micmac.foregrounds.mixingmatrix module

class MixingMatrix(frequency_array, n_components, spv_nodes_b, nside, params=None, pos_special_freqs=[0, -1])

Bases: object

Note: units are K_CMB.

Parameters:

frequency_array : array

Array of frequencies

n_components : int

Number of components

spv_nodes_b : list

List of nodes for b containing info patches to build spv_templates

nside : int

Healpix nside of the expected input maps

params : array (optional)

Initial values of the parameters of the mixing matrix, default None (then initialized with zeros)

pos_special_freqs : list (optional)

List of indexes of special frequencies (e.g. 0 for synchrotron, -1 for dust)

__init__(frequency_array, n_components, spv_nodes_b, nside, params=None, pos_special_freqs=[0, -1])

Note: units are K_CMB.

Parameters:

frequency_array : array

Array of frequencies

n_components : int

Number of components

spv_nodes_b : list

List of nodes for b containing info patches to build spv_templates

nside : int

Healpix nside of the expected input maps

params : array (optional)

Initial values of the parameters of the mixing matrix, default None (then initialized with zeros)

pos_special_freqs : list (optional)

List of indexes of special frequencies (e.g. 0 for synchrotron, -1 for dust)

get_B(jax_use=False)

Full mixing matrix, (n_frequencies*n_components). CMB is given as the first component.

Parameters:

jax_use : bool (optional)

If True, returned as JAX Array, default False

Returns:

B_mat – Full mixing matrix

Return type:

array[float] of dimensions [n_frequencies, n_components, n_pix]

get_B_cmb(jax_use=False)

CMB column of the mixing matrix.

Parameters:

jax_use : bool (optional)

If True, returned as JAX Array, default False

Returns:

B_cmb – CMB column of the mixing matrix, filled with ones

Return type:

array[float] of dimensions [n_frequencies, 1, n_pix]

get_B_db(jax_use=False)

STATUS: Not used currently, to be adjusted with spv

List of derivatives of the Mixing Matrix (wrt to each entry of first comp and then each entry of second comp) Note: w/o pixel dimension

get_B_fgs(jax_use=False)

Foreground part of the mixing matrix.

Parameters:

jax_use : bool (optional)

If True, params are expected as JAX Array, default False

Returns:

B_fgs – Foreground part of the mixing matrix (including special frequencies)

Return type:

array[float] of dimensions [n_frequencies, n_components - 1, n_pix]

get_B_fgs_from_params(params, jax_use=False)

Foreground part of the mixing matrix obtained from the parameters.

Parameters:

params : array[float]

Flattened version of all free parameters of the mixing matrix per patch expected to be stored as [Bf1_comp1_patch1, Bf1_comp1_patch2, …, Bf2_comp1_patch1, …, Bf1_comp2_patch1, …, Bfn_comp2_patchn, …]

jax_use : bool (optional)

If True, params are expected as JAX Array and B_fgs will be returned as JAX Array, default False

Returns:

B_fgs – Foreground part of the mixing matrix (including special frequencies)

Return type:

array[float] of dimensions [n_frequencies, n_components - 1, n_pix]

get_B_from_params(params, jax_use=False)

Full mixing matrix, (n_frequencies*n_components), obtained from the parameters. CMB is given as the first component.

Parameters:

params : array[float]

Flattened version of all free parameters of the mixing matrix per patch expected to be stored as [Bf1_comp1_patch1, Bf1_comp1_patch2, …, Bf2_comp1_patch1, …, Bf1_comp2_patch1, …, Bfn_comp2_patchn, …]

jax_use : bool (optional)

If True, params are expected as JAX Array and B_mat will be returned as JAX Array, default False

Returns:

B_mat – Full mixing matrix

Return type:

array[float] of dimensions [n_frequencies, n_components, n_pix]

get_all_templates()

Retrieve all templates maps whose values correspond to the indices of params, and indexed per frequency and component

Returns:

all_templates – All templates indexes maps whose values correspond to the indices of params for all the patches distributions per frequency and component

Return type:

array[int] of dimensions [n_frequencies - n_components + 1, n_components - 1, 12*nside**2]

get_idx_template_params_long_python(idx_template, params, print_bool=False)

From the params to all the entries of the mixing matrix For a given template index, retrieve the corresponding template

Parameters:

idx_template : array[int]

index of params of the corresponding template which will be saved

params : array[float]

flatttened compressed array of the free params of the mixing matrix expected to be stored as [Bf1_comp1_patch1, Bf1_comp1_patch2, …, Bf2_comp1_patch1, …, Bf1_comp2_patch1, …, Bfn_comp2_patchn, …]

Returns:

idx_template – Full parameters of the mixing matrix re-flattened as [(n_frequencies - n_components + 1)*(n_components - 1), n_pix] stacked with one patch distribution template indicated by idx_template

Return type:

array[int]

get_one_template(nside_patch)

Retrieve all templates maps whose values correspond to the indices of params, and indexed per frequency and component

Parameters:

nside_patch : int

Healpix nside of one patch distribution

Returns:

template – One template indexes map whose values correspond to the indices of params for one patch distribution according to Healpix pixelization

Return type:

array[int] of dimensions [12*nside_patch**2]

get_params_db(jax_use=False)

STATUS: Not used currently, to be adjusted with spv

Derivatives of the part of the Mixing Matrix w params (wrt to each entry of first comp and then each entry of second comp) Note: w/o pixel dimension

get_params_long(jax_use=False)

From the params to all the entries of the mixing matrix

Parameters:

jax_use : bool (optional)

If True, params are expected as JAX Array, default False

Returns:

params_long – Reshaped free parameters of the mixing matrix

Return type:

array[float] of dimensions [n_frequencies - n_components + 1, n_components - 1, n_pix]

get_params_long_python(params, print_bool=False)

From the params (Bf) to all the entries of the mixing matrix only with Python (numpy) without JAX

Parameters:

params : array[float]

Flattened version of all free parameters of the mixing matrix per patch expected to be stored as [Bf1_comp1_patch1, Bf1_comp1_patch2, …, Bf2_comp1_patch1, …, Bf1_comp2_patch1, …, Bfn_comp2_patchn, …]

print_bool : bool (optional)

If True, print the node names

Returns:

params_long – Reshaped free parameters of the mixing matrix

Return type:

array[float] of dimensions [n_frequencies - n_components + 1, n_components - 1, n_pix]

get_patch_B_from_params(nside_patch, params, jax_use=False)

Full mixing matrix, (n_frequencies*n_components) from params and one patch distribution template. cmb is given as the first component.

Parameters:

nside_patch : int

Healpix nside of one patch distribution, expected lower than nside of the input maps

params : array[float]

Flattened version of all free parameters of the mixing matrix per patch expected to be stored as [Bf1_comp1_patch1, Bf1_comp1_patch2, …, Bf2_comp1_patch1, …, Bf1_comp2_patch1, …, Bfn_comp2_patchn, …]

jax_use : bool (optional)

If True, params are expected as JAX Array and the results will be returned as JAX Array, default False

Returns:

• B_mat (array[float] of dimensions [n_frequencies, n_components, n_pix]) – Full mixing matrix

• template (array[int] of dimensions [12*nside_patch**2]) – One template indexes map whose values correspond to the indices of params for one patch distribution according to Healpix pixelization

get_template_B_fgs_from_params(nside_patch, params, jax_use=False)

Foreground (fgs) part of the mixing matrix and one patch distribution template obtained from nside_patch expected lower than nside of the input maps.

Parameters:

nside_patch : int

Healpix nside of one patch distribution, expected lower than nside of the input maps

params : array[float]

Flattened version of all free parameters of the mixing matrix per patch expected to be stored as [Bf1_comp1_patch1, Bf1_comp1_patch2, …, Bf2_comp1_patch1, …, Bf1_comp2_patch1, …, Bfn_comp2_patchn, …]

jax_use : bool (optional)

If True, params are expected as JAX Array and the results will be returned as JAX Array, default False

Returns:

• B_fgs (array[float] of dimensions [n_frequencies, n_components - 1, n_pix]) – Foreground part of the mixing matrix (including special frequencies)

• template (array[int] of dimensions [12*nside_patch**2]) – One template indexes map whose values correspond to the indices of params for one patch distribution according to Healpix pixelization

property n_pix

Number of pixels of one input freq map

update_params(new_params, jax_use=False)

Update values of the params in the mixing matrix.

Parameters:

new_params : array

New values of the parameters of the mixing matrix

jax_use : bool (optional)

If True, use JAX to update it as JAX Array, default False

get_indexes_b(n_frequencies, n_components, spv_nodes_b)

Return indexes of params for all frequencies and components

Parameters:

n_frequencies : int

Number of frequencies

n_components : int

Number of components

spv_nodes_b : list

List of nodes for b containing info patches to build spv_templates

Returns:

indexes – Indexes of params for all frequencies and components

Return type:

array

get_indexes_patches(indexes_b, len_params)

Return indexes of params for all patches of a b

Parameters:

indexes_b : array

Indexes of params for all frequencies and components

len_params : int

Total number of free parameters (per frequency, component, patch)

Returns:

indexes_patches_list – List of indexes of params for all patches of a b

Return type:

list

get_len_params(spv_nodes_b)

Return total number of free parameters (per frequency, component, patch)

Parameters:

spv_nodes_b : list

List of nodes for b containing info patches to build spv_templates

Returns:

len_params – Total number of free parameters (per frequency, component, patch)

Return type:

int