66011491]
#label2 = 'ld fixed'
outputf = 'model.avi'
outputf2 = 'periodigram.avi'
#x1 = [2.16157235,-1.72168638,6.94563836,-0.02163093,0.23282932,0.53659169,9.87500318e+44,4.09058055]
#label1 = 'rotation'
#x2 = [2.16157235,-1.72168638,3.462003715,3.462003715,0.23282932,0.53659169,9.87500318e+44,4.09058055]
#label2 = 'constant offset'
#outputf = 'compare_rotation.avi'
dir = 'everything/'
files = os.listdir(dir)
sodium_d = [5889.950,5895.924]
time = []
wvl = []
spectra = []
spectra_errors = []
spectra2 = []
spectra_errors2 = []
min_wvl = 5887
max_wvl = 5899
min_wvl = 5869
max_wvl = 5909
telluric = pickle.load(open('example_telluric.p','rb'))
for f in files:
time += [float(f[:-3])]
data = pickle.load(open(dir+f,'rb'))
index = [(data['wvl'] > min_wvl) & (data['wvl'] < max_wvl)]
wvl += [data['wvl'][index]]
spectra += [data['spec'][index]]
spectra_errors += [data['spec_error'][index]]
wvl = wvl[0]
time = array(time)
spectra = array(spectra)
spectra_errors = array(spectra_errors)
planet_K = 154
sodium_d = [5889.950,5895.924]
midtransit = 54341.056842
period = 2.21857312
planet_K = 154
star_K = 0.20196
line_centers = sodium_d
plotting = False
nproc = 4
best_fit=False
save_star = False
i1 = 9
i2 = 29
spectra = spectra[argsort(time)]
spectra = spectra[np.arange(i1,i2)]
spectra_errors = spectra_errors[argsort(time)]
spectra_errors = spectra_errors[np.arange(i1,i2)]
time = time[argsort(time)]
time = time[np.arange(i1,i2)]
new_errors = array([std(spectra, axis=0)]*20)
print spectra_errors
print new_errors
spectra_errors = new_errors
load = True
save = True
save_m1 = 'model1.p'
save_m2 = 'model2.p'
if load == False:
#model1 = model_3d(x1,time,period,planet_K,star_K,midtransit,wvl,line_centers,plotting,nproc=nproc,best_fit=best_fit,save_star=save_star,star_vsini=3.1)
#model2 = model_3d(x2,time,period,planet_K,star_K,midtransit,wvl,line_centers,plotting,nproc=nproc,best_fit=best_fit,save_star=save_star,star_vsini=3.1)
model1 = model_3d(x1,time,period,planet_K,star_K,midtransit,wvl,line_centers,plotting,nproc=nproc,best_fit=best_fit,save_star=save_star,star_vsini=0)
model2 = model_3d(x2,time,period,planet_K,star_K,midtransit,wvl,line_centers,plotting,nproc=nproc,best_fit=best_fit,save_star=save_star,star_vsini=0)
if save == True:
pickle.dump(model1, open(save_m1, 'wb'))
pickle.dump(model2, open(save_m2, 'wb'))
else:
model1 = pickle.load(open(save_m1))
model2 = pickle.load(open(save_m2))
os.system('rm model_plot/*.png')
#spectra_errors = array(spectra_errors)*sqrt(0.5)
#for i in range(0,len(model1)):
#plot(wvl,model1[i],'r-')
#plot(wvl,model2[i],'g-')
#show()
flist = ''
flist2 = ''
delta_chi2 = []
bad_chi2 = sum(((spectra-1)/spectra_errors)**2,axis=0)
good_chi2 = sum(((spectra-model1)/spectra_errors)**2,axis=0)
ffour, axone = plt.subplots(2, 1,sharex=True)
good_scaled_chi = sum(((spectra-model1)/spectra_errors),axis=0)
axone[0].errorbar(wvl,good_scaled_chi/sqrt(20),yerr=1,fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
#axone[1].errorbar(wvl,good_chi2-bad_chi2,yerr=1,fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
axone[1].plot(wvl,good_chi2-bad_chi2,'k-')
axone[1].set_xlim(5888.0,5898.0)
axone[0].set_ylim(-13,13)
axone[1].set_ylim(-23,13)
xlabel('Wavelength ($\AA$)')
axone[0].set_ylabel('Normalised residuals')
axone[1].set_ylabel('$\Delta {\chi}^2$')
#axone[0].yaxis.set_ticks([])
gcf().subplots_adjust(bottom=0.15)
ffour.subplots_adjust(hspace=0,wspace=0)
savefig('deltachi_comparison.png')
savefig('deltachi_comparison.pdf')
close()
bad_absolute = sum(abs((spectra-1)/spectra_errors),axis=0)
good_absolute = sum(abs((spectra-model1)/spectra_errors),axis=0)
bad_chi = sum(((spectra-1)),axis=0)
good_chi = sum(((spectra-model1)),axis=0)
bad_scaled_chi = sum(((spectra-1)/spectra_errors),axis=0)
good_scaled_chi = sum(((spectra-model1)/spectra_errors),axis=0)
#ffour, axfour = plt.subplots(3, 3,sharex=True)
#axfour = axfour.reshape(-1)
#axfour[0].plot(wvl,bad_absolute)
#axfour[3].plot(wvl,good_absolute)
#axfour[6].plot(wvl,good_absolute-bad_absolute)
#axfour[1].plot(wvl,bad_chi/sqrt(20))
#axfour[4].plot(wvl,good_chi/sqrt(20))
#axfour[7].plot(wvl,(bad_chi-good_chi)/sqrt(20))
#axfour[2].plot(wvl,bad_scaled_chi/sqrt(20))
#axfour[5].plot(wvl,good_scaled_chi/sqrt(20))
#axfour[8].plot(wvl,(bad_scaled_chi-good_scaled_chi)/sqrt(20))
#axfour[4].set_xlim(5880.0,5905.0)
#show()
for plotn in range(0,2):
fbig, axbig = plt.subplots(5, 4)
fbig.set_size_inches(8.27,11.69)
fbig.subplots_adjust(hspace=0.1,wspace=0.1)
axbig = axbig.reshape(-1)
resamp_factor = 6
residuals = []
null_residuals = []
for i in range(0,len(model1)/2):
si = i +(10*plotn)
binned_wvl, binned_spectra, binned_spectra_errors = bin_curve(wvl,spectra[si],spectra_errors[si],resamp_factor)
chiresid = (spectra[si]-model1[si])/spectra_errors[si]
residuals += [(spectra[si]-model1[si])]
null_residuals += [(spectra[si]-(model1[si]/model1[si]))]
axbig[i*2].set_xlim(5888.0,5892.0)
axbig[i*2].errorbar(wvl,spectra[si],yerr=spectra_errors[si],fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
#axbig[i*2].errorbar(binned_wvl,binned_spectra,yerr=binned_spectra_errors,fmt='k.',capsize=0)
axbig[i*2].plot(wvl,model1[si],'r')
axbig[i*2].set_ylim(0.940,1.060)
axbig[i*2].set_ylim(0.880,1.060)
axbig2 = axbig[i*2].twinx()
#axbig2.plot(wvl,chiresid,'k,')
axbig2.errorbar(wvl,chiresid,yerr=chiresid/chiresid,fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
axbig2.plot(wvl,-1 + (model1[si]/model1[si]),'r')
axbig2.set_ylim(-5,20)
axbig[i*2].set_xlim(5888.0,5892.0)
axbig2.set_xlim(5888.0,5892.0)
axbig[(i*2)+1].set_xlim(5894.0,5898.0)
axbig[(i*2)+1].errorbar(wvl,spectra[si],yerr=spectra_errors[si],fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
#axbig[(i*2)+1].errorbar(binned_wvl,binned_spectra,yerr=binned_spectra_errors,fmt='k.',capsize=0)
axbig[(i*2)+1].plot(wvl,model1[si],'r')
axbig[(i*2)+1].set_ylim(0.880,1.060)
axbig3 = axbig[(i*2)+1].twinx()
#axbig3.plot(wvl,chiresid,'k,')
axbig3.errorbar(wvl,chiresid,yerr=chiresid/chiresid,fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
axbig3.plot(wvl,-1 + (model1[si]/model1[si]),'r')
axbig3.set_ylim(-5,20)
axbig[(i*2)+1].set_xlim(5894.0,5898.0)
axbig3.set_xlim(5894.0,5898.0)
axbig[i*2].spines['right'].set_visible(False)
axbig[i*2].yaxis.tick_left()
axbig2.tick_params(labelright='off')
axbig2.set_yticks([])
axbig[(i*2)+1].set_yticks([])
if i % 2 != 0:
axbig[i*2].tick_params(labelright='off')
axbig[i*2].tick_params(labelleft='off')
if i == 5:
axbig3.set_ylabel('Normalised residuals',rotation=270,labelpad=15)
if i % 2 == 0:
if i == 4:
axbig[(i*2)].set_ylabel('Differential spectra')
axbig2.tick_params(labelleft='off')
axbig3.tick_params(labelright='off')
axbig3.tick_params(labelleft='off')
range1 = [5888,5889,5890,5891]
range2 = [5894,5895,5896,5897]
axbig2.xaxis.set_ticks(range1)
axbig3.xaxis.set_ticks(range2)
axbig[i*2].xaxis.set_ticks(range1)
axbig[(i*2)+1].xaxis.set_ticks(range2)
if (10 - i) > 2:
axbig2.xaxis.set_ticklabels([])
axbig3.xaxis.set_ticklabels([])
axbig[i*2].xaxis.set_ticklabels([])
axbig[(i*2)+1].xaxis.set_ticklabels([])
else:
axbig2.xaxis.set_ticklabels(range1)
axbig3.xaxis.set_ticklabels(range2)
axbig[i*2].xaxis.set_ticklabels(range1)
axbig[(i*2)+1].xaxis.set_ticklabels(range2)
axbig[i*2].set_xlabel(' Wavelength ($\AA$)')
axbig[i*2].xaxis.set_label_coords(1,-0.11)
axbig2.spines['right'].set_visible(False)
axbig[(i*2)+1].spines['left'].set_visible(False)
axbig[(i*2)+1].tick_params(labelleft='off')
axbig3.spines['left'].set_visible(False)
for item in (axbig2.get_xticklabels() + axbig2.get_yticklabels() + axbig3.get_xticklabels() + axbig3.get_yticklabels() + axbig[(i*2)+1].get_xticklabels() + axbig[(i*2)+1].get_yticklabels() + axbig[i*2].get_xticklabels() + axbig[i*2].get_yticklabels()):
item.set_fontsize(12)
d = .03
kwargs = dict(transform=axbig[(i*2)].transAxes, color='k', clip_on=False)
axbig[(i*2)].plot((1-d,1+d),(-d,+d), **kwargs)
axbig[(i*2)].plot((1-d,1+d),(1-d,1+d), **kwargs)
kwargs = dict(transform=axbig[(i*2)+1].transAxes, color='k', clip_on=False)
axbig[(i*2)+1].plot((-d,+d),(-d,+d), **kwargs)
axbig[(i*2)+1].plot((-d,+d),(1-d,1+d), **kwargs)
savefig('bigfig_'+str(plotn)+'.pdf')
savefig('bigfig_'+str(plotn)+'.png')
close()
# stop here for now
return
sum_null = mean(null_residuals, axis=0)/(sqrt(3)*spectra_errors[si]/sqrt(si))
sum_residuals = mean(residuals, axis=0)/(sqrt(3)*spectra_errors[si]/sqrt(si))
sum_null = sum((null_residuals/spectra_errors[si])**2, axis=0)
sum_residuals = sum((residuals/spectra_errors[si])**2, axis=0)
sum_null = sum(null_residuals/spectra_errors[si], axis=0)
sum_residuals = sum(residuals/spectra_errors[si], axis=0)
#sum_null = mean(null_residuals, axis=0)
#sum_residuals = mean(residuals, axis=0)
#errorbar(wvl,sum_null,yerr=spectra_errors[si]/spectra_errors[si],fmt='k,',alpha=1.0,capsize=0,elinewidth=0.5,marker=None)
#stack = mean(((spectra[:10])),axis=0)
#stack_model = mean(((model1[:10])),axis=0)
#binned_wvl, binned_spectra, binned_spectra_errors = bin_curve(wvl,stack,spectra_errors[i]/sqrt(10/3),5)
#errorbar(binned_wvl,binned_spectra,yerr=binned_spectra_errors,fmt='ko')
#plot(wvl,stack_model,'r-')
#stack = mean(((spectra[10:])),axis=0)
#stack_model = mean(((model1[10:])),axis=0)
#binned_wvl, binned_spectra, binned_spectra_errors = bin_curve(wvl,stack,spectra_errors[i]/sqrt(10/3),5)
#errorbar(binned_wvl,binned_spectra-0.03,yerr=binned_spectra_errors,fmt='ko')
#plot(wvl,stack_model-0.03,'r-')
show()
for i in range(0,len(model1)):
ni = str(i)
if len(ni) < 2:
ni = '0'+str(i)
chi2diffmap = ((model1[i] - spectra[i])/spectra_errors[i])**2 - ((model2[i] - spectra[i])/spectra_errors[i])**2
#plot(wvl,np.cumsum(chi2diffmap),label='difference')
#close()
chi2_1 = sum(((model1[i] - spectra[i])/spectra_errors[i])**2)
chi2_2 = sum(((model2[i] - spectra[i])/spectra_errors[i])**2)
delta_chi2 += [chi2_1 - chi2_2]
print label1, label2
print chi2_1, chi2_2
print chi2_1/len(model1[i]), chi2_2/len(model1[i])
print delta_chi2[-1]
chiresid = (spectra[i]-model1[i])/spectra_errors[i]
plot_freq =False
if plot_freq == True:
freqs = linspace(0.01,20,1000)
x = arange(0,len(model1[i]))
pgram = signal.lombscargle(x.astype('float64'),chiresid.astype('float64'),freqs.astype('float64'))
f2 , ax5 = plt.subplots(1, 1)
ax5.plot(freqs,pgram)
fname = 'model_plot/'+ni+'_periodigram.png'
savefig(fname)
close()
flist2 += fname+','
f, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2)
binned_wvl, binned_spectra, binned_spectra_errors = bin_curve(wvl,spectra[i],spectra_errors[i],resamp_factor)
for ax in [ax1, ax2]:
ax.errorbar(wvl,spectra[i],yerr=spectra_errors[i],fmt='k,',alpha=0.2,capsize=0)
ax.errorbar(binned_wvl,binned_spectra,yerr=binned_spectra_errors,fmt='k.',capsize=0)
ax.plot(wvl,model1[i],'r',label=label1)
#ax.plot(wvl,model2[i],'g--',label=label2)
ax.set_ylim(0.940,1.070)
for ax in [ax3, ax4]:
ax.errorbar(wvl,chiresid,yerr=spectra_errors[i]/spectra_errors[i],fmt='k,',alpha=0.2,capsize=0)
binned_wvl, binned_spectra, binned_spectra_errors = bin_curve(wvl,(spectra[i]-model1[i])/spectra_errors[i],spectra_errors[i]/spectra_errors[i],resamp_factor)
ax.errorbar(binned_wvl,binned_spectra,yerr=binned_spectra_errors,fmt='k.',capsize=0)
ax.plot(wvl,-1 + model1[i]/model1[i],'r',label=label1)
ax.set_ylim(-5,5)
ax1.set_xlabel('Wavelength ($\AA$)')
ax1.set_xlim(5880,5905)
ax1.set_xlim(5887.5,5897.5)
ax1.set_xlim(5888.0,5892.0)
ax3.set_xlim(5888.0,5892.0)
ax2.set_xlim(5894,5898)
ax4.set_xlim(5894,5898)
f.suptitle(str(round(delta_chi2[-1],2))+' delta chi2 '+' rchi1 '+str(round(chi2_1/len(model1[i]),2))+' rchi2 '+str(round(chi2_2/len(model1[i]),2)))
#legend()
f.suptitle('rchi2 '+str(round(chi2_1/len(model1[i]),2)))
gcf().subplots_adjust(bottom=0.15)
fname = 'model_plot/'+ni+'.png'
savefig(fname)
flist += fname+','
close()
flist = flist.rstrip(',')
print sum(delta_chi2), 'total delta chi2'
print flist
fps = 1
command = 'mencoder mf://'+flist+' -mf w=800:h=600:fps='+str(fps)+':type=png -ovc raw -oac copy -o '+outputf
print command
os.system(command)
os.system('firefox '+ outputf)
command = 'mencoder mf://'+flist2+' -mf w=800:h=600:fps='+str(fps)+':type=png -ovc raw -oac copy -o '+outputf2
print command
os.system(command)
def calc_profile(clipping_index,midtransit,period,planet_K,star_K,line_centers,plotting=True,nproc=4):
best_fit = [-1.14601379e+02,3.67424540e+00,6.97253518e-03,4.89808018e-01]
# load the master spectrum
master_dat = np.loadtxt('sodium_spectrum.dat')
master_wvl = master_dat[:,0]
master_flux = master_dat[:,1]
# the old way - probably contaminated by my mistake with the rvs
#dir_old = 'sodium_diff_bin/'
dir_old = 'no_poly_correct_rv/'
dir = 'no_poly_correct_rv/'
dir_old = 'everything/'
dir_old = 'errors/'
dir = 'model_rv/'
dir = 'test_rv/'
dir = 'everything/'
#dir = 'errors/'
#dir = 'ccf_rvs/'
#dir = 'diff_only/'
files = os.listdir(dir)
sodium_d = [5889.950,5895.924]
time = []
wvl = []
spectra = []
spectra_errors = []
spectra2 = []
spectra_errors2 = []
min_wvl = 5887
max_wvl = 5899
min_wvl = 5869
max_wvl = 5909
telluric = pickle.load(open('example_telluric.p','rb'))
for f in files:
time += [float(f[:-3])]
data = pickle.load(open(dir+f,'rb'))
data2 = pickle.load(open(dir_old+f,'rb'))
index = [(data['wvl'] > min_wvl) & (data['wvl'] < max_wvl)]
wvl += [data['wvl'][index]]
spectra += [data['spec'][index]]
spectra_errors += [data['spec_error'][index]]
spectra2 += [data2['spec'][index]]
spectra_errors2 += [data2['spec_error'][index]]
bin_wvl, bin_combined, bin_combined_error = bin_curve(wvl[-1],spectra[-1],spectra_errors[-1],1)
bin_wvl2, bin_combined2, bin_combined_error2 = bin_curve(wvl[-1],spectra2[-1],spectra_errors2[-1],1)
#p0 = 0.1572
#radiustotal = (1.0 + p0)/8.92
#gamma_0 = 0.94426940
#gamma_1 = -0.41811691
#inc = 85.68
#star_vsini = 3.1
#planet_K = 154
#vel_offset = 2.3
#atm_radius = 0.1
#input = [p0,radiustotal,gamma_0,gamma_1,inc,star_vsini,planet_K,vel_offset,atm_radius]
#fwhm = 0.5
#ratio = 300
#data_x = array([(time[-1]-midtransit)/period])
#master_dat = loadtxt('sodium_spectrum.dat')
#model_wvl = array(master_dat[:,0])
#master_flux = array(master_dat[:,1])
#profile = {'wvl':model_wvl,'spectrum':master_flux}
#results = vel_prism(data_x, input, profile, time, spot_data=False, type_data='FLUX',plotting=False,result_wvl=wvl[-1])
#errorbar(bin_wvl,bin_combined,yerr=bin_combined_error,fmt='r.')
#plot(wvl[-1],results[0])
#show()
#plot(bin_wvl,bin_combined,'.')
#errorbar(bin_wvl,bin_combined2,yerr=bin_combined_error2,fmt='r.')
#errorbar(bin_wvl,bin_combined,yerr=bin_combined_error,fmt='r.')
#errorbar(bin_wvl,bin_combined2,yerr=bin_combined_error2,fmt='g.')
#plot(bin_wvl,1.0+bin_combined2-bin_combined,'r')
#plot(telluric['wvl'],1.0 + (telluric['spectra']-1.0)/39,'k')
#plot(master_wvl,1+((master_flux/median(master_flux))-1.0)/40,'b')
#show()
#plot(wvl[0],sum(spectra,axis=0),'bo')
#show()
min_wvl = 5887
max_wvl = 5899
min_wvl = wvl[0][argmin(wvl[0])+2]
max_wvl = wvl[0][argmax(wvl[0])-2]
min_wvl = 5887 - 9
max_wvl = 5899 + 9
time = array(time)
spectra = array(spectra)
spectra_errors = array(spectra_errors)
wvl = array(wvl)
# don't try to cross correlate... is nasty
#rv = measure_rvs(wvl,spectra,min_vel=-30000,max_vel=30000,steps=10)
#plot(time,rv,'bo')
#show()
index = [argsort(time)]
#index = [arange(0,len(time))]
time = time[index]
spectra = spectra[index]
spectra_errors = spectra_errors[index]
#spectra_errors = spectra_errors[index]*0 + 1.0
wvl = array(wvl[index])
wvl = array(wvl[0])
#stellar_rv = -201.96*sin(2*pi*(midtransit-time)/period)
#spectra = align_spectra(wvl,spectra,stellar_rv)
#spectra_errors = align_spectra(wvl,spectra_errors,stellar_rv)
spectra = spectra[clipping_index]
spectra_errors = spectra_errors[clipping_index]
time = time[clipping_index]
index = [argsort(time)]
#index = [arange(0,len(time))]
time = time[index]
spectra = spectra[index]
spectra_errors = spectra_errors[index]
#spectra_errors = spectra_errors[index]*0 + 1.0
#print abs(time - 54341.0568358)*24 > 0.48613901131
# sanity check to make sure we're alligning properly
#vel_model = 0*best_fit[1] + -1*best_fit[0]*sin(2.0*pi*(time-midtransit)/period)
#t = (time - midtransit)*24
#norm = max(abs(t))
#rv = 1000*0.0*sin((pi*t/norm)/2.0)
#rv = 1000*0.0*sin((pi*((t/norm)+0.15))/2.0)
#vel_model = 1000*vel_model + rv
#spectra = align_spectra(wvl,spectra.copy(),vel_model)
c = 299792458.0
velocity_line1 = -((wvl - sodium_d[0])/sodium_d[0])*c
velocity_line2 = -((wvl - sodium_d[1])/sodium_d[1])*c
combined_vel = array(list(velocity_line1) + list(velocity_line2))
vel_index = argsort(combined_vel)
combined_vel = combined_vel[vel_index]
# show()
resamp_factor = 10
dummy_wvl = linspace(min(wvl),max(wvl),len(wvl)/resamp_factor)
# time to try the 3d model
prior_3d = []
for i in range(0,len(spectra)):
spectra[i][0] = 1.0
spectra[i][-1] = 1.0
gamma_0 = 0.94426940
gamma_1 = -0.41811691
planet_K = 154
vel_offset = 2.3
atm_radius = 0.1
fwhm = 0.5
ratio = 300
prior_3d = [gamma_0,gamma_1,planet_K,vel_offset,atm_radius,fwhm,ratio]
# for when you're fixed to 0.1
prior_3d = [1.65308844e+00,-1.04812914e+00,1.00491621e+02,5.71777342e+00,1.00000000e-01,2.13167534e+00,-4.56788205e+07]
# for when you're fixed to 0.2
prior_3d = [1.97972829e+00,-1.76607321e+00,1.19430590e+02,4.84616011e+00,2.00000000e-01,1.20117236e+00,-2.75251623e+14]
# added the east/west offset split
prior_3d = [1.98496666e+00,-1.78689117e+00,1.54000000e+02,7.35833821e+00,1.49665948e+00,3.00000000e-01,7.71559205e-01,3.68166783e+28]
# careful! from now on we allow K to be fixed. (could make it a gaussian prior?)
prior_3d = [1.98496666e+00,-1.78689117e+00,7.35833821e+00,1.49665948e+00,3.00000000e-01,7.71559205e-01,3.68166783e+28]
best_fit=False
# careful! from now on we leave radius fixed until we figure out the genetic algorithm
prior_3d = [1.98496666e+00,-1.78689117e+00,7.35833821e+00,1.49665948e+00,7.71559205e-01,3.68166783e+28]
prior_3d = [1.95876320e+00,-1.73341318e+00,6.56348290e+00,2.29320705e+00,9.45055108e-01,-6.48816563e+33]
prior_3d = [2.73989049e+00,-2.86942945e+00,6.74199449e+00,1.99626115e+00,9.52902509e-01,9.04716892e+40]
best_fit={'atm_radius':0.275}
#prior_3d = [2.23924678e+00,-2.02719249e+00,5.48955581e+00,3.45310090e+00,2.39205669e+00,6.38920070e+39]
#best_fit={'atm_radius':0.1}
# now we've added saturation! line profiles much more realistic
prior_3d = [2.23924678e+00,-2.02719249e+00,5.48955581e+00,3.45310090e+00,2.39205669e+00,6.38920070e+39,2.0]
prior_3d = [2.26339698e+00,-2.05583762e+00,8.55614089e+00,-7.98945634e-02,1.07223682e+00,9.48587317e+44,3.66446647e+00]
prior_3d = [1.8,-0.8,8.55614089e+00,-7.98945634e-02,1.07223682e+00,9.48587317e+44,3.66446647e+00]
best_fit={'atm_radius':0.15}
prior_3d = [0.51374991, 0.18869174, 7.3171305975049998, -0.083571968753300002, 0.44724670399400002, 1.24645711824e+45, 4.2957133770149998]
best_fit={'atm_radius':0.235380980973}
# prior_3d = [7.3171305975049998, -0.083571968753300002, 0.44724670399400002, 1.24645711824e+45, 4.2957133770149998]
# best_fit={'ld1':0.51374991,'ld2':0.18869174,'atm_radius':0.235380980973}
output = opt.leastsq(fit_model_3d,prior_3d,args=(time,period,planet_K,star_K,midtransit,spectra,spectra_errors,wvl,line_centers,plotting,nproc,best_fit),full_output=True)
final = output[0]
cov = output[1]
return final
fit_model_3d(final,time,period,planet_K,star_K,midtransit,spectra,spectra_errors,wvl,line_centers,True,1,best_fit=best_fit)
#linetype = 'lorentz'
#linetype = 'gauss'
linetype = 'voigt'
if linetype == 'voigt':
prior = [130,5,5.0e-3,0.5,0.5]
else:
prior = [13,0.5,5.0e-3,0.5]
#poly_prior = [1,1e-5,1e-9,1e-13,1e-16]
poly_prior = [1]
for i in range(0,len(spectra)):
prior += poly_prior
len_pol = len(poly_prior)
#output = opt.leastsq(sin_func,prior,args=(time,period,midtransit,spectra,spectra_errors,wvl,len_pol,False,linetype),full_output=True)
free_prior = [0.5,5.0e-3,300] + [10]*20
poly_prior = [1]
for i in range(0,len(spectra)):
free_prior += poly_prior
len_pol = len(poly_prior)
output = opt.leastsq(free_func,free_prior,args=(time,period,midtransit,spectra,spectra_errors,wvl,len_pol,plotting,linetype),full_output=True)
#output = opt.minimize(sin_func,prior,args=(time,period,midtransit,spectra,spectra_errors,wvl,len_pol))
final = output[0]
cov = output[1]
print final
vel_model = final[3:23]
print 'mean, median, std'
print 'ingress ',mean(vel_model[0:5]),median(vel_model[0:5]),std(vel_model[0:5])
print 'full ',mean(vel_model[5:15]),median(vel_model[5:15]),std(vel_model[5:15])
print 'egress ',mean(vel_model[15:20]),median(vel_model[15:20]),std(vel_model[15:20])
plot(vel_model,'bo')
show()
quit()
# let's see what we have
if plotting != True:
return final
print final
model = sin_model(final,time,period,midtransit,wvl,len_pol,linetype)
# right, we need to get in here and remove all the baselines
#
baseline_vals = final.copy()
baseline_vals[2] = 0
#uncomment this if you want to see the individual plots
#sin_func(final,time,period,midtransit,spectra,spectra_errors,wvl,len_pol,plotting=True,linetype=linetype)
model = array(model)
baselines = sin_model(baseline_vals,time,period,midtransit,wvl,len_pol,linetype,baseline=True)
spectra = spectra/baselines
npoints = len(model)*len(model[0])
nfree = npoints - 1 - len(final)
chi2 = sum(((spectra-model)/spectra_errors)**2.0)
BIC = chi2 + len(final)*log(npoints)
Rchi2 = chi2 / nfree
try:
s_errors = sqrt(abs(cov)*Rchi2)
fit_errors = [s_errors[i][i] for i in range(0,len(s_errors))]
except:
fit_errors = final
print (final[0]),'+-',fit_errors[0],'km/s projected orbital velocity'
print (final[1]),'+-',fit_errors[1],'km/s offset velocity'
print abs(final[2]),'+-',fit_errors[2],'Fractional depth'
print abs(final[3]),'+-',fit_errors[3],'FWHM'
vel_model2 = final[1] + final[0]*sin(2.0*pi*(midtransit-time)/period)
vel_model2 = final[1] + 154*sin(2.0*pi*(midtransit-time)/period)
print vel_model2[0], 'velocity of frame'
print vel_model2[0]-final[1], 'orbital velocity of frame'
vel_model = 0*final[1] + final[0]*sin(2.0*pi*(midtransit-time)/period)
vel_model2 = final[1] + 0*final[0]*sin(2.0*pi*(midtransit-time)/period)
vel_model = 0*final[1] + 154*sin(2.0*pi*(midtransit-time)/period)
vel_model2 = final[1] + 0*final[0]*sin(2.0*pi*(midtransit-time)/period)
d1 = 5895.924
d2 = 5889.950
velocity = 10e3
c = 3e8
alligned = align_spectra(wvl,spectra.copy(),vel_model)
sod_model = [final[2],final[1],final[3],1.0]
if linetype == 'voigt':
sod_model = [final[2],final[1],final[3],final[4],1.0]
smodel = constrained_sodium_model(sod_model,wvl,wvl.copy(),linetype)
make_velspace_plot(alligned,smodel,wvl,d1,d2)
fig, (ax1,ax2) = plt.subplots(2, sharex=True)
ax1.minorticks_on()
ax2.minorticks_on()
t = (time - midtransit)*24
norm = max(abs(t))
rv = 1000*0.0*sin((pi*t/norm)/2.0)
#rv = 1000*0.0*sin((pi*((t/norm)+0.15))/2.0)
print (max(vel_model) - min(vel_model)), 'change from start to end of transit'
vel_model2 = 1000*vel_model2
vel_model = 1000*vel_model + rv
sodium_d = [5889.950,5895.924]
linewidth = 3.0
print (max(vel_model) - min(vel_model))/1000, 'change from start to end of transit'
print smodel
ax1.set_ylim(0.0,1.2)
ax1.set_yticks([0.2,0.4,0.6,0.8,1.0])
ax1.plot(master_wvl,master_flux/median(master_flux),'k-')
narrow_region = [5887,5899]
#ax1.axvline(x=narrow_region[0],color='k',ls='--')
#ax1.axvline(x=narrow_region[1],color='k',ls='--')
ax1.axvline(x=sodium_d[0]-linewidth/2.0,color='k',ls=':')
ax1.axvline(x=sodium_d[0]+linewidth/2.0,color='k',ls=':')
ax1.axvline(x=sodium_d[1]-linewidth/2.0,color='k',ls=':')
ax1.axvline(x=sodium_d[1]+linewidth/2.0,color='k',ls=':')
shifted_d1 = d1 - d1*velocity/c
shifted_d2 = d2 - d2*velocity/c
print shifted_d1
print shifted_d2
#ax1.axvline(x=shifted_d1,color='k',ls=':')
#ax1.axvline(x=shifted_d2,color='k',ls=':')
ax1.plot(telluric['wvl'],1.1 + 3*(telluric['spectra']-1.0)/39,'r')
ax1.set_ylabel('Normalised intensity',labelpad=20)
alligned = align_spectra(wvl,spectra.copy(),vel_model)
#spectra = spectra/array([std(spectra,axis=1)]*2537).T
#alligned = spectra.copy()
#combined = median(alligned,axis=0)
#combined = mean(alligned,axis=0)
#errors = std(alligned,axis=0)/sqrt(len(spectra))
#combined = rebin_spec(wvl,combined,dummy_wvl)
#errors = rebin_spec(wvl,errors,dummy_wvl)/sqrt(resamp_factor)
scale = sqrt(3.0)
bin_wvl, bin_combined, bin_combined_error = bin_curve(wvl,median(alligned,axis=0),scale*std(alligned,axis=0)/sqrt(len(spectra)),resamp_factor)
#bin_wvl, bin_combined, bin_combined_error = bin_curve(wvl,median(alligned,axis=0),std(alligned,axis=0)/sqrt(len(spectra)),resamp_factor)
bin_wvl, bin_smodel, bin_smodel_error = bin_curve(wvl,smodel,std(alligned,axis=0)/sqrt(len(spectra)),resamp_factor)
#ax2.errorbar(dummy_wvl[1:-2],combined[1:-2],yerr=errors[1:-2],fmt='bo')
#ylim(0.992-1.0,1.003-1.0)
#ax2.plot(wvl,smodel-smodel,'r-')
#ax2.errorbar(bin_wvl[1:-2],bin_combined[1:-2]-bin_smodel[1:-2],yerr=bin_combined_error[1:-2],fmt='k.')
ax2.set_ylim(0.987,1.005)
ax2.set_ylim(0.987,1.010)
ax2.plot(wvl,smodel,'r-')
ax2.errorbar(bin_wvl[5:-5],bin_combined[5:-5],yerr=bin_combined_error[5:-5],fmt='k.')
#axvline(x=narrow_region[0],color='k',ls='--')
#axvline(x=narrow_region[1],color='k',ls='--')
axvline(x=sodium_d[0]-linewidth/2.0,color='k',ls=':')
axvline(x=sodium_d[0]+linewidth/2.0,color='k',ls=':')
axvline(x=sodium_d[1]-linewidth/2.0,color='k',ls=':')
axvline(x=sodium_d[1]+linewidth/2.0,color='k',ls=':')
xlabel("Wavelength (\AA)")
xlabel("Wavelength (nm)")
ylabel('Relative transmission')
yticks([0.988,0.992,0.996,1.000,1.004,1.008])
xlim(min_wvl,max_wvl)
fig.tight_layout(pad=0.1)
fig.subplots_adjust(hspace=0)
tickys = array([5880,5885,5890,5895,5900,5905])
xticks(tickys)
ax2.set_xticklabels(tickys/10.0)
savefig('combined_profile.jpg')
savefig('combined_profile.pdf')
os.system('pdftops combined_profile.pdf combined_profile.eps')
os.system('rm combined_profile.pdf')
show()
close()
fig = plt.figure()
new_vel = linspace(-500000,500000,116/2)
vel_specs = []
spectra = spectra
vel_resamp_factor = 20
model = model/baselines
#imshow(spectra,aspect='auto',cmap='gray',interpolation='none')
#show()
#imshow(model,aspect='auto',cmap='gray',interpolation='none')
#show()
#imshow((model-spectra),aspect='auto',cmap='gray',interpolation='none')
#show()
resamp_factor = 1
binned_spectra = []
binned_residuals = []
binned_model = []
for i in range(0,len(spectra)):
bin_wvl, bin_spectra, bin_combined_error = bin_curve(wvl,spectra[i]-1.0,spectra[i],resamp_factor)
bin_wvl, bin_resid, bin_combined_error = bin_curve(wvl,spectra[i]-model[i],spectra[i],resamp_factor)
bin_wvl, bin_model, bin_combined_error = bin_curve(wvl,model[i]-1.0,spectra[i],resamp_factor)
binned_spectra += [super_sample(bin_spectra,resamp_factor)]
binned_residuals += [super_sample(bin_resid,resamp_factor)]
binned_model += [super_sample(bin_model,resamp_factor)]
wvl = wvl[0:len(binned_spectra[0])]
binned_spectra = array(binned_spectra)
binned_residuals = array(binned_residuals)
binned_model = array(binned_model)
#imshow(binned_spectra,aspect='auto',cmap='gray',interpolation='none',vmin=0.99,vmax=1.01)
#show()
#imshow(binned_residuals,aspect='auto',cmap='gray',interpolation='none',vmin=0.99,vmax=1.01)
#show()
#spectra = binned_residuals
spectra = binned_spectra
#spectra = binned_model
#for i in range(0,len(spectra)):
#resamp = rebin_spec(wvl,spectra[i],dummy_wvl)
#combined = array(list(spectra[i]) + list(spectra[i]))
#combined_err = array(list(spectra_errors[i]) + list(spectra_errors[i]))
#combined = combined[vel_index]
#combined_err = combined_err[vel_index]
#trial = rebin_spec(combined_vel+10000000,combined,new_vel+10000000)
#trial_err = rebin_spec(combined_vel+10000000,combined_err,new_vel+10000000)
#vel_resamp_factor = len(combined_vel[(combined_vel > -500000) & (combined_vel < 500000)])/len(trial)
#trial = rebin_spec(new_vel+10000000,trial,combined_vel+10000000)
#trial_err = rebin_spec(new_vel+10000000,trial_err,combined_vel+10000000)
#bin_vel, bin_combined, bin_combined_error = bin_curve(combined_vel,combined,combined_err,vel_resamp_factor)
#bin_vel = combined_vel
#bin_combined = super_sample(bin_combined,vel_resamp_factor)
#bin_combined_error = super_sample(bin_combined_error,vel_resamp_factor)
#plot(bin_vel,bin_combined)
#plot(new_vel,trial)
#show()
#trial = bin_combined
#trial_err = bin_combined_error
#vel_specs += [((trial-1.0)/(trial_err/sqrt(vel_resamp_factor)))]
#vel_specs += [((trial-1.0)/(trial_err))]
#vel_specs += [(trial-1.0)]
#resamp = rebin_spec(wvl,spectra[i],dummy_wvl)
#resamp_err = rebin_spec(wvl,spectra_errors[i],dummy_wvl)
#resamp = rebin_spec(wvl,((spectra[i]-1.0)/(spectra_errors[i]/sqrt(resamp_factor))),dummy_wvl)
#spectra[i] = rebin_spec(dummy_wvl,resamp,wvl)
#spectra_errors[i] = rebin_spec(dummy_wvl,resamp_err,wvl)
#spectra[i] = ((spectra[i]-1.0)/(spectra_errors[i]/sqrt(resamp_factor)))
#new_vel = combined_vel/1000
#new_vel = bin_vel/1000.0
#min_vel = -200
#max_vel = 200
newspec = []
#newvelspec = []
#vel_index = (new_vel > min_vel) & (new_vel < max_vel)
#print len(vel_index)
#new_vel = new_vel[vel_index]
#print len(vel_specs)
#print len(vel_specs[0])
#print len(new_vel)
#print len(vel_index)
min_wvl = 5887
max_wvl = 5899
for i in range(0,len(spectra)):
index = [(wvl > min_wvl) & (wvl < max_wvl)]
newspec += [spectra[i][index]]
#newvelspec += [vel_specs[i][vel_index]]
#plot(new_vel,newvelspec[-1])
#show()
wvl = wvl[index]
spectra = array(newspec)
#vel_specs = array(newvelspec)
#imshow(spectra,aspect='auto',cmap='gray',interpolation='none')
index = arange(0,len(t))
#xval = add_2dline(vel_model/1000,new_vel)
#plot(xval,index,'r--')
#imshow(vel_specs,aspect='auto',cmap='gray',interpolation='Bicubic')
#imshow(vel_specs,aspect='auto',cmap='gray',interpolation='none')
#ax = plt.gca()
#index = np.arange(99,len(new_vel),200)
#labels = np.array([int(x) for x in new_vel[index]])
#xticks(index,labels)
#t = (time - midtransit)*24
#index = np.arange(2,len(t),4)
#labels = np.array([round(x,2) for x in t[index]])
#yticks(index,labels)
#ylabel('Time from midtransit (hours)')
#xlabel("Velocity (km/s)")
#fig.tight_layout(pad=0.1)
#cb = colorbar()
#cb.set_label('Deviation from median')
#show()
cmax = amax(binned_spectra)
cmin= amin(binned_spectra)
imshow(spectra,aspect='auto',cmap='gray',interpolation='none',vmin=cmin,vmax=cmax)
# imshow(spectra,aspect='auto',cmap='gray',interpolation='spline36')
ax = plt.gca()
index = np.arange(99,len(wvl),200)
labels = np.array([int(x) for x in wvl[index]])
xticks(index,labels)
t = (time - midtransit)*24
index = np.arange(2,len(t),4)
labels = np.array([round(x,2) for x in t[index]])
yticks(index,labels)
ylabel('Time from midtransit (hours)')
xlabel("Wavelength (\AA)")
fig.tight_layout(pad=0.1)
index = arange(0,len(t))
line = sin_line(final,time,period,midtransit,wvl,sodium_d[0])
xval1 = add_2dline(line,wvl)
line = sin_line(final,time,period,midtransit,wvl,sodium_d[1])
xval2 = add_2dline(line,wvl)
cb = colorbar()
cb.set_label('Deviation from median')
plot(xval1,index,'r--')
plot(xval2,index,'r--')
savefig('2dplot.pdf')
show()
def calc_profile(clipping_index,
midtransit,
period,
planet_K,
star_K,
line_centers,
plotting=True,
nproc=4):
best_fit = [
-1.14601379e+02, 3.67424540e+00, 6.97253518e-03, 4.89808018e-01
]
# load the master spectrum
master_dat = np.loadtxt('sodium_spectrum.dat')
master_wvl = master_dat[:, 0]
master_flux = master_dat[:, 1]
# the old way - probably contaminated by my mistake with the rvs
#dir_old = 'sodium_diff_bin/'
dir_old = 'no_poly_correct_rv/'
dir = 'no_poly_correct_rv/'
dir_old = 'everything/'
dir_old = 'errors/'
dir = 'model_rv/'
dir = 'test_rv/'
dir = 'everything/'
#dir = 'errors/'
#dir = 'ccf_rvs/'
#dir = 'diff_only/'
files = os.listdir(dir)
sodium_d = [5889.950, 5895.924]
time = []
wvl = []
spectra = []
spectra_errors = []
spectra2 = []
spectra_errors2 = []
min_wvl = 5887
max_wvl = 5899
min_wvl = 5869
max_wvl = 5909
telluric = pickle.load(open('example_telluric.p', 'rb'))
for f in files:
time += [float(f[:-3])]
data = pickle.load(open(dir + f, 'rb'))
data2 = pickle.load(open(dir_old + f, 'rb'))
index = [(data['wvl'] > min_wvl) & (data['wvl'] < max_wvl)]
wvl += [data['wvl'][index]]
spectra += [data['spec'][index]]
spectra_errors += [data['spec_error'][index]]
spectra2 += [data2['spec'][index]]
spectra_errors2 += [data2['spec_error'][index]]
bin_wvl, bin_combined, bin_combined_error = bin_curve(
wvl[-1], spectra[-1], spectra_errors[-1], 1)
bin_wvl2, bin_combined2, bin_combined_error2 = bin_curve(
wvl[-1], spectra2[-1], spectra_errors2[-1], 1)
#p0 = 0.1572
#radiustotal = (1.0 + p0)/8.92
#gamma_0 = 0.94426940
#gamma_1 = -0.41811691
#inc = 85.68
#star_vsini = 3.1
#planet_K = 154
#vel_offset = 2.3
#atm_radius = 0.1
#input = [p0,radiustotal,gamma_0,gamma_1,inc,star_vsini,planet_K,vel_offset,atm_radius]
#fwhm = 0.5
#ratio = 300
#data_x = array([(time[-1]-midtransit)/period])
#master_dat = loadtxt('sodium_spectrum.dat')
#model_wvl = array(master_dat[:,0])
#master_flux = array(master_dat[:,1])
#profile = {'wvl':model_wvl,'spectrum':master_flux}
#results = vel_prism(data_x, input, profile, time, spot_data=False, type_data='FLUX',plotting=False,result_wvl=wvl[-1])
#errorbar(bin_wvl,bin_combined,yerr=bin_combined_error,fmt='r.')
#plot(wvl[-1],results[0])
#show()
#plot(bin_wvl,bin_combined,'.')
#errorbar(bin_wvl,bin_combined2,yerr=bin_combined_error2,fmt='r.')
#errorbar(bin_wvl,bin_combined,yerr=bin_combined_error,fmt='r.')
#errorbar(bin_wvl,bin_combined2,yerr=bin_combined_error2,fmt='g.')
#plot(bin_wvl,1.0+bin_combined2-bin_combined,'r')
#plot(telluric['wvl'],1.0 + (telluric['spectra']-1.0)/39,'k')
#plot(master_wvl,1+((master_flux/median(master_flux))-1.0)/40,'b')
#show()
#plot(wvl[0],sum(spectra,axis=0),'bo')
#show()
min_wvl = 5887
max_wvl = 5899
min_wvl = wvl[0][argmin(wvl[0]) + 2]
max_wvl = wvl[0][argmax(wvl[0]) - 2]
min_wvl = 5887 - 9
max_wvl = 5899 + 9
time = array(time)
spectra = array(spectra)
spectra_errors = array(spectra_errors)
wvl = array(wvl)
# don't try to cross correlate... is nasty
#rv = measure_rvs(wvl,spectra,min_vel=-30000,max_vel=30000,steps=10)
#plot(time,rv,'bo')
#show()
index = [argsort(time)]
#index = [arange(0,len(time))]
time = time[index]
spectra = spectra[index]
spectra_errors = spectra_errors[index]
#spectra_errors = spectra_errors[index]*0 + 1.0
wvl = array(wvl[index])
wvl = array(wvl[0])
#stellar_rv = -201.96*sin(2*pi*(midtransit-time)/period)
#spectra = align_spectra(wvl,spectra,stellar_rv)
#spectra_errors = align_spectra(wvl,spectra_errors,stellar_rv)
spectra = spectra[clipping_index]
spectra_errors = spectra_errors[clipping_index]
time = time[clipping_index]
index = [argsort(time)]
#index = [arange(0,len(time))]
time = time[index]
spectra = spectra[index]
spectra_errors = spectra_errors[index]
#spectra_errors = spectra_errors[index]*0 + 1.0
#print abs(time - 54341.0568358)*24 > 0.48613901131
# sanity check to make sure we're alligning properly
#vel_model = 0*best_fit[1] + -1*best_fit[0]*sin(2.0*pi*(time-midtransit)/period)
#t = (time - midtransit)*24
#norm = max(abs(t))
#rv = 1000*0.0*sin((pi*t/norm)/2.0)
#rv = 1000*0.0*sin((pi*((t/norm)+0.15))/2.0)
#vel_model = 1000*vel_model + rv
#spectra = align_spectra(wvl,spectra.copy(),vel_model)
c = 299792458.0
velocity_line1 = -((wvl - sodium_d[0]) / sodium_d[0]) * c
velocity_line2 = -((wvl - sodium_d[1]) / sodium_d[1]) * c
combined_vel = array(list(velocity_line1) + list(velocity_line2))
vel_index = argsort(combined_vel)
combined_vel = combined_vel[vel_index]
# show()
resamp_factor = 10
dummy_wvl = linspace(min(wvl), max(wvl), len(wvl) / resamp_factor)
# time to try the 3d model
prior_3d = []
for i in range(0, len(spectra)):
spectra[i][0] = 1.0
spectra[i][-1] = 1.0
gamma_0 = 0.94426940
gamma_1 = -0.41811691
planet_K = 154
vel_offset = 2.3
atm_radius = 0.1
fwhm = 0.5
ratio = 300
prior_3d = [
gamma_0, gamma_1, planet_K, vel_offset, atm_radius, fwhm, ratio
]
# for when you're fixed to 0.1
prior_3d = [
1.65308844e+00, -1.04812914e+00, 1.00491621e+02, 5.71777342e+00,
1.00000000e-01, 2.13167534e+00, -4.56788205e+07
]
# for when you're fixed to 0.2
prior_3d = [
1.97972829e+00, -1.76607321e+00, 1.19430590e+02, 4.84616011e+00,
2.00000000e-01, 1.20117236e+00, -2.75251623e+14
]
# added the east/west offset split
prior_3d = [
1.98496666e+00, -1.78689117e+00, 1.54000000e+02, 7.35833821e+00,
1.49665948e+00, 3.00000000e-01, 7.71559205e-01, 3.68166783e+28
]
# careful! from now on we allow K to be fixed. (could make it a gaussian prior?)
prior_3d = [
1.98496666e+00, -1.78689117e+00, 7.35833821e+00, 1.49665948e+00,
3.00000000e-01, 7.71559205e-01, 3.68166783e+28
]
best_fit = False
# careful! from now on we leave radius fixed until we figure out the genetic algorithm
prior_3d = [
1.98496666e+00, -1.78689117e+00, 7.35833821e+00, 1.49665948e+00,
7.71559205e-01, 3.68166783e+28
]
prior_3d = [
1.95876320e+00, -1.73341318e+00, 6.56348290e+00, 2.29320705e+00,
9.45055108e-01, -6.48816563e+33
]
prior_3d = [
2.73989049e+00, -2.86942945e+00, 6.74199449e+00, 1.99626115e+00,
9.52902509e-01, 9.04716892e+40
]
best_fit = {'atm_radius': 0.275}
#prior_3d = [2.23924678e+00,-2.02719249e+00,5.48955581e+00,3.45310090e+00,2.39205669e+00,6.38920070e+39]
#best_fit={'atm_radius':0.1}
# now we've added saturation! line profiles much more realistic
prior_3d = [
2.23924678e+00, -2.02719249e+00, 5.48955581e+00, 3.45310090e+00,
2.39205669e+00, 6.38920070e+39, 2.0
]
prior_3d = [
2.26339698e+00, -2.05583762e+00, 8.55614089e+00, -7.98945634e-02,
1.07223682e+00, 9.48587317e+44, 3.66446647e+00
]
prior_3d = [
1.8, -0.8, 8.55614089e+00, -7.98945634e-02, 1.07223682e+00,
9.48587317e+44, 3.66446647e+00
]
best_fit = {'atm_radius': 0.15}
prior_3d = [
0.51374991, 0.18869174, 7.3171305975049998, -0.083571968753300002,
0.44724670399400002, 1.24645711824e+45, 4.2957133770149998
]
best_fit = {'atm_radius': 0.235380980973}
# prior_3d = [7.3171305975049998, -0.083571968753300002, 0.44724670399400002, 1.24645711824e+45, 4.2957133770149998]
# best_fit={'ld1':0.51374991,'ld2':0.18869174,'atm_radius':0.235380980973}
output = opt.leastsq(fit_model_3d,
prior_3d,
args=(time, period, planet_K, star_K, midtransit,
spectra, spectra_errors, wvl, line_centers,
plotting, nproc, best_fit),
full_output=True)
final = output[0]
cov = output[1]
return final
fit_model_3d(final,
time,
period,
planet_K,
star_K,
midtransit,
spectra,
spectra_errors,
wvl,
line_centers,
True,
1,
best_fit=best_fit)
#linetype = 'lorentz'
#linetype = 'gauss'
linetype = 'voigt'
if linetype == 'voigt':
prior = [130, 5, 5.0e-3, 0.5, 0.5]
else:
prior = [13, 0.5, 5.0e-3, 0.5]
#poly_prior = [1,1e-5,1e-9,1e-13,1e-16]
poly_prior = [1]
for i in range(0, len(spectra)):
prior += poly_prior
len_pol = len(poly_prior)
#output = opt.leastsq(sin_func,prior,args=(time,period,midtransit,spectra,spectra_errors,wvl,len_pol,False,linetype),full_output=True)
free_prior = [0.5, 5.0e-3, 300] + [10] * 20
poly_prior = [1]
for i in range(0, len(spectra)):
free_prior += poly_prior
len_pol = len(poly_prior)
output = opt.leastsq(free_func,
free_prior,
args=(time, period, midtransit, spectra,
spectra_errors, wvl, len_pol, plotting,
linetype),
full_output=True)
#output = opt.minimize(sin_func,prior,args=(time,period,midtransit,spectra,spectra_errors,wvl,len_pol))
final = output[0]
cov = output[1]
print final
vel_model = final[3:23]
print 'mean, median, std'
print 'ingress ', mean(vel_model[0:5]), median(vel_model[0:5]), std(
vel_model[0:5])
print 'full ', mean(vel_model[5:15]), median(vel_model[5:15]), std(
vel_model[5:15])
print 'egress ', mean(vel_model[15:20]), median(vel_model[15:20]), std(
vel_model[15:20])
plot(vel_model, 'bo')
show()
quit()
# let's see what we have
if plotting != True:
return final
print final
model = sin_model(final, time, period, midtransit, wvl, len_pol, linetype)
# right, we need to get in here and remove all the baselines
#
baseline_vals = final.copy()
baseline_vals[2] = 0
#uncomment this if you want to see the individual plots
#sin_func(final,time,period,midtransit,spectra,spectra_errors,wvl,len_pol,plotting=True,linetype=linetype)
model = array(model)
baselines = sin_model(baseline_vals,
time,
period,
midtransit,
wvl,
len_pol,
linetype,
baseline=True)
spectra = spectra / baselines
npoints = len(model) * len(model[0])
nfree = npoints - 1 - len(final)
chi2 = sum(((spectra - model) / spectra_errors)**2.0)
BIC = chi2 + len(final) * log(npoints)
Rchi2 = chi2 / nfree
try:
s_errors = sqrt(abs(cov) * Rchi2)
fit_errors = [s_errors[i][i] for i in range(0, len(s_errors))]
except:
fit_errors = final
print(final[0]), '+-', fit_errors[0], 'km/s projected orbital velocity'
print(final[1]), '+-', fit_errors[1], 'km/s offset velocity'
print abs(final[2]), '+-', fit_errors[2], 'Fractional depth'
print abs(final[3]), '+-', fit_errors[3], 'FWHM'
vel_model2 = final[1] + final[0] * sin(2.0 * pi *
(midtransit - time) / period)
vel_model2 = final[1] + 154 * sin(2.0 * pi * (midtransit - time) / period)
print vel_model2[0], 'velocity of frame'
print vel_model2[0] - final[1], 'orbital velocity of frame'
vel_model = 0 * final[1] + final[0] * sin(2.0 * pi *
(midtransit - time) / period)
vel_model2 = final[1] + 0 * final[0] * sin(2.0 * pi *
(midtransit - time) / period)
vel_model = 0 * final[1] + 154 * sin(2.0 * pi *
(midtransit - time) / period)
vel_model2 = final[1] + 0 * final[0] * sin(2.0 * pi *
(midtransit - time) / period)
d1 = 5895.924
d2 = 5889.950
velocity = 10e3
c = 3e8
alligned = align_spectra(wvl, spectra.copy(), vel_model)
sod_model = [final[2], final[1], final[3], 1.0]
if linetype == 'voigt':
sod_model = [final[2], final[1], final[3], final[4], 1.0]
smodel = constrained_sodium_model(sod_model, wvl, wvl.copy(), linetype)
make_velspace_plot(alligned, smodel, wvl, d1, d2)
fig, (ax1, ax2) = plt.subplots(2, sharex=True)
ax1.minorticks_on()
ax2.minorticks_on()
t = (time - midtransit) * 24
norm = max(abs(t))
rv = 1000 * 0.0 * sin((pi * t / norm) / 2.0)
#rv = 1000*0.0*sin((pi*((t/norm)+0.15))/2.0)
print(max(vel_model) -
min(vel_model)), 'change from start to end of transit'
vel_model2 = 1000 * vel_model2
vel_model = 1000 * vel_model + rv
sodium_d = [5889.950, 5895.924]
linewidth = 3.0
print(max(vel_model) -
min(vel_model)) / 1000, 'change from start to end of transit'
print smodel
ax1.set_ylim(0.0, 1.2)
ax1.set_yticks([0.2, 0.4, 0.6, 0.8, 1.0])
ax1.plot(master_wvl, master_flux / median(master_flux), 'k-')
narrow_region = [5887, 5899]
#ax1.axvline(x=narrow_region[0],color='k',ls='--')
#ax1.axvline(x=narrow_region[1],color='k',ls='--')
ax1.axvline(x=sodium_d[0] - linewidth / 2.0, color='k', ls=':')
ax1.axvline(x=sodium_d[0] + linewidth / 2.0, color='k', ls=':')
ax1.axvline(x=sodium_d[1] - linewidth / 2.0, color='k', ls=':')
ax1.axvline(x=sodium_d[1] + linewidth / 2.0, color='k', ls=':')
shifted_d1 = d1 - d1 * velocity / c
shifted_d2 = d2 - d2 * velocity / c
print shifted_d1
print shifted_d2
#ax1.axvline(x=shifted_d1,color='k',ls=':')
#ax1.axvline(x=shifted_d2,color='k',ls=':')
ax1.plot(telluric['wvl'], 1.1 + 3 * (telluric['spectra'] - 1.0) / 39, 'r')
ax1.set_ylabel('Normalised intensity', labelpad=20)
alligned = align_spectra(wvl, spectra.copy(), vel_model)
#spectra = spectra/array([std(spectra,axis=1)]*2537).T
#alligned = spectra.copy()
#combined = median(alligned,axis=0)
#combined = mean(alligned,axis=0)
#errors = std(alligned,axis=0)/sqrt(len(spectra))
#combined = rebin_spec(wvl,combined,dummy_wvl)
#errors = rebin_spec(wvl,errors,dummy_wvl)/sqrt(resamp_factor)
scale = sqrt(3.0)
bin_wvl, bin_combined, bin_combined_error = bin_curve(
wvl, median(alligned, axis=0),
scale * std(alligned, axis=0) / sqrt(len(spectra)), resamp_factor)
#bin_wvl, bin_combined, bin_combined_error = bin_curve(wvl,median(alligned,axis=0),std(alligned,axis=0)/sqrt(len(spectra)),resamp_factor)
bin_wvl, bin_smodel, bin_smodel_error = bin_curve(
wvl, smodel,
std(alligned, axis=0) / sqrt(len(spectra)), resamp_factor)
#ax2.errorbar(dummy_wvl[1:-2],combined[1:-2],yerr=errors[1:-2],fmt='bo')
#ylim(0.992-1.0,1.003-1.0)
#ax2.plot(wvl,smodel-smodel,'r-')
#ax2.errorbar(bin_wvl[1:-2],bin_combined[1:-2]-bin_smodel[1:-2],yerr=bin_combined_error[1:-2],fmt='k.')
ax2.set_ylim(0.987, 1.005)
ax2.set_ylim(0.987, 1.010)
ax2.plot(wvl, smodel, 'r-')
ax2.errorbar(bin_wvl[5:-5],
bin_combined[5:-5],
yerr=bin_combined_error[5:-5],
fmt='k.')
#axvline(x=narrow_region[0],color='k',ls='--')
#axvline(x=narrow_region[1],color='k',ls='--')
axvline(x=sodium_d[0] - linewidth / 2.0, color='k', ls=':')
axvline(x=sodium_d[0] + linewidth / 2.0, color='k', ls=':')
axvline(x=sodium_d[1] - linewidth / 2.0, color='k', ls=':')
axvline(x=sodium_d[1] + linewidth / 2.0, color='k', ls=':')
xlabel("Wavelength (\AA)")
xlabel("Wavelength (nm)")
ylabel('Relative transmission')
yticks([0.988, 0.992, 0.996, 1.000, 1.004, 1.008])
xlim(min_wvl, max_wvl)
fig.tight_layout(pad=0.1)
fig.subplots_adjust(hspace=0)
tickys = array([5880, 5885, 5890, 5895, 5900, 5905])
xticks(tickys)
ax2.set_xticklabels(tickys / 10.0)
savefig('combined_profile.jpg')
savefig('combined_profile.pdf')
os.system('pdftops combined_profile.pdf combined_profile.eps')
os.system('rm combined_profile.pdf')
show()
close()
fig = plt.figure()
new_vel = linspace(-500000, 500000, 116 / 2)
vel_specs = []
spectra = spectra
vel_resamp_factor = 20
model = model / baselines
#imshow(spectra,aspect='auto',cmap='gray',interpolation='none')
#show()
#imshow(model,aspect='auto',cmap='gray',interpolation='none')
#show()
#imshow((model-spectra),aspect='auto',cmap='gray',interpolation='none')
#show()
resamp_factor = 1
binned_spectra = []
binned_residuals = []
binned_model = []
for i in range(0, len(spectra)):
bin_wvl, bin_spectra, bin_combined_error = bin_curve(
wvl, spectra[i] - 1.0, spectra[i], resamp_factor)
bin_wvl, bin_resid, bin_combined_error = bin_curve(
wvl, spectra[i] - model[i], spectra[i], resamp_factor)
bin_wvl, bin_model, bin_combined_error = bin_curve(
wvl, model[i] - 1.0, spectra[i], resamp_factor)
binned_spectra += [super_sample(bin_spectra, resamp_factor)]
binned_residuals += [super_sample(bin_resid, resamp_factor)]
binned_model += [super_sample(bin_model, resamp_factor)]
wvl = wvl[0:len(binned_spectra[0])]
binned_spectra = array(binned_spectra)
binned_residuals = array(binned_residuals)
binned_model = array(binned_model)
#imshow(binned_spectra,aspect='auto',cmap='gray',interpolation='none',vmin=0.99,vmax=1.01)
#show()
#imshow(binned_residuals,aspect='auto',cmap='gray',interpolation='none',vmin=0.99,vmax=1.01)
#show()
#spectra = binned_residuals
spectra = binned_spectra
#spectra = binned_model
#for i in range(0,len(spectra)):
#resamp = rebin_spec(wvl,spectra[i],dummy_wvl)
#combined = array(list(spectra[i]) + list(spectra[i]))
#combined_err = array(list(spectra_errors[i]) + list(spectra_errors[i]))
#combined = combined[vel_index]
#combined_err = combined_err[vel_index]
#trial = rebin_spec(combined_vel+10000000,combined,new_vel+10000000)
#trial_err = rebin_spec(combined_vel+10000000,combined_err,new_vel+10000000)
#vel_resamp_factor = len(combined_vel[(combined_vel > -500000) & (combined_vel < 500000)])/len(trial)
#trial = rebin_spec(new_vel+10000000,trial,combined_vel+10000000)
#trial_err = rebin_spec(new_vel+10000000,trial_err,combined_vel+10000000)
#bin_vel, bin_combined, bin_combined_error = bin_curve(combined_vel,combined,combined_err,vel_resamp_factor)
#bin_vel = combined_vel
#bin_combined = super_sample(bin_combined,vel_resamp_factor)
#bin_combined_error = super_sample(bin_combined_error,vel_resamp_factor)
#plot(bin_vel,bin_combined)
#plot(new_vel,trial)
#show()
#trial = bin_combined
#trial_err = bin_combined_error
#vel_specs += [((trial-1.0)/(trial_err/sqrt(vel_resamp_factor)))]
#vel_specs += [((trial-1.0)/(trial_err))]
#vel_specs += [(trial-1.0)]
#resamp = rebin_spec(wvl,spectra[i],dummy_wvl)
#resamp_err = rebin_spec(wvl,spectra_errors[i],dummy_wvl)
#resamp = rebin_spec(wvl,((spectra[i]-1.0)/(spectra_errors[i]/sqrt(resamp_factor))),dummy_wvl)
#spectra[i] = rebin_spec(dummy_wvl,resamp,wvl)
#spectra_errors[i] = rebin_spec(dummy_wvl,resamp_err,wvl)
#spectra[i] = ((spectra[i]-1.0)/(spectra_errors[i]/sqrt(resamp_factor)))
#new_vel = combined_vel/1000
#new_vel = bin_vel/1000.0
#min_vel = -200
#max_vel = 200
newspec = []
#newvelspec = []
#vel_index = (new_vel > min_vel) & (new_vel < max_vel)
#print len(vel_index)
#new_vel = new_vel[vel_index]
#print len(vel_specs)
#print len(vel_specs[0])
#print len(new_vel)
#print len(vel_index)
min_wvl = 5887
max_wvl = 5899
for i in range(0, len(spectra)):
index = [(wvl > min_wvl) & (wvl < max_wvl)]
newspec += [spectra[i][index]]
#newvelspec += [vel_specs[i][vel_index]]
#plot(new_vel,newvelspec[-1])
#show()
wvl = wvl[index]
spectra = array(newspec)
#vel_specs = array(newvelspec)
#imshow(spectra,aspect='auto',cmap='gray',interpolation='none')
index = arange(0, len(t))
#xval = add_2dline(vel_model/1000,new_vel)
#plot(xval,index,'r--')
#imshow(vel_specs,aspect='auto',cmap='gray',interpolation='Bicubic')
#imshow(vel_specs,aspect='auto',cmap='gray',interpolation='none')
#ax = plt.gca()
#index = np.arange(99,len(new_vel),200)
#labels = np.array([int(x) for x in new_vel[index]])
#xticks(index,labels)
#t = (time - midtransit)*24
#index = np.arange(2,len(t),4)
#labels = np.array([round(x,2) for x in t[index]])
#yticks(index,labels)
#ylabel('Time from midtransit (hours)')
#xlabel("Velocity (km/s)")
#fig.tight_layout(pad=0.1)
#cb = colorbar()
#cb.set_label('Deviation from median')
#show()
cmax = amax(binned_spectra)
cmin = amin(binned_spectra)
imshow(spectra,
aspect='auto',
cmap='gray',
interpolation='none',
vmin=cmin,
vmax=cmax)
# imshow(spectra,aspect='auto',cmap='gray',interpolation='spline36')
ax = plt.gca()
index = np.arange(99, len(wvl), 200)
labels = np.array([int(x) for x in wvl[index]])
xticks(index, labels)
t = (time - midtransit) * 24
index = np.arange(2, len(t), 4)
labels = np.array([round(x, 2) for x in t[index]])
yticks(index, labels)
ylabel('Time from midtransit (hours)')
xlabel("Wavelength (\AA)")
fig.tight_layout(pad=0.1)
index = arange(0, len(t))
line = sin_line(final, time, period, midtransit, wvl, sodium_d[0])
xval1 = add_2dline(line, wvl)
line = sin_line(final, time, period, midtransit, wvl, sodium_d[1])
xval2 = add_2dline(line, wvl)
cb = colorbar()
cb.set_label('Deviation from median')
plot(xval1, index, 'r--')
plot(xval2, index, 'r--')
savefig('2dplot.pdf')
show()
