bartorch.optim.NIHT

bartorch.optim.NIHT#

class bartorch.optim.NIHT(regularizers, *, maxiter=30, cclambda=0.0, precond=None)#

Normalized iterative hard thresholding.

Parameters:

  • regularizers (WaveletNIHT or ImageNIHT, or an iterable of them) – The hard-thresholding terms from bartorch.prox.

  • maxiter (int)

  • cclambda (float) – Weight of an identity added to the normal operator (pics -q).

__init__(regularizers, *, maxiter=30, cclambda=0.0, precond=None)#

Methods

__init__(regularizers, *[, maxiter, ...])

fixed_point(image_shape, *[, trainable])

This solver as a deep-equilibrium model: the step's fixed point.

in_library(y, A[, x0])

Solve with BART's own loop, without crossing back into Python.

unrolled(image_shape, *[, trainable])

This solver as a network of maxiter steps, trained end to end.