def cheb1Dfit(x, y, order=5, weight=None, niteration=0, \
high_nsig=0.0, low_nsig=0.0):
stat = True
n = len(x)
if weight is None:
weight1 = np.ones(n)
else:
weight1 = copy.deepcopy(weight)
c = chebfit(x,y,order,w=weight1)
for k in range(niteration):
residualdata = y - chebval(x,c)
# Calculating weighted standard deviation
mean, sig = weighted_average_std(residualdata, weight1)
# Clipiing the data
if mean != 0:
if sig/mean > 1.0e-10 or sig/mean < -1.0e-10:
highlimit = high_nsig * sig
lowlimit = -low_nsig * sig
for j in range(n):
if residualdata[j] > highlimit or residualdata[j] < lowlimit:
weight1[j] = 0.0
# Fitting again
with warnings.catch_warnings():
warnings.filterwarnings("error")
try:
c = chebfit(x,y,order,w=weight1)
except np.polynomial.polyutils.RankWarning:
print('RankWarning')
stat = False
return c, weight1, stat
def normalize_spectra(self):
"""
Normalize the data and model spectra
"""
self.spectra['m_flux_norm'] = deepcopy(self.spectra['m_flux'])
self.spectra['d_flux_norm'] = deepcopy(self.spectra['d_flux'])
self.spectra['unc_norm'] = deepcopy(self.spectra['unc'])
chunks = 1000
min_ = min(self.spectra['wave'])
max_ = max(self.spectra['wave'])
num = int((max_ - min_) / chunks) + 1
for i in range(num):
k = ((self.spectra['wave'] >= min_ + chunks * i) &
(self.spectra['wave'] <= min_ + chunks * (i + 1)))
if len(self.spectra['d_flux_norm'][k]) < 10:
continue
coeffs = chebfit(self.spectra['wave'][k],
self.spectra['d_flux_norm'][k], 2)
poly = chebval(self.spectra['wave'][k], coeffs)
self.spectra['d_flux_norm'][
k] = self.spectra['d_flux_norm'][k] / poly
self.spectra['unc_norm'][k] = self.spectra['unc_norm'][k] / poly
coeffs = chebfit(self.spectra['wave'][k],
self.spectra['m_flux_norm'][k], 2)
poly = chebval(self.spectra['wave'][k], coeffs)
self.spectra['m_flux_norm'][
k] = self.spectra['m_flux_norm'][k] / poly
def polyLine(startPt, endPt, polyIndex, t, sgf, data):
x = []
y = []
vt = []
for j in range(startPt, endPt): #for loop 特性,需要到endpt 因此要+1
x.append(t[j])
y.append(float(sgf[j]))
vt.append(float(data[j]))
#=================Polyfit 回歸多項式 =========
# =============================================================================
# tp=np.polyfit(x,y,polyIndex)
# ys_line=np.polyval(tp,x)
# =============================================================================
#=================chebyshev 回歸多項式 =========
coeffRaw = chy.chebfit(x, vt, polyIndex)
coeffSGF = chy.chebfit(x, y, polyIndex)
ys_lineRaw = chy.chebval(x, coeffRaw)
ys_lineSGF = chy.chebval(x, coeffSGF)
#=================rsq: 決定係數=========
rsqSGF = round(
coeff_of_determination(np.array(y), ys_lineRaw, startPt, endPt), 2)
rsqRW = round(
coeff_of_determination(np.array(vt), ys_lineRaw, startPt, endPt), 2)
return coeffRaw, ys_lineRaw, rsqSGF, rsqRW
def define_background(self, q, I, k, plot=False, az_idx=0):
""" Background profile - q, I points with Chebdev poly fit.
Can supply 2d array for q and I, which define the background
intensity as a function of azimuthal position. This is predominantly
for use with the pyxe package!
Args:
q (ndarray): 1d or 2d array with q positions
I (ndarray): 1d or 2d array with intensity values
k (int): Order for Cheyshev polynomial
plot (bool): True/False
az_idx (int): Azimuthal slice to plot
"""
# Calculate fit and apply to all slices
if np.array(q).ndim == 1:
f = chebfit(q, I, k)
# Calculate fit for each slice
else:
assert self.phi.size == q.shape[0]
f = np.zeros((self.q.shape[0], k + 1))
for az in range(q.shape[0]):
finite = np.isfinite(q[az])
f[az] = chebfit(q[az][finite], I[az][finite], k)
if plot:
plt.plot(self.q, chebval(self.q, f[az_idx]), 'k-')
x = q[az_idx] if q.ndim == 2 else q
y = I[az_idx] if q.ndim == 2 else I
plt.plot(x, y, 'r+')
plt.show()
self._back = f
def handle_summary(outname=None, filelist=[]):
"""Extract all std-correction vectors and create ensemble (obsolete)"""
if outname is None:
outname = 'correction.npy'
# Get a list of good correction vectors
keepers = []
for ifile in filelist:
print ifile
f = np.load(ifile)[0]
# Correction values should be small
# if "std-correction" not in data.keys():
if np.nanmin(f['std-correction']) < 9:
keepers.append(f)
print "Keeping %s" % ifile
# Set fiducial wavelengths from first correction vector
corl = keepers[0]['nm'].copy()
# Insert first vector
cor = np.zeros((len(corl), len(keepers)))
cor[:, 0] = keepers[0]['std-correction'].copy()
# Insert the rest of the vectors
for ix, keeper in enumerate(keepers[1:]):
f = interp1d(keeper['nm'], keeper['std-correction'],
bounds_error=False, fill_value=np.nan)
cor[:, ix] = f(corl)
# Create mean correction
cs = np.nanmean(cor, 1)
# Fit wl coefficients
ccs = chebfit(corl, cs, 6)
# Create output
cor = [{"nm": corl, "cor": cs, "coeff": ccs}]
np.save(outname, cor)
def main(plot=False):
warnings.filterwarnings('error', category=RuntimeWarning)
days, fuel_mass, uglovi, snaga = pop_init(gen=1)
print(days.shape)
print(fuel_mass.shape)
print(uglovi.shape)
print(snaga.shape)
pop_fit = np.empty(broj_jedinki)
koeficijenti = np.empty((broj_jedinki, podaci.chebdeg + 1))
fitness = np.loadtxt('gen_99.txt', dtype=float, usecols=364)
ind = np.argmin(fitness)
pop_elita = None
print('Fitness: ', fitness[ind])
for i in range(broj_gen):
for j in range(30, 40):
print(j)
print('File fitness: ', fitness[j])
print('Launch offset:', days[j])
print('Fuel mass:', fuel_mass[j] / 255 * podaci.max_fuel_mass)
#print('Uglovi:', (uglovi[j]%(2*pi))*360/(2*pi))
_r, _, _, min_dist_dest = simulacija_pogon.simulacija(
days[j], fuel_mass[j], uglovi[j], snaga[j], y_max)
# if pop_fit[j] == -1:
pop_fit[j] = fitnes(min_dist_dest)
koeficijenti[j] = chebfit(x_osa(broj_segmenata), uglovi[j],
podaci.chebdeg)
if plot:
x, y = crtanje_planeta(plot=False)
plt.plot(np.array(x[2]) / au,
np.array(y[2]) / au,
'b--',
np.array(x[3]) / au,
np.array(y[3]) / au,
'r-.',
_r[:, 0] / au,
_r[:, 1] / au,
'g:',
linewidth=0.9)
plt.plot(0.0, 0.0, 'k*', markersize=7)
plt.axis('scaled')
plt.xlabel('x-osa [astronomska jedinica]')
plt.ylabel('y-osa [astronomska jedinica]')
plt.title('Flyby Marsa')
plt.show()
# print(podaci.trajanje)
# pop_bit = np.array([[0, 1, 1, 0, 1, 0, 1, 0, 0, 1, 15.1],
# [1, 1, 0, 1, 0, 0, 1, 0, 1, 0, 5.6],
# [0, 0, 0, 1, 0, 1, 0, 0, 1, 1, 10.8],
# [1, 0, 1, 0, 1, 0, 1, 1, 0, 1, 2.6],
# [0, 1, 1, 0, 1, 0, 0, 0, 1, 0, 4.9]])
# pop_elita = np.array([[0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 1.5]])
pop_bit = float_to_bit(days[:, np.newaxis], fuel_mass, koeficijenti,
snaga)
# print(pop_bit)
pop_bit_new, pop_elita = genetski_algoritam.genetski_algoritam(
pop_bit, pop_elita, podaci.p_elit, podaci.p_mut)
# print(pop_bit_new)
# print(pop_elita)
# date_time, fuel_mass, uglovi, snaga = bit_to_float(pop_bit_new)
print(np.min(pop_fit))
def fit_poly(w, r, type, order):
if type == "legendre":
coef = legfit(w, r, order)
return coef
elif type == "chebyshev":
coef = chebfit(w, r, order)
return coef
def handle_summary(outname=None, filelist=[]):
if outname is None: outname = 'correction.npy'
keepers = []
for file in filelist:
print file
f = np.load(file)[0]
if np.nanmin(f['std-correction']) < 9:
keepers.append(f)
print "Keeping %s" % file
corl = keepers[0]['nm'].copy()
cor = np.zeros((len(corl), len(keepers)))
cor[:,0] = keepers[0]['std-correction'].copy()
for ix, keeper in enumerate(keepers[1:]):
f = interp1d(keeper['nm'], keeper['std-correction'], bounds_error=False,
fill_value = np.nan)
cor[:,ix] = f(corl)
cs = np.nanmean(cor,1)
ccs = chebfit(corl, cs, 6)
cor = [{"nm": corl, "cor": cs, "coeff": ccs}]
np.save(outname, cor)
def cheb_fitcurve(x, y, order):
x = cheb.chebpts2(len(x))
order = 64
coef = legend.legfit(x, y, order)
assert_equal(len(coef), order + 1)
y1 = legend.legval(x, coef)
err_1 = np.linalg.norm(y1 - y) / np.linalg.norm(y)
coef = cheb.chebfit(x, y, order)
assert_equal(len(coef), order + 1)
thrsh = abs(coef[0] / 1000)
for i in range(len(coef)):
if abs(coef[i]) < thrsh:
coef = coef[0:i + 1]
break
y2 = cheb.chebval(x, coef)
err_2 = np.linalg.norm(y2 - y) / np.linalg.norm(y)
plt.plot(x, y2, '.')
plt.plot(x, y, '-')
plt.title("nPt={} order={} err_cheby={:.6g} err_legend={:.6g}".format(
len(x), order, err_2, err_1))
plt.show()
assert_almost_equal(cheb.chebval(x, coef), y)
#
return coef
def polyLine(startPt, endPt, polyIndex, t, sgf, data):
x = []
y = []
vt = []
for j in range(startPt, endPt): #for loop 特性,需要到endpt 因此要+1
x.append(t[j])
y.append(float(sgf[j]))
vt.append(float(data[j]))
#=================Polyfit 回歸多項式 =========
# =============================================================================
# tp=np.polyfit(x,y,polyIndex)
# ys_line=np.polyval(tp,x)
# =============================================================================
#=================chebyshev 回歸多項式 =========
coeff = chy.chebfit(x, y, polyIndex)
# =============================================================================
# print(chy.Chebyshev.fit(x,y,polyIndex))
# print(coeff)
# print(np.poly1d(coeff))
# =============================================================================
ys_line = chy.chebval(x, coeff)
# =============================================================================
# approximated_values = np.poly1d(coeff)(x)
# print(ys_line,approximated_values)
# =============================================================================
#=================rsq: 決定係數=========
rsqSGF = round(
coeff_of_determination(np.array(y), ys_line, startPt, endPt), 3)
rsqTT = round(
coeff_of_determination(np.array(vt), ys_line, startPt, endPt), 3)
return coeff, ys_line, rsqSGF, rsqTT
def che_fit(txt_path, deg, fit_save_dir):
'''
input : a txt of label, imgw, imgh, x, y, w, h, centerx, centery, 360deg
return label, imgw, imgh, x, y, w, h, centerx, centery, coef(16,24)
'''
with open(txt_path, 'r') as f:
img_info = np.loadtxt(txt_path)
img_info = img_info.reshape(-1, 369)
new_path = os.path.join(fit_save_dir, txt_path.split('/')[-1])
results = []
for objects_info in img_info:
# 1,360
objects_new = np.zeros(9 + deg + 1)
objects_new[0:9] = objects_info[0:9]
bboxw = objects_info[5]
bboxh = objects_info[6]
bbox_len = np.sqrt(bboxw * bboxw + bboxh * bboxh)
r = objects_info[9:] / float(bbox_len)
theta = np.linspace(-1, 1, 360)
coefficient, Res = chebyshev.chebfit(theta, r, deg, full=True)
objects_new[9:] = np.array(coefficient)
results.append(objects_new)
results = np.array(results)
np.savetxt(new_path, results)
def handle_summary(outname=None, filelist=[]):
if outname is None: outname = 'correction.npy'
keepers = []
for file in filelist:
print file
f = np.load(file)[0]
if np.nanmin(f['std-correction']) < 9:
keepers.append(f)
print "Keeping %s" % file
corl = keepers[0]['nm'].copy()
cor = np.zeros((len(corl), len(keepers)))
cor[:, 0] = keepers[0]['std-correction'].copy()
for ix, keeper in enumerate(keepers[1:]):
f = interp1d(keeper['nm'],
keeper['std-correction'],
bounds_error=False,
fill_value=np.nan)
cor[:, ix] = f(corl)
cs = np.nanmean(cor, 1)
ccs = chebfit(corl, cs, 6)
cor = [{"nm": corl, "cor": cs, "coeff": ccs}]
np.save(outname, cor)
def test_recursivechebyshevfunction(self):
'''Test routines to compute using Chebyshev polynomials
recursively.'''
from scipy.linalg import funm
from numpy.polynomial.chebyshev import chebfit, chebval
from numpy import exp, linspace
# Starting Matrix
matrix1 = self.create_matrix(scaled=True)
self.write_matrix(matrix1, self.input_file)
# Function
x = linspace(-1.0, 1.0, 200)
y = [exp(i) for i in x]
coef = chebfit(x, y, 16 - 1)
# Check Matrix
dense_check = funm(matrix1.todense(), lambda x: chebval(x, coef))
self.CheckMat = csr_matrix(dense_check)
# Result Matrix
input_matrix = nt.Matrix_ps(self.input_file, False)
poly_matrix = nt.Matrix_ps(self.mat_dim)
polynomial = nt.ChebyshevPolynomial(len(coef))
for j in range(0, len(coef)):
polynomial.SetCoefficient(j, coef[j])
permutation = nt.Permutation(input_matrix.GetLogicalDimension())
permutation.SetRandomPermutation()
self.fsp.SetLoadBalance(permutation)
polynomial.ComputeFactorized(input_matrix, poly_matrix, self.fsp)
poly_matrix.WriteToMatrixMarket(result_file)
comm.barrier()
self.check_result()
def Create(x, y, itm_mask, degree):
# filter out-of-money paths
x_filtered = list(itertools.compress(x, itm_mask))
y_filtered = list(itertools.compress(y, itm_mask))
interpolator = ChebyshevInterpolator(
chebyshev.chebfit(x_filtered, y_filtered, degree))
return interpolator
def get_coff(hull, degs):
dist_degs = np.array(compute_dist(degs, hull))
dist_len = np.sqrt(dist_degs[:, 0]**2 + dist_degs[:, 1]**2)
coefficient, Res = chebyshev.chebfit(degs, dist_len, 8, full=True)
return coefficient
def polynomial_max_like(self):
order = int(self.param["order"])
ratio = self.y_model / self.y
errs = np.abs(self.y_err * self.y_model / self.y**2)
coefs = chebfit(self.x, ratio, order, w=1. / errs)
self.poly_coefs = np.array(coefs)
self.model = chebval(self.x, coefs)
def main1():
days, fuel_mass, uglovi, snaga = pop_init()
print(days)
koeficijenti = chebfit(x_osa(broj_segmenata), uglovi.T, podaci.chebdeg).T
# print(uglovi, snaga)
pop_bit = float_to_bit(days, fuel_mass, koeficijenti, snaga)
days2, fuel_mass2, uglovi2, snaga2 = bit_to_float(pop_bit)
print(days2 - days)
def getCont(wave, flam, ss=21.):
"""Fits and returns continuum"""
cont = maximum_filter(savgol_filter(flam, 5, 2, mode='interp'), ss)
chebx = wave - wave.min()
chebx *= 2. / chebx.max()
chebx -= 1.
contfit = chebfit(chebx, cont, 6)
cont = chebval(chebx, contfit)
return cont
def spline_interpolation(pointx,
pointy,
wave,
wave_temp,
flux,
flux_temp,
chebfitval,
linewidth=2.0,
endpoints='y',
endpoint_order=4):
"""Sort spline points and interpolate between marked continuum points"""
from numpy.polynomial import chebyshev
from scipy.interpolate import splrep, splev
# Insert endpoints
if endpoints == "y" or endpoints == "t":
sort_array = np.argsort(pointx)
x = np.array(pointx)[sort_array]
y = np.array(pointy)[sort_array]
chebfit = chebyshev.chebfit(x, y, deg=endpoint_order)
chebfitval = chebyshev.chebval(wave, chebfit)
i = wave[150], wave[-150]
window1, window2 = ((i[0] - 70) <= wave) & (wave <= (i[0] + 70)), (
(i[1] - 70) <= wave) & (wave <= (i[1] + 70))
y_val = np.median(chebfitval[window1]).astype(np.float64), np.median(
chebfitval[window2]).astype(np.float64)
pointx = np.concatenate([pointx, i])
pointy = np.concatenate([pointy, y_val])
ind_uni = f8(pointx)
pointx = np.array(pointx)[ind_uni]
pointy = np.array(pointy)[ind_uni]
# Sort numerically
# print(pointx)
sort_array = np.argsort(pointx)
# print(sort_array, pointx)
# x, y = pointx[sort_array], pointy[sort_array]
x = np.array(pointx)[sort_array]
y = np.array(pointy)[sort_array]
# Interpolate
spline = splrep(x, y, k=3)
def tap(s, x_temp):
# offset wavelength array to ensure that wavelength does not become negative in the mirror
diff = max(x_temp) - min(x_temp)
k = np.tanh((x_temp - min(x_temp)) / (diff / 10.0))
return k
if endpoints == "t":
continuum = splev(wave, spline) * (tap(10, wave)) * (tap(
10, 2 * np.mean(wave) -
wave)) + chebfitval * (1 - tap(10, wave)) + chebfitval * (
1 - tap(10, 2 * np.mean(wave) - wave))
else:
continuum = splev(wave, spline)
return continuum
def cheby_newton_root(z, f, z0=None, degree=512):
import numpy.polynomial.chebyshev as npcheb
import scipy.optimize as scpop
Lz = np.max(z) - np.min(z)
if z0 is None:
z0 = Lz / 2
def to_x(z, Lz):
# convert back to [-1,1]
return (2 / Lz) * z - 1
def to_z(x, Lz):
# convert back from [-1,1]
return (x + 1) * Lz / 2
logger.info("searching for roots starting from z={}".format(z0))
x = to_x(z, Lz)
x0 = to_x(z0, Lz)
cheb_coeffs = npcheb.chebfit(x, f, degree)
cheb_interp = npcheb.Chebyshev(cheb_coeffs)
cheb_der = npcheb.chebder(cheb_coeffs)
def newton_func(x_newton):
return npcheb.chebval(x_newton, cheb_coeffs)
def newton_derivative_func(x_newton):
return npcheb.chebval(x_newton, cheb_der)
try:
x_root = scpop.newton(newton_func,
x0,
fprime=newton_derivative_func,
tol=1e-10)
z_root = to_z(x_root, Lz)
except:
logger.info("error in root find")
x_root = np.nan
z_root = np.nan
logger.info("newton: found root z={} (x0:{} -> {})".format(
z_root, x0, x_root))
for x0 in x:
print(x0, newton_func(x0))
a = Lz / 4
b = Lz * 3 / 4
logger.info("bisecting between z=[{},{}] (x=[{},{}])".format(
a, b, to_x(a, Lz), to_x(b, Lz)))
logger.info("f(a) = {} and f(b) = {}".format(newton_func(to_x(a, Lz)),
newton_func(to_x(b, Lz))))
x_root_2 = scpop.bisect(newton_func, to_x(a, Lz), to_x(b, Lz))
z_root_2 = to_z(x_root_2, Lz)
logger.info("bisect: found root z={} (x={})".format(z_root_2, x_root_2))
return z_root_2
def to_coefficients(self, f):
from numpy.polynomial.chebyshev import chebfit
import numpy as np
# Convert infinite grid to xg = [-1, 1]
xg = self.zg / np.sqrt(self.C ** 2 + self.zg ** 2)
# Get coefficients for standard Chebyshev polynomials
c, res = chebfit(xg, f, deg=self.N, full=True)
return c
def spline_interpolation(pointx, pointy, wave, wave_temp, flux, flux_temp, axis,
leg_instance, con_instance, linewidth= 2.0, endpoints = 'y',
endpoint_order = 4):
"""Sort spline points and interpolate between marked continuum points"""
from numpy.polynomial import chebyshev
from scipy.interpolate import splrep,splev
#Insert endpoints
if endpoints == 'y':
sort_array = np.argsort(pointx)
x = np.array(pointx)[sort_array]
y = np.array(pointy)[sort_array]
chebfit = chebyshev.chebfit(x, y, deg = endpoint_order)
chebfitval = chebyshev.chebval(wave, chebfit)
i =wave[150], wave[-150]
window1, window2 = ((i[0]-70)<=wave) & (wave<=(i[0]+70)), ((i[1]-70)<=wave) & (wave<=(i[1]+70))
y_val = np.median(chebfitval[window1]).astype(np.float64),np.median(chebfitval[window2]).astype(np.float64)
pointx = np.concatenate([pointx,i])
pointy = np.concatenate([pointy,y_val])
ind_uni = f8(pointx)
pointx = np.array(pointx)[ind_uni]
pointy = np.array(pointy)[ind_uni]
try:
leg_instance.remove()
except AttributeError:
pass
finally:
leg_instance, = axis.plot(wave, chebfitval, color='black', lw = linewidth,
label = 'legendre fit', zorder = 10)
# print(pointx,pointy)
#Sort numerically
sort_array = np.argsort(pointx)
x = np.array(pointx)[sort_array]
y = np.array(pointy)[sort_array]
from gen_methods import smooth
#Interpolate
spline = splrep(x,y, k=3)
continuum = splev(wave,spline)
continuum = smooth(continuum, window_len=1, window='hanning')
#Plot
try:
con_instance.remove()
except AttributeError:
pass
finally:
con_instance, = axis.plot(wave,continuum, color='red', lw = linewidth,
label = 'continuum', zorder = 10)
return continuum, leg_instance, con_instance
def to_coefficients(self, f):
from numpy.polynomial.chebyshev import chebfit
import numpy as np
# Convert grid to standard xg = [-1, 1]
xg = (self.zg - self.zmin) / self.L * 2. - 1.
# Get coefficients for standard Chebyshev polynomials
c, res = chebfit(xg, f, deg=self.N, full=True)
return c
def to_coefficients(self, f):
"""Convert from grid values to coefficients"""
from numpy.polynomial.chebyshev import chebfit
# Convert semi-infinite grid to xg = [-1, 1]
xg = (self.zg - self.C) / (self.zg + self.C)
# Get coefficients for standard Chebyshev polynomials
c, res = chebfit(xg, f, deg=self.N, full=True)
return c
def test_chebfit(self) :
def f(x) :
return x*(x - 1)*(x - 2)
# Test exceptions
assert_raises(ValueError, ch.chebfit, [1], [1], -1)
assert_raises(TypeError, ch.chebfit, [[1]], [1], 0)
assert_raises(TypeError, ch.chebfit, [], [1], 0)
assert_raises(TypeError, ch.chebfit, [1], [[[1]]], 0)
assert_raises(TypeError, ch.chebfit, [1, 2], [1], 0)
assert_raises(TypeError, ch.chebfit, [1], [1, 2], 0)
# Test fit
x = np.linspace(0,2)
y = f(x)
coef = ch.chebfit(x, y, 3)
assert_equal(len(coef), 4)
assert_almost_equal(ch.chebval(x, coef), y)
coef = ch.chebfit(x, y, 4)
assert_equal(len(coef), 5)
assert_almost_equal(ch.chebval(x, coef), y)
coef2d = ch.chebfit(x, np.array([y,y]).T, 4)
assert_almost_equal(coef2d, np.array([coef,coef]).T)
def test_chebfit(self):
def f(x):
return x * (x - 1) * (x - 2)
# Test exceptions
assert_raises(ValueError, ch.chebfit, [1], [1], -1)
assert_raises(TypeError, ch.chebfit, [[1]], [1], 0)
assert_raises(TypeError, ch.chebfit, [], [1], 0)
assert_raises(TypeError, ch.chebfit, [1], [[[1]]], 0)
assert_raises(TypeError, ch.chebfit, [1, 2], [1], 0)
assert_raises(TypeError, ch.chebfit, [1], [1, 2], 0)
# Test fit
x = np.linspace(0, 2)
y = f(x)
coef = ch.chebfit(x, y, 3)
assert_equal(len(coef), 4)
assert_almost_equal(ch.chebval(x, coef), y)
coef = ch.chebfit(x, y, 4)
assert_equal(len(coef), 5)
assert_almost_equal(ch.chebval(x, coef), y)
coef2d = ch.chebfit(x, np.array([y, y]).T, 4)
assert_almost_equal(coef2d, np.array([coef, coef]).T)
def fit_range(R, T, calrange, redo_lims=True):
# try binning in T to estimate noise for each point
inds = np.where((T < calrange.upper) & (T > calrange.lower))
Z = np.log10(R[inds])
if redo_lims:
ZL = np.min(Z)
ZU = np.max(Z)
else:
ZL = calrange.ZL
ZU = calrange.ZU
x = ((Z - ZL) - (ZU - Z)) / (ZU - ZL)
w = get_weights(x, T[inds])
return x, cheby.chebfit(x, T[inds], calrange.order, w=w)
def test_chebfit(self):
def f(x):
return x * (x - 1) * (x - 2)
# Test exceptions
assert_raises(ValueError, ch.chebfit, [1], [1], -1)
assert_raises(TypeError, ch.chebfit, [[1]], [1], 0)
assert_raises(TypeError, ch.chebfit, [], [1], 0)
assert_raises(TypeError, ch.chebfit, [1], [[[1]]], 0)
assert_raises(TypeError, ch.chebfit, [1, 2], [1], 0)
assert_raises(TypeError, ch.chebfit, [1], [1, 2], 0)
assert_raises(TypeError, ch.chebfit, [1], [1], 0, w=[[1]])
assert_raises(TypeError, ch.chebfit, [1], [1], 0, w=[1, 1])
# Test fit
x = np.linspace(0, 2)
y = f(x)
#
coef3 = ch.chebfit(x, y, 3)
assert_equal(len(coef3), 4)
assert_almost_equal(ch.chebval(x, coef3), y)
#
coef4 = ch.chebfit(x, y, 4)
assert_equal(len(coef4), 5)
assert_almost_equal(ch.chebval(x, coef4), y)
#
coef2d = ch.chebfit(x, np.array([y, y]).T, 3)
assert_almost_equal(coef2d, np.array([coef3, coef3]).T)
# test weighting
w = np.zeros_like(x)
yw = y.copy()
w[1::2] = 1
y[0::2] = 0
wcoef3 = ch.chebfit(x, yw, 3, w=w)
assert_almost_equal(wcoef3, coef3)
#
wcoef2d = ch.chebfit(x, np.array([yw, yw]).T, 3, w=w)
assert_almost_equal(wcoef2d, np.array([coef3, coef3]).T)
def test_chebfit(self) :
def f(x) :
return x*(x - 1)*(x - 2)
# Test exceptions
assert_raises(ValueError, cheb.chebfit, [1], [1], -1)
assert_raises(TypeError, cheb.chebfit, [[1]], [1], 0)
assert_raises(TypeError, cheb.chebfit, [], [1], 0)
assert_raises(TypeError, cheb.chebfit, [1], [[[1]]], 0)
assert_raises(TypeError, cheb.chebfit, [1, 2], [1], 0)
assert_raises(TypeError, cheb.chebfit, [1], [1, 2], 0)
assert_raises(TypeError, cheb.chebfit, [1], [1], 0, w=[[1]])
assert_raises(TypeError, cheb.chebfit, [1], [1], 0, w=[1,1])
# Test fit
x = np.linspace(0,2)
y = f(x)
#
coef3 = cheb.chebfit(x, y, 3)
assert_equal(len(coef3), 4)
assert_almost_equal(cheb.chebval(x, coef3), y)
#
coef4 = cheb.chebfit(x, y, 4)
assert_equal(len(coef4), 5)
assert_almost_equal(cheb.chebval(x, coef4), y)
#
coef2d = cheb.chebfit(x, np.array([y,y]).T, 3)
assert_almost_equal(coef2d, np.array([coef3,coef3]).T)
# test weighting
w = np.zeros_like(x)
yw = y.copy()
w[1::2] = 1
y[0::2] = 0
wcoef3 = cheb.chebfit(x, yw, 3, w=w)
assert_almost_equal(wcoef3, coef3)
#
wcoef2d = cheb.chebfit(x, np.array([yw,yw]).T, 3, w=w)
assert_almost_equal(wcoef2d, np.array([coef3,coef3]).T)
def find_func(x, yreal):
import numpy as np
from numpy.polynomial.chebyshev import chebfit, chebval
f = np.polyfit(x, yreal, len(yreal))
c = chebfit(x, yreal, len(yreal))
ytest_poly = [round(np.polyval(f, xi), 3) for xi in x]
ytest_cheb = [round(chebval(xi, c), 3) for xi in x]
ytest = ytest_poly
function_coeff = f
return ytest, function_coeff
def main():
warnings.filterwarnings("error", category=RuntimeWarning)
comm = MPI.COMM_WORLD
rank = comm.Get_rank()
n = comm.Get_size()
pop_elita = None
if rank == 0:
days, fuel_mass, uglovi, snaga = pop_init()
pop_fit = np.empty(broj_jedinki)
koeficijenti = np.empty((broj_jedinki, podaci.chebdeg + 1))
for i in range(broj_gen):
for j in range(broj_jedinki):
proc_id = comm.recv(source=MPI.ANY_SOURCE, tag=1)
comm.send(
(j, days[j], fuel_mass[j], uglovi[j], snaga[j], y_max),
dest=proc_id)
waiting = n
while waiting > 0:
data = comm.recv(source=MPI.ANY_SOURCE, tag=2)
(j, r_, _, _, min_dist_dest) = data
pop_fit[j] = fitnes(min_dist_dest)
koeficijenti[j] = chebfit(x_osa(broj_segmenata), uglovi[j],
podaci.chebdeg)
waiting = waiting - 1
pop_bit = float_to_bit(days[:, np.newaxis], fuel_mass,
koeficijenti, snaga, pop_fit)
# print(pop_bit)
pop_bit_new, pop_elita = genetski_algoritam.genetski_algoritam(
pop_bit, pop_elita, podaci.p_elit, podaci.p_mut)
days, fuel_mass, uglovi, snaga = bit_to_float(pop_bit_new)
for it in range(broj_jedinki):
days[it] = days[it] % max_time_span
if i % 10 == 0:
np.savetxt('gen_' + str(i) + '.txt', pop_bit_new, fmt='%d')
for i in range(1, n):
proc_id = comm.recv(source=MPI.ANY_SOURCE)
comm.send(-1, dest=proc_id)
else:
while True:
comm.send(comm.Get_rank(), dest=0, tag=1)
data = comm.recv(source=0)
if data != -1:
_r, _v, _step, min_dist_dest = simulacija_pogon.simulacija(
data[1], data[2], data[3], data[4], data[5])
comm.send((data[0], _r, _v, _step, min_dist_dest),
dest=0,
tag=2)
else:
print('Proc ' + str(comm.Get_rank()) + ' logs out')
break
def interpCby(self, x_in, t_in, deg=20):
# print("t_in",t_in)
# print("x_in",x_in)
self.timeb = [t_in.min(), t_in.max()]
self.chebdeg = deg
x_in = np.transpose(x_in)
t_fit = t_in - self.timeb[0]
cheb_coef = cheby.chebfit(x=t_fit, y=x_in, deg=deg, full=False)
# print(cheby.chebfit(x=t_in,y=x_in,deg=deg,full=True))
# print("cheb_coef",cheb_coef)
# cheb_coef = cheb_coef[0]
self.cheb = cheb_coef
def cheby_fit(contours, num_coefs):
coefs, r_maxs = [], []
for contour in contours:
contour = np.array(contour)
theta = np.linspace(-1, 1, 360, endpoint=False)
r = contour[:, 1]
r_max = np.max(r)
r = r / r_max
coef, res = chebyshev.chebfit(theta, r, num_coefs, full=True)
coefs.append(coef)
r_maxs.append(r_max)
coefs = np.array(coefs)
r_maxs = np.array(r_maxs)
cheby_coef = np.hstack((coefs, r_maxs[:, np.newaxis]))
return cheby_coef
def continuum_fit(wl, flux, fluxerr, edge_mask_len=50):
"""
Small function to estimate continuum. Takes spectrum and uses only the edges, as specified by edge_mask_len, fits
a polynomial and returns the fit.
:param wl: Wavelenght array in which to estimate continuum
:param flux: Corresponding flux array
:param fluxerr: Corresponding array containing flux errors
:param edge_mask_len: Size of edges to use to interpolate continuum
:return: Interpolated continuum
"""
from numpy.polynomial import chebyshev
wl_chebfit = np.hstack((wl[:edge_mask_len],wl[-edge_mask_len:]))
mask_cont = np.array([i for i,n in enumerate(wl) if n in wl_chebfit])
chebfit = chebyshev.chebfit(wl_chebfit, flux[mask_cont], deg = 1, w=1/fluxerr[mask_cont]**2)
chebfitval = chebyshev.chebval(wl, chebfit)
return chebfitval, chebfit
def th_plot(data, tt):
fields = 'concurence', 'iops_mediana', 'lat_mediana'
conc_4k = filter_data('concurrence_test_' + tt, fields, blocksize='4k')
filtered_data = sorted(list(conc_4k(data)))
x, iops, lat = zip(*filtered_data)
_, ax1 = plt.subplots()
xnew = np.linspace(min(x), max(x), 50)
# plt.plot(xnew, power_smooth, 'b-', label='iops')
ax1.plot(x, iops, 'b*')
for degree in (3,):
c = chebfit(x, iops, degree)
vals = chebval(xnew, c)
ax1.plot(xnew, vals, 'g--')
def th_plot(data, tt):
fields = 'concurence', 'iops_mediana', 'lat_mediana'
conc_4k = filter_data('concurrence_test_' + tt, fields, blocksize='4k')
filtered_data = sorted(list(conc_4k(data)))
x, iops, lat = zip(*filtered_data)
_, ax1 = plt.subplots()
xnew = np.linspace(min(x), max(x), 50)
# plt.plot(xnew, power_smooth, 'b-', label='iops')
ax1.plot(x, iops, 'b*')
for degree in (3, ):
c = chebfit(x, iops, degree)
vals = chebval(xnew, c)
ax1.plot(xnew, vals, 'g--')
def chebForwardTransform(orders, locations, functionVals):
if len(locations.shape) == 1:
return np.array(cheb.chebfit(locations, functionVals, orders[0]))
else:
if locations.shape[1] == 2:
V = cheb.chebvander2d(locations[:, 0], locations[:, 1], orders)
elif locations.shape[1] == 3:
V = cheb.chebvander3d(locations[:, 0], locations[:, 1],
locations[:, 2], orders)
elif locations.shape[1] == 4:
V = chebVander4d(locations, orders)
elif locations.shape[1] == 5:
V = chebVander5d(locations, orders)
else:
raise NotImplementedError # there's a bad startup joke about this being good enough for the paper.
ret, _, _, _ = npl.lstsq(V, functionVals, rcond=None)
return np.reshape(ret, (np.array(orders) + 1).flatten())
def handle_A(A, fine, outname=None, standard=None, corrfile=None,
Aoffset=None, radius=2, flat_corrections=None, nosky=False,
lmin=650, lmax=700):
'''Loads 2k x 2k IFU frame "A" and extracts spectra from the locations
in "fine".
Args:
A (string): filename of ifu FITS file to extract from.
fine (string): filename of NumPy file with locations + wavelength
soln
outname (string): filename to write results to
Aoffset (2tuple): X (nm)/Y (pix) shift to apply for flexure correction
radius (float): Extraction radius in arcsecond
flat_corrections (list): A list of FlatCorrection objects for
correcting the extraction
nosky (Boolean): if True don't subtract sky, merely sum in aperture
Returns:
The extracted spectrum, a dictionary:
{'ph_10m_nm': Flux in photon / 10 m / nanometer integrated
'nm': Wavelength solution in nm
'N_spax': Total number of spaxels that created ph_10m_nm
'skyph': Sky flux in photon / 10 m / nanometer / spaxel
'radius_as': Extraction radius in arcsec
'pos': X/Y extraction location of spectrum in arcsec}
Raises:
None
'''
fine = np.load(fine)
if outname is None:
outname = "%s" % (A)
if Aoffset is not None:
ff = np.load(Aoffset)
flexure_x_corr_nm = ff[0]['dXnm']
flexure_y_corr_pix = ff[0]['dYpix']
print "Dx %2.1f nm | Dy %2.1f px" % (ff[0]['dXnm'], ff[0]['dYpix'])
else:
flexure_x_corr_nm = 0
flexure_y_corr_pix = 0
if os.path.isfile(outname+".npy"):
print "USING extractions in %s.npy!" % outname
print "rm %s.npy # if you want to recreate extractions" % outname
E, meta = np.load(outname+".npy")
E_var, meta_var = np.load("var_" + outname + ".npy")
else:
print "\nCREATING extractions ..."
spec = pf.open(A)
adcspeed = spec[0].header["ADCSPEED"]
if adcspeed == 2: read_var = 22*22
else: read_var = 5*5
var = addcon(A, str(read_var), "var_" + outname + ".fits")
print "\nExtracting object spectra"
E, meta = Wavelength.wavelength_extract(spec, fine, filename=outname,
flexure_x_corr_nm=flexure_x_corr_nm,
flexure_y_corr_pix=flexure_y_corr_pix,
flat_corrections = flat_corrections)
meta['airmass'] = spec[0].header['airmass']
header = {}
for k,v in spec[0].header.iteritems():
try: header[k] = v
except: pass
meta['HA'] = spec[0].header['HA']
meta['Dec'] = spec[0].header['Dec']
meta['RA'] = spec[0].header['RA']
meta['PRLLTC'] = spec[0].header['PRLLTC']
meta['equinox'] = spec[0].header['Equinox']
meta['utc'] = spec[0].header['utc']
meta['header'] = header
meta['exptime'] = spec[0].header['exptime']
np.save(outname, [E, meta])
print "\nExtracting variance spectra"
E_var, meta_var = Wavelength.wavelength_extract(var, fine,
filename=outname,
flexure_x_corr_nm = flexure_x_corr_nm,
flexure_y_corr_pix = flexure_y_corr_pix,
flat_corrections=flat_corrections)
np.save("var_" + outname, [E_var, meta_var])
object = meta['header']['OBJECT'].split()[0]
sixA, posA, adcpos, radius_used = identify_spectra_gui(E, radius=radius,
PRLLTC=Angle(meta['PRLLTC'], unit='deg'),
lmin=lmin, lmax=lmax, object=object, airmass=meta['airmass'])
to_image(E, meta, outname, posA=posA, adcpos=adcpos)
if standard is None:
kixA = identify_bgd_spectra(E, posA, inner=radius_used*1.1)
else:
kixA = identify_sky_spectra(E, posA, inner=radius_used*1.1)
# get the mean spectrum over the selected spaxels
resA = interp_spectra(E, sixA, outname=outname+".pdf", corrfile=corrfile)
skyA = interp_spectra(E, kixA, outname=outname+"_sky.pdf", corrfile=corrfile)
varA = interp_spectra(E_var, sixA, outname=outname+"_var.pdf", corrfile=corrfile)
## Plot out the X/Y positions of the selected spaxels
XSA = []
YSA = []
XSK = []
YSK = []
for ix in sixA:
XSA.append(E[ix].X_as)
YSA.append(E[ix].Y_as)
for ix in kixA:
XSK.append(E[ix].X_as)
YSK.append(E[ix].Y_as)
pl.figure()
pl.clf()
pl.ylim(-30,30)
pl.xlim(-30,30)
pl.scatter(XSA,YSA, color='red', marker='H', linewidth=.1)
pl.scatter(XSK,YSK, color='green', marker='H', linewidth=.1)
pl.savefig("XYs_%s.pdf" % outname)
pl.close()
# / End Plot
# Define our standard wavelength grid
ll = Wavelength.fiducial_spectrum()
# Resample sky onto standard wavelength grid
sky_A = interp1d(skyA[0]['nm'], skyA[0]['ph_10m_nm'], bounds_error=False)
sky = sky_A(ll)
# Resample variance onto standard wavelength grid
var_A = interp1d(varA[0]['nm'], varA[0]['ph_10m_nm'], bounds_error=False)
varspec = var_A(ll)
# Copy and resample object spectrum onto standard wavelength grid
res = np.copy(resA)
res = [{"doc": resA[0]["doc"], "ph_10m_nm": np.copy(resA[0]["ph_10m_nm"]),
"spectra": np.copy(resA[0]["spectra"]),
"coefficients": np.copy(resA[0]["coefficients"]),
"nm": np.copy(resA[0]["ph_10m_nm"])}]
res[0]['nm'] = np.copy(ll)
f1 = interp1d(resA[0]['nm'], resA[0]['ph_10m_nm'], bounds_error=False)
# Calculate airmass correction
airmass = meta['airmass']
extCorr = 10**(Atm.ext(ll*10) * airmass/2.5)
print "Median airmass corr: %.4f" % np.median(extCorr)
# Calculate output corrected spectrum
if nosky:
# Account for airmass and aperture
res[0]['ph_10m_nm'] = f1(ll) * extCorr * len(sixA)
else:
# Account for sky, airmass and aperture
res[0]['ph_10m_nm'] = (f1(ll)-sky_A(ll)) * extCorr * len(sixA)
# Process standard star objects
if standard is not None:
print "STANDARD"
# Extract reference data
wav = standard[:,0]/10.0
flux = standard[:,1]
# Calculate/Interpolate correction onto object wavelengths
fun = interp1d(wav, flux, bounds_error=False, fill_value = np.nan)
correction0 = fun(res[0]['nm'])/res[0]['ph_10m_nm']
# Filter for resolution
flxf = filters.gaussian_filter(flux,19.)
# Calculate/Interpolate filtered correction
fun = interp1d(wav, flxf, bounds_error=False, fill_value = np.nan)
correction = fun(res[0]['nm'])/res[0]['ph_10m_nm']
# Use unfiltered for H-beta region
ROI = (res[0]['nm'] > 470.) & (res[0]['nm'] < 600.)
correction[ROI] = correction0[ROI]
res[0]['std-correction'] = correction
res[0]['std-maxnm'] = np.max(wav)
res[0]['exptime'] = meta['exptime']
res[0]['Extinction Correction'] = 'Applied using Hayes & Latham'
res[0]['extinction_corr'] = extCorr
res[0]['skyph'] = sky * len(sixA)
res[0]['skynm'] = ll
res[0]['var'] = varspec
res[0]['radius_as'] = radius_used
res[0]['position'] = posA
res[0]['N_spax'] = len(sixA)
res[0]['meta'] = meta
res[0]['object_spaxel_ids'] = sixA
res[0]['sky_spaxel_ids'] = kixA
res[0]['sky_spectra'] = skyA[0]['spectra']
coef = chebfit(np.arange(len(ll)), ll, 4)
xs = np.arange(len(ll)+1)
newll = chebval(xs, coef)
res[0]['dlam'] = np.diff(newll)
np.save("sp_" + outname, res)
print "Wrote sp_"+outname+".npy"
start=0
for i, index in enumerate(index_ranges):
#Where the telluric features are
t_start=int(index[0][0])
t_end=int(index[1][0])
#Get the indices away from the telluric features
end=t_start
#Rescale lamda array to be between -1 and 1
l=np.array([2*(w-lamdas[start:end].min())/(lamdas[start:end].max()-lamdas[start:end].min())-1 for w in lamdas[start:end]])
polyfit=chebfit(l, spec[start:end], order)
polynomial=chebval(l, polyfit)
print("Fitting Polynomials from lamda={}A to lamda={}A".format(lamdas[start], lamdas[end]))
fit_spec[start:end]=polynomial
fit_spec_no_smooth[start:end]=polynomial
fit_spec[start:start+pixels]=start_blend(start, pixels, fit_spec, spec)
fit_spec[end-pixels:end]=end_blend(end, pixels, fit_spec, spec)
#Update the start value for the next section of the spectrum to fit
start=t_end
#Do the last section of the fit
s=index_ranges[-1][1][0]
#Rescale the wavelength array to be between -1 and 1.
def wavelegth_fitter(wavelength_fits, cube, var, aperture_mask, lamdas, order, aperture_indices, plot=False, verbose=False):
"""
Fit polynomials in the wavelength direction of each pixel in an aperture. This makes the P(x,lamda) values used in Horne 1986
Returns a cube of model values, called mcube
Wavelength_fits is an cube of un-normalised values, initially all 0s
cube is the datacube in question
var is the variance cube
aperture mask is a 3D mask of True and False. Good values=True, Bad ones =False
lamdas is an array of lamda values along the wavelength axis, formed from CRPIX,CRVAL,CDELT in the fits header
order is the order of fitting
aperture indices are indices of the cube corresponding to our circular aperture, in an Nx2 array. This is looped over.
plot=True shows a plot of the polynomials (both before and after normalising spatially)
verbose=True prints more to the terminal
"""
#Lamdas must be scaled to between -1 and 1 for Chebyshev fitting
min_val=lamdas.min()
max_val=lamdas.max()
scaled_l=np.array([2*(lamda-min_val)/(max_val-min_val) -1 for lamda in lamdas])
for n, (i, j) in enumerate(zip(aperture_indices[0], aperture_indices[1])):
"""Take each spaxel and fit a polynomial along the wavelength direction, each pixel weighted by one over its variance"""
if verbose==True:
print("Fitting Pixel {}".format(n))
weights=1.0/var[:, i, j]
#Fit the Chebyshev coefficients
coefficients=chebfit(scaled_l, cube[:, i, j], order, w=weights)
#Make a polynomial from these
polynomial=chebval(scaled_l, coefficients)
#Ensure all values are positve
polynomial[polynomial<0]=0
wavelength_fits[:, i, j]=polynomial
if plot==True:
for i, j in zip(aperture_indices[0], aperture_indices[1]):
plt.plot(lamdas, wavelength_fits[:, :])
plt.show()
if verbose==True:
print("Normalising Spatially")
#Normalise spatially
spatial_norm=np.sum(np.sum(wavelength_fits, axis=2), axis=1)
mcube=wavelength_fits/spatial_norm[:, np.newaxis, np.newaxis]
if plot==True:
for i, j in zip(aperture_indices[0], aperture_indices[1]):
plt.plot(lamdas, mcube[:,i, j])
plt.show()
return mcube
def handle_AB(A, B, fine, outname=None, corrfile=None,
Aoffset=None, Boffset=None, radius=2, flat_corrections=None,
nosky=False, lmin=650, lmax=700):
'''Loads 2k x 2k IFU frame "A" and "B" and extracts A-B and A+B spectra
from the "fine" location.
Args:
A (string): filename of ifu FITS file to extract from.
B (string): filename of ifu FITS file to extract from.
fine (string): filename of NumPy file with locations + wavelength
soln
outname (string): filename to write results to
Aoffset (2tuple): X (nm)/Y (pix) shift to apply for flexure correction
Boffset (2tuple): X (nm)/Y (pix) shift to apply for flexure correction
radius (float): Extraction radius in arcsecond
flat_corrections (list): A list of FlatCorrection objects for
correcting the extraction
nosky (Boolean): if True don't subtract sky, merely sum in aperture
Returns:
The extracted spectrum, a dictionary:
{'ph_10m_nm': Flux in photon / 10 m / nanometer integrated
'var'
'nm': Wavelength solution in nm
'N_spaxA': Total number of "A" spaxels
'N_spaxB': Total number of "B" spaxels
'skyph': Sky flux in photon / 10 m / nanometer / spaxel
'radius_as': Extraction radius in arcsec
'pos': X/Y extraction location of spectrum in arcsec}
Raises:
None
'''
fine = np.load(fine)
if outname is None:
outname = "%sm%s" % (A,B)
if Aoffset is not None:
ff = np.load(Aoffset)
flexure_x_corr_nm = ff[0]['dXnm']
flexure_y_corr_pix = -ff[0]['dYpix']
print "Dx %2.1f | Dy %2.1f" % (ff[0]['dXnm'], ff[0]['dYpix'])
else:
flexure_x_corr_nm = 0
flexure_y_corr_pix = 0
read_var = 5*5
if os.path.isfile(outname + ".fits.npy"):
print "USING extractions in %s!" % outname
E, meta = np.load(outname + ".fits.npy")
E_var, meta_var = np.load("var_" + outname + ".fits.npy")
else:
if not outname.endswith(".fits"):
outname = outname + ".fits"
diff = subtract(A,B, outname)
add(A,B, "tmpvar_" + outname)
adcspeed = diff[0].header["ADCSPEED"]
if adcspeed == 2: read_var = 22*22
else: read_var = 5*5
var = add("tmpvar_" + outname, str(read_var), "var_" + outname)
os.remove("tmpvar_" + outname + ".gz")
E, meta = Wavelength.wavelength_extract(diff, fine,
filename=outname,
flexure_x_corr_nm = flexure_x_corr_nm,
flexure_y_corr_pix = flexure_y_corr_pix,
flat_corrections=flat_corrections)
meta['airmass1'] = diff[0].header['airmass1']
meta['airmass2'] = diff[0].header['airmass2']
meta['airmass'] = diff[0].header['airmass']
header = {}
for k,v in diff[0].header.iteritems():
try: header[k] = v
except: pass
meta['HA'] = diff[0].header['HA']
meta['Dec'] = diff[0].header['Dec']
meta['RA'] = diff[0].header['RA']
meta['PRLLTC'] = diff[0].header['PRLLTC']
meta['equinox'] = diff[0].header['Equinox']
meta['utc'] = diff[0].header['utc']
meta['header'] = header
meta['exptime'] = diff[0].header['exptime']
np.save(outname, [E, meta])
exfile = "extracted_var_%s.npy" % outname
E_var, meta_var = Wavelength.wavelength_extract(var, fine,
filename=outname,
flexure_x_corr_nm = flexure_x_corr_nm,
flexure_y_corr_pix = flexure_y_corr_pix,
flat_corrections=flat_corrections)
np.save("var_" + outname, [E_var, meta_var])
sixA, posA, all_A = identify_spectra_gui(E, radius=radius,
PRLLTC=Angle(meta['PRLLTC'], unit='deg'),
lmin=lmin, lmax=lmax, airmass=meta['airmass'])
sixB, posB, all_B = identify_spectra_gui(E, radius=radius,
PRLLTC=Angle(meta['PRLLTC'], unit='deg'),
lmin=lmin, lmax=lmax, airmass=meta['airmass'])
to_image(E, meta, outname, posA=posA, posB=posB, adcpos=all_A)
skyA = identify_bgd_spectra(E, posA)
skyB = identify_bgd_spectra(E, posB)
allix = np.concatenate([sixA, sixB])
resA = interp_spectra(E, sixA, sign=1, outname=outname+"_A.pdf", corrfile=corrfile)
resB = interp_spectra(E, sixB, sign=-1, outname=outname+"_B.pdf", corrfile=corrfile)
skyA = interp_spectra(E, skyA, sign=1, outname=outname+"_skyA.pdf", corrfile=corrfile)
skyB = interp_spectra(E, skyB, sign=-1, outname=outname+"_skYB.pdf", corrfile=corrfile)
varA = interp_spectra(E_var, sixA, sign=1, outname=outname+"_A_var.pdf", corrfile=corrfile)
varB = interp_spectra(E_var, sixB, sign=1, outname=outname+"_B_var.pdf", corrfile=corrfile)
## Plot out the X/Y selected spectra
XSA = []
YSA = []
XSB = []
YSB = []
for ix in sixA:
XSA.append(E[ix].X_as)
YSA.append(E[ix].Y_as)
for ix in sixB:
XSB.append(E[ix].X_as)
YSB.append(E[ix].Y_as)
pl.figure()
pl.clf()
pl.ylim(-30,30)
pl.xlim(-30,30)
pl.scatter(XSA,YSA, color='blue', marker='H', linewidth=.1)
pl.scatter(XSB,YSB, color='red', marker='H', linewidth=.1)
pl.savefig("XYs_%s.pdf" % outname)
pl.close()
# / End Plot
np.save("sp_A_" + outname, resA)
np.save("sp_B_" + outname, resB)
np.save("var_A_" + outname, varA)
np.save("var_B_" + outname, varB)
ll = Wavelength.fiducial_spectrum()
sky_A = interp1d(skyA[0]['nm'], skyA[0]['ph_10m_nm'], bounds_error=False)
sky_B = interp1d(skyB[0]['nm'], skyB[0]['ph_10m_nm'], bounds_error=False)
sky = np.nanmean([sky_A(ll), sky_B(ll)], axis=0)
var_A = interp1d(varA[0]['nm'], varA[0]['ph_10m_nm'], bounds_error=False)
var_B = interp1d(varB[0]['nm'], varB[0]['ph_10m_nm'], bounds_error=False)
varspec = np.nanmean([var_A(ll), var_B(ll)], axis=0) * (len(sixA) + len(sixB))
res = np.copy(resA)
res = [{"doc": resA[0]["doc"], "ph_10m_nm": np.copy(resA[0]["ph_10m_nm"]),
"nm": np.copy(resA[0]["ph_10m_nm"])}]
res[0]['nm'] = np.copy(ll)
f1 = interp1d(resA[0]['nm'], resA[0]['ph_10m_nm'], bounds_error=False)
f2 = interp1d(resB[0]['nm'], resB[0]['ph_10m_nm'], bounds_error=False)
airmassA = meta['airmass1']
airmassB = meta['airmass2']
extCorrA = 10**(Atm.ext(ll*10)*airmassA/2.5)
extCorrB = 10**(Atm.ext(ll*10)*airmassB/2.5)
print "Median airmass corr: ", np.median(extCorrA), np.median(extCorrB)
# If requested merely sum in aperture, otherwise subtract sky
if nosky:
res[0]['ph_10m_nm'] = \
np.nansum([
f1(ll) * extCorrA,
f2(ll) * extCorrB], axis=0) * \
(len(sixA) + len(sixB))
else:
res[0]['ph_10m_nm'] = \
np.nansum([
(f1(ll)-sky_A(ll)) * extCorrA,
(f2(ll)-sky_B(ll)) * extCorrB], axis=0) * \
(len(sixA) + len(sixB))
res[0]['exptime'] = meta['exptime']
res[0]['Extinction Correction'] = 'Applied using Hayes & Latham'
res[0]['extinction_corr_A'] = extCorrA
res[0]['extinction_corr_B'] = extCorrB
res[0]['skyph'] = sky
res[0]['var'] = varspec
res[0]['radius_as'] = radius
res[0]['positionA'] = posA
res[0]['positionB'] = posA
res[0]['N_spaxA'] = len(sixA)
res[0]['N_spaxB'] = len(sixB)
res[0]['meta'] = meta
res[0]['object_spaxel_ids_A'] = sixA
res[0]['sky_spaxel_ids_A'] = skyA
res[0]['object_spaxel_ids_B'] = sixB
res[0]['sky_spaxel_ids_B'] = skyB
coef = chebfit(np.arange(len(ll)), ll, 4)
xs = np.arange(len(ll)+1)
newll = chebval(xs, coef)
res[0]['dlam'] = np.diff(newll)
np.save("sp_" + outname, res)
def main():
M = 3 * pow(2, 5)
N = 15
# good values
a1 = np.pi / 8.
a2 = np.pi / 4.
alphas = np.array([i / float(M) for i in range(1, M + 3)])
r = np.zeros((len(alphas), 2))
err = np.zeros((len(alphas), 2))
A1 = np.linspace(0, a1, 50)
T1 = getTheta(A1)
Phitrue1 = phi(T1)
A2 = np.linspace(a1, a2, 50)
T2 = getTheta(A2)
Phitrue2 = phi(T2)
x = getChebNodes(N)
for k in range(len(alphas)):
p = alphas[k]
print x
#ax1 = [np.pi/8.*pow((xi+1.)/2,1./p) for xi in x]
ax1 = fA(x, p, 0, a1)
ax2 = fA(x, p, a2, a1)
#ax2 = [np.pi/8.+np.pi/8.*pow((xi+1.)/2,1./p) for xi in x]
theta1 = getTheta(ax1)
theta2 = getTheta(ax2)
# print "ax2"
#print (ax2-1./8.*(theta2-np.sin(theta2)))
# print "Theta"
# print theta
#print (1./8*(theta-np.sin(theta))-ax)
phi1 = phi(theta1)
phi2 = phi(theta2)
fp1 = cheb.chebfit(x, phi1, N)
fp2 = cheb.chebfit(x, phi2, N)
print "Chebyshev Coeffs"
print fp2
r[k, 0] = abs(fp1[-1])
r[k, 1] = abs(fp2[-1])
err[k, 0] = np.linalg.norm(
Phitrue1 - cheb.chebval(fX(A1, p, 0, a1), fp1))
err[k, 1] = np.linalg.norm(
Phitrue2 - cheb.chebval(fX(A2, p, a2, a1), fp2))
print "Nth coefficient"
print r
print "Error"
print err
# print alphas
p1 = 1. / 3.
p2 = 5. / 12.
ax1 = fA(x, p1, 0., a1)
ax2 = fA(x, p2, a2, a1)
theta1 = getTheta(ax1)
theta2 = getTheta(ax2)
phi1 = phi(theta1)
phi2 = phi(theta2)
fp1 = cheb.chebfit(x, phi1, N)
fp2 = cheb.chebfit(x, phi2, N)
print fp1
plt.rc('text', usetex=True)
plt.rc('font', family='serif')
plt.subplots_adjust(hspace=0.4)
plt.subplot(211)
plt.semilogy(alphas, r)
plt.legend([r"$\phi_1(A)$", r"$\phi_2(A)$"])
xlab = ['1/6', '1/4', '1/3', '5/12', '1/2', '2/3', '3/4', '5/6', '1']
ticks = [1. / 6., 1. / 4., 1. / 3., 5. / 12.,
1. / 2., 2. / 3., 3. / 4., 5. / 6., 1.]
plt.xticks(ticks, xlab)
plt.grid(True)
plt.title(
r'$\phi(x) \approx \tilde{\phi}(x) =\sum_k a_k T_k(cx^{\alpha}-1)$')
plt.ylabel('coefficient decay')
# plt.ylabel(r'a_N')
plt.xlabel(r'$\alpha$')
plt.subplot(212)
plt.xlabel(r'$\alpha$')
# plt.ylabel(r'$\phi_i(x)-\phi_i^N(x)$')
plt.semilogy(alphas, err)
plt.xticks(ticks, xlab)
plt.grid(True)
# plt.title('Error')
plt.ylabel('Error')
plt.legend([r"$\phi_1(A)$", r"$\phi_2(A)$"])
f = open("blah.txt", "w")
MM = 200
t = np.linspace(0, 2 * np.pi, MM)
aa = 1. / 8. * (t - np.sin(t))
phit = phi(t)
phit1 = cheb.chebval(fX(aa[0:MM / 2], p1, 0., a1), fp1)
phit2 = cheb.chebval(fX(aa[MM / 2:], p2, a2, a1), fp2)
for k in range(len(t)):
if k < MM / 2:
ha = phit1[k]
else:
ha = phit2[k - MM / 2]
f.write("%s %s %s\n" % (aa[k], phit[k] - ha, ha))
f.close()
plt.show()
print phi([2 * np.pi])
def chebfit(x, y, deg, rcond=None, full=False, w=None):
from numpy.polynomial.chebyshev import chebfit
return chebfit(x, y, deg, rcond, full, w)
# Get sorted element indices for x-axis variable. It is the requirement of the
# function being used for Spline Fitting that the X be in an increasing order.
# But the way the rotation curve was extracted was to stepping away from the centre
# to the outside and then back to centre and the other side. Hence the need for
# sorting the same.
sortind = np.argsort( rotcurve["Row"] )
fig1 = plt.figure(1)
xs = np.linspace( np.min(rotcurve["Row"]), np.max(rotcurve["Row"]), 1000)
residuals = []
change=0.0
for i in range(15):
interp_coeff = chebfit( rotcurve["Row"], rotcurve["lambda"], w=1/rotcurve["lambda_err"], deg=i)
interp_coeff_z = chebfit( rotcurve["Row"], rotcurve["z"], w=1/rotcurve["z_err"], deg=i)
# Below, we are just generating a sample curve for the user to visualize the fit.
ys = chebval(xs, interp_coeff)
# Residual calculation in redshift space.
predicted = chebval( rotcurve["Row"], interp_coeff_z )
rms = np.sqrt( np.sum( (predicted - rotcurve["z"])**2 ) / len(predicted) )
print rms
residuals.append(rms)
# Now, present the plot to the user.
plt.errorbar( rotcurve["Row"], rotcurve["lambda"], rotcurve["lambda_err"], fmt="o" )
plt.plot(xs, ys)
plt.xlabel("Row Number", fontsize=18)
def main():
pi = np.pi
N = 20
a1 = pi / 8.
a2 = pi / 4.
p1 = 1. / 3.
p2 = 5. / 12.
x = getChebNodes(N)
ax1 = fA(x, p1, 0., a1)
ax2 = fA(x, p2, a2, a1)
theta1 = getTheta(ax1)
theta2 = getTheta(ax2)
phi1 = phi(theta1)
phi2 = phi(theta2)
fp1 = cheb.chebfit(x, phi1, N)
fp2 = cheb.chebfit(x, phi2, N)
# print fp1
# print fp2
print "Disagreement at pi/8 is %.15f" % (cheb.chebval(1, fp1) - cheb.chebval(1, fp2))
phim1 = cheb.chebval(1, fp2)
phim2 = cheb.chebval(-1, fp2)
# print phim1;
# print phim2;
K = 4
M = 3 * pow(2, K)
M = 60
K = 5
alphas = np.array([i / float(M) for i in range(2 * K, M + 3)])
Atrue1 = np.linspace(0, pi / 8., 200)
Atrue2 = np.linspace(pi / 8., pi / 4., 200)
phis1 = cheb.chebval(fX(Atrue1, p1, 0, a1), fp1)
phis2 = cheb.chebval(fX(Atrue2, p2, a2, a1), fp2)
#alphas = [1./3.]
# print cheb.chebval(fX(0,p1,0,a1),fp1)
# print cheb.chebval(fX(pi/8,p1,0,a1),fp1)-phim1
r = np.zeros((len(alphas), 2))
err = np.zeros((len(alphas), 2))
for k in range(len(alphas)):
p = alphas[k]
sx1 = fA(x, p, 0, phim1)
sx2 = fA(x, p, phim2, phim1)
A1 = np.zeros(len(sx1))
A2 = np.zeros(len(sx1))
# print "p = %f" %p
# print sx1
# print sx2
for i in range(1, len(sx1) - 1):
A1[i] = optimize.ridder(
lambda x: -cheb.chebval(fX(x, p1, 0, a1), fp1) + sx1[i], 0, pi / 8. + .1)
A1[-1] = pi / 8.
A2 = [optimize.ridder(lambda x: cheb.chebval(
fX(x, p2, a2, a1), fp2) - sxi, pi / 8., pi / 4.) for sxi in sx2]
# print A1
# print A2
fa1 = cheb.chebfit(x, A1, N)
fa2 = cheb.chebfit(x, A2, N)
# print fa1
# print fa2
r[k, 0] = abs(fa1[-1])
r[k, 1] = abs(fa2[-1])
err[k, 0] = np.linalg.norm(
Atrue1 - cheb.chebval(fX(phis1, p, 0, phim1), fa1))
err[k, 1] = np.linalg.norm(
Atrue2 - cheb.chebval(fX(phis2, p, phim2, phim1), fa2))
print "alpha r1 r2 err1 err2"
for j in range(len(r)):
print " %f %e %e %e %e" % (alphas[j], r[j, 0], r[j, 1], err[j, 0], err[j, 1])
plt.rc('text', usetex=True)
plt.rc('font', family='serif')
plt.subplots_adjust(hspace=0.4)
plt.subplot(211)
plt.semilogy(alphas, r)
plt.legend([r"$\phi^{-1}_1$", r"$\phi^{-1}_2$"])
xlab = ['1/6', '1/4', '1/3', '2/5', '1/2', '3/5', '2/3', '3/4', '5/6', '1']
ticks = [1. / 6., 1. / 4., 1. / 3., 2. / 5., 1. /
2., 3. / 5., 2. / 3., 3. / 4., 5. / 6., 1.]
plt.xticks(ticks, xlab)
plt.grid(True)
plt.title(r'$\phi^{-1}(x) \approx \sum_k a_k T_k(cx^{\alpha}-1)$')
# plt.ylabel(r'a_N')
plt.ylabel('coefficient decay')
plt.xlabel(r'$\alpha$')
plt.subplot(212)
plt.xlabel(r'$\alpha$')
plt.ylabel('Error')
# plt.ylabel(r'$\phi^{-1}_i(x)-\tilde{\phi}_i^{-1}(x)$')
plt.semilogy(alphas, err)
plt.xticks(ticks, xlab)
plt.grid(True)
plt.title('Error')
plt.legend([r"$\phi^{-1}_1$", r"$\phi^{-1}_2$"])
# plt.show()
p3 = 1.
p4 = 3. / 5
sx1 = fA(x, p3, 0, phim1)
sx2 = fA(x, p4, phim2, phim1)
A1 = np.zeros(len(sx1))
A2 = np.zeros(len(sx1))
# print sx1
# print sx2
for i in range(1, len(sx1) - 1):
A1[i] = optimize.ridder(
lambda x: -cheb.chebval(fX(x, p1, 0, a1), fp1) + sx1[i], 0, pi / 8. + .1)
A1[-1] = pi / 8.
A2 = [optimize.ridder(lambda x: cheb.chebval(
fX(x, p2, a2, a1), fp2) - sxi, pi / 8., pi / 4.) for sxi in sx2]
# print A1
# print A2
fa1 = cheb.chebfit(x, A1, N)
fa2 = cheb.chebfit(x, A2, N)
# print fa1
# print fa2
print "Power for phi1 is %f" % p1
print "Power for phi2 is %f" % p2
print "Power for phi1^(-1) is %f" % p3
print "Power for phi2^(-1) is %f" % p4
fnames = ["phiofA1.txt", "phiofA2.txt", "Aofphi1.txt", "Aofphi2.txt"]
coeffs = np.zeros((N + 1, 4))
coeffs[:, 0] = fp1
coeffs[:, 1] = fp2
coeffs[:, 2] = fa1
coeffs[:, 3] = fa2
for i in range(4):
f = open(fnames[i], "w")
for j in range(len(fa1)):
f.write("%.16f, " % (coeffs[j, i]))
f.close()
for i in range(N + 1):
print "%.16f, " % x[i]
# For spline measurement
s_bot, s_cnt, s_fwhm, s_as12, s_fw13, s_as13, s_fw23, s_as23, s_err, s_ew, s_cont = np.arange(0, 11)
reg = __import__("6405")
flat1 = list(range(169, 184))
flat2 = list(range(913, 932))
flat3 = list(range(979, 995))
flat4 = list(range(1147, 1239))
flt1 = reg.qu1m[flat1] / reg.qu1m[flat1].mean()
flt2 = reg.qu1m[flat2] / reg.qu1m[flat2].mean()
flt3 = reg.qu1m[flat3] / reg.qu1m[flat3].mean()
flt4 = reg.qu1m[flat4] / reg.qu1m[flat4].mean()
flt = np.hstack((flt1, flt2, flt3, flt4))
pos = np.arange(0, len(flt))
chb = ch.chebfit(pos, flt, 5) # Use for normalization later
bak1 = reg.qu1[:, flat1] / reg.qu1[:, flat1].mean(axis=1).reshape(-1, 1)
bak2 = reg.qu1[:, flat2] / reg.qu1[:, flat2].mean(axis=1).reshape(-1, 1)
bak3 = reg.qu1[:, flat3] / reg.qu1[:, flat3].mean(axis=1).reshape(-1, 1)
bak4 = reg.qu1[:, flat4] / reg.qu1[:, flat4].mean(axis=1).reshape(-1, 1)
bak = np.hstack((bak1, bak2, bak3, bak4)) / ch.chebval(pos, chb).T
width = 1420 # Magic number found by trial
bak = bak.reshape(-1, width)
blmbd = reg.qu1.lmbd[:width]
bmet = reg.qu1.meta
bmet.lmbd = blmbd
bmet.ref = None
bmet.cont = (0, np.ones(bak.shape[0]))
def chebyshev_fit(fit_me, xi, xf, weights,order=7, *args, **kwargs): chebfunc = chebyshev.chebfit(xi,fit_me,deg=order,w=weights) fitted = chebyshev.chebval(xf,chebfunc) residual = chebyshev.chebval(xi, chebfunc) - fit_me return (fitted, residual)
def _fit_builtin(self,func,N):
""" Return the chebyshev coefficients using the builtin chebfit """
pts = cheb.chebpts2(N)
y = func(pts)
coeffs = cheb.chebfit(pts,y,N)
return coeffs
def test_chebfit(self):
def f(x):
return x*(x - 1)*(x - 2)
def f2(x):
return x**4 + x**2 + 1
# Test exceptions
assert_raises(ValueError, cheb.chebfit, [1], [1], -1)
assert_raises(TypeError, cheb.chebfit, [[1]], [1], 0)
assert_raises(TypeError, cheb.chebfit, [], [1], 0)
assert_raises(TypeError, cheb.chebfit, [1], [[[1]]], 0)
assert_raises(TypeError, cheb.chebfit, [1, 2], [1], 0)
assert_raises(TypeError, cheb.chebfit, [1], [1, 2], 0)
assert_raises(TypeError, cheb.chebfit, [1], [1], 0, w=[[1]])
assert_raises(TypeError, cheb.chebfit, [1], [1], 0, w=[1, 1])
assert_raises(ValueError, cheb.chebfit, [1], [1], [-1,])
assert_raises(ValueError, cheb.chebfit, [1], [1], [2, -1, 6])
assert_raises(TypeError, cheb.chebfit, [1], [1], [])
# Test fit
x = np.linspace(0, 2)
y = f(x)
#
coef3 = cheb.chebfit(x, y, 3)
assert_equal(len(coef3), 4)
assert_almost_equal(cheb.chebval(x, coef3), y)
coef3 = cheb.chebfit(x, y, [0, 1, 2, 3])
assert_equal(len(coef3), 4)
assert_almost_equal(cheb.chebval(x, coef3), y)
#
coef4 = cheb.chebfit(x, y, 4)
assert_equal(len(coef4), 5)
assert_almost_equal(cheb.chebval(x, coef4), y)
coef4 = cheb.chebfit(x, y, [0, 1, 2, 3, 4])
assert_equal(len(coef4), 5)
assert_almost_equal(cheb.chebval(x, coef4), y)
# check things still work if deg is not in strict increasing
coef4 = cheb.chebfit(x, y, [2, 3, 4, 1, 0])
assert_equal(len(coef4), 5)
assert_almost_equal(cheb.chebval(x, coef4), y)
#
coef2d = cheb.chebfit(x, np.array([y, y]).T, 3)
assert_almost_equal(coef2d, np.array([coef3, coef3]).T)
coef2d = cheb.chebfit(x, np.array([y, y]).T, [0, 1, 2, 3])
assert_almost_equal(coef2d, np.array([coef3, coef3]).T)
# test weighting
w = np.zeros_like(x)
yw = y.copy()
w[1::2] = 1
y[0::2] = 0
wcoef3 = cheb.chebfit(x, yw, 3, w=w)
assert_almost_equal(wcoef3, coef3)
wcoef3 = cheb.chebfit(x, yw, [0, 1, 2, 3], w=w)
assert_almost_equal(wcoef3, coef3)
#
wcoef2d = cheb.chebfit(x, np.array([yw, yw]).T, 3, w=w)
assert_almost_equal(wcoef2d, np.array([coef3, coef3]).T)
wcoef2d = cheb.chebfit(x, np.array([yw, yw]).T, [0, 1, 2, 3], w=w)
assert_almost_equal(wcoef2d, np.array([coef3, coef3]).T)
# test scaling with complex values x points whose square
# is zero when summed.
x = [1, 1j, -1, -1j]
assert_almost_equal(cheb.chebfit(x, x, 1), [0, 1])
assert_almost_equal(cheb.chebfit(x, x, [0, 1]), [0, 1])
# test fitting only even polynomials
x = np.linspace(-1, 1)
y = f2(x)
coef1 = cheb.chebfit(x, y, 4)
assert_almost_equal(cheb.chebval(x, coef1), y)
coef2 = cheb.chebfit(x, y, [0, 2, 4])
assert_almost_equal(cheb.chebval(x, coef2), y)
assert_almost_equal(coef1, coef2)
def test_chebfit(self) :
def f(x) :
return x*(x - 1)*(x - 2)
# Test exceptions
assert_raises(ValueError, cheb.chebfit, [1], [1], -1)
assert_raises(TypeError, cheb.chebfit, [[1]], [1], 0)
assert_raises(TypeError, cheb.chebfit, [], [1], 0)
assert_raises(TypeError, cheb.chebfit, [1], [[[1]]], 0)
assert_raises(TypeError, cheb.chebfit, [1, 2], [1], 0)
assert_raises(TypeError, cheb.chebfit, [1], [1, 2], 0)
assert_raises(TypeError, cheb.chebfit, [1], [1], 0, w=[[1]])
assert_raises(TypeError, cheb.chebfit, [1], [1], 0, w=[1,1])
# Test fit
x = np.linspace(0,2)
y = f(x)
#
coef3 = cheb.chebfit(x, y, 3)
assert_equal(len(coef3), 4)
assert_almost_equal(cheb.chebval(x, coef3), y)
#
coef4 = cheb.chebfit(x, y, 4)
assert_equal(len(coef4), 5)
assert_almost_equal(cheb.chebval(x, coef4), y)
#
coef2d = cheb.chebfit(x, np.array([y,y]).T, 3)
assert_almost_equal(coef2d, np.array([coef3,coef3]).T)
# test weighting
w = np.zeros_like(x)
yw = y.copy()
w[1::2] = 1
y[0::2] = 0
wcoef3 = cheb.chebfit(x, yw, 3, w=w)
assert_almost_equal(wcoef3, coef3)
#
wcoef2d = cheb.chebfit(x, np.array([yw,yw]).T, 3, w=w)
assert_almost_equal(wcoef2d, np.array([coef3,coef3]).T)
#test NA
y = f(x)
y[10] = 100
xm = x.view(maskna=1)
xm[10] = np.NA
res = cheb.chebfit(xm, y, 3)
assert_almost_equal(res, coef3)
ym = y.view(maskna=1)
ym[10] = np.NA
res = cheb.chebfit(x, ym, 3)
assert_almost_equal(res, coef3)
y2 = np.vstack((y,y)).T
y2[10,0] = 100
y2[15,1] = 100
y2m = y2.view(maskna=1)
y2m[10,0] = np.NA
y2m[15,1] = np.NA
res = cheb.chebfit(x, y2m, 3).T
assert_almost_equal(res[0], coef3)
assert_almost_equal(res[1], coef3)
wm = np.ones_like(x, maskna=1)
wm[10] = np.NA
res = cheb.chebfit(x, y, 3, w=wm)
assert_almost_equal(res, coef3)
def fit_continuum(self, deg): inds=np.where(self.mask !=0 ) x_to_fit=self.xdata[inds] y_to_fit=self.ydata[inds] self.cont_fit=chebyshev.chebval(self.xdata, chebyshev.chebfit(x_to_fit,y_to_fit,deg))
def approximate_curve(x, y, xnew, curved_coef):
"""returns ynew - y values of some curve approximation"""
if no_numpy:
return None
return chebval(xnew, chebfit(x, y, curved_coef))
def approximate_curve(x, y, xnew, curved_coef):
"""returns ynew - y values of some curve approximation"""
if no_numpy:
raise ValueError("No numpy found")
return chebval(xnew, chebfit(x, y, curved_coef))
def GetChebyshevCoeffs(data, pvals, order=5):
pmid = pvals[pvals.size / 2]
prange = pvals[-1] - pvals[0]
nvals = (pvals - pmid) / (prange / 2.0)
return chebyshev.chebfit(nvals, data.x, order)
raise RuntimeError("%d rows is not a multiple of 64" % nrows)
num_spectra = nrows/64
use_spectra = user_input("Select spectra",range(num_spectra)).split(',')
avg = zeros(64)
for spectrum in use_spectra:
base_index = int(spectrum)*64
for index in range(64):
avg[index] += data[base_index+index,1]
averages = avg/num_spectra
# normalize the spectrum
mean = averages.mean()
averages /= mean
# fit the normalized spectrum
freqs = data[:64,0] - data[:64,0].mean()
coefs = chebfit(freqs,averages,11)
model = chebval(freqs,coefs)
freqstep = freqs[1]-freqs[0]
bandwidth = freqs[-1]-freqs[0]+freqstep
coeffile = open("baseline_coefs.pkl","rb")
try:
coef_dict = load(coeffile)
print "Loaded:",coef_dict.keys()
except EOFError:
print "Empty coefficients file"
coeffile.close()
try:
coef_dict[bandwidth] = coefs
except NameError:
coef_dict = {bandwidth: coefs}
