starred.utils package

Submodules

starred.utils.ds9_reg module

starred.utils.generic_utils.Downsample(image, factor=1, conserve_flux=False)[source]

Resizes an image from dimensions (nx, ny) to (nx/factor, ny/factor).

Parameters:

  • image – 2D array with shape (nx, ny)

  • factor (int) – downsampling factor, which must be greater than or equal to 1

Returns:

2D array

starred.utils.generic_utils.Upsample(image, factor=1)[source]

Resizes an image without interpolation.

Parameters:

  • image – 2D array with shape (nx, ny)

  • factor (int) – upsampling factor, which must be greater than or equal to 1

Returns:

2D array

starred.utils.generic_utils.convert_list_to_numpy_array(kwargs)[source]
starred.utils.generic_utils.convert_numpy_array_to_list(kwargs)[source]
starred.utils.generic_utils.fft_convolve2d(x, y)[source]

2D convolution using the Fast Fourier Transform (FFT).

Parameters:

  • x – first 2D array

  • y – second 2D array

Returns:

2D array

starred.utils.generic_utils.fourier_division(a, b)[source]

Divides two arrays in Fourier space. Work much better with even kernel. To obtain the narrow PSF, a should be the full PSF and b a gaussian with 2pix FWHM. This function makes use of scipy if the kernel are even, and is not yet jaxified

Parameters:

  • a – first array

  • b – second array

starred.utils.generic_utils.fwhm2sigma(fwhm)[source]

Converts the FWHM (Full Width at Half Maximum) to the Gaussian sigma.

Parameters:

fwhm (float) – the full width at half maximum value

Returns:

Gaussian standard deviation, i.e., sqrt(var)

starred.utils.generic_utils.gaussian_function(x, y, amp, sigma_x, sigma_y, center_x, center_y)[source]
Parameters:

  • x – 1D array of x positions

  • y – 1D array of y positions

  • amp (float) – the amplitude coefficient

  • sigma_x (float) – x-spread of the Gaussian blob

  • sigma_y (float) – y-spread of the Gaussian blob

  • center_x (float) – x position of the center

  • center_y (float) – y position of the center

Returns:

2D Gaussian

starred.utils.generic_utils.gaussian_function_batched(x, y, amp, sigma_x, sigma_y, center_x, center_y)[source]

adjusted gaussian_function to accept and process batches for center_x and center_y. amp, sigma_x, sigma_y can also be vectors. but won’t be needed in starred, as we scale the PSF at the moffat step.

x and y: 2d arrays of coordinates made by make_grid and reshaped.

starred.utils.generic_utils.make_grid(numPix, deltapix, subgrid_res=1)[source]

Creates pixel grid as 1D arrays of x and y positions. The default coordinate frame is such that (0,0) is at the center of the coordinate grid.

Parameters:

  • numPix – number of pixels per axis. Provide an integer for a square grid

  • deltapix – pixel size

  • subgrid_res – sub-pixel resolution

Returns:

x and y position information given as two 1D arrays

starred.utils.generic_utils.moffat_elliptical_function(x, y, amp, fwhm_x, fwhm_y, phi, beta, center_x, center_y)[source]
Parameters:

  • x – 1D array of x positions

  • y – 1D array of y positions

  • amp (float) – the amplitude coefficient

  • fwhm_x (float) – the full width at half maximum value in the x direction

  • fwhm_y (float) – the full width at half maximum value in the y direction

  • phi (float) – orientation angle

  • beta (float) – the Moffat beta parameter

  • center_x (float) – x position of the center

  • center_y (float) – y position of the center

Returns:

2D Elliptical Moffat

starred.utils.generic_utils.moffat_function(x, y, amp, fwhm, beta, center_x, center_y)[source]
Parameters:

  • x – 1D array of x positions

  • y – 1D array of y positions

  • amp (float) – the amplitude coefficient

  • fwhm (float) – the full width at half maximum value

  • beta (float) – the Moffat beta parameter

  • center_x (float) – x position of the center

  • center_y (float) – y position of the center

Returns:

2D Moffat

starred.utils.generic_utils.pad_and_convolve(x, y, padding=True)[source]

Lax convolution is performed after an optional padding.

Parameters:

  • x – first array

  • y – second array

  • padding (bool) – padding

Returns:

output array

starred.utils.generic_utils.pad_and_convolve_fft(x, y, padding=True)[source]

FFT (Fast Fourier Transform) convolution is performed after an optional padding.

Parameters:

  • x – first array

  • y – second array

  • padding (bool) – padding

Returns:

Output array

starred.utils.generic_utils.save_fits(array, path, header=None)[source]

Saves .fits file of an array to a specified path.

starred.utils.generic_utils.save_npy(array, path, header=None)[source]

Saves .npy file of an array to a specified path. Header is not used in this format.

starred.utils.generic_utils.scipy_convolve(data, kernel)[source]

FFT-based Scipy convolution.

Parameters:

  • data – first array

  • kernel – second array

Returns:

output array

starred.utils.generic_utils.timer_func(func)[source]

This function shows the execution time of the provided function object.

starred.utils.generic_utils.twoD_Gaussian(x, y, amplitude, xo, yo, sigma_x, sigma_y, theta)[source]

Analytical 2D Gaussian function

Parameters:

  • x – array containing x-axis variable

  • y – array containing y-axis variable

  • amplitude – amplitude of the Gaussian

  • xo – center x coordinate

  • yo – center y coordinate

  • sigma_x – standard deviation of x variable

  • sigma_y – standard deviation of y variable

  • theta – ellongation angle

starred.utils.generic_utils module

starred.utils.generic_utils.Downsample(image, factor=1, conserve_flux=False)[source]

Resizes an image from dimensions (nx, ny) to (nx/factor, ny/factor).

Parameters:

  • image – 2D array with shape (nx, ny)

  • factor (int) – downsampling factor, which must be greater than or equal to 1

Returns:

2D array

starred.utils.generic_utils.Upsample(image, factor=1)[source]

Resizes an image without interpolation.

Parameters:

  • image – 2D array with shape (nx, ny)

  • factor (int) – upsampling factor, which must be greater than or equal to 1

Returns:

2D array

starred.utils.generic_utils.convert_list_to_numpy_array(kwargs)[source]
starred.utils.generic_utils.convert_numpy_array_to_list(kwargs)[source]
starred.utils.generic_utils.fft_convolve2d(x, y)[source]

2D convolution using the Fast Fourier Transform (FFT).

Parameters:

  • x – first 2D array

  • y – second 2D array

Returns:

2D array

starred.utils.generic_utils.fourier_division(a, b)[source]

Divides two arrays in Fourier space. Work much better with even kernel. To obtain the narrow PSF, a should be the full PSF and b a gaussian with 2pix FWHM. This function makes use of scipy if the kernel are even, and is not yet jaxified

Parameters:

  • a – first array

  • b – second array

starred.utils.generic_utils.fwhm2sigma(fwhm)[source]

Converts the FWHM (Full Width at Half Maximum) to the Gaussian sigma.

Parameters:

fwhm (float) – the full width at half maximum value

Returns:

Gaussian standard deviation, i.e., sqrt(var)

starred.utils.generic_utils.gaussian_function(x, y, amp, sigma_x, sigma_y, center_x, center_y)[source]
Parameters:

  • x – 1D array of x positions

  • y – 1D array of y positions

  • amp (float) – the amplitude coefficient

  • sigma_x (float) – x-spread of the Gaussian blob

  • sigma_y (float) – y-spread of the Gaussian blob

  • center_x (float) – x position of the center

  • center_y (float) – y position of the center

Returns:

2D Gaussian

starred.utils.generic_utils.gaussian_function_batched(x, y, amp, sigma_x, sigma_y, center_x, center_y)[source]

adjusted gaussian_function to accept and process batches for center_x and center_y. amp, sigma_x, sigma_y can also be vectors. but won’t be needed in starred, as we scale the PSF at the moffat step.

x and y: 2d arrays of coordinates made by make_grid and reshaped.

starred.utils.generic_utils.make_grid(numPix, deltapix, subgrid_res=1)[source]

Creates pixel grid as 1D arrays of x and y positions. The default coordinate frame is such that (0,0) is at the center of the coordinate grid.

Parameters:

  • numPix – number of pixels per axis. Provide an integer for a square grid

  • deltapix – pixel size

  • subgrid_res – sub-pixel resolution

Returns:

x and y position information given as two 1D arrays

starred.utils.generic_utils.moffat_elliptical_function(x, y, amp, fwhm_x, fwhm_y, phi, beta, center_x, center_y)[source]
Parameters:

  • x – 1D array of x positions

  • y – 1D array of y positions

  • amp (float) – the amplitude coefficient

  • fwhm_x (float) – the full width at half maximum value in the x direction

  • fwhm_y (float) – the full width at half maximum value in the y direction

  • phi (float) – orientation angle

  • beta (float) – the Moffat beta parameter

  • center_x (float) – x position of the center

  • center_y (float) – y position of the center

Returns:

2D Elliptical Moffat

starred.utils.generic_utils.moffat_function(x, y, amp, fwhm, beta, center_x, center_y)[source]
Parameters:

  • x – 1D array of x positions

  • y – 1D array of y positions

  • amp (float) – the amplitude coefficient

  • fwhm (float) – the full width at half maximum value

  • beta (float) – the Moffat beta parameter

  • center_x (float) – x position of the center

  • center_y (float) – y position of the center

Returns:

2D Moffat

starred.utils.generic_utils.pad_and_convolve(x, y, padding=True)[source]

Lax convolution is performed after an optional padding.

Parameters:

  • x – first array

  • y – second array

  • padding (bool) – padding

Returns:

output array

starred.utils.generic_utils.pad_and_convolve_fft(x, y, padding=True)[source]

FFT (Fast Fourier Transform) convolution is performed after an optional padding.

Parameters:

  • x – first array

  • y – second array

  • padding (bool) – padding

Returns:

Output array

starred.utils.generic_utils.save_fits(array, path, header=None)[source]

Saves .fits file of an array to a specified path.

starred.utils.generic_utils.save_npy(array, path, header=None)[source]

Saves .npy file of an array to a specified path. Header is not used in this format.

starred.utils.generic_utils.scipy_convolve(data, kernel)[source]

FFT-based Scipy convolution.

Parameters:

  • data – first array

  • kernel – second array

Returns:

output array

starred.utils.generic_utils.timer_func(func)[source]

This function shows the execution time of the provided function object.

starred.utils.generic_utils.twoD_Gaussian(x, y, amplitude, xo, yo, sigma_x, sigma_y, theta)[source]

Analytical 2D Gaussian function

Parameters:

  • x – array containing x-axis variable

  • y – array containing y-axis variable

  • amplitude – amplitude of the Gaussian

  • xo – center x coordinate

  • yo – center y coordinate

  • sigma_x – standard deviation of x variable

  • sigma_y – standard deviation of y variable

  • theta – ellongation angle

starred.utils.jax_utils module

starred.utils.jax_utils.convolveSeparableDilated(image2D, kernel1D, dilation=1)[source]

Convolves an image contained in image2D with the 1D kernel kernel1D. The operation is blured2D = image2D * (kernel1D ∧ kernel1D), where ∧ is a wedge product, here a tensor product.

Parameters:

  • kernel1D – 1D array to convolve the image with

  • image2D – 2D array

  • dilation – makes the spacial extent of the kernel bigger. The default is 1.

Returns:

2D array

starred.utils.jax_utils.decompose(image, n_scales)[source]

Decomposes an image into a chosen wavelet basis.

starred.utils.jax_utils.get_l1_norm(image, W, n_scales)[source]
starred.utils.jax_utils.reconstruct(coeffs)[source]

Reconstructs an image from wavelet decomposition coefficients.

starred.utils.jax_utils.rotate_array(array, angle_deg)[source]

rotates a 2D array by angle_deg degrees. We are using bilinear interpolation (higher orders not implemented in jax yet)

Parameters:

  • array – 2D array to be rotated

  • angle_deg – float, rotation in degrees.

Returns:

2D array - a rotated version of the input array

starred.utils.jax_utils.scale_norms(n_scales)[source]
starred.utils.jax_utils.translate_array(array, dx, dy)[source]

Using first order interpolation, translates the array by (dx, dy) pixels.

Parameters:

  • array – 2D array to be translated

  • dx (float) – translation in the x direction (in pixels)

  • dy (float) – translation in the y direction (in pixels)

Returns:

2D array - A translated version of the input array

starred.utils.light_profile_analytical module

starred.utils.light_profile_analytical.R_stable(R)[source]

Floor R_ at self._smoothing for numerical stability :param R: radius :return: smoothed and stabilized radius

starred.utils.light_profile_analytical.b_n(n)[source]

b(n) computation. This is the approximation of the exact solution to the relation, 2*incomplete_gamma_function(2n; b_n) = Gamma_function(2*n). :param n: the sersic index :return:

starred.utils.light_profile_analytical.ellipticity2phi_q(e1, e2)[source]

Transform complex ellipticity components to position angle and axis ratio.

Parameters:

  • e1 (float or array_like) – Ellipticity components.

  • e2 (float or array_like) – Ellipticity components.

Returns:

phi, q – Position angle (rad) and axis ratio (semi-minor / semi-major axis)

Return type:

same type as e1, e2

starred.utils.light_profile_analytical.get_distance_from_center(x, y, phi_G, q, center_x, center_y)[source]

Get the distance from the center of Sersic, accounting for orientation and axis ratio :param x: :param y: :param phi_G: orientation angle in rad :param q: axis ratio :param center_x: center x of sersic :param center_y: center y of sersic

starred.utils.light_profile_analytical.phi_q2_ellipticity(phi, q)[source]

transforms orientation angle and axis ratio into complex ellipticity moduli e1, e2

Parameters:

  • phi – angle of orientation (in radian)

  • q – axis ratio minor axis / major axis

Returns:

eccentricities e1 and e2 in complex ellipticity moduli

starred.utils.light_profile_analytical.r_sersic(R, R_sersic, n_sersic, max_R_frac=100.0)[source]
Parameters:

  • R – radius (array or float)

  • R_sersic – Sersic radius (half-light radius)

  • n_sersic – Sersic index (float)

  • max_R_frac – maximum window outside of which the mass is zeroed, in units of R_sersic (float)

Returns:

kernel of the Sersic surface brightness at R

starred.utils.light_profile_analytical.sersic_ellipse(x, y, R_sersic, n_sersic, e1, e2, center_x, center_y, amp, max_R_frac=100.0)[source]
Parameters:

  • x

  • y

  • amp – surface brightness/amplitude value at the half light radius

  • R_sersic – semi-major axis half light radius, in pixel

  • n_sersic – Sersic index

  • e1 – eccentricity parameter

  • e2 – eccentricity parameter

  • center_x – center in x-coordinate

  • center_y – center in y-coordinate

  • max_R_frac – truncation radius, in units of R_sersic (float)

Returns:

Sersic profile value at (x, y)

starred.utils.noise_utils module

starred.utils.noise_utils.dirac_impulse(num_pix)[source]

Returns a 2D array with a Dirac impulse at its center.

Parameters:

num_pix (int) – number of pixels per axis

Returns:

2D array

starred.utils.noise_utils.propagate_noise(model, noise_maps, kwargs, masks=None, wavelet_type_list=['starlet'], method='MC', num_samples=100, seed=None, likelihood_type='chi2', verbose=False, upsampling_factor=1, scaling_noise_ref=None)[source]

Performs noise propagation using MC or SLIT methods.

Parameters:

  • model – array containing the model

  • noise_maps – array containing the noise maps

  • kwargs – dictionary containing the parameters of the model

  • masks – array containing the masks

  • wavelet_type_list – list with the appropriate wavelet families

  • method – method for noise propagation. Choose ‘MC’ for an empirical propagation of the noise or ‘SLIT’ for analytical propagation.

  • num_samples (int) – number of realizations for the MC method. Not used if `method`=’SLIT’.

  • upsampling_factor (int) – the upsampling factor

  • scaling_noise_ref – index position of the reference noise map (lambda is given in the unit of standard deviation for this noise map).

Leave it to None to take the mean of all provided noise maps (recommended). This only works for the PSF reconstruction. For the deconvolution, the reference image’s noise map is always used, so lambda is given in unit of noise level, according to this noise map. :type scaling_noise_ref: integer

starred.utils.parameters module

class starred.utils.parameters.Parameters(kwargs_init, kwargs_fixed, kwargs_up=None, kwargs_down=None)[source]

Bases: object

Parameters class.

Parameters:

  • kwargs_init – dictionary with information on the initial values of the parameters

  • kwargs_fixed – dictionary containing the fixed parameters

  • kwargs_up – dictionary with information on the upper bounds of the parameters

  • kwargs_down – dictionary with information on the lower bounds of the parameters

best_fit_values(as_kwargs=False, copy=False)[source]

Maximum-a-postriori estimate.

convert_in_jnp_array(kwargs)[source]
current_values(as_kwargs=False, restart=False, copy=False)[source]

Returns the current values of the parameters.

get_bounds()[source]

Returns the upper and lower bounds of the parameters.

get_nan_indices()[source]
initial_values(as_kwargs=False, copy=False)[source]

Returns the initial values of the parameters.

property optimized

Checks whether a function is optimized.

set_best_fit(args)[source]

Sets the maximum-a-postriori estimate as the parameter values.

update_kwargs(kwargs_init=None, kwargs_fixed=None, kwargs_up=None, kwargs_down=None)[source]

Updates the kwargs with provided values.

Module contents

This subpackage contains generic and optimization functions