foscat.FoCUS

Attributes

TMPFILE_VERSION

Classes

FoCUS

Module Contents

foscat.FoCUS.TMPFILE_VERSION = 'V14_0'

class foscat.FoCUS.FoCUS(NORIENT=4, LAMBDA=1.2, KERNELSZ=3, slope=1.0, all_type='float32', nstep_max=20, padding='SAME', gpupos=0, mask_thres=None, mask_norm=False, isMPI=False, TEMPLATE_PATH=None, BACKEND='torch', use_2D=False, use_1D=False, return_data=False, DODIV=False, use_median=False, InitWave=None, silent=True, mpi_size=1, mpi_rank=0)

TMPFILE_VERSION = 'V14_0'

P1_dic = None

P2_dic = None

isMPI = False

mask_thres = None

mask_norm = False

InitWave = None

mask_mask = None

mpi_size = 1

mpi_rank = 0

return_data = False

silent = True

use_median = False

kernel_smooth

padding_smooth

kernelR_conv

kernelI_conv

padding_conv

down

up

TEMPLATE_PATH = None

number_of_loss = 0

history

nlog = 0

padding = 'SAME'

use_2D = False

use_1D = False

all_type = 'float32'

BACKEND = 'torch'

all_bk_type = Ellipsis

all_cbk_type = Ellipsis

gpulist

ngpu = 1

rank = 0

gpupos

NORIENT = 4

LAMBDA = 1.2

slope = 1.0

R_off = 1

ww_Real

ww_Imag

ww_CNN_Transpose

ww_CNN

X_CNN

Y_CNN

Z_CNN

Idx_CNN

Idx_WCNN

filters_set

edge_masks

KERNELSZ = 3

Idx_Neighbours

w_smooth

pix_interp_val

weight_interp_val

ring2nest

ampnorm

loss

dtype_dcode_map

dtype_code_map

get_dtype_code(dtype)

get_type()

get_mpi_type()

conv_to_FoCUS(x, axis=0)

diffang(a, b)

corr_idx_wXX(x, y)

calc_indices_convol(nside, kernel, rotation=None)

save_index(filepath, data, offset=0, count=None)

Save an N-dimensional NumPy array with shape (N, …) to binary file. A 12x int64 header is written, describing dtype and shape beyond axis 0.

Header layout (12 x int64): [0] = dtype code (0=int64, 1=int32, 2=float32, 3=float64, 4=complex64, 5=complex128) [1] = number of extra dimensions (i.e., data.ndim - 1) [2:12] = shape[1:] padded with zeros

Parameters: - filepath: target binary file path - data: NumPy array with shape (N, …) - offset: number of items to skip on axis 0 - count: number of items to write on axis 0 (default: rest of the array)

read_index(filepath, offset=0, count=None)

Load a NumPy array from a binary file with a 12x int64 header.

Header layout: [0] = dtype code [1] = number of extra dimensions (D) [2:2+D] = shape[1:] of each sample (shape after axis 0)

Parameters: - filepath: path to the binary file - offset: number of samples to skip on axis 0 - count: number of samples to read (default: all remaining)

Returns: - data: NumPy array with shape (count, …) and correct dtype

healpix_layer(im, ww, indices=None, weights=None)

get_rank()

get_size()

barrier()

toring(image, axis=0)

ud_grade(im, j, axis=0, cell_ids=None, nside=None)

ud_grade_2(im, axis=0, cell_ids=None, nside=None, max_poll=False)

up_grade(im, nout, axis=-1, nouty=None, cell_ids=None, o_cell_ids=None, force_init_index=False, nside=None)

fill_1d(i_arr, nullval=0)

fill_2d(i_arr, nullval=0)

fill_healpy(i_map, nmax=10, nullval=hp.UNSEEN)

ud_grade_1d(im, nout, axis=0)

up_grade_2_1d(im, axis=0)

convol_1d(im, axis=0)

smooth_1d(im, axis=0)

up_grade_1d(im, nout, axis=0)

init_index(nside, kernel=-1, cell_ids=None, spin=0)

init_index_cnn(nside, NORIENT=4, kernel=-1, cell_ids=None)

swapaxes(x, axis1, axis2)

masked_mean(x, mask, rank=0, calc_var=False)

reduce_dim(x, axis=0)

conv2d(image, ww, axis=0)

diff_data(x, y, is_complex=True, sigma=None)

convol(in_image, axis=0, cell_ids=None, nside=None, spin=0)

smooth(in_image, axis=0, cell_ids=None, nside=None, spin=0)

get_kernel_size()

get_nb_orient()

get_ww(nside=1)

plot_ww()