Python Module Documentation

class funct

Description of the class.

__init__(NORIENT=4,

KERNELSZ=3,

all_type='float64',

gpupos=0,

TEMPLATE_PATH='data',

BACKEND='tensorflow',

JmaxDelta=0,

isMPI=False,

mpi_size=1,

mpi_rank=0):

Parameters

NORIENT=4 Number of Wavelet orientation.
KERNELSZ=3 Size of the kernel used to define the wavelet
all_type='float64' Internal data format. Select float32 to earn memory and speed.
gpupos=0 Select the default GPU id if several are availble. Note that foscat always use all GPU if several losses are defined while compute synthesis.
TEMPLATE_PATH='data' Directory where default foscat information are stored.
BACKEND='tensorflow' Select the backend. numpy and pytorch are beta testedonly.
JmaxDelta=0 Define the delta of the last scale computed compare to the max possible scale considering the nside. For instance, the maximum number of scale usable for scat_cov of a nside=16 image is 4. JmaxDelta=2 computes only the 2 first scales.
isMPI=False Set to True to use the inside MPI job.``mpi_size`` and mpi_rank should be set.
mpi_size=1 Set the size of the MPI run.
mpi_rank=0 Set the rank of the MPI run.

ud_grade(image, jscale): return an image with a corresponding nside divided by a factor 2^jscale compare to the input image

up_grade(image, out_nside): return an image with a corresponding nside = out_nside using bilinear interpolation on image data.

convol(image, axis=0): convol the image by default wavelet defined in __init__. Input dimension is [..,Npix,..], output dimension is [..,Npix,Norient,..]. Npix is the number of pixels of the image.

smooth(image, axis=0): smooth the image by default symetric wavelet defined in __init__. Input dimension is [..,Npix,..], output dimension is [..,Mpix,..]. Npix is the number of pixels of the image.

fill(im, nullval): Fill the im samples equal to nullval in such way that the scat_cov computation as less affected by unknown data. Be aware that a mask should used to get the proper statistic while doing

moments(list_scat): Calculate the mean and the standard deviation for a list of scat_cov objects provided by list_scat. The return value is scat_mean, scat_std, which are two scat_cov objects representing the mean and the standard deviation values, respectively.

eval(image1, image2, mask, norm, Auto, calc_var)

Calculates the scattering correlations for a batch of images. Mean are done over pixels. mean of modulus:

S1 = <|I * Psi_j3|> Normalization : take the log

power spectrum:

P00 = <|I * Psi_j3|^2> Normalization : take the log

orig. x modulus:
C01 = < (I * Psi)_j3 x (|I * Psi_j2| * Psi_j3)^* >

Normalization : divide by (P00_j2 * P00_j3)^0.5 modulus x modulus:

C11 = <(|I * psi1| * psi3)(|I * psi2| * psi3)^*>

Normalization : divide by (P00_j1 * P00_j2)^0.5

Parameters

image1: tensor Image on which we compute the scattering coefficients [Nbatch, Npix, 1, 1]
image2: tensor Second image. If not None, we compute cross-scattering covariance coefficients.
mask:
norm: None or str If None no normalization is applied, if ‘auto’ normalize by the reference P00, if ‘self’ normalize by the current P00.

all_cross: False or True

If False compute all the coefficient even the Imaginary part, If True return only the terms computable in the auto case.

Returns

S1, P00, C01, C11 normalized

clean_norm(self): Internal method not to be used.

_compute_C01(j2, conv, vmask, M_dic, MconvPsi_dic, calc_var, return_data): Internal method not to be used.

_compute_C11(j1, j2, vmask, M1convPsi_dic, M2convPsi_dic, calc_var, return_data): Internal method not to be used.

square(x): Compute all coefficients (S1, P00, C01, C11, …) attached to the scat_cov x with the square of their values.

sqrt(x): Compute all coefficients (S1, P00, C01, C11, …) attached to the scat_cov x with the square root of their values.

reduce_mean(x): Compute the mean values of all the coefficients.

reduce_sum(x): Compute the sum values of all the coefficients.

ldiff(sig, x): Description of the method.

log(x): Compute all coefficients (S1, P00, C01, C11, …) attached to the scat_cov x with the logarithm of their values.

std(list_of_sc): Do the standard deviation of all the coefficients

eval_comp_fast(image1, image2, mask, norm, Auto): Internal method not to be used .

eval_fast(image1, image2, mask, norm, Auto): Same method than eval but run in Graph Execution mode fastest while doing lot of eval_fast. The first execution could be long.

backend.bk_real(x): return the real part of the x data.

backend.bk_imag(x): return the imaginary part of the x data.

backend.bk_conjugate(x): return the conjugate value of the x data.

backend.bk_norm(x): return the complex norm value of the x data.

class scat_cov

Description of the class.

__init__(s0, p00, c01, c11, s1, c10, backend): Description of the method.

numpy(self): Description of the method.

constant(self): Description of the method.

flatten(self): Description of the method.

flattenMask(self): Description of the method.

get_S0(self): Description of the method.

get_S1(self): Description of the method.

get_P00(self): Description of the method.

reset_P00(self): Description of the method.

get_C01(self): Description of the method.

get_C10(self): Description of the method.

get_C11(self): Description of the method.

get_j_idx(self): Description of the method.

get_jc11_idx(self): Description of the method.

__add__(other): Description of the method.

relu(self): Description of the method.

__radd__(other): Description of the method.

__truediv__(other): Description of the method.

__rtruediv__(other): Description of the method.

__rsub__(other): Description of the method.

__sub__(other): Description of the method.

domult(x, y): Description of the method.

dodiv(x, y): Description of the method.

domin(x, y): Description of the method.

doadd(x, y): Description of the method.

__mul__(other): Description of the method.

__rmul__(other): Description of the method.

interp(nscale, extend, constant): Description of the method.

plot(name, hold, color, lw, legend): Description of the method.

get_np(x): Description of the method.

save(filename): Description of the method.

read(filename): Description of the method.

std(self): Description of the method.

mean(self): Description of the method.

initdx(norient): Description of the method.

sqrt(self): Description of the method.

L1(self): Description of the method.

square_comp(self): Description of the method.

iso_mean(repeat): Description of the method.

fft_ang(nharm): Description of the method.

iso_std(repeat): Description of the method.

get_nscale(self): Description of the method.

get_norient(self): Description of the method.

add_data_from_log_slope(y, n, ds): Description of the method.

add_data_from_slope(y, n, ds): Description of the method.

up_grade(nscale, ds): Description of the method.