def draw_spectra(self):
''' Draw nearest spectra '''
if self.picked is None: return
print "Drawing spectra"
#xl = self.ax_spec.get_xlim()
#yl = self.ax_spec.get_ylim()
self.ax_spec.cla()
#self.ax_spec.set_xlim(xl)
#self.ax_spec.set_ylim(yl)
x,y = self.X2[self.picked], self.Y1[self.picked]
objs = self.KT.query_ball_point((x,y), 70)
print "Query around %s found: %s" % ((x,y), objs)
spec = self.cube[self.picked]
ix = np.arange(*spec.xrange)
fiducial_ll = chebval(ix, spec.lamcoeff)
#self.ax_spec.plot(ix-spec.xrange[0], fiducial_ll, linewidth=3)
self.ax_spec.step(fiducial_ll, spec.spec, linewidth=3)
for spec in self.good_cube[objs]:
try:
ix = np.arange(*spec.xrange)
ll = chebval(ix, spec.lamcoeff)
#self.ax_spec.plot(ix-spec.xrange[0], ll-fiducial_ll)
self.ax_spec.step(ll, spec.spec)
except: pass
#self.ax_spec.set_ylim(-30,30)
self.ax_spec.set_xlim(370,700)
self.fig.show()
def __init__(self, cube=None, figure=None, pointsize=65, bgd_sub=False, radius_as=3):
''' Create spectum picking gui.
Args:
cube: Data cube list
figure: Figure to draw to [default is None or open a new figure
window '''
self.actions = {"m": self.mode_switch}
self.cube = cube
self.pointsize = pointsize
for ix, s in enumerate(self.cube):
if s.xrange is None: continue
xs = np.arange(s.xrange[0], s.xrange[1], .1)
if s.lamcoeff is not None:
lls = chebval(xs, s.lamcoeff)
ha1 = np.argmin(np.abs(lls - 656.3))/10.0 + s.xrange[0]
if s.mdn_coeff is not None:
lls = chebval(xs, s.mdn_coeff)
ha2 = np.argmin(np.abs(lls - 656.3))/10.0 + s.xrange[0]
self.X1.append(ha1)
self.X2.append(ha2)
self.Y1.append(s.yrange[0])
self.X1, self.X2, self.Y1 = map(np.array, [self.X1, self.X2,
self.Y1])
OK = (np.abs(self.X1 - self.X2) < 2) & np.isfinite(self.X2) & \
np.isfinite(self.Y1)
self.good_cube = self.cube[OK]
locs = np.array([self.X2[OK], self.Y1[OK]]).T
self.KT = scipy.spatial.KDTree(locs)
assert(len(locs) == len(self.good_cube))
# Setup drawing
gs = gridspec.GridSpec(1,2, width_ratios=[1,2.5])
self.fig = pl.figure(1, figsize=(22,5.5))
self.ax_cube = pl.subplot(gs[0])
self.ax_spec = pl.subplot(gs[1])
self.ax_cube.set_xlim(-100, 2200)
self.ax_cube.set_ylim(-100, 2200)
pl.ion()
self.draw_cube()
print "Registering events"
self.fig.canvas.mpl_connect("button_press_event", self)
self.fig.canvas.mpl_connect("key_press_event", self)
pl.show()
def referenceFunc_i(point):
"""Reference inverse implementation; point must be in range [-1, 1]
"""
c1 = np.array([0, 0, 0, 1.6], dtype=float)
c2 = np.array([0, -3.6], dtype=float)
x1 = point
return (
chebval(x1, c1),
chebval(x1, c2),
)
def get_lambda_nm(self):
"""Returns lambda spectrum in nm"""
xs = np.arange(*self.xrange)
if self.lamcoeff is not None:
lam = chebval(xs, self.lamcoeff)
else:
lam = chebval(xs, self.mdn_coeff)
return lam
def test_chebval(self) :
def f(x) :
return x*(x**2 - 1)
#check empty input
assert_equal(ch.chebval([], 1).size, 0)
#check normal input)
for i in range(5) :
tgt = 1
res = ch.chebval(1, [0]*i + [1])
assert_almost_equal(res, tgt)
tgt = (-1)**i
res = ch.chebval(-1, [0]*i + [1])
assert_almost_equal(res, tgt)
zeros = np.cos(np.linspace(-np.pi, 0, 2*i + 1)[1::2])
tgt = 0
res = ch.chebval(zeros, [0]*i + [1])
assert_almost_equal(res, tgt)
x = np.linspace(-1,1)
tgt = f(x)
res = ch.chebval(x, [0, -.25, 0, .25])
assert_almost_equal(res, tgt)
#check that shape is preserved
for i in range(3) :
dims = [2]*i
x = np.zeros(dims)
assert_equal(ch.chebval(x, [1]).shape, dims)
assert_equal(ch.chebval(x, [1,0]).shape, dims)
assert_equal(ch.chebval(x, [1,0,0]).shape, dims)
def test_chebvander(self) :
# check for 1d x
x = np.arange(3)
v = ch.chebvander(x, 3)
assert_(v.shape == (3,4))
for i in range(4) :
coef = [0]*i + [1]
assert_almost_equal(v[...,i], ch.chebval(x, coef))
# check for 2d x
x = np.array([[1,2],[3,4],[5,6]])
v = ch.chebvander(x, 3)
assert_(v.shape == (3,2,4))
for i in range(4) :
coef = [0]*i + [1]
assert_almost_equal(v[...,i], ch.chebval(x, coef))
def single_pawley(detector, q, I, back, p_fw=None):
""" Full pawley fitting for a specific q, I combination.
Option to use different initial parameters for FWHM fit. This allows a
lower order polynomial to be specified.
Args:
detector: pyxpb detector object
q (ndarray): Reciprocal lattice
I (ndarray): Intensity values
az_idx (int): Azimuthal slice index
p_fw (list, tuple): New initial estimate
"""
nmat, nf = len(detector.materials), len(detector._fwhm)
background = chebval(q, back)
q_lim = [np.min(q), np.max(q)]
p0 = extract_parameters(detector, q_lim, np.nanmax(I))
if p_fw is not None:
p_fw0_idx = range(nmat, nmat + nf)
p0_new = [i for idx, i in enumerate(p0) if idx not in p_fw0_idx]
for idx, i in enumerate(p_fw):
p0_new.insert(nmat + idx, i)
nf = len(p_fw)
else:
p0_new = p0
pawley = pawley_hkl(detector, background, nf - 1)
return curve_fit(pawley, q, I, p0=p0_new)
def spec_calibration(self, theta=None, obs=None, **kwargs):
"""Implements a Chebyshev polynomial calibration model. If
``"pivot_wave"`` is not present in ``obs`` then 1.0 is returned.
:returns cal:
If ``params["cal_type"]`` is ``"poly"``, a polynomial given by
'spec_norm' * (1 + \Sum_{m=1}^M 'poly_coeffs'[m-1] T_n(x)).
Otherwise, the exponential of a Chebyshev polynomial.
"""
if theta is not None:
self.set_parameters(theta)
if ('pivot_wave' in obs) and ('poly_coeffs' in self.params):
# map wavelengths to the interval -1, 1
x = obs['wavelength'] - obs['wavelength'].min()
x = 2.0 * (x / x.max()) - 1.0
# get coefficients. Here we are setting the first term to 0 so we
# can deal with it separately for the exponential and regular
# multiplicative cases
c = np.insert(self.params['poly_coeffs'], 0, 0)
poly = chebval(x, c)
#switch to have spec_norm be multiplicative or additive
#depending on whether the calibration model is
#multiplicative in exp^poly or just poly
if self.params.get('cal_type', 'exp_poly') is 'poly':
return (1.0 + poly) * self.params['spec_norm']
else:
return np.exp(self.params['spec_norm'] + poly)
else:
return 1.0
def chebinterp(x,f,y):
""" Interpolate a function f on a set of points x onto a set of points y
x and y should must be in [-1,1]"""
n = len(x)
Tx = chebvander(x,n-1)
a = solve(Tx,f)
p = chebval(y,a)
return p
def Poly(pars, middle, low, high, x):
"""
Generates a polynomial with the given parameters
for all of the x-values.
x is assumed to be a np.ndarray!
Not meant to be called directly by the user!
"""
xgrid = (x - middle) / (high - low) # re-scale
return chebyshev.chebval(xgrid, pars)
def plot_result():
n = [2**x for x in range(1,8)]
for j in n:
x = np.linspace(-1,1,num=100,endpoint=True)
y = np.array([f(i) for i in x])
chebpoly = chebval(x,cheb_interp(f,j))
plt.plot(x,y)
plt.plot(x,chebpoly)
plt.show()
def initModel(self,):
HtIndx=min(max(int(self.zjet_max-50)/20-1,0),20)
scoef=spd_coefs[HtIndx]
dcoef=dir_coefs[HtIndx]
prms=np.array([0,self.Ri,self.UStar,1])
if self.zjet_max==self.RefHt:
prms[0]=self.URef
else:
utmp1=chebval(self.RefHt,np.dot(scoef[:,1:],prms[1:][:,None]))
utmp2=chebval(self.RefHt,scoef[:,0])
prms[0]=(self.URef-utmp1)/utmp2
scoef=np.dot(scoef,prms[:,None]) # These are now vectors
dcoef=np.dot(dcoef,prms[:,None]) # These are now vectors
ang=chebval(self.grid.z,dcoef)
ang-=ang[self.grid.ihub[0]] # The hub-height angle should be zero.
ang[ang<-45.],ang[ang>45.]=-45.,45. # No angle should be more than 45 degrees.
tmpdat=chebval(self.grid.z,scoef)*np.exp(1j*np.pi/180.*ang)[:,None]
self._u[:1]=tmpdat.real,tmpdat.imag
def test_approximation(self):
def powx(x, p):
return x**p
x = np.linspace(-1, 1, 10)
for deg in range(0, 10):
for p in range(0, deg + 1):
c = cheb.chebinterpolate(powx, deg, (p,))
assert_almost_equal(cheb.chebval(x, c), powx(x, p), decimal=12)
def pawley_plot(q, I, detector, az_idx, ax):
""" Plots q against measured intensity overlaid with Pawley fit.
Includes highlighting of anticipated Bragg peak locations and
difference between measured intensity and Pawley fit.
Args:
q (ndarray): Reciprocal lattice
I (ndarray): Intensity
detector: pyxpb detector instance
az_idx (int): Azimuthal slice index
ax: Axis to apply plot to
"""
background = chebval(q, detector._back[az_idx])
q_lim = [np.min(q), np.max(q)]
p0 = extract_parameters(detector, q_lim, np.nanmax(I))
pawley = pawley_hkl(detector, background)
coeff, var_mat = curve_fit(pawley, q, I, p0=p0)
I_pawley = pawley(q, *coeff)
# Plot raw data and Pawley fit to data
ax.plot(q, I, 'o', markeredgecolor='0.3', markersize=4,
markerfacecolor='none', label=r'$\mathregular{I_{obs}}$')
ax.plot(q, I_pawley, 'r-', linewidth=0.75,
label=r'$\mathregular{I_{calc}}$')
# Plot Bragg positions - locate relative to max intensity
ymin = -ax.get_ylim()[1] / 10
materials = detector.materials
for idx, mat in enumerate(materials):
offset = (1 + idx) * ymin / 2
for q0 in detector.q0[mat]:
bragg_line = [offset + ymin / 8, offset - ymin / 8]
ax.plot([q0, q0], bragg_line, 'g-', linewidth=2)
# Use error bars to fudge vertical lines in legend
ax.errorbar(0, 0, yerr=1, fmt='none', capsize=0, ecolor='g',
elinewidth=1.5, label=r'Bragg ({})'.format(mat))
# Plot difference between raw and Pawley fit (shifted below Bragg)
I_diff = I - I_pawley
max_diff = np.max(I_diff)
shifted_error = I - I_pawley + (idx + 2) * ymin / 2 - max_diff
ax.plot(q, shifted_error, 'b-', linewidth=0.75,
label=r'$\mathregular{I_{diff}}$')
# Remove ticks below 0 intensity
ylocs = ax.yaxis.get_majorticklocs()
yticks = ax.yaxis.get_major_ticks()
for idx, yloc in enumerate(ylocs):
if yloc < 0:
yticks[idx].set_visible(False)
legend = ax.legend(numpoints=1)
frame = legend.get_frame()
frame.set_facecolor('w')
def define_background(self, seg=50, k=12, pnt=None, fwhm=None, plot=True,
az_idx=0, auto=True, x=None, y=None):
""" Background profile fitting - segment data and auto find points.
Fits a Chebyshev polynomial (of order k) to automatically acquired
q, I points. These points are selected wrt. the materials present
(defined via add_materials), with the Bragg peaks being avoided.
If auto is False, q v I points can be manually specified.
Args:
seg (int): Number of points to split data into/extract
k (int): Order for Chebyshev polynomial
pnt (tuple): Point co_ords or None for averaged data
fwhm (float): Peak fwhm (to exclude more data around peaks)
plot (bool): True/False
az_idx (int): Azimuthal slice to plot
auto (bool): Automatically extract q v I data
x (ndarray): If not auto maunally defined q values
y (ndarray): If not auto maunally defined I values
"""
# Automatically averages over all points unless point is specified
if pnt is None:
I = np.mean(self.I, tuple(range(self.I.ndim - 2)))
else:
I = self.I[pnt]
if auto:
az = self.q.shape[0]
x, y = np.zeros((az, seg)), np.zeros((az, seg))
for a in range(az):
split_q = np.array_split(self.q[a], seg)
split_I = np.array_split(I[a], seg)
for idx, q in enumerate(split_q):
q_min, q_max = np.min(q), np.max(q)
for mat in self.detector.materials:
q0 = self.detector.q0[mat]
fw = self.detector.fwhm[mat] if fwhm is None else fwhm
clash = np.any(np.logical_and(q0 > q_min - 2 * fw,
q0 < q_max + 2 * fw))
x[a, idx] = np.nan if clash else np.mean(split_q[idx])
y[a, idx] = np.nan if clash else np.mean(split_I[idx])
self.detector.define_background(x, y, k)
f = self.detector._back
if plot:
plt.plot(self.q[az_idx], I[az_idx], 'k')
plt.plot(self.q[az_idx], chebval(self.q[az_idx], f[az_idx]), 'r-')
plt.plot(x[az_idx], y[az_idx], 'r+')
plt.ylim([0, np.nanmax(y) * 1.5])
plt.xlabel(r'$\mathregular{q (A^{-1})}$')
plt.ylabel(r'Intensity')
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 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])
def test_chebval(self) :
#check empty input
assert_equal(cheb.chebval([], [1]).size, 0)
#check normal input)
x = np.linspace(-1,1)
y = [polyval(x, c) for c in Tlist]
for i in range(10) :
msg = "At i=%d" % i
tgt = y[i]
res = cheb.chebval(x, [0]*i + [1])
assert_almost_equal(res, tgt, err_msg=msg)
#check that shape is preserved
for i in range(3) :
dims = [2]*i
x = np.zeros(dims)
assert_equal(cheb.chebval(x, [1]).shape, dims)
assert_equal(cheb.chebval(x, [1,0]).shape, dims)
assert_equal(cheb.chebval(x, [1,0,0]).shape, dims)
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 get_dlambda(self):
"""Returns dlambda spectrum """
# Note the +1 below to handle the eating of an element
# by diff
xs = np.arange(self.xrange[0], self.xrange[1] + 1)
lam = chebval(xs, self.lamcoeff)
dlam = np.abs(np.diff(lam))
return dlam
def Strength(x):
if x.xrange is None: return None
if x.lamcoeff is None: return None
if x.specw is None: return None
ix = np.arange(*x.xrange)
ll = chebval(ix, x.lamcoeff)
OK = (ll > 450) & (ll < 700)
if OK.any():
return np.sum(x.specw[OK])
else:
return None
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 interpolate(self, z, f):
from numpy.polynomial.chebyshev import chebval
import numpy as np
msg = "Can't interpolate outside grid domain"
assert np.array([z]).min() >= self.zmin, msg
assert np.array([z]).max() <= self.zmax, msg
c = self.to_coefficients(f)
# Convert z-values to standard xg = [-1, 1]
x = (z - self.zmin) / self.L * 2. - 1.
return chebval(x, c)
def fit_func(dummy_variable, *labels):
nn_spec = self.network.get_spectrum_scaled(
scaled_labels=labels[:nnl])
nn_spec = doppler_shift(self.network.wave, nn_spec, labels[nnl])
if hasattr(self, 'lsf'):
nn_resampl = self.lsf.apply(nn_spec)
else:
nn_resampl = np.interp(wavelength, self.network.wave, nn_spec)
Cheb_coefs = labels[nnl + 1:nnl + 1 + self.Cheb_order]
Cheb_x = np.linspace(-1, 1, len(nn_resampl))
Cheb_poly = chebval(Cheb_x, Cheb_coefs)
spec_with_resp = nn_resampl * Cheb_poly
return spec_with_resp
def eval(nr, a, b, r):
# evaluates the projection matrix matrix for the chebyshev basis
xx = (np.array(r)-b)/a
coeffs = np.eye(nr)*2
#coeffs = np.eye(nr)
coeffs[0,0]=1.
# return the coo type matrix, because it is needed for integration (evaluation of antiderivative) purposes
#return spsp.lil_matrix(cheb.chebval(xx, coeffs).transpose())
return np.mat(cheb.chebval(xx, coeffs).transpose())
def __call__(self, point):
'''
Evaluate the polynomial at 'point'.
Parameters
----------
point : array-like
point at which to evaluate the polynomial
Returns
-------
c : complex
value of the polynomial at the given point
'''
super(MultiCheb, self).__call__(point)
c = self.coeff
n = len(c.shape)
c = cheb.chebval(point[0], c)
for i in range(1, n):
c = cheb.chebval(point[i], c, tensor=False)
return c
def bit_to_float(matrica_bit):
matrica_bit, fitness = np.split(matrica_bit, [-1], axis=1)
days_bit, brod_bit, koeficijenti_bit, snaga = np.split(matrica_bit,
64*np.array((1, 3, 4+podaci.chebdeg)),
axis=1)
days = np.packbits(days_bit.astype(bool), axis=1).view(np.float64)
# date = np.empty(broj_jedinki)
# for i in range(broj_jedinki):
# date[i] = datetime.timedelta(days=days[i]) + podaci.beg_of_time
brod = np.packbits(brod_bit.astype(bool), axis=1).view(np.float64)
koeficijenti = np.packbits(koeficijenti_bit.astype(bool), axis=1).view(np.float64)
uglovi = chebval(x_osa(broj_segmenata), koeficijenti)
return days, brod, uglovi, snaga, fitness
def test_cheb_poly_single():
deg = 21
npts = 100
pts = np.linspace(0, 1, npts, dtype=np.float32)
out = np.zeros((deg + 1) * npts, dtype=np.float32)
cheb_utils.cheb_poly_float(deg, pts, npts, out)
out = out.reshape([deg + 1, npts])
cheb_coefs = np.zeros(deg + 1)
for m in range(deg + 1):
cheb_coefs.fill(0)
cheb_coefs[m] = 1
cheb_vals_np = chebval(pts, cheb_coefs)
assert np.allclose(out[m], cheb_vals_np, rtol=1e-6, atol=1e-6)
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 bit_to_float(matrica_bit):
warnings.filterwarnings("error", category=RuntimeWarning)
days_bit, fuel_bit, koeficijenti_bit, snaga = np.split(matrica_bit,
np.array((16, 24, (1+podaci.chebdeg)*64 + 24)),
axis=1)
days = np.packbits(days_bit.astype(bool), axis=1).view(np.uint16)
brod = np.packbits(fuel_bit.astype(bool), axis=1).view(np.uint8)
koeficijenti = np.packbits(koeficijenti_bit.astype(bool), axis=1).view(np.float64)
try:
uglovi = chebval(x_osa(broj_segmenata), koeficijenti.T)
uglovi = np.mod(uglovi, 2*pi)
except RuntimeWarning:
uglovi = np.random.random_sample(size=(broj_jedinki, broj_segmenata)) * 2*pi
print(days.shape, brod.shape, uglovi.shape, snaga.shape)
return days[:, 0], brod[:, 0], uglovi, snaga
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 measure_flexure(sky):
''' Measure expected (589.3 nm) - measured emission line in nm'''
ll, ss = sky['nm'], sky['ph_10m_nm']
pl.figure()
pl.step(ll, ss)
pix = SG.argrelmax(ss, order=20)[0]
skynms = chebval(pix, sky['coefficients'])
for s in skynms: pl.axvline(s)
ixmin = np.argmin(np.abs(skynms - 589.3))
dnm = 589.3 - skynms[ixmin]
return dnm
def _get_temperature(self, R):
"""Return temperature from input resistance or voltage based on
calibration"""
# Find the right resistance range
for n in range(len(self.fitdata)):
Z = log10(R)
ZL = self.fitdata[n]["Zlower for fit range"]
ZU = self.fitdata[n]["Zupper for fit range"]
if Z > ZL and Z < ZU:
# Apply the formula
k = ((Z - ZL) - (ZU - Z)) / (ZU - ZL)
T = chebval(k, self.fitdata[n]["coef"])
return T
print("Temperature out of calibration range.")
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 proj_radial(nr, a, b, x = None):
# evaluate the radial basis functions of degree <nr
# to the projection of the poloidal field in r direction (Q-part of a qst decomposition
if x is None:
xx,ww = cheb.chebgauss(nr)
else:
xx = np.array(x)
# use chebyshev polynomials normalized for FCT
coeffs = np.eye(nr)*2
#coeffs = np.eye(nr)
coeffs[0,0]=1.
#return spsp.lil_matrix((cheb.chebval(xx,coeffs)).transpose())
return np.mat((cheb.chebval(xx,coeffs)).transpose())
def measure_flexure(sky):
''' Measure expected (589.3 nm) - measured emission line in nm'''
ll, ss = sky['nm'], sky['ph_10m_nm']
pl.figure()
pl.step(ll, ss)
pix = SG.argrelmax(ss, order=20)[0]
skynms = chebval(pix, sky['coefficients'])
for s in skynms:
pl.axvline(s)
ixmin = np.argmin(np.abs(skynms - 589.3))
dnm = 589.3 - skynms[ixmin]
return dnm
def chebfit2(x, y, deg, alpha=1e-4):
"""
This is a reimplementaion of a subset of chebyshev.chebfit
https://numpy.org/doc/stable/reference/generated/numpy.polynomial.chebyshev.chebfit.html
using Tichnov regularization (Ridge regression)
"""
cheb_values = []
for pos in range(deg + 1):
coefs = np.zeros(deg + 1)
coefs[pos] = 1
vals = cheb.chebval(x, coefs)
cheb_values.append(vals)
cheb_values = np.array(cheb_values)
regr = Ridge(alpha=alpha, fit_intercept=False)
_ = regr.fit(cheb_values.T, y)
return regr.coef_
def ifct(f):
# =============================================================================
# INPUT: Chebyshev coefficients
# OUTPUT: Inverse discrete cosine tranform of f (f is in spectral space). This function takes f from spectral space to physical space
#==============================================================================
# ##Calculate chebyshev coefficients with DCT
N = 1001
n = np.arange(0, N, 1)
theta = np.pi * (2 * n + 1) / ((2 * (N)))
x = np.cos(theta)
#Correct to coincide with Moore 2017
#f[0] = f[0]*2
#F = idct(f)*len(f)
F = chebval(x, f)
return F
def chebstate_to_forwards(state, do_normalize=False):
"""
we return fwd values that correspond to scaled state (= coefs starting from the 2nd one, first two assume to be zero)
Let n=len(state). We set coefs = [0,0,state/norm(state)] and fv = chebval(range(n+2), coefs)
If we want len(fv) = nf then we need len(state) = nf-2
details
https://numpy.org/doc/stable/reference/generated/numpy.polynomial.chebyshev.chebfit.html
"""
nf = len(state) + 2
xs = cheb_fwd_xs(nf)
state_norm = np.linalg.norm(state)
if do_normalize:
state = state / state_norm
return cheb.chebval(xs, c=[0, 0, *state]), state_norm
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 interpolate(self, z, f):
"""See equations 17.37 and 17.38 in Boyd"""
from numpy.polynomial.chebyshev import chebval
import numpy as np
msg = "Can't interpolate outside grid domain"
assert np.array([z]).min() >= self.zmin, msg
assert np.array([z]).max() <= self.zmax, msg
# Get coefficients for standard Chebyshev polynomials
c = self.to_coefficients(f)
# Convert infinite grid to xg = [-1, 1]
x = z / np.sqrt(self.C ** 2 + z ** 2)
# Evaluate the Chebyshev polynomial
return chebval(x, c)
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 stationarityTest(x):
windowSize = 10
xStats = movingWindowStats(x, windowSize, arange(1, 5))
nSamples = min(128, x.shape[0] // windowSize)
t = linspace(-1, 1, nSamples)
i = array(linspace(0, xStats.shape[0] - 1, nSamples), int)
ct = chebval(t, eye(2)).T
minPvalues = 1
for j in range(xStats.shape[1]):
for k in range(xStats.shape[2]):
est = sm.OLS(xStats[i, j, k], ct)
est2 = est.fit()
print(j, k, est2.pvalues[1:])
minPvalues = min(minPvalues, est2.pvalues[1:].min())
# cla()
# plot(t, xStats[i,j,k], '.-')
# stop
return minPvalues
def proj_radial_r2(nr, a, b, x = None):
# evaluate the radial basis functions of degree <nr over r
# this projection operator includes the 1/r**2 part and corresponds
# to the a projection from spectral to physical for the toroidal field (T-part of a qst decomposition)
if x is None:
xx,ww = cheb.chebgauss(nr)
else:
xx = np.array(x)
# use chebyshev polynomials normalized for FCT
coeffs = np.eye(nr)*2
#coeffs = np.eye(nr)
coeffs[0,0]=1.
temp = (cheb.chebval(xx,coeffs)/(a*xx+b)**2).transpose()
#return spsp.coo_matrix(temp)
return np.mat(temp)
def object_func(*params):
params = params[0]
J = 0
fit = chebval(Q, params[:4])
E = params[4]
fit = fit + E / Q
for i in range(len(intensity)):
if Q[i] < 1:
if intensity[i] < fit[i]:
J = J + (intensity[i] - fit[i])**4
elif intensity[i] >= fit[i]:
J = J + (intensity[i] - fit[i])**4
else:
if intensity[i] < fit[i]:
J = J + (intensity[i] - fit[i])**4
elif intensity[i] >= fit[i]:
J = J + (intensity[i] - fit[i])**2
return J
def evaluate_at(self, point):
super(MultiCheb, self).evaluate_at(point)
poly_value = complex(0)
for mon in self.monomialList():
mon_value = 1
for i in range(len(point)):
cheb_deg = mon[i]
cheb_coeff = [0. for i in range(cheb_deg)]
cheb_coeff.append(1.)
cheb_val = cheb.chebval([point[i]], cheb_coeff)[0]
mon_value *= cheb_val
mon_value *= self.coeff[mon]
poly_value += mon_value
if abs(poly_value) < 1.e-10:
return 0
else:
return poly_value
def GaussHermiteIntegrate(degree, A, c, k):
"""This returns the integral of the expected Value function when
s is a log-normal distribution."""
# Note that f(x) = V( (s**rho)*exp(sigma*x*sqrt(2))*k**alpha - c, (s**rho) *
# exp( sigma*x*sqrt(2))) * 1/sqrt(pi)
integral = 0
for i in range(degree):
# Traditonally we used: V( k^alpha - c)
# Now it has taken the form of: V( s*k^alpha - c, s)
sNew = np.exp(mu + np.sqrt(2) * sigma * x[i])
#xPos = np.clip(sNew * k**alpha - c, a, b)
xPos = sNew * k**alpha - c
if (xPos < a):
print(xPos)
if (xPos > b):
print(xPos)
chebValue = cheb.chebval(xPos, A)
integral += y[i] * chebValue
return integral * (1. / np.sqrt(math.pi))
def cheb_eval(nr, a, b, r):
# INPUT:
# nr: int, number of radial functions
# a: double, stretch factor for the mapping to shells
# b: double, shift factor for the mapping to shells
# r: np.array or list, list of collocation points
# evaluates the projection matrix for the chebyshev basis
xx = (np.array(r) - b) / a
coeffs = np.eye(nr) * 2
coeffs[0, 0] = 1. # set 1 on the first diag entry because of DCT
# proptierties
# return the dense matrix, because it is needed for integration
# (evaluation of antiderivative) purposes
mat = np.mat(cheb.chebval(xx, coeffs).transpose())
return mat
def find_ha(cc):
"""Find out where in X position Halpha falls
Args:
cc (float vector): coefficients of fit giving
wavelength versus spatial position
Returns:
float: spatial position corresponding to Halpha
Note:
assumes spatial position falls between 30 and 100
arcsec and has an accuracy limit of 0.1 arcsec
"""
ix = np.arange(30, 100, .1)
haix = np.argmin(np.abs(chebval(ix, cc) - 656.3))
return ix[haix]
def spec_calibration(self, theta=None, obs=None, **kwargs):
"""Implements a Chebyshev polynomial calibration model. This only
occurs if ``"poly_coeffs"`` is present in the :py:attr:`params`
dictionary, otherwise the value of ``params["spec_norm"]`` is returned.
:param theta: (optional)
If given, set :py:attr:`params` using this vector before
calculating the calibration polynomial. ndarray of shape
``(ndim,)``
:param obs:
A dictionary of observational data, must contain the key
``"wavelength"``
:returns cal:
If ``params["cal_type"]`` is ``"poly"``, a polynomial given by
``'spec_norm'`` :math:`\times (1 + \Sum_{m=1}^M```'poly_coeffs'[m-1]``:math:` \times T_n(x))`.
Otherwise, the exponential of a Chebyshev polynomial.
"""
if theta is not None:
self.set_parameters(theta)
if ('poly_coeffs' in self.params):
mask = obs.get('mask', slice(None))
# map unmasked wavelengths to the interval -1, 1
# masked wavelengths may have x>1, x<-1
x = self.wave_to_x(obs["wavelength"], mask)
# get coefficients. Here we are setting the first term to 0 so we
# can deal with it separately for the exponential and regular
# multiplicative cases
c = np.insert(self.params['poly_coeffs'], 0, 0)
poly = chebval(x, c)
# switch to have spec_norm be multiplicative or additive depending
# on whether the calibration model is multiplicative in exp^poly or
# just poly
if self.params.get('cal_type', 'exp_poly') is 'poly':
return (1.0 + poly) * self.params.get('spec_norm', 1.0)
else:
return np.exp(self.params.get('spec_norm', 0) + poly)
else:
return 1.0 * self.params.get('spec_norm', 1.0)
def convert_spectra_to_img(spectra, CRVAL1, CRPIX1):
lfid = exp_fid_wave(CRVAL1=CRVAL1, CRPIX1=CRPIX1)
img = np.zeros((len(spectra), len(lfid)))
img2 = np.zeros((len(spectra), len(lfid)))
img[:] = np.nan
img2[:] = np.nan
for ix, spectrum in enumerate(spectra):
spec = spectrum.specw
if spec is None: continue
img[ix, 0:len(spec)] = spec
try: lam = chebval(np.arange(*spectrum.xrange), spectrum.lamcoeff)
except: continue
IF = interp1d(lam, spec, bounds_error=False, fill_value=np.nan)
img2[ix, :] = IF(lfid)
return img, img2
def power_monitor_calibration(wavelength, voltage):
ps = [ -1.954872514465467656097352744382078527e-02,
-9.858164527032511337267806084128096700e-01,
-9.030551056544730559316747076081810519e-02,
-1.176690008380113228181329532162635587e-01,
-7.280035069541325454256508464823127724e-02,
-3.945345662306017076037534252463956363e-02,
-3.615273166450208575106728403625311330e-02,
9.788152479007751483042198969997116365e-03,
-1.456367977705685066991403431302387617e-02,
-1.786831712872795699387218348874739604e-02,
2.597734938056468420586320178244932322e-02,
-2.107763828923307075635662499735190067e-03,
7.008138732863849161558444933461942128e-03,
2.152771594326421261689219477375445422e-02,
-1.331119139869020931432608279010310071e-02,
-1.377397203478183430880310567090418772e-02,
5.892610556615005337754986669551726663e-03,
4.013333930879096900223856891898321919e-03,
-1.388605295508058999273681699548887991e-03,
-6.461661481385170711574938984256277763e-04,
1.965802838584800071083241723712831117e-04,
5.949925144541367866713593715033425724e-05,
-1.676361665718153744295687568310881943e-05,
-2.944515435510320108369441277629263709e-06,
7.941387605415402886735497592352039931e-07,
6.088879431605282524011537540614691366e-08,
-1.605397480549405216611181363393912047e-08]
#The values below are determined from the dataset associated with the coefficients
#The fitting is scaled for numerical stability
w_std = 20.579900281127852
w_mean = 763.93403214285706
r_std = 2.9169654028890814e-05
r_mean = 0.00048691497166276345
scale_wavelength = (float(wavelength) - w_mean)/w_std
conversion = chebval(scale_wavelength, ps)
unscale_r = conversion*r_std + r_mean
return str(unscale_r*float(voltage))
def test_chebint(self) :
# check exceptions
assert_raises(ValueError, cheb.chebint, [0], .5)
assert_raises(ValueError, cheb.chebint, [0], -1)
assert_raises(ValueError, cheb.chebint, [0], 1, [0,0])
# test integration of zero polynomial
for i in range(2, 5):
k = [0]*(i - 2) + [1]
res = cheb.chebint([0], m=i, k=k)
assert_almost_equal(res, [0, 1])
# check single integration with integration constant
for i in range(5) :
scl = i + 1
pol = [0]*i + [1]
tgt = [i] + [0]*i + [1/scl]
chebpol = cheb.poly2cheb(pol)
chebint = cheb.chebint(chebpol, m=1, k=[i])
res = cheb.cheb2poly(chebint)
assert_almost_equal(trim(res), trim(tgt))
# check single integration with integration constant and lbnd
for i in range(5) :
scl = i + 1
pol = [0]*i + [1]
chebpol = cheb.poly2cheb(pol)
chebint = cheb.chebint(chebpol, m=1, k=[i], lbnd=-1)
assert_almost_equal(cheb.chebval(-1, chebint), i)
# check single integration with integration constant and scaling
for i in range(5) :
scl = i + 1
pol = [0]*i + [1]
tgt = [i] + [0]*i + [2/scl]
chebpol = cheb.poly2cheb(pol)
chebint = cheb.chebint(chebpol, m=1, k=[i], scl=2)
res = cheb.cheb2poly(chebint)
assert_almost_equal(trim(res), trim(tgt))
# check multiple integrations with default k
for i in range(5) :
for j in range(2,5) :
pol = [0]*i + [1]
tgt = pol[:]
for k in range(j) :
tgt = cheb.chebint(tgt, m=1)
res = cheb.chebint(pol, m=j)
assert_almost_equal(trim(res), trim(tgt))
# check multiple integrations with defined k
for i in range(5) :
for j in range(2,5) :
pol = [0]*i + [1]
tgt = pol[:]
for k in range(j) :
tgt = cheb.chebint(tgt, m=1, k=[k])
res = cheb.chebint(pol, m=j, k=list(range(j)))
assert_almost_equal(trim(res), trim(tgt))
# check multiple integrations with lbnd
for i in range(5) :
for j in range(2,5) :
pol = [0]*i + [1]
tgt = pol[:]
for k in range(j) :
tgt = cheb.chebint(tgt, m=1, k=[k], lbnd=-1)
res = cheb.chebint(pol, m=j, k=list(range(j)), lbnd=-1)
assert_almost_equal(trim(res), trim(tgt))
# check multiple integrations with scaling
for i in range(5) :
for j in range(2,5) :
pol = [0]*i + [1]
tgt = pol[:]
for k in range(j) :
tgt = cheb.chebint(tgt, m=1, k=[k], scl=2)
res = cheb.chebint(pol, m=j, k=list(range(j)), scl=2)
assert_almost_equal(trim(res), trim(tgt))
# 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)
plt.ylabel("Wavelength Center", fontsize=18)
if i!=0:
change = (residuals[i] - residuals[i-1])/residuals[i-1]
metric = residuals[i]/residuals[i-1] - 1
print('Metric of Change = ',metric)
