def find_shift(data_type,data_spectrum,teff,logg,feh,start_lambda,end_lambda):
### Import the template spectrum
template_spectrum = "template_" + str(teff) + "_" + str(logg) + "_" + str(feh) + ".dat"
if data_type=="flux":
template_spectrum = functions.read_ascii(model_path_flux + template_spectrum)
if data_type=="norm":
template_spectrum = functions.read_ascii(model_path_norm + template_spectrum)
template_spectrum = functions.read_table(template_spectrum)
template_spectrum = transpose(array(template_spectrum))
if data_type=="flux":
template_spectrum =spectype_functions.normalise(template_spectrum,flux_normalise_w1,flux_normalise_w2)
### Chop both spectra
data_region = spectype_numerical_functions.chop_spectrum(data_spectrum,start_lambda,end_lambda)
template_region = spectype_numerical_functions.chop_spectrum(template_spectrum,start_lambda,end_lambda)
### Conform template spectrum to data spectrum -> same wavelength scale
template_region = spectype_numerical_functions.conform_spectrum(data_region,template_region)
### Find shift
chisq_shift = []
shift_limit = 20
shift = -1*shift_limit
while shift <= shift_limit:
data_region_shifted,template_region_shifted = spectype_functions.shift_spectrum(data_region,template_region,shift)
chisq_shift.append(spectype_numerical_functions.chisq(data_region_shifted,template_region_shifted))
shift = shift + 1
chisq_min = spectype_functions.find_min(chisq_shift)
best_shift = chisq_min - shift_limit
return best_shift
def test_template(data_type,data_spectrum,teff,logg,feh,start_lambda,end_lambda,shift):
### Import a template spectrum - test
template_spectrum = "template_" + str(teff) + "_" + str(logg) + "_" + str(feh) + ".dat"
if data_type=="flux":
template_spectrum = functions.read_ascii(model_path_flux + template_spectrum)
if data_type=="norm":
template_spectrum = functions.read_ascii(model_path_norm + template_spectrum)
template_spectrum = functions.read_table(template_spectrum)
template_spectrum = transpose(array(template_spectrum))
if data_type=="flux":
template_spectrum = spectype_functions.normalise(template_spectrum,flux_normalise_w1,flux_normalise_w2)
### Chop both spectra
data_region = spectype_numerical_functions.chop_spectrum(data_spectrum,start_lambda,end_lambda)
template_region = spectype_numerical_functions.chop_spectrum(template_spectrum,start_lambda,end_lambda)
### Conform template spectrum to data spectrum -> same wavelength scale
template_region = spectype_numerical_functions.conform_spectrum(data_region,template_region)
### Find shift
data_region_shifted,template_region_shifted = spectype_functions.shift_spectrum(data_region,template_region,shift)
chisq = spectype_numerical_functions.chisq(data_region_shifted,template_region_shifted)
# plt.clf()
# plt.plot(data_region_shifted[0],data_region_shifted[1])
# plt.plot(template_region_shifted[0],template_region_shifted[1])
# plt.title(data_type + " " + str(teff) + " " + str(logg) + " " + str(feh) +" " + str(chisq))
# plt.show()
return chisq
def loop_input_spectrum(input_wave,input_flux,folder,teff_space,logg_space,feh_space,w1,w2,perform_normalise):
data = []
for teff in teff_space:
for logg in logg_space:
for feh in feh_space:
template_spectrum = "template_" + str(teff) + "_" + str(logg) + "_" + str(feh)+".dat"
#print folder + template_spectrum
template_spectrum = functions.read_ascii(folder+template_spectrum)
template_spectrum = functions.read_table(template_spectrum)
template_spectrum = transpose(array(template_spectrum))
if folder == model_path_flux:
template_spectrum = spectype_functions.normalise(template_spectrum,flux_normalise_w1,flux_normalise_w2)
i1 = w1 - min(input_wave)
i2 = w2 - min(input_wave)
input_wave_cropped = input_wave[i1:i2]
input_flux_cropped = input_flux[i1:i2]
template_spectrum = spectype_numerical_functions.chop_spectrum(template_spectrum,w1-10,w2+10)
template_interp = interpolate.splrep(template_spectrum[0],template_spectrum[1],s=0)
template_flux = interpolate.splev(input_wave_cropped,template_interp,der=0)
sigma = 3.0
if perform_normalise:
diff_flux = input_flux_cropped/median(input_flux_cropped) - template_flux/median(template_flux)
else:
diff_flux = input_flux_cropped - template_flux
diff_flux = clip(diff_flux,median(diff_flux) - sigma*std(diff_flux),median(diff_flux)+sigma*std(diff_flux))
rms = sqrt(sum(diff_flux**2) /float(len(input_wave_cropped)))
# plt.clf()
# plt.plot(input_wave_cropped,input_flux_cropped/median(input_flux_cropped))
# plt.plot(input_wave_cropped,template_flux/median(template_flux))
# plt.show()
# #sys.exit()
#print rms
data.append(rms)
return data
