def plot_model(self):
"""
Plot the star model.
"""
x = np.arange(self.x0 - 5 * self.fwhm, self.x0 + 5 * self.fwhm)
y = np.arange(self.y0 - 5 * self.fwhm, self.y0 + 5 * self.fwhm)
xx, yy = np.meshgrid(x, y)
star = self.psf.evaluate(np.array([xx, yy]))
fig, ax = plt.subplots(1, 1)
plot_image_simple(
ax,
star,
title=f'Star model: A={self.amplitude:.2f}, fwhm={self.fwhm:.2f}',
units='Arbitrary units')
if parameters.DISPLAY:
plt.show()
if parameters.PdfPages:
parameters.PdfPages.savefig()
def plot_spectrogram_comparison_simple(self, ax, title='', extent=None, dispersion=False):
"""Method to plot a spectrogram issued from data and compare it with simulations.
Parameters
----------
ax: Axes
Axes instance of shape (4, 2).
title: str, optional
Title for the simulation plot (default: '').
extent: array_like, optional
Extent argument for imshow to crop plots (default: None).
dispersion: bool, optional
If True, plot a colored bar to see the associated wavelength color along the x axis (default: False).
"""
lambdas = self.spectrum.lambdas
sub = np.where((lambdas > parameters.LAMBDA_MIN) & (lambdas < parameters.LAMBDA_MAX))[0]
sub = np.where(sub < self.spectrum.spectrogram_Nx)[0]
if extent is not None:
sub = np.where((lambdas > extent[0]) & (lambdas < extent[1]))[0]
if len(sub) > 0:
norm = np.max(self.data[:, sub])
plot_image_simple(ax[0, 0], data=self.data[:, sub] / norm, title='Data', aspect='auto',
cax=ax[0, 1], vmin=0, vmax=1, units='1/max(data)')
ax[0, 0].set_title('Data', fontsize=10, loc='center', color='white', y=0.8)
plot_image_simple(ax[1, 0], data=self.model[:, sub] / norm, aspect='auto', cax=ax[1, 1], vmin=0, vmax=1,
units='1/max(data)')
if dispersion:
x = self.spectrum.chromatic_psf.table['Dx'][sub[5:-5]] + self.spectrum.spectrogram_x0 - sub[0]
y = np.ones_like(x)
ax[1, 0].scatter(x, y, cmap=from_lambda_to_colormap(self.lambdas[sub[5:-5]]), edgecolors='None',
c=self.lambdas[sub[5:-5]],
label='', marker='o', s=10)
# p0 = ax.plot(lambdas, self.model(lambdas), label='model')
# # ax.plot(self.lambdas, self.model_noconv, label='before conv')
if title != '':
ax[1, 0].set_title(title, fontsize=10, loc='center', color='white', y=0.8)
residuals = (self.data - self.model)
# residuals_err = self.spectrum.spectrogram_err / self.model
norm = self.err
residuals /= norm
std = float(np.std(residuals[:, sub]))
plot_image_simple(ax[2, 0], data=residuals[:, sub], vmin=-5 * std, vmax=5 * std, title='(Data-Model)/Err',
aspect='auto', cax=ax[2, 1], units='', cmap="bwr")
ax[2, 0].set_title('(Data-Model)/Err', fontsize=10, loc='center', color='black', y=0.8)
ax[2, 0].text(0.05, 0.05, f'mean={np.mean(residuals[:, sub]):.3f}\nstd={np.std(residuals[:, sub]):.3f}',
horizontalalignment='left', verticalalignment='bottom',
color='black', transform=ax[2, 0].transAxes)
ax[0, 0].set_xticks(ax[2, 0].get_xticks()[1:-1])
ax[0, 1].get_yaxis().set_label_coords(3.5, 0.5)
ax[1, 1].get_yaxis().set_label_coords(3.5, 0.5)
ax[2, 1].get_yaxis().set_label_coords(3.5, 0.5)
ax[3, 1].remove()
ax[3, 0].plot(self.lambdas[sub], self.data.sum(axis=0)[sub], label='Data')
ax[3, 0].plot(self.lambdas[sub], self.model.sum(axis=0)[sub], label='Model')
ax[3, 0].set_ylabel('Cross spectrum')
ax[3, 0].set_xlabel(r'$\lambda$ [nm]')
ax[3, 0].legend(fontsize=7)
ax[3, 0].grid(True)
def plot_model(self):
xx, yy = np.mgrid[0:parameters.CCD_IMSIZE:1, 0:parameters.CCD_IMSIZE:1]
starfield = self.model(xx, yy)
fig, ax = plt.subplots(1, 1)
plot_image_simple(ax,
starfield,
scale="log10",
target_pixcoords=self.pixcoords)
# im = plt.imshow(starfield, origin='lower', cmap='jet')
# ax.grid(color='white', ls='solid')
# ax.grid(True)
# ax.set_xlabel('X [pixels]')
# ax.set_ylabel('Y [pixels]')
# ax.set_title(f'Star field model: fwhm={self.fwhm.value:.2f}')
# cb = plt.colorbar(im, ax=ax)
# cb.formatter.set_powerlimits((0, 0))
# cb.locator = MaxNLocator(7, prune=None)
# cb.update_ticks()
# cb.set_label('Arbitrary units') # ,fontsize=16)
if parameters.DISPLAY:
plt.show()
if parameters.PdfPages:
parameters.PdfPages.savefig()
def remove_image_background_sextractor(data,
sigma=3.0,
box_size=(50, 50),
filter_size=(3, 3),
positive=False):
sigma_clip = SigmaClip(sigma=sigma)
bkg_estimator = SExtractorBackground()
bkg = Background2D(data,
box_size,
filter_size=filter_size,
sigma_clip=sigma_clip,
bkg_estimator=bkg_estimator)
data_wo_bkg = data - bkg.background
if positive:
data_wo_bkg -= np.min(data_wo_bkg)
if parameters.DEBUG:
fig, ax = plt.subplots(1, 2, figsize=(11, 5))
plot_image_simple(ax[0], bkg.background, scale="lin")
plot_image_simple(ax[1], data_wo_bkg, scale="symlog")
fig.tight_layout()
plt.show()
if parameters.PdfPages:
parameters.PdfPages.savefig()
return data_wo_bkg
def plot_fit(self):
if self.data.ndim == 1:
fig, ax = plt.subplots(2, 1, figsize=(6, 6), sharex='all', gridspec_kw={'height_ratios': [5, 1]})
data = np.copy(self.data)
if self.bgd_model_func is not None:
data = data + self.bgd_model_func(self.pixels)
ax[0].errorbar(self.pixels, data, yerr=self.err, fmt='ro', label="Data")
if len(self.outliers) > 0:
ax[0].errorbar(self.outliers, data[self.outliers], yerr=self.err[self.outliers], fmt='go',
label=rf"Outliers ({self.sigma_clip}$\sigma$)")
if self.bgd_model_func is not None:
ax[0].plot(self.pixels, self.bgd_model_func(self.pixels), 'b--', label="fitted bgd")
if self.guess is not None:
if self.bgd_model_func is not None:
ax[0].plot(self.pixels, self.psf.evaluate(self.pixels, p=self.guess)
+ self.bgd_model_func(self.pixels), 'k--', label="Guess")
else:
ax[0].plot(self.pixels, self.psf.evaluate(self.pixels, p=self.guess),
'k--', label="Guess")
self.psf.p = np.copy(self.p)
model = np.copy(self.model)
# if self.bgd_model_func is not None:
# model = self.model + self.bgd_model_func(self.pixels)
ax[0].plot(self.pixels, model, 'b-', label="Model")
ylim = list(ax[0].get_ylim())
ylim[1] = 1.2 * np.max(self.model)
ax[0].set_ylim(ylim)
ax[0].set_ylabel('Transverse profile')
ax[0].legend(loc=2, numpoints=1)
ax[0].grid(True)
txt = ""
for ip, p in enumerate(self.input_labels):
txt += f'{p}: {self.p[ip]:.4g}\n'
ax[0].text(0.95, 0.95, txt, horizontalalignment='right', verticalalignment='top', transform=ax[0].transAxes)
# residuals
residuals = (data - model) / self.err
residuals_err = np.ones_like(self.err)
ax[1].errorbar(self.pixels, residuals, yerr=residuals_err, fmt='ro')
if len(self.outliers) > 0:
residuals_outliers = (data[self.outliers] - model[self.outliers]) / self.err[self.outliers]
residuals_outliers_err = np.ones_like(residuals_outliers)
ax[1].errorbar(self.outliers, residuals_outliers, yerr=residuals_outliers_err, fmt='go')
ax[1].axhline(0, color='b')
ax[1].grid(True)
std = np.std(residuals)
ax[1].set_ylim([-3. * std, 3. * std])
ax[1].set_xlabel(ax[0].get_xlabel())
ax[1].set_ylabel('(data-fit)/err')
ax[0].set_xticks(ax[1].get_xticks()[1:-1])
ax[0].get_yaxis().set_label_coords(-0.1, 0.5)
ax[1].get_yaxis().set_label_coords(-0.1, 0.5)
# fig.tight_layout()
# fig.subplots_adjust(wspace=0, hspace=0)
elif self.data.ndim == 2:
gs_kw = dict(width_ratios=[3, 0.15], height_ratios=[1, 1, 1, 1])
fig, ax = plt.subplots(nrows=4, ncols=2, figsize=(5, 7), gridspec_kw=gs_kw)
norm = np.nanmax(self.data)
plot_image_simple(ax[0, 0], data=self.model / norm, aspect='auto', cax=ax[0, 1], vmin=0, vmax=1,
units='1/max(data)')
ax[0, 0].set_title("Model", fontsize=10, loc='center', color='white', y=0.8)
plot_image_simple(ax[1, 0], data=self.data / norm, title='Data', aspect='auto',
cax=ax[1, 1], vmin=0, vmax=1, units='1/max(data)')
ax[1, 0].set_title('Data', fontsize=10, loc='center', color='white', y=0.8)
residuals = (self.data - self.model)
# residuals_err = self.spectrum.spectrogram_err / self.model
norm = self.err
residuals /= norm
std = float(np.std(residuals))
plot_image_simple(ax[2, 0], data=residuals, vmin=-5 * std, vmax=5 * std, title='(Data-Model)/Err',
aspect='auto', cax=ax[2, 1], units='', cmap="bwr")
ax[2, 0].set_title('(Data-Model)/Err', fontsize=10, loc='center', color='black', y=0.8)
ax[2, 0].text(0.05, 0.05, f'mean={np.mean(residuals):.3f}\nstd={np.std(residuals):.3f}',
horizontalalignment='left', verticalalignment='bottom',
color='black', transform=ax[2, 0].transAxes)
ax[0, 0].set_xticks(ax[2, 0].get_xticks()[1:-1])
ax[0, 1].get_yaxis().set_label_coords(3.5, 0.5)
ax[1, 1].get_yaxis().set_label_coords(3.5, 0.5)
ax[2, 1].get_yaxis().set_label_coords(3.5, 0.5)
ax[3, 1].remove()
ax[3, 0].plot(np.arange(self.Nx), self.data.sum(axis=0), label='Data')
ax[3, 0].plot(np.arange(self.Nx), self.model.sum(axis=0), label='Model')
ax[3, 0].set_ylabel('Transverse sum')
ax[3, 0].set_xlabel(r'X [pixels]')
ax[3, 0].legend(fontsize=7)
ax[3, 0].grid(True)
else:
raise ValueError(f"Data array must have dimension 1 or 2. Here data.ndim={self.data.ndim}.")
if self.live_fit: # pragma: no cover
plt.draw()
plt.pause(1e-8)
plt.close()
else:
if parameters.DISPLAY:
plt.show()
else:
plt.close(fig)
if parameters.SAVE: # pragma: no cover
figname = os.path.splitext(self.filename)[0] + "_bestfit.pdf"
self.my_logger.info(f"\n\tSave figure {figname}.")
fig.savefig(figname, dpi=100, bbox_inches='tight')
if parameters.PdfPages:
parameters.PdfPages.savefig()