def create_poly(coeff):
"""Create a polynomial model from coefficients.
Parameters
----------
coeff: list of float
The coefficients of the polynomial, constant term first, highest
order term last.
Returns
-------
astropy.modeling.polynomial.Polynomial1D object, or None if `coeff`
is empty.
"""
n = len(coeff)
if n < 1:
return None
coeff_dict = {}
for i in range(n):
key = "c{}".format(i)
coeff_dict[key] = coeff[i]
return polynomial.Polynomial1D(degree=n - 1, **coeff_dict)
def test_g_average():
"""Assign an array of values to tab_wl; these are strictly increasing,
but they're not uniformly spaced.
The abscissas and weights in combine_fast_slow are for three-point
Gaussian integration, which will (in principle) integrate a fifth-order
polynomial exactly. So for a test, we use a set of six polynomial
coefficients to create the tab_flat array by evaluating the polynomial
with those coefficients over the tab_wl array.
"""
# Generate an array (tab_wl) of wavelengths, not uniformly spaced.
wl_coeff = {'c0': 5., 'c1': 1., 'c2': -0.05}
wl_poly = polynomial.Polynomial1D(degree=2, **wl_coeff)
tab_index = np.arange(6000, dtype=np.float64) / 1000.
tab_wl = wl_poly(tab_index)
del tab_index
coeff = {
'c0': -41.9,
'c1': 30.7,
'c2': -8.3,
'c3': 1.15,
'c4': -7.8e-2,
'c5': 2.0e-3
}
poly = polynomial.Polynomial1D(degree=5, **coeff)
tab_flat = poly(tab_wl)
wl0 = 7.37
dwl0 = 2.99
lower = wl0 - dwl0 / 2.
upper = wl0 + dwl0 / 2.
# Compute the integral of the polynomial over wavelength.
# This function returns a tuple, the integral and an error estimate.
integral = quad(poly, lower, upper)[0]
# We want the average over the interval, not the integral itself.
correct_value = integral / (upper - lower)
# These three lines were copied from combine_fast_slow in flat_field.py
d = math.sqrt(0.6) / 2.
dx = np.array([-d, 0., d])
wgt = np.array([5., 8., 5.]) / 18.
value = g_average(wl0, dwl0, tab_wl, tab_flat, dx, wgt)
assert math.isclose(value, correct_value, rel_tol=1.e-8, abs_tol=1.e-8)
def fit_gaussians_to_spectrum(obs_flx, obs_wvl, idx_peaks):
print ' Fitting multiple ({:.0f}) gaussinas to the extracted arc spectrum'.format(len(idx_peaks))
#
median_val = np.median(obs_flx)
mean_wvl = 0 # np.mean(obs_wvl)
fit_model = models.Const1D(amplitude=np.median(obs_flx)) + polynomial.Polynomial1D(4)
for i_p, idx_p in enumerate(idx_peaks):
peak_val = obs_flx[idx_p] - median_val
# dela boljs brez nastavljenih boundov
fit_model += models.Gaussian1D(amplitude=peak_val, mean=obs_wvl[idx_p]-mean_wvl, stddev=0.1)#,
# bounds={'mean': (obs_wvl[idx_p]-0.5, obs_wvl[idx_p]+0.5),
# 'amplitude': (peak_val*0.8, peak_val*1.2)})
fit_t = fitting.LevMarLSQFitter()
fitted_model = fit_t(fit_model, obs_wvl-mean_wvl, obs_flx)
return fitted_model
def poly_fit(xp, yp, grado_pol):
t_init = polynomial.Polynomial1D(degree=int(grado_pol))
fit_t = fitting.LevMarLSQFitter()
t = fit_t(t_init, xp, yp)
return t
#Calculando o potencial pra uma galáxia gaussiana fake p = Potential() coords = np.random.normal(0, 3, 1000) mass = np.random.choice([1., 2.], 1000) p.coords = coords p.mass = mass c, pot, n = p.evaluate_potential(n_bins=10) #Ajustando uma gaussiana nessa galáxia com o astropy g_init = models.Gaussian1D(amplitude=-1., mean=0, stddev=1.) fit_g = fitting.LevMarLSQFitter() g = fit_g(g_init, c, pot) #Ajustando um polinômio p_init = polynomial.Polynomial1D(degree=6) p = fit_g(p_init, c, pot) #Minimizando m = minimize(g, 0) #plotando plt.figure(figsize=(8, 5)) plt.plot(c, pot, 'ko') plt.plot(c, g(c), label='best gaussian fit') plt.plot(c, p(c), label='best sixth degree polynomial fit') plt.plot(m.x, m.fun, 'X', label='minimum') plt.legend() plt.show()
