Normalized IHT

Normalized iterative hard thresholding algorithm.

mr_utils.cs.thresholding.normalized_iht.nIHT(A, y, k, c=0.1, kappa=None, x=None, maxiter=200, tol=1e-08, disp=False)[source]

Normalized iterative hard thresholding.

Parameters:
  • A (array_like) – Measurement matrix
  • y (array_like) – Measurements (i.e., y = Ax)
  • k (int) – Number of nonzero coefficients preserved after thresholding.
  • c (float, optional) – Small, fixed constant. Tunable.
  • kappa (float, optional) – Constant, > 1/(1 - c).
  • x (array_like, optional) – True signal we want to estimate.
  • maxiter (int, optional) – Maximum number of iterations (of the outer loop).
  • tol (float, optional) – Stopping criteria.
  • disp (bool, optional) – Whether or not to display iteration info.
Returns:

x_hat – Estimate of x.
Return type:

array_like

Notes

Implements Algorithm 8.6 from [1].

References

[1]Eldar, Yonina C., and Gitta Kutyniok, eds. Compressed sensing: theory and applications. Cambridge University Press, 2012.