def voigt_fit(theta, x, y, yerr, sat_lim=0.05):
peak1 = Voigt1D(x_0=theta[0],
amplitude_L=theta[1],
fwhm_L=theta[2],
fwhm_G=theta[3])(x)
peak2 = np.zeros(len(peak1))
if len(theta) > 5:
peak2 = Voigt1D(x_0=theta[4],
amplitude_L=theta[5],
fwhm_L=theta[6],
fwhm_G=theta[7])(x)
model = 1.0 - (peak1 + peak2)
sat = model < sat_lim
model[sat] = np.mean(sat_lim)
return model
def FitLine(self, params):
'''
#Calulates the fitted line at given params and wavenumbers
'''
upperFWavenumbers = np.zeros_like(self.relIntensities)
for i, F in enumerate(self.listF[:, 0]):
upperFWavenumbers[i] += dE(params[0], params[2],
K(F, self.upperJ, self.I), F,
self.upperJ, self.I)
lowerFWavenumbers = np.zeros_like(self.relIntensities)
for i, F in enumerate(self.listF[:, 1]):
lowerFWavenumbers[i] += dE(params[1], params[3],
K(F, self.lowerJ, self.I), F,
self.lowerJ, self.I)
self.hfsLines = upperFWavenumbers - lowerFWavenumbers + params[7]
line_fit = np.zeros_like(self.w)
self.components = []
for i, v in enumerate(self.relIntensities):
voigt = Voigt1D(self.hfsLines[i], v, params[5], params[4])
line_fit += voigt(self.w) / (np.pi * params[5] / 2
) #normalisation of area
self.components.append(voigt(self.w) / (np.pi * params[5] / 2))
line_fit = line_fit * params[6] / np.sum(self.relIntensities)
self.components = np.array(self.components) * params[6] / np.sum(
self.relIntensities)
line_fit = satFit(line_fit, params[8])
self.components = satFit(self.components, params[8])
self.noApoFit = cp.copy(line_fit)
line_fit = self.ApoFit(line_fit)
return line_fit
def __init__(self,
center=0.0,
intensity=1.0,
fwhm_L=1.0,
fwhm_G=1.0,
x_min=-5.0,
x_max=5.0,
n_points=100,
snr=30):
super(FakeVoigt, self).__init__(center=center,
intensity=intensity,
x_min=x_min,
x_max=x_max,
n_points=n_points,
snr=snr)
self.fwhm_L = fwhm_L
self.fwhm_G = fwhm_G
v = Voigt1D(
x_0=self.center,
amplitude_L=self.intensity,
fwhm_L=self.fwhm_L,
fwhm_G=self.fwhm_G,
)
self.voigt = v(self.x)
self.noisy_voigt = self.voigt + self.noise
def model_from_frequencies(self, frequencies):
v = Voigt1D(
x_0=self.center,
amplitude_L=self.intensity,
fwhm_L=self.fwhm_L,
fwhm_G=self.fwhm_G,
)
return 1. - v(frequencies)
def VoigtFunction(x, centroid, amplitude, L_fwhm, G_fwhm):
"""
Return the Voigt line shape at x with Lorentzian component HWHM gamma
and Gaussian component HWHM alpha.
Source: http://scipython.com/book/chapter-8-scipy/examples/the-voigt-profile/
Args:
x (numpy array): wavelength array
centroid (float): center of profile
alpha (float): Gaussian HWHM
gamma (float): Lorentzian HWHM
"""
#sigma = alpha / np.sqrt(2 * np.log(2))
#return np.real(wofz((x - centroid + 1j*gamma)/sigma/np.sqrt(2))) / sigma / np.sqrt(2*np.pi)
v = Voigt1D(x_0=centroid,
amplitude_L=amplitude,
fwhm_L=L_fwhm,
fwhm_G=G_fwhm)
return v(x)
#NH = NH_list[9]
#NH = 1 * 10**12.5 * unit.cm**-2
b_list = np.linspace(10, 50, 5) * unit.km / unit.s
#b = b_list[1]
lam_0 = 1215.6701 * unit.Angstrom
om_0 = 2 * np.pi * const.c / lam_0
f = 0.4164
gamma = 6.265 * 10**8 * unit.s**-1 #* np.pi
for b in b_list:
W = []
a = np.float((gamma / (4 * om_0) * const.c / b).decompose())
H2 = Voigt1D(x_0=0,
amplitude_L=.57 / a,
fwhm_L=2 * a,
fwhm_G=2 * np.sqrt(np.log(2)))
lam = np.linspace(1200., 1230, 4000) * unit.Angstrom
om = 2 * np.pi * const.c / lam
u = np.array(((om - om_0) / om_0 * const.c / b).decompose())
sigma = 2 * np.sqrt(np.pi) * lam * f * sigma_0 / b * H(a, u)
for NH in NH_list:
tau = sigma * NH
flux = np.exp(-tau)
# plt.plot(lam,flux,label="b={},N={:.3e}".format(b,NH))
import numpy as np
from astropy.modeling.models import Voigt1D
import matplotlib.pyplot as plt
import pandas as pd
plt.figure()
df=pd.read_csv('7.csv')
x=df.X*5.692082111+(656-9.705)
y=df.Y
v1 = Voigt1D(x_0=657.2, amplitude_L=100, fwhm_L=0.02, fwhm_G=0.02)
plt.plot(x, v1(x),lw=0.7)
#plt.plot(x,y-12)
plt.show()
plt.errorbar(vel_2803, col_2803, yerr = sigma_tau2803, linewidth = 0.1, color = '#99ccff', label = '_nolegend_')
ax.plot(vel_2796, col_2796, color = '#2CA14B', label = names[0])
ax.plot(vel_2803, col_2803, color = '#2C6EA1', label = names[1])
plt.title("Uncertainty of Column Density in %s" %(gal[h]))
plt.xlabel("Velocity (km/s)")
plt.ylabel("Column Density (Particle/cm^2)")
ax.set_ylim(0, 5E12)
ax.set_xlim(-3000,500)
plt.legend(loc = 1)
f = interp1d(vel_kms[0], flux)
vel_new = np.linspace(-3000, 500, num = 3501, endpoint = True)
flux_king = f(vel_new)
voi = Voigt1D(amplitude_L=-0.5, x_0=0.0, fwhm_L=5.0, fwhm_G=5.0)
#xarr = np.linspace(,5.0,num=40)
#xarr = vel_kms[0]
xarr = vel_new
#yarr = voi(xarr)
#yarr = flux[good]
yarr = flux_king - 1.
voi_init = Voigt1D(amplitude_L=-1.0, x_0=-1000, fwhm_L=200, fwhm_G=200)
fitter = fitting.LevMarLSQFitter()
voi_fit = fitter(voi_init, xarr, yarr)
print(voi_fit)
ax = fig.add_subplot(3,1,3)
ax.plot(xarr,yarr+1, color='magenta')
ax.plot(xarr,voi_fit(xarr)+1, color='red')
vel_median = np.median(vel_kms[0][test_2796])
vel_new = np.linspace(vel_median-1000, vel_median+1000, num=2001, endpoint=True)
boom = len(vel_kms[0][test_2796])
one = vel_kms[0][test_2796][round(boom*0.2)]
two = vel_kms[0][test_2796][round(boom*0.4)]
thr = vel_kms[0][test_2796][round(boom*0.6)]
fou = vel_kms[0][test_2796][round(boom*0.8)]
flux_king = f(vel_new)
xarr = vel_new
yarr = flux_king - 1.
voi_init = Voigt1D(amplitude_L=-1.0, x_0=one, fwhm_L=two-one, fwhm_G=two-one)+Voigt1D(amplitude_L=-1.0, x_0=two, fwhm_L=thr-two, fwhm_G=thr-two)+Voigt1D(amplitude_L=-1.0, x_0=thr, fwhm_L=fou-thr, fwhm_G=fou-thr)+Voigt1D(amplitude_L=-1.0, x_0=fou, fwhm_L=fou-thr, fwhm_G=fou-thr)+Voigt1D(amplitude_L=-1.0, x_0=vel_median, fwhm_L=200, fwhm_G=200)
## Write function that combines cover fraction code with Voigt profile code ??????????? ##
##Correlation between amplitude and cover frac could be key##
fitter = fitting.LevMarLSQFitter()
voi_fit = fitter(voi_init, xarr, yarr)
ax.plot(xarr,voi_fit(xarr)+1, color='red')
plt.title("Mg 2796 Voigt Profile for Galaxy %s" %(gal[h]), **title_font)
plt.xlabel("Velocity(km/s)", **axis_font)
plt.ylabel("C.N. Flux", **axis_font)
plt.rc('xtick', labelsize=10)
plt.rc('ytick', labelsize=10)
ax.set_ylim (0,2)
ax.set_xlim(-3000,500)
tied_RVa = {'x_0': tie_RVa }
tied_RV = {'x_0': tie_RV }
#need good guess
#gg_init =( Voigt1D(x_0=lineB[0], amplitude_L=lineB[1], fwhm_L=lineB[2], fwhm_G=lineB[2] ) +
# Voigt1D(x_0=0, amplitude_L=lineC[1], fwhm_L=lineC[2], fwhm_G=lineC[2], tied=tied_RV) +
# Chebyshev1D(3, c0=C0, c1=C1, c2=C2, c3=C3, fixed={'c0': True, 'c1': True, 'c2': True, 'c3': True}) )
# 3 LINES MODEL CAT
gg_init =( Voigt1D(x_0=lineB[0], amplitude_L=lineB[1], fwhm_L=lineB[2], fwhm_G=lineB[2] ,bounds={"amplitude_L": (-0.1, 50)} ) +
Voigt1D(x_0=0, amplitude_L=lineC[1], fwhm_L=lineC[2], fwhm_G=lineC[2], tied=tied_RV, bounds={"amplitude_L": (-0.1, 50)}) +
Chebyshev1D(3, c0=C0, c1=C1, c2=C2, c3=C3, fixed={'c0': True, 'c1': True, 'c2': True, 'c3': True}) +
Voigt1D(x_0=lineA[0], amplitude_L=lineA[1], fwhm_L=lineA[2], fwhm_G=lineA[2], tied=tied_RVa, bounds={"amplitude_L": (-0.1, 50)} )
)
#Levenberg-Marquardt algorithm
#SLSQPLSQFitter() does not work for strong lines sequential quadratic programming
#fitter = fitting.SLSQPLSQFitter()
fitter = fitting.LevMarLSQFitter()
gg_fit = fitter(gg_init, x ,y)
#gg_fit.mean.tied = tie_center
import numpy as np
import os
from astropy.modeling.models import Voigt1D
import matplotlib.pyplot as plt
from astropy.modeling import fitting
from matplotlib.backends.backend_pdf import PdfPages
filename = 'ad_voigt_test.pdf'
with PdfPages(filename) as pdf:
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
x = np.arange(-3500, 500, 1)
v1 = Voigt1D(x_0=-1000, amplitude_L=-1, fwhm_L=100, fwhm_G=100)
ax.plot(x, v1(x) + 1.)
plt.ylabel('Continuum-normalized Flux')
plt.xlabel('Velocity [km/s]')
pdf.savefig()
plt.close()
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)
voi = Voigt1D(amplitude_L=-0.5, x_0=0.0, fwhm_L=5.0, fwhm_G=5.0)
xarr = np.linspace(-5.0, 5.0, num=40)
yarr = voi(xarr)
def __voigtPeak__(self, spectrum, x0, amplitudeL, fwhmG, fwhmL):
voigt_peak = Voigt1D(x_0=x0,
amplitude_L=amplitudeL,
fwhm_L=fwhmL,
fwhm_G=fwhmG)
return voigt_peak
def voigtfull(xs, mean, aL, sigmaL, sigmaG, offset):
p = [mean, aL, sigmaL, sigmaG, offset]
v = Voigt1D(x_0=p[0], amplitude_L=p[1], fwhm_L=p[2], fwhm_G=p[3])
full = p[4] + v(xs)
return full
def voigt(xs, p0):
'''Returns a gaussian function with inputs p0 over xs. p0 = [sigma, mean]'''
v = Voigt1D(x_0=p0[0], amplitude_L=p0[1], fwhm_L=p0[2], fwhm_G=p0[3])
full = p0[4] + v(xs)
return full
def voigt2(x, p):
decay = (np.exp(-x / p[4]))
v1 = Voigt1D(x_0=p[3], amplitude_L=0.18, fwhm_L=p[2], fwhm_G=p[5])
V = p[0] + decay * p[1] * v1(x) #voigt function
return V
def voigt(x, x0, a_l, fwhm_l, fwhm_g, offset):
v = Voigt1D(x_0=x0, amplitude_L=a_l, fwhm_L=fwhm_l, fwhm_G=fwhm_g)
return v(x) + offset
