foscat.xarray.parameters#

Classes#

Parameters

parameters for the scattering covariance transform

Module Contents#

class foscat.xarray.parameters.Parameters#

parameters for the scattering covariance transform

Parameters:

  • n_orientations (int) – The number of orientations of the kernel

  • kernel_size (int) – The size of the kernel in cells / pixels

  • jmax_delta (float) – Compute the scattering covariance coefficients for this many refinement levels, starting with the data level (level, which is defined as log2(nside)) in decreasing order. This means that the levels for which coefficients are computed are in the range

    \[[level - relative_level, level]\]

    If relative_level is None, compute all levels until level=0.

  • dtype (str or numpy.dtype, default: "float64") – The dtype to use for the transform.

  • backend ({"numpy", "torch", "tensorflow"}, default: "tensorflow") – The compute backend.

n_orientations: int#
kernel_size: int#
jmax_delta: float#
dtype: str | numpy.dtype = 'float64'#
backend: str = 'tensorflow'#
cache: foscat.scat_cov.funct | None = None#